linux-stable/drivers/tty/n_gsm.c

4533 lines
111 KiB
C
Raw Normal View History

tty: add SPDX identifiers to all remaining files in drivers/tty/ It's good to have SPDX identifiers in all files to make it easier to audit the kernel tree for correct licenses. Update the drivers/tty files files with the correct SPDX license identifier based on the license text in the file itself. The SPDX identifier is a legally binding shorthand, which can be used instead of the full boiler plate text. This work is based on a script and data from Thomas Gleixner, Philippe Ombredanne, and Kate Stewart. Cc: Jiri Slaby <jslaby@suse.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Chris Metcalf <cmetcalf@mellanox.com> Cc: Jiri Kosina <jikos@kernel.org> Cc: David Sterba <dsterba@suse.com> Cc: James Hogan <jhogan@kernel.org> Cc: Rob Herring <robh@kernel.org> Cc: Eric Anholt <eric@anholt.net> Cc: Stefan Wahren <stefan.wahren@i2se.com> Cc: Florian Fainelli <f.fainelli@gmail.com> Cc: Ray Jui <rjui@broadcom.com> Cc: Scott Branden <sbranden@broadcom.com> Cc: bcm-kernel-feedback-list@broadcom.com Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: Helge Deller <deller@gmx.de> Cc: Joachim Eastwood <manabian@gmail.com> Cc: Matthias Brugger <matthias.bgg@gmail.com> Cc: Masahiro Yamada <yamada.masahiro@socionext.com> Cc: Tobias Klauser <tklauser@distanz.ch> Cc: Russell King <linux@armlinux.org.uk> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Richard Genoud <richard.genoud@gmail.com> Cc: Alexander Shiyan <shc_work@mail.ru> Cc: Baruch Siach <baruch@tkos.co.il> Cc: "Maciej W. Rozycki" <macro@linux-mips.org> Cc: "Uwe Kleine-König" <kernel@pengutronix.de> Cc: Pat Gefre <pfg@sgi.com> Cc: "Guilherme G. Piccoli" <gpiccoli@linux.vnet.ibm.com> Cc: Jason Wessel <jason.wessel@windriver.com> Cc: Vladimir Zapolskiy <vz@mleia.com> Cc: Sylvain Lemieux <slemieux.tyco@gmail.com> Cc: Carlo Caione <carlo@caione.org> Cc: Kevin Hilman <khilman@baylibre.com> Cc: Liviu Dudau <liviu.dudau@arm.com> Cc: Sudeep Holla <sudeep.holla@arm.com> Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Cc: Andy Gross <andy.gross@linaro.org> Cc: David Brown <david.brown@linaro.org> Cc: "Andreas Färber" <afaerber@suse.de> Cc: Kevin Cernekee <cernekee@gmail.com> Cc: Laxman Dewangan <ldewangan@nvidia.com> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: Jonathan Hunter <jonathanh@nvidia.com> Cc: Barry Song <baohua@kernel.org> Cc: Patrice Chotard <patrice.chotard@st.com> Cc: Maxime Coquelin <mcoquelin.stm32@gmail.com> Cc: Alexandre Torgue <alexandre.torgue@st.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: Peter Korsgaard <jacmet@sunsite.dk> Cc: Timur Tabi <timur@tabi.org> Cc: Tony Prisk <linux@prisktech.co.nz> Cc: Michal Simek <michal.simek@xilinx.com> Cc: "Sören Brinkmann" <soren.brinkmann@xilinx.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Kate Stewart <kstewart@linuxfoundation.org> Cc: Philippe Ombredanne <pombredanne@nexb.com> Cc: Jiri Slaby <jslaby@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-06 17:11:51 +00:00
// SPDX-License-Identifier: GPL-2.0
/*
* n_gsm.c GSM 0710 tty multiplexor
* Copyright (c) 2009/10 Intel Corporation
*
* * THIS IS A DEVELOPMENT SNAPSHOT IT IS NOT A FINAL RELEASE *
*
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
* Outgoing path:
* tty -> DLCI fifo -> scheduler -> GSM MUX data queue ---o-> ldisc
* control message -> GSM MUX control queue --´
*
* Incoming path:
* ldisc -> gsm_queue() -o--> tty
* `-> gsm_control_response()
*
* TO DO:
* Mostly done: ioctls for setting modes/timing
* Partly done: hooks so you can pull off frames to non tty devs
* Restart DLCI 0 when it closes ?
* Improve the tx engine
* Resolve tx side locking by adding a queue_head and routing
* all control traffic via it
* General tidy/document
* Review the locking/move to refcounts more (mux now moved to an
* alloc/free model ready)
* Use newest tty open/close port helpers and install hooks
* What to do about power functions ?
* Termios setting and negotiation
* Do we need a 'which mux are you' ioctl to correlate mux and tty sets
*
*/
#include <linux/types.h>
#include <linux/major.h>
#include <linux/errno.h>
#include <linux/signal.h>
#include <linux/fcntl.h>
#include <linux/sched/signal.h>
#include <linux/interrupt.h>
#include <linux/tty.h>
#include <linux/bitfield.h>
#include <linux/ctype.h>
#include <linux/mm.h>
#include <linux/math.h>
#include <linux/nospec.h>
#include <linux/string.h>
#include <linux/slab.h>
#include <linux/poll.h>
#include <linux/bitops.h>
#include <linux/file.h>
#include <linux/uaccess.h>
#include <linux/module.h>
#include <linux/timer.h>
#include <linux/tty_flip.h>
#include <linux/tty_driver.h>
#include <linux/serial.h>
#include <linux/kfifo.h>
#include <linux/skbuff.h>
#include <net/arp.h>
#include <linux/ip.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/gsmmux.h>
#include "tty.h"
static int debug;
module_param(debug, int, 0600);
/* Module debug bits */
#define DBG_DUMP BIT(0) /* Data transmission dump. */
#define DBG_CD_ON BIT(1) /* Always assume CD line on. */
#define DBG_DATA BIT(2) /* Data transmission details. */
#define DBG_ERRORS BIT(3) /* Details for fail conditions. */
#define DBG_TTY BIT(4) /* Transmission statistics for DLCI TTYs. */
#define DBG_PAYLOAD BIT(5) /* Limits DBG_DUMP to payload frames. */
/* Defaults: these are from the specification */
#define T1 10 /* 100mS */
#define T2 34 /* 333mS */
#define T3 10 /* 10s */
#define N2 3 /* Retry 3 times */
#define K 2 /* outstanding I frames */
#define MAX_T3 255 /* In seconds. */
#define MAX_WINDOW_SIZE 7 /* Limit of K in error recovery mode. */
/* Use long timers for testing at low speed with debug on */
#ifdef DEBUG_TIMING
#define T1 100
#define T2 200
#endif
/*
* Semi-arbitrary buffer size limits. 0710 is normally run with 32-64 byte
* limits so this is plenty
*/
#define MAX_MRU 1500
#define MAX_MTU 1500
#define MIN_MTU (PROT_OVERHEAD + 1)
/* SOF, ADDR, CTRL, LEN1, LEN2, ..., FCS, EOF */
#define PROT_OVERHEAD 7
#define GSM_NET_TX_TIMEOUT (HZ*10)
/*
* struct gsm_mux_net - network interface
*
* Created when net interface is initialized.
*/
struct gsm_mux_net {
struct kref ref;
struct gsm_dlci *dlci;
};
/*
* Each block of data we have queued to go out is in the form of
* a gsm_msg which holds everything we need in a link layer independent
* format
*/
struct gsm_msg {
struct list_head list;
u8 addr; /* DLCI address + flags */
u8 ctrl; /* Control byte + flags */
unsigned int len; /* Length of data block (can be zero) */
unsigned char *data; /* Points into buffer but not at the start */
unsigned char buffer[];
};
enum gsm_dlci_state {
DLCI_CLOSED,
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
DLCI_WAITING_CONFIG, /* Waiting for DLCI configuration from user */
DLCI_CONFIGURE, /* Sending PN (for adaption > 1) */
DLCI_OPENING, /* Sending SABM not seen UA */
DLCI_OPEN, /* SABM/UA complete */
DLCI_CLOSING, /* Sending DISC not seen UA/DM */
};
enum gsm_dlci_mode {
DLCI_MODE_ABM, /* Normal Asynchronous Balanced Mode */
DLCI_MODE_ADM, /* Asynchronous Disconnected Mode */
};
/*
* Each active data link has a gsm_dlci structure associated which ties
* the link layer to an optional tty (if the tty side is open). To avoid
* complexity right now these are only ever freed up when the mux is
* shut down.
*
* At the moment we don't free DLCI objects until the mux is torn down
* this avoid object life time issues but might be worth review later.
*/
struct gsm_dlci {
struct gsm_mux *gsm;
int addr;
enum gsm_dlci_state state;
struct mutex mutex;
/* Link layer */
enum gsm_dlci_mode mode;
spinlock_t lock; /* Protects the internal state */
struct timer_list t1; /* Retransmit timer for SABM and UA */
int retries;
/* Uplink tty if active */
struct tty_port port; /* The tty bound to this DLCI if there is one */
#define TX_SIZE 4096 /* Must be power of 2. */
struct kfifo fifo; /* Queue fifo for the DLCI */
int adaption; /* Adaption layer in use */
int prev_adaption;
u32 modem_rx; /* Our incoming virtual modem lines */
u32 modem_tx; /* Our outgoing modem lines */
unsigned int mtu;
bool dead; /* Refuse re-open */
/* Configuration */
u8 prio; /* Priority */
u8 ftype; /* Frame type */
u8 k; /* Window size */
/* Flow control */
bool throttled; /* Private copy of throttle state */
bool constipated; /* Throttle status for outgoing */
/* Packetised I/O */
struct sk_buff *skb; /* Frame being sent */
struct sk_buff_head skb_list; /* Queued frames */
/* Data handling callback */
void (*data)(struct gsm_dlci *dlci, const u8 *data, int len);
void (*prev_data)(struct gsm_dlci *dlci, const u8 *data, int len);
struct net_device *net; /* network interface, if created */
};
/*
* Parameter bits used for parameter negotiation according to 3GPP 27.010
* chapter 5.4.6.3.1.
*/
struct gsm_dlci_param_bits {
u8 d_bits;
u8 i_cl_bits;
u8 p_bits;
u8 t_bits;
__le16 n_bits;
u8 na_bits;
u8 k_bits;
};
static_assert(sizeof(struct gsm_dlci_param_bits) == 8);
#define PN_D_FIELD_DLCI GENMASK(5, 0)
#define PN_I_CL_FIELD_FTYPE GENMASK(3, 0)
#define PN_I_CL_FIELD_ADAPTION GENMASK(7, 4)
#define PN_P_FIELD_PRIO GENMASK(5, 0)
#define PN_T_FIELD_T1 GENMASK(7, 0)
#define PN_N_FIELD_N1 GENMASK(15, 0)
#define PN_NA_FIELD_N2 GENMASK(7, 0)
#define PN_K_FIELD_K GENMASK(2, 0)
/* Total number of supported devices */
#define GSM_TTY_MINORS 256
/* DLCI 0, 62/63 are special or reserved see gsmtty_open */
#define NUM_DLCI 64
/*
* DLCI 0 is used to pass control blocks out of band of the data
* flow (and with a higher link priority). One command can be outstanding
* at a time and we use this structure to manage them. They are created
* and destroyed by the user context, and updated by the receive paths
* and timers
*/
struct gsm_control {
u8 cmd; /* Command we are issuing */
u8 *data; /* Data for the command in case we retransmit */
int len; /* Length of block for retransmission */
int done; /* Done flag */
int error; /* Error if any */
};
enum gsm_encoding {
GSM_BASIC_OPT,
GSM_ADV_OPT,
};
enum gsm_mux_state {
GSM_SEARCH,
GSM_START,
GSM_ADDRESS,
GSM_CONTROL,
GSM_LEN,
GSM_DATA,
GSM_FCS,
GSM_OVERRUN,
GSM_LEN0,
GSM_LEN1,
GSM_SSOF,
};
/*
* Each GSM mux we have is represented by this structure. If we are
* operating as an ldisc then we use this structure as our ldisc
* state. We need to sort out lifetimes and locking with respect
* to the gsm mux array. For now we don't free DLCI objects that
* have been instantiated until the mux itself is terminated.
*
* To consider further: tty open versus mux shutdown.
*/
struct gsm_mux {
struct tty_struct *tty; /* The tty our ldisc is bound to */
spinlock_t lock;
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
struct mutex mutex;
unsigned int num;
struct kref ref;
/* Events on the GSM channel */
wait_queue_head_t event;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
/* ldisc send work */
struct work_struct tx_work;
/* Bits for GSM mode decoding */
/* Framing Layer */
unsigned char *buf;
enum gsm_mux_state state;
unsigned int len;
unsigned int address;
unsigned int count;
bool escape;
enum gsm_encoding encoding;
u8 control;
u8 fcs;
u8 *txframe; /* TX framing buffer */
/* Method for the receiver side */
void (*receive)(struct gsm_mux *gsm, u8 ch);
/* Link Layer */
unsigned int mru;
unsigned int mtu;
int initiator; /* Did we initiate connection */
bool dead; /* Has the mux been shut down */
struct gsm_dlci *dlci[NUM_DLCI];
tty: n_gsm: fix software flow control handling n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.8.1 states that XON/XOFF characters shall be used instead of Fcon/Fcoff command in advanced option mode to handle flow control. Chapter 5.4.8.2 describes how XON/XOFF characters shall be handled. Basic option mode only used Fcon/Fcoff commands and no XON/XOFF characters. These are treated as data bytes here. The current implementation uses the gsm_mux field 'constipated' to handle flow control from the remote peer and the gsm_dlci field 'constipated' to handle flow control from each DLCI. The later is unrelated to this patch. The gsm_mux field is correctly set for Fcon/Fcoff commands in gsm_control_message(). However, the same is not true for XON/XOFF characters in gsm1_receive(). Disable software flow control handling in the tty to allow explicit handling by n_gsm. Add the missing handling in advanced option mode for gsm_mux in gsm1_receive() to comply with the standard. This patch depends on the following commit: Commit 8838b2af23ca ("tty: n_gsm: fix SW flow control encoding/handling") Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-3-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:25 +00:00
int old_c_iflag; /* termios c_iflag value before attach */
bool constipated; /* Asked by remote to shut up */
bool has_devices; /* Devices were registered */
spinlock_t tx_lock;
unsigned int tx_bytes; /* TX data outstanding */
#define TX_THRESH_HI 8192
#define TX_THRESH_LO 2048
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
struct list_head tx_ctrl_list; /* Pending control packets */
struct list_head tx_data_list; /* Pending data packets */
/* Control messages */
struct timer_list kick_timer; /* Kick TX queuing on timeout */
struct timer_list t2_timer; /* Retransmit timer for commands */
int cretries; /* Command retry counter */
struct gsm_control *pending_cmd;/* Our current pending command */
spinlock_t control_lock; /* Protects the pending command */
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
/* Keep-alive */
struct timer_list ka_timer; /* Keep-alive response timer */
u8 ka_num; /* Keep-alive match pattern */
signed int ka_retries; /* Keep-alive retry counter, -1 if not yet initialized */
/* Configuration */
int adaption; /* 1 or 2 supported */
u8 ftype; /* UI or UIH */
int t1, t2; /* Timers in 1/100th of a sec */
unsigned int t3; /* Power wake-up timer in seconds. */
int n2; /* Retry count */
u8 k; /* Window size */
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
bool wait_config; /* Wait for configuration by ioctl before DLCI open */
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
u32 keep_alive; /* Control channel keep-alive in 10ms */
/* Statistics (not currently exposed) */
unsigned long bad_fcs;
unsigned long malformed;
unsigned long io_error;
unsigned long bad_size;
unsigned long unsupported;
};
/*
* Mux objects - needed so that we can translate a tty index into the
* relevant mux and DLCI.
*/
#define MAX_MUX 4 /* 256 minors */
static struct gsm_mux *gsm_mux[MAX_MUX]; /* GSM muxes */
static DEFINE_SPINLOCK(gsm_mux_lock);
static struct tty_driver *gsm_tty_driver;
/*
* This section of the driver logic implements the GSM encodings
* both the basic and the 'advanced'. Reliable transport is not
* supported.
*/
#define CR 0x02
#define EA 0x01
#define PF 0x10
/* I is special: the rest are ..*/
#define RR 0x01
#define UI 0x03
#define RNR 0x05
#define REJ 0x09
#define DM 0x0F
#define SABM 0x2F
#define DISC 0x43
#define UA 0x63
#define UIH 0xEF
/* Channel commands */
#define CMD_NSC 0x09
#define CMD_TEST 0x11
#define CMD_PSC 0x21
#define CMD_RLS 0x29
#define CMD_FCOFF 0x31
#define CMD_PN 0x41
#define CMD_RPN 0x49
#define CMD_FCON 0x51
#define CMD_CLD 0x61
#define CMD_SNC 0x69
#define CMD_MSC 0x71
/* Virtual modem bits */
#define MDM_FC 0x01
#define MDM_RTC 0x02
#define MDM_RTR 0x04
#define MDM_IC 0x20
#define MDM_DV 0x40
#define GSM0_SOF 0xF9
#define GSM1_SOF 0x7E
#define GSM1_ESCAPE 0x7D
#define GSM1_ESCAPE_BITS 0x20
#define XON 0x11
#define XOFF 0x13
tty: n_gsm: fix SW flow control encoding/handling n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.2.7.3 states that DC1 (XON) and DC3 (XOFF) are the control characters defined in ISO/IEC 646. These shall be quoted if seen in the data stream to avoid interpretation as flow control characters. ISO/IEC 646 refers to the set of ISO standards described as the ISO 7-bit coded character set for information interchange. Its final version is also known as ITU T.50. See https://www.itu.int/rec/T-REC-T.50-199209-I/en To abide the standard it is needed to quote DC1 and DC3 correctly if these are seen as data bytes and not as control characters. The current implementation already tries to enforce this but fails to catch all defined cases. 3GPP 27.010 chapter 5.2.7.3 clearly states that the most significant bit shall be ignored for DC1 and DC3 handling. The current implementation handles only the case with the most significant bit set 0. Cases in which DC1 and DC3 have the most significant bit set 1 are left unhandled. This patch fixes this by masking the data bytes with ISO_IEC_646_MASK (only the 7 least significant bits set 1) before comparing them with XON (a.k.a. DC1) and XOFF (a.k.a. DC3) when testing which byte values need quotation via byte stuffing. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220120101857.2509-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-01-20 10:18:57 +00:00
#define ISO_IEC_646_MASK 0x7F
static const struct tty_port_operations gsm_port_ops;
/*
* CRC table for GSM 0710
*/
static const u8 gsm_fcs8[256] = {
0x00, 0x91, 0xE3, 0x72, 0x07, 0x96, 0xE4, 0x75,
0x0E, 0x9F, 0xED, 0x7C, 0x09, 0x98, 0xEA, 0x7B,
0x1C, 0x8D, 0xFF, 0x6E, 0x1B, 0x8A, 0xF8, 0x69,
0x12, 0x83, 0xF1, 0x60, 0x15, 0x84, 0xF6, 0x67,
0x38, 0xA9, 0xDB, 0x4A, 0x3F, 0xAE, 0xDC, 0x4D,
0x36, 0xA7, 0xD5, 0x44, 0x31, 0xA0, 0xD2, 0x43,
0x24, 0xB5, 0xC7, 0x56, 0x23, 0xB2, 0xC0, 0x51,
0x2A, 0xBB, 0xC9, 0x58, 0x2D, 0xBC, 0xCE, 0x5F,
0x70, 0xE1, 0x93, 0x02, 0x77, 0xE6, 0x94, 0x05,
0x7E, 0xEF, 0x9D, 0x0C, 0x79, 0xE8, 0x9A, 0x0B,
0x6C, 0xFD, 0x8F, 0x1E, 0x6B, 0xFA, 0x88, 0x19,
0x62, 0xF3, 0x81, 0x10, 0x65, 0xF4, 0x86, 0x17,
0x48, 0xD9, 0xAB, 0x3A, 0x4F, 0xDE, 0xAC, 0x3D,
0x46, 0xD7, 0xA5, 0x34, 0x41, 0xD0, 0xA2, 0x33,
0x54, 0xC5, 0xB7, 0x26, 0x53, 0xC2, 0xB0, 0x21,
0x5A, 0xCB, 0xB9, 0x28, 0x5D, 0xCC, 0xBE, 0x2F,
0xE0, 0x71, 0x03, 0x92, 0xE7, 0x76, 0x04, 0x95,
0xEE, 0x7F, 0x0D, 0x9C, 0xE9, 0x78, 0x0A, 0x9B,
0xFC, 0x6D, 0x1F, 0x8E, 0xFB, 0x6A, 0x18, 0x89,
0xF2, 0x63, 0x11, 0x80, 0xF5, 0x64, 0x16, 0x87,
0xD8, 0x49, 0x3B, 0xAA, 0xDF, 0x4E, 0x3C, 0xAD,
0xD6, 0x47, 0x35, 0xA4, 0xD1, 0x40, 0x32, 0xA3,
0xC4, 0x55, 0x27, 0xB6, 0xC3, 0x52, 0x20, 0xB1,
0xCA, 0x5B, 0x29, 0xB8, 0xCD, 0x5C, 0x2E, 0xBF,
0x90, 0x01, 0x73, 0xE2, 0x97, 0x06, 0x74, 0xE5,
0x9E, 0x0F, 0x7D, 0xEC, 0x99, 0x08, 0x7A, 0xEB,
0x8C, 0x1D, 0x6F, 0xFE, 0x8B, 0x1A, 0x68, 0xF9,
0x82, 0x13, 0x61, 0xF0, 0x85, 0x14, 0x66, 0xF7,
0xA8, 0x39, 0x4B, 0xDA, 0xAF, 0x3E, 0x4C, 0xDD,
0xA6, 0x37, 0x45, 0xD4, 0xA1, 0x30, 0x42, 0xD3,
0xB4, 0x25, 0x57, 0xC6, 0xB3, 0x22, 0x50, 0xC1,
0xBA, 0x2B, 0x59, 0xC8, 0xBD, 0x2C, 0x5E, 0xCF
};
#define INIT_FCS 0xFF
#define GOOD_FCS 0xCF
static void gsm_dlci_close(struct gsm_dlci *dlci);
static int gsmld_output(struct gsm_mux *gsm, u8 *data, int len);
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
static int gsm_modem_update(struct gsm_dlci *dlci, u8 brk);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
static struct gsm_msg *gsm_data_alloc(struct gsm_mux *gsm, u8 addr, int len,
u8 ctrl);
static int gsm_send_packet(struct gsm_mux *gsm, struct gsm_msg *msg);
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
static struct gsm_dlci *gsm_dlci_alloc(struct gsm_mux *gsm, int addr);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
static void gsmld_write_trigger(struct gsm_mux *gsm);
static void gsmld_write_task(struct work_struct *work);
/**
* gsm_fcs_add - update FCS
* @fcs: Current FCS
* @c: Next data
*
* Update the FCS to include c. Uses the algorithm in the specification
* notes.
*/
static inline u8 gsm_fcs_add(u8 fcs, u8 c)
{
return gsm_fcs8[fcs ^ c];
}
/**
* gsm_fcs_add_block - update FCS for a block
* @fcs: Current FCS
* @c: buffer of data
* @len: length of buffer
*
* Update the FCS to include c. Uses the algorithm in the specification
* notes.
*/
static inline u8 gsm_fcs_add_block(u8 fcs, u8 *c, int len)
{
while (len--)
fcs = gsm_fcs8[fcs ^ *c++];
return fcs;
}
/**
* gsm_read_ea - read a byte into an EA
* @val: variable holding value
* @c: byte going into the EA
*
* Processes one byte of an EA. Updates the passed variable
* and returns 1 if the EA is now completely read
*/
static int gsm_read_ea(unsigned int *val, u8 c)
{
/* Add the next 7 bits into the value */
*val <<= 7;
*val |= c >> 1;
/* Was this the last byte of the EA 1 = yes*/
return c & EA;
}
/**
* gsm_read_ea_val - read a value until EA
* @val: variable holding value
* @data: buffer of data
* @dlen: length of data
*
* Processes an EA value. Updates the passed variable and
* returns the processed data length.
*/
static unsigned int gsm_read_ea_val(unsigned int *val, const u8 *data, int dlen)
{
unsigned int len = 0;
for (; dlen > 0; dlen--) {
len++;
if (gsm_read_ea(val, *data++))
break;
}
return len;
}
/**
* gsm_encode_modem - encode modem data bits
* @dlci: DLCI to encode from
*
* Returns the correct GSM encoded modem status bits (6 bit field) for
* the current status of the DLCI and attached tty object
*/
static u8 gsm_encode_modem(const struct gsm_dlci *dlci)
{
u8 modembits = 0;
/* FC is true flow control not modem bits */
if (dlci->throttled)
modembits |= MDM_FC;
if (dlci->modem_tx & TIOCM_DTR)
modembits |= MDM_RTC;
if (dlci->modem_tx & TIOCM_RTS)
modembits |= MDM_RTR;
if (dlci->modem_tx & TIOCM_RI)
modembits |= MDM_IC;
if (dlci->modem_tx & TIOCM_CD || dlci->gsm->initiator)
modembits |= MDM_DV;
/* special mappings for passive side to operate as UE */
if (dlci->modem_tx & TIOCM_OUT1)
modembits |= MDM_IC;
if (dlci->modem_tx & TIOCM_OUT2)
modembits |= MDM_DV;
return modembits;
}
static void gsm_hex_dump_bytes(const char *fname, const u8 *data,
unsigned long len)
{
char *prefix;
if (!fname) {
print_hex_dump(KERN_INFO, "", DUMP_PREFIX_NONE, 16, 1, data, len,
true);
return;
}
prefix = kasprintf(GFP_ATOMIC, "%s: ", fname);
if (!prefix)
return;
print_hex_dump(KERN_INFO, prefix, DUMP_PREFIX_OFFSET, 16, 1, data, len,
true);
kfree(prefix);
}
/**
* gsm_encode_params - encode DLCI parameters
* @dlci: DLCI to encode from
* @params: buffer to fill with the encoded parameters
*
* Encodes the parameters according to GSM 07.10 section 5.4.6.3.1
* table 3.
*/
static int gsm_encode_params(const struct gsm_dlci *dlci,
struct gsm_dlci_param_bits *params)
{
const struct gsm_mux *gsm = dlci->gsm;
unsigned int i, cl;
switch (dlci->ftype) {
case UIH:
i = 0; /* UIH */
break;
case UI:
i = 1; /* UI */
break;
default:
pr_debug("unsupported frame type %d\n", dlci->ftype);
return -EINVAL;
}
switch (dlci->adaption) {
case 1: /* Unstructured */
cl = 0; /* convergence layer type 1 */
break;
case 2: /* Unstructured with modem bits. */
cl = 1; /* convergence layer type 2 */
break;
default:
pr_debug("unsupported adaption %d\n", dlci->adaption);
return -EINVAL;
}
params->d_bits = FIELD_PREP(PN_D_FIELD_DLCI, dlci->addr);
/* UIH, convergence layer type 1 */
params->i_cl_bits = FIELD_PREP(PN_I_CL_FIELD_FTYPE, i) |
FIELD_PREP(PN_I_CL_FIELD_ADAPTION, cl);
params->p_bits = FIELD_PREP(PN_P_FIELD_PRIO, dlci->prio);
params->t_bits = FIELD_PREP(PN_T_FIELD_T1, gsm->t1);
params->n_bits = cpu_to_le16(FIELD_PREP(PN_N_FIELD_N1, dlci->mtu));
params->na_bits = FIELD_PREP(PN_NA_FIELD_N2, gsm->n2);
params->k_bits = FIELD_PREP(PN_K_FIELD_K, dlci->k);
return 0;
}
/**
* gsm_register_devices - register all tty devices for a given mux index
*
* @driver: the tty driver that describes the tty devices
* @index: the mux number is used to calculate the minor numbers of the
* ttys for this mux and may differ from the position in the
* mux array.
*/
static int gsm_register_devices(struct tty_driver *driver, unsigned int index)
{
struct device *dev;
int i;
unsigned int base;
if (!driver || index >= MAX_MUX)
return -EINVAL;
base = index * NUM_DLCI; /* first minor for this index */
for (i = 1; i < NUM_DLCI; i++) {
/* Don't register device 0 - this is the control channel
* and not a usable tty interface
*/
dev = tty_register_device(gsm_tty_driver, base + i, NULL);
if (IS_ERR(dev)) {
if (debug & DBG_ERRORS)
pr_info("%s failed to register device minor %u",
__func__, base + i);
for (i--; i >= 1; i--)
tty_unregister_device(gsm_tty_driver, base + i);
return PTR_ERR(dev);
}
}
return 0;
}
/**
* gsm_unregister_devices - unregister all tty devices for a given mux index
*
* @driver: the tty driver that describes the tty devices
* @index: the mux number is used to calculate the minor numbers of the
* ttys for this mux and may differ from the position in the
* mux array.
*/
static void gsm_unregister_devices(struct tty_driver *driver,
unsigned int index)
{
int i;
unsigned int base;
if (!driver || index >= MAX_MUX)
return;
base = index * NUM_DLCI; /* first minor for this index */
for (i = 1; i < NUM_DLCI; i++) {
/* Don't unregister device 0 - this is the control
* channel and not a usable tty interface
*/
tty_unregister_device(gsm_tty_driver, base + i);
}
}
/**
* gsm_print_packet - display a frame for debug
* @hdr: header to print before decode
* @addr: address EA from the frame
* @cr: C/R bit seen as initiator
* @control: control including PF bit
* @data: following data bytes
* @dlen: length of data
*
* Displays a packet in human readable format for debugging purposes. The
* style is based on amateur radio LAP-B dump display.
*/
static void gsm_print_packet(const char *hdr, int addr, int cr,
u8 control, const u8 *data, int dlen)
{
if (!(debug & DBG_DUMP))
return;
/* Only show user payload frames if debug & DBG_PAYLOAD */
if (!(debug & DBG_PAYLOAD) && addr != 0)
if ((control & ~PF) == UI || (control & ~PF) == UIH)
return;
pr_info("%s %d) %c: ", hdr, addr, "RC"[cr]);
switch (control & ~PF) {
case SABM:
pr_cont("SABM");
break;
case UA:
pr_cont("UA");
break;
case DISC:
pr_cont("DISC");
break;
case DM:
pr_cont("DM");
break;
case UI:
pr_cont("UI");
break;
case UIH:
pr_cont("UIH");
break;
default:
if (!(control & 0x01)) {
pr_cont("I N(S)%d N(R)%d",
(control & 0x0E) >> 1, (control & 0xE0) >> 5);
} else switch (control & 0x0F) {
case RR:
pr_cont("RR(%d)", (control & 0xE0) >> 5);
break;
case RNR:
pr_cont("RNR(%d)", (control & 0xE0) >> 5);
break;
case REJ:
pr_cont("REJ(%d)", (control & 0xE0) >> 5);
break;
default:
pr_cont("[%02X]", control);
}
}
if (control & PF)
pr_cont("(P)");
else
pr_cont("(F)");
gsm_hex_dump_bytes(NULL, data, dlen);
}
/*
* Link level transmission side
*/
/**
* gsm_stuff_frame - bytestuff a packet
* @input: input buffer
* @output: output buffer
* @len: length of input
*
* Expand a buffer by bytestuffing it. The worst case size change
* is doubling and the caller is responsible for handing out
* suitable sized buffers.
*/
static int gsm_stuff_frame(const u8 *input, u8 *output, int len)
{
int olen = 0;
while (len--) {
if (*input == GSM1_SOF || *input == GSM1_ESCAPE
tty: n_gsm: fix SW flow control encoding/handling n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.2.7.3 states that DC1 (XON) and DC3 (XOFF) are the control characters defined in ISO/IEC 646. These shall be quoted if seen in the data stream to avoid interpretation as flow control characters. ISO/IEC 646 refers to the set of ISO standards described as the ISO 7-bit coded character set for information interchange. Its final version is also known as ITU T.50. See https://www.itu.int/rec/T-REC-T.50-199209-I/en To abide the standard it is needed to quote DC1 and DC3 correctly if these are seen as data bytes and not as control characters. The current implementation already tries to enforce this but fails to catch all defined cases. 3GPP 27.010 chapter 5.2.7.3 clearly states that the most significant bit shall be ignored for DC1 and DC3 handling. The current implementation handles only the case with the most significant bit set 0. Cases in which DC1 and DC3 have the most significant bit set 1 are left unhandled. This patch fixes this by masking the data bytes with ISO_IEC_646_MASK (only the 7 least significant bits set 1) before comparing them with XON (a.k.a. DC1) and XOFF (a.k.a. DC3) when testing which byte values need quotation via byte stuffing. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220120101857.2509-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-01-20 10:18:57 +00:00
|| (*input & ISO_IEC_646_MASK) == XON
|| (*input & ISO_IEC_646_MASK) == XOFF) {
*output++ = GSM1_ESCAPE;
*output++ = *input++ ^ GSM1_ESCAPE_BITS;
olen++;
} else
*output++ = *input++;
olen++;
}
return olen;
}
/**
* gsm_send - send a control frame
* @gsm: our GSM mux
* @addr: address for control frame
* @cr: command/response bit seen as initiator
* @control: control byte including PF bit
*
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
* Format up and transmit a control frame. These should be transmitted
* ahead of data when they are needed.
*/
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
static int gsm_send(struct gsm_mux *gsm, int addr, int cr, int control)
{
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
struct gsm_msg *msg;
u8 *dp;
int ocr;
unsigned long flags;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
msg = gsm_data_alloc(gsm, addr, 0, control);
if (!msg)
return -ENOMEM;
/* toggle C/R coding if not initiator */
ocr = cr ^ (gsm->initiator ? 0 : 1);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
msg->data -= 3;
dp = msg->data;
*dp++ = (addr << 2) | (ocr << 1) | EA;
*dp++ = control;
if (gsm->encoding == GSM_BASIC_OPT)
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
*dp++ = EA; /* Length of data = 0 */
*dp = 0xFF - gsm_fcs_add_block(INIT_FCS, msg->data, dp - msg->data);
msg->len = (dp - msg->data) + 1;
gsm_print_packet("Q->", addr, cr, control, NULL, 0);
spin_lock_irqsave(&gsm->tx_lock, flags);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
list_add_tail(&msg->list, &gsm->tx_ctrl_list);
gsm->tx_bytes += msg->len;
spin_unlock_irqrestore(&gsm->tx_lock, flags);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsmld_write_trigger(gsm);
return 0;
}
/**
* gsm_dlci_clear_queues - remove outstanding data for a DLCI
* @gsm: mux
* @dlci: clear for this DLCI
*
* Clears the data queues for a given DLCI.
*/
static void gsm_dlci_clear_queues(struct gsm_mux *gsm, struct gsm_dlci *dlci)
{
struct gsm_msg *msg, *nmsg;
int addr = dlci->addr;
unsigned long flags;
/* Clear DLCI write fifo first */
spin_lock_irqsave(&dlci->lock, flags);
kfifo_reset(&dlci->fifo);
spin_unlock_irqrestore(&dlci->lock, flags);
/* Clear data packets in MUX write queue */
spin_lock_irqsave(&gsm->tx_lock, flags);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
list_for_each_entry_safe(msg, nmsg, &gsm->tx_data_list, list) {
if (msg->addr != addr)
continue;
gsm->tx_bytes -= msg->len;
list_del(&msg->list);
kfree(msg);
}
spin_unlock_irqrestore(&gsm->tx_lock, flags);
}
/**
* gsm_response - send a control response
* @gsm: our GSM mux
* @addr: address for control frame
* @control: control byte including PF bit
*
* Format up and transmit a link level response frame.
*/
static inline void gsm_response(struct gsm_mux *gsm, int addr, int control)
{
gsm_send(gsm, addr, 0, control);
}
/**
* gsm_command - send a control command
* @gsm: our GSM mux
* @addr: address for control frame
* @control: control byte including PF bit
*
* Format up and transmit a link level command frame.
*/
static inline void gsm_command(struct gsm_mux *gsm, int addr, int control)
{
gsm_send(gsm, addr, 1, control);
}
/* Data transmission */
#define HDR_LEN 6 /* ADDR CTRL [LEN.2] DATA FCS */
/**
* gsm_data_alloc - allocate data frame
* @gsm: GSM mux
* @addr: DLCI address
* @len: length excluding header and FCS
* @ctrl: control byte
*
* Allocate a new data buffer for sending frames with data. Space is left
* at the front for header bytes but that is treated as an implementation
* detail and not for the high level code to use
*/
static struct gsm_msg *gsm_data_alloc(struct gsm_mux *gsm, u8 addr, int len,
u8 ctrl)
{
struct gsm_msg *m = kmalloc(sizeof(struct gsm_msg) + len + HDR_LEN,
GFP_ATOMIC);
if (m == NULL)
return NULL;
m->data = m->buffer + HDR_LEN - 1; /* Allow for FCS */
m->len = len;
m->addr = addr;
m->ctrl = ctrl;
INIT_LIST_HEAD(&m->list);
return m;
}
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
/**
* gsm_send_packet - sends a single packet
* @gsm: GSM Mux
* @msg: packet to send
*
* The given packet is encoded and sent out. No memory is freed.
* The caller must hold the gsm tx lock.
*/
static int gsm_send_packet(struct gsm_mux *gsm, struct gsm_msg *msg)
{
int len, ret;
if (gsm->encoding == GSM_BASIC_OPT) {
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsm->txframe[0] = GSM0_SOF;
memcpy(gsm->txframe + 1, msg->data, msg->len);
gsm->txframe[msg->len + 1] = GSM0_SOF;
len = msg->len + 2;
} else {
gsm->txframe[0] = GSM1_SOF;
len = gsm_stuff_frame(msg->data, gsm->txframe + 1, msg->len);
gsm->txframe[len + 1] = GSM1_SOF;
len += 2;
}
if (debug & DBG_DATA)
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsm_hex_dump_bytes(__func__, gsm->txframe, len);
gsm_print_packet("-->", msg->addr, gsm->initiator, msg->ctrl, msg->data,
msg->len);
ret = gsmld_output(gsm, gsm->txframe, len);
if (ret <= 0)
return ret;
/* FIXME: Can eliminate one SOF in many more cases */
gsm->tx_bytes -= msg->len;
return 0;
}
/**
* gsm_is_flow_ctrl_msg - checks if flow control message
* @msg: message to check
*
* Returns true if the given message is a flow control command of the
* control channel. False is returned in any other case.
*/
static bool gsm_is_flow_ctrl_msg(struct gsm_msg *msg)
{
unsigned int cmd;
if (msg->addr > 0)
return false;
switch (msg->ctrl & ~PF) {
case UI:
case UIH:
cmd = 0;
if (gsm_read_ea_val(&cmd, msg->data + 2, msg->len - 2) < 1)
break;
switch (cmd & ~PF) {
case CMD_FCOFF:
case CMD_FCON:
return true;
}
break;
}
return false;
}
/**
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
* gsm_data_kick - poke the queue
* @gsm: GSM Mux
*
* The tty device has called us to indicate that room has appeared in
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
* the transmit queue. Ram more data into the pipe if we have any.
* If we have been flow-stopped by a CMD_FCOFF, then we can only
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
* send messages on DLCI0 until CMD_FCON. The caller must hold
* the gsm tx lock.
*/
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
static int gsm_data_kick(struct gsm_mux *gsm)
{
struct gsm_msg *msg, *nmsg;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
struct gsm_dlci *dlci;
int ret;
clear_bit(TTY_DO_WRITE_WAKEUP, &gsm->tty->flags);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
/* Serialize control messages and control channel messages first */
list_for_each_entry_safe(msg, nmsg, &gsm->tx_ctrl_list, list) {
if (gsm->constipated && !gsm_is_flow_ctrl_msg(msg))
continue;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
ret = gsm_send_packet(gsm, msg);
switch (ret) {
case -ENOSPC:
return -ENOSPC;
case -ENODEV:
/* ldisc not open */
gsm->tx_bytes -= msg->len;
list_del(&msg->list);
kfree(msg);
continue;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
default:
if (ret >= 0) {
list_del(&msg->list);
kfree(msg);
}
break;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
}
}
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
if (gsm->constipated)
return -EAGAIN;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
/* Serialize other channels */
if (list_empty(&gsm->tx_data_list))
return 0;
list_for_each_entry_safe(msg, nmsg, &gsm->tx_data_list, list) {
dlci = gsm->dlci[msg->addr];
/* Send only messages for DLCIs with valid state */
if (dlci->state != DLCI_OPEN) {
gsm->tx_bytes -= msg->len;
list_del(&msg->list);
kfree(msg);
continue;
}
ret = gsm_send_packet(gsm, msg);
switch (ret) {
case -ENOSPC:
return -ENOSPC;
case -ENODEV:
/* ldisc not open */
gsm->tx_bytes -= msg->len;
list_del(&msg->list);
kfree(msg);
continue;
default:
if (ret >= 0) {
list_del(&msg->list);
kfree(msg);
}
break;
}
}
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
return 1;
}
/**
* __gsm_data_queue - queue a UI or UIH frame
* @dlci: DLCI sending the data
* @msg: message queued
*
* Add data to the transmit queue and try and get stuff moving
* out of the mux tty if not already doing so. The Caller must hold
* the gsm tx lock.
*/
static void __gsm_data_queue(struct gsm_dlci *dlci, struct gsm_msg *msg)
{
struct gsm_mux *gsm = dlci->gsm;
u8 *dp = msg->data;
u8 *fcs = dp + msg->len;
/* Fill in the header */
if (gsm->encoding == GSM_BASIC_OPT) {
if (msg->len < 128)
*--dp = (msg->len << 1) | EA;
else {
*--dp = (msg->len >> 7); /* bits 7 - 15 */
*--dp = (msg->len & 127) << 1; /* bits 0 - 6 */
}
}
*--dp = msg->ctrl;
if (gsm->initiator)
*--dp = (msg->addr << 2) | CR | EA;
else
*--dp = (msg->addr << 2) | EA;
*fcs = gsm_fcs_add_block(INIT_FCS, dp , msg->data - dp);
/* Ugly protocol layering violation */
if (msg->ctrl == UI || msg->ctrl == (UI|PF))
*fcs = gsm_fcs_add_block(*fcs, msg->data, msg->len);
*fcs = 0xFF - *fcs;
gsm_print_packet("Q> ", msg->addr, gsm->initiator, msg->ctrl,
msg->data, msg->len);
/* Move the header back and adjust the length, also allow for the FCS
now tacked on the end */
msg->len += (msg->data - dp) + 1;
msg->data = dp;
/* Add to the actual output queue */
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
switch (msg->ctrl & ~PF) {
case UI:
case UIH:
if (msg->addr > 0) {
list_add_tail(&msg->list, &gsm->tx_data_list);
break;
}
fallthrough;
default:
list_add_tail(&msg->list, &gsm->tx_ctrl_list);
break;
}
gsm->tx_bytes += msg->len;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsmld_write_trigger(gsm);
mod_timer(&gsm->kick_timer, jiffies + 10 * gsm->t1 * HZ / 100);
}
/**
* gsm_data_queue - queue a UI or UIH frame
* @dlci: DLCI sending the data
* @msg: message queued
*
* Add data to the transmit queue and try and get stuff moving
* out of the mux tty if not already doing so. Take the
* the gsm tx lock and dlci lock.
*/
static void gsm_data_queue(struct gsm_dlci *dlci, struct gsm_msg *msg)
{
unsigned long flags;
spin_lock_irqsave(&dlci->gsm->tx_lock, flags);
__gsm_data_queue(dlci, msg);
spin_unlock_irqrestore(&dlci->gsm->tx_lock, flags);
}
/**
* gsm_dlci_data_output - try and push data out of a DLCI
* @gsm: mux
* @dlci: the DLCI to pull data from
*
* Pull data from a DLCI and send it into the transmit queue if there
* is data. Keep to the MRU of the mux. This path handles the usual tty
* interface which is a byte stream with optional modem data.
*
* Caller must hold the tx_lock of the mux.
*/
static int gsm_dlci_data_output(struct gsm_mux *gsm, struct gsm_dlci *dlci)
{
struct gsm_msg *msg;
u8 *dp;
int h, len, size;
/* for modem bits without break data */
h = ((dlci->adaption == 1) ? 0 : 1);
len = kfifo_len(&dlci->fifo);
if (len == 0)
return 0;
/* MTU/MRU count only the data bits but watch adaption mode */
if ((len + h) > dlci->mtu)
len = dlci->mtu - h;
size = len + h;
msg = gsm_data_alloc(gsm, dlci->addr, size, dlci->ftype);
if (!msg)
return -ENOMEM;
dp = msg->data;
switch (dlci->adaption) {
case 1: /* Unstructured */
break;
case 2: /* Unstructured with modem bits.
* Always one byte as we never send inline break data
*/
*dp++ = (gsm_encode_modem(dlci) << 1) | EA;
break;
default:
pr_err("%s: unsupported adaption %d\n", __func__,
dlci->adaption);
break;
}
WARN_ON(len != kfifo_out_locked(&dlci->fifo, dp, len,
&dlci->lock));
/* Notify upper layer about available send space. */
tty_port_tty_wakeup(&dlci->port);
__gsm_data_queue(dlci, msg);
/* Bytes of data we used up */
return size;
}
/**
* gsm_dlci_data_output_framed - try and push data out of a DLCI
* @gsm: mux
* @dlci: the DLCI to pull data from
*
* Pull data from a DLCI and send it into the transmit queue if there
* is data. Keep to the MRU of the mux. This path handles framed data
* queued as skbuffs to the DLCI.
*
* Caller must hold the tx_lock of the mux.
*/
static int gsm_dlci_data_output_framed(struct gsm_mux *gsm,
struct gsm_dlci *dlci)
{
struct gsm_msg *msg;
u8 *dp;
int len, size;
int last = 0, first = 0;
int overhead = 0;
/* One byte per frame is used for B/F flags */
if (dlci->adaption == 4)
overhead = 1;
/* dlci->skb is locked by tx_lock */
if (dlci->skb == NULL) {
dlci->skb = skb_dequeue_tail(&dlci->skb_list);
if (dlci->skb == NULL)
return 0;
first = 1;
}
len = dlci->skb->len + overhead;
/* MTU/MRU count only the data bits */
if (len > dlci->mtu) {
if (dlci->adaption == 3) {
/* Over long frame, bin it */
dev_kfree_skb_any(dlci->skb);
dlci->skb = NULL;
return 0;
}
len = dlci->mtu;
} else
last = 1;
size = len + overhead;
msg = gsm_data_alloc(gsm, dlci->addr, size, dlci->ftype);
if (msg == NULL) {
skb_queue_tail(&dlci->skb_list, dlci->skb);
dlci->skb = NULL;
return -ENOMEM;
}
dp = msg->data;
if (dlci->adaption == 4) { /* Interruptible framed (Packetised Data) */
/* Flag byte to carry the start/end info */
*dp++ = last << 7 | first << 6 | 1; /* EA */
len--;
}
memcpy(dp, dlci->skb->data, len);
skb_pull(dlci->skb, len);
__gsm_data_queue(dlci, msg);
if (last) {
dev_kfree_skb_any(dlci->skb);
dlci->skb = NULL;
}
return size;
}
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
/**
* gsm_dlci_modem_output - try and push modem status out of a DLCI
* @gsm: mux
* @dlci: the DLCI to pull modem status from
* @brk: break signal
*
* Push an empty frame in to the transmit queue to update the modem status
* bits and to transmit an optional break.
*
* Caller must hold the tx_lock of the mux.
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
*/
static int gsm_dlci_modem_output(struct gsm_mux *gsm, struct gsm_dlci *dlci,
u8 brk)
{
u8 *dp = NULL;
struct gsm_msg *msg;
int size = 0;
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
/* for modem bits without break data */
switch (dlci->adaption) {
case 1: /* Unstructured */
break;
case 2: /* Unstructured with modem bits. */
size++;
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
if (brk > 0)
size++;
break;
default:
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
pr_err("%s: unsupported adaption %d\n", __func__,
dlci->adaption);
return -EINVAL;
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
}
msg = gsm_data_alloc(gsm, dlci->addr, size, dlci->ftype);
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
if (!msg) {
pr_err("%s: gsm_data_alloc error", __func__);
return -ENOMEM;
}
dp = msg->data;
switch (dlci->adaption) {
case 1: /* Unstructured */
break;
case 2: /* Unstructured with modem bits. */
if (brk == 0) {
*dp++ = (gsm_encode_modem(dlci) << 1) | EA;
} else {
*dp++ = gsm_encode_modem(dlci) << 1;
*dp++ = (brk << 4) | 2 | EA; /* Length, Break, EA */
}
break;
default:
/* Handled above */
break;
}
__gsm_data_queue(dlci, msg);
return size;
}
/**
* gsm_dlci_data_sweep - look for data to send
* @gsm: the GSM mux
*
* Sweep the GSM mux channels in priority order looking for ones with
* data to send. We could do with optimising this scan a bit. We aim
* to fill the queue totally or up to TX_THRESH_HI bytes. Once we hit
* TX_THRESH_LO we get called again
*
* FIXME: We should round robin between groups and in theory you can
* renegotiate DLCI priorities with optional stuff. Needs optimising.
*/
static int gsm_dlci_data_sweep(struct gsm_mux *gsm)
{
/* Priority ordering: We should do priority with RR of the groups */
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
int i, len, ret = 0;
bool sent;
struct gsm_dlci *dlci;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
while (gsm->tx_bytes < TX_THRESH_HI) {
for (sent = false, i = 1; i < NUM_DLCI; i++) {
dlci = gsm->dlci[i];
/* skip unused or blocked channel */
if (!dlci || dlci->constipated)
continue;
/* skip channels with invalid state */
if (dlci->state != DLCI_OPEN)
continue;
/* count the sent data per adaption */
if (dlci->adaption < 3 && !dlci->net)
len = gsm_dlci_data_output(gsm, dlci);
else
len = gsm_dlci_data_output_framed(gsm, dlci);
/* on error exit */
if (len < 0)
return ret;
if (len > 0) {
ret++;
sent = true;
/* The lower DLCs can starve the higher DLCs! */
break;
}
/* try next */
}
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
if (!sent)
break;
}
return ret;
}
/**
* gsm_dlci_data_kick - transmit if possible
* @dlci: DLCI to kick
*
* Transmit data from this DLCI if the queue is empty. We can't rely on
* a tty wakeup except when we filled the pipe so we need to fire off
* new data ourselves in other cases.
*/
static void gsm_dlci_data_kick(struct gsm_dlci *dlci)
{
unsigned long flags;
int sweep;
if (dlci->constipated)
return;
spin_lock_irqsave(&dlci->gsm->tx_lock, flags);
/* If we have nothing running then we need to fire up */
sweep = (dlci->gsm->tx_bytes < TX_THRESH_LO);
if (dlci->gsm->tx_bytes == 0) {
if (dlci->net)
gsm_dlci_data_output_framed(dlci->gsm, dlci);
else
gsm_dlci_data_output(dlci->gsm, dlci);
}
if (sweep)
gsm_dlci_data_sweep(dlci->gsm);
spin_unlock_irqrestore(&dlci->gsm->tx_lock, flags);
}
/*
* Control message processing
*/
/**
* gsm_control_command - send a command frame to a control
* @gsm: gsm channel
* @cmd: the command to use
* @data: data to follow encoded info
* @dlen: length of data
*
* Encode up and queue a UI/UIH frame containing our command.
*/
static int gsm_control_command(struct gsm_mux *gsm, int cmd, const u8 *data,
int dlen)
{
struct gsm_msg *msg;
msg = gsm_data_alloc(gsm, 0, dlen + 2, gsm->dlci[0]->ftype);
if (msg == NULL)
return -ENOMEM;
msg->data[0] = (cmd << 1) | CR | EA; /* Set C/R */
msg->data[1] = (dlen << 1) | EA;
memcpy(msg->data + 2, data, dlen);
gsm_data_queue(gsm->dlci[0], msg);
return 0;
}
/**
* gsm_control_reply - send a response frame to a control
* @gsm: gsm channel
* @cmd: the command to use
* @data: data to follow encoded info
* @dlen: length of data
*
* Encode up and queue a UI/UIH frame containing our response.
*/
static void gsm_control_reply(struct gsm_mux *gsm, int cmd, const u8 *data,
int dlen)
{
struct gsm_msg *msg;
msg = gsm_data_alloc(gsm, 0, dlen + 2, gsm->dlci[0]->ftype);
if (msg == NULL)
return;
msg->data[0] = (cmd & 0xFE) << 1 | EA; /* Clear C/R */
msg->data[1] = (dlen << 1) | EA;
memcpy(msg->data + 2, data, dlen);
gsm_data_queue(gsm->dlci[0], msg);
}
/**
* gsm_process_modem - process received modem status
* @tty: virtual tty bound to the DLCI
* @dlci: DLCI to affect
* @modem: modem bits (full EA)
* @slen: number of signal octets
*
* Used when a modem control message or line state inline in adaption
* layer 2 is processed. Sort out the local modem state and throttles
*/
static void gsm_process_modem(struct tty_struct *tty, struct gsm_dlci *dlci,
u32 modem, int slen)
{
int mlines = 0;
u8 brk = 0;
int fc;
/* The modem status command can either contain one octet (V.24 signals)
* or two octets (V.24 signals + break signals). This is specified in
* section 5.4.6.3.7 of the 07.10 mux spec.
*/
if (slen == 1)
modem = modem & 0x7f;
else {
brk = modem & 0x7f;
modem = (modem >> 7) & 0x7f;
}
/* Flow control/ready to communicate */
fc = (modem & MDM_FC) || !(modem & MDM_RTR);
if (fc && !dlci->constipated) {
/* Need to throttle our output on this device */
dlci->constipated = true;
} else if (!fc && dlci->constipated) {
dlci->constipated = false;
gsm_dlci_data_kick(dlci);
}
/* Map modem bits */
if (modem & MDM_RTC)
mlines |= TIOCM_DSR | TIOCM_DTR;
if (modem & MDM_RTR)
mlines |= TIOCM_RTS | TIOCM_CTS;
if (modem & MDM_IC)
mlines |= TIOCM_RI;
if (modem & MDM_DV)
mlines |= TIOCM_CD;
/* Carrier drop -> hangup */
if (tty) {
if ((mlines & TIOCM_CD) == 0 && (dlci->modem_rx & TIOCM_CD))
if (!C_CLOCAL(tty))
tty_hangup(tty);
}
if (brk & 0x01)
tty_insert_flip_char(&dlci->port, 0, TTY_BREAK);
dlci->modem_rx = mlines;
wake_up_interruptible(&dlci->gsm->event);
}
/**
* gsm_process_negotiation - process received parameters
* @gsm: GSM channel
* @addr: DLCI address
* @cr: command/response
* @params: encoded parameters from the parameter negotiation message
*
* Used when the response for our parameter negotiation command was
* received.
*/
static int gsm_process_negotiation(struct gsm_mux *gsm, unsigned int addr,
unsigned int cr,
const struct gsm_dlci_param_bits *params)
{
struct gsm_dlci *dlci = gsm->dlci[addr];
unsigned int ftype, i, adaption, prio, n1, k;
i = FIELD_GET(PN_I_CL_FIELD_FTYPE, params->i_cl_bits);
adaption = FIELD_GET(PN_I_CL_FIELD_ADAPTION, params->i_cl_bits) + 1;
prio = FIELD_GET(PN_P_FIELD_PRIO, params->p_bits);
n1 = FIELD_GET(PN_N_FIELD_N1, get_unaligned_le16(&params->n_bits));
k = FIELD_GET(PN_K_FIELD_K, params->k_bits);
if (n1 < MIN_MTU) {
if (debug & DBG_ERRORS)
pr_info("%s N1 out of range in PN\n", __func__);
return -EINVAL;
}
switch (i) {
case 0x00:
ftype = UIH;
break;
case 0x01:
ftype = UI;
break;
case 0x02: /* I frames are not supported */
if (debug & DBG_ERRORS)
pr_info("%s unsupported I frame request in PN\n",
__func__);
return -EINVAL;
default:
if (debug & DBG_ERRORS)
pr_info("%s i out of range in PN\n", __func__);
return -EINVAL;
}
if (!cr && gsm->initiator) {
if (adaption != dlci->adaption) {
if (debug & DBG_ERRORS)
pr_info("%s invalid adaption %d in PN\n",
__func__, adaption);
return -EINVAL;
}
if (prio != dlci->prio) {
if (debug & DBG_ERRORS)
pr_info("%s invalid priority %d in PN",
__func__, prio);
return -EINVAL;
}
if (n1 > gsm->mru || n1 > dlci->mtu) {
/* We requested a frame size but the other party wants
* to send larger frames. The standard allows only a
* smaller response value than requested (5.4.6.3.1).
*/
if (debug & DBG_ERRORS)
pr_info("%s invalid N1 %d in PN\n", __func__,
n1);
return -EINVAL;
}
dlci->mtu = n1;
if (ftype != dlci->ftype) {
if (debug & DBG_ERRORS)
pr_info("%s invalid i %d in PN\n", __func__, i);
return -EINVAL;
}
if (ftype != UI && ftype != UIH && k > dlci->k) {
if (debug & DBG_ERRORS)
pr_info("%s invalid k %d in PN\n", __func__, k);
return -EINVAL;
}
dlci->k = k;
} else if (cr && !gsm->initiator) {
/* Only convergence layer type 1 and 2 are supported. */
if (adaption != 1 && adaption != 2) {
if (debug & DBG_ERRORS)
pr_info("%s invalid adaption %d in PN\n",
__func__, adaption);
return -EINVAL;
}
dlci->adaption = adaption;
if (n1 > gsm->mru) {
/* Propose a smaller value */
dlci->mtu = gsm->mru;
} else if (n1 > MAX_MTU) {
/* Propose a smaller value */
dlci->mtu = MAX_MTU;
} else {
dlci->mtu = n1;
}
dlci->prio = prio;
dlci->ftype = ftype;
dlci->k = k;
} else {
return -EINVAL;
}
return 0;
}
/**
* gsm_control_modem - modem status received
* @gsm: GSM channel
* @data: data following command
* @clen: command length
*
* We have received a modem status control message. This is used by
* the GSM mux protocol to pass virtual modem line status and optionally
* to indicate break signals. Unpack it, convert to Linux representation
* and if need be stuff a break message down the tty.
*/
static void gsm_control_modem(struct gsm_mux *gsm, const u8 *data, int clen)
{
unsigned int addr = 0;
unsigned int modem = 0;
struct gsm_dlci *dlci;
int len = clen;
int cl = clen;
const u8 *dp = data;
struct tty_struct *tty;
len = gsm_read_ea_val(&addr, data, cl);
if (len < 1)
return;
addr >>= 1;
/* Closed port, or invalid ? */
if (addr == 0 || addr >= NUM_DLCI || gsm->dlci[addr] == NULL)
return;
dlci = gsm->dlci[addr];
/* Must be at least one byte following the EA */
if ((cl - len) < 1)
return;
dp += len;
cl -= len;
/* get the modem status */
len = gsm_read_ea_val(&modem, dp, cl);
if (len < 1)
return;
tty = tty_port_tty_get(&dlci->port);
gsm_process_modem(tty, dlci, modem, cl);
if (tty) {
tty_wakeup(tty);
tty_kref_put(tty);
}
gsm_control_reply(gsm, CMD_MSC, data, clen);
}
/**
* gsm_control_negotiation - parameter negotiation received
* @gsm: GSM channel
* @cr: command/response flag
* @data: data following command
* @dlen: data length
*
* We have received a parameter negotiation message. This is used by
* the GSM mux protocol to configure protocol parameters for a new DLCI.
*/
static void gsm_control_negotiation(struct gsm_mux *gsm, unsigned int cr,
const u8 *data, unsigned int dlen)
{
unsigned int addr;
struct gsm_dlci_param_bits pn_reply;
struct gsm_dlci *dlci;
struct gsm_dlci_param_bits *params;
if (dlen < sizeof(struct gsm_dlci_param_bits))
return;
/* Invalid DLCI? */
params = (struct gsm_dlci_param_bits *)data;
addr = FIELD_GET(PN_D_FIELD_DLCI, params->d_bits);
if (addr == 0 || addr >= NUM_DLCI || !gsm->dlci[addr])
return;
dlci = gsm->dlci[addr];
/* Too late for parameter negotiation? */
if ((!cr && dlci->state == DLCI_OPENING) || dlci->state == DLCI_OPEN)
return;
/* Process the received parameters */
if (gsm_process_negotiation(gsm, addr, cr, params) != 0) {
/* Negotiation failed. Close the link. */
if (debug & DBG_ERRORS)
pr_info("%s PN failed\n", __func__);
gsm_dlci_close(dlci);
return;
}
if (cr) {
/* Reply command with accepted parameters. */
if (gsm_encode_params(dlci, &pn_reply) == 0)
gsm_control_reply(gsm, CMD_PN, (const u8 *)&pn_reply,
sizeof(pn_reply));
else if (debug & DBG_ERRORS)
pr_info("%s PN invalid\n", __func__);
} else if (dlci->state == DLCI_CONFIGURE) {
/* Proceed with link setup by sending SABM before UA */
dlci->state = DLCI_OPENING;
gsm_command(gsm, dlci->addr, SABM|PF);
mod_timer(&dlci->t1, jiffies + gsm->t1 * HZ / 100);
} else {
if (debug & DBG_ERRORS)
pr_info("%s PN in invalid state\n", __func__);
}
}
/**
* gsm_control_rls - remote line status
* @gsm: GSM channel
* @data: data bytes
* @clen: data length
*
* The modem sends us a two byte message on the control channel whenever
* it wishes to send us an error state from the virtual link. Stuff
* this into the uplink tty if present
*/
static void gsm_control_rls(struct gsm_mux *gsm, const u8 *data, int clen)
{
struct tty_port *port;
unsigned int addr = 0;
u8 bits;
int len = clen;
const u8 *dp = data;
while (gsm_read_ea(&addr, *dp++) == 0) {
len--;
if (len == 0)
return;
}
/* Must be at least one byte following ea */
len--;
if (len <= 0)
return;
addr >>= 1;
/* Closed port, or invalid ? */
if (addr == 0 || addr >= NUM_DLCI || gsm->dlci[addr] == NULL)
return;
/* No error ? */
bits = *dp;
if ((bits & 1) == 0)
return;
port = &gsm->dlci[addr]->port;
if (bits & 2)
tty_insert_flip_char(port, 0, TTY_OVERRUN);
if (bits & 4)
tty_insert_flip_char(port, 0, TTY_PARITY);
if (bits & 8)
tty_insert_flip_char(port, 0, TTY_FRAME);
tty_flip_buffer_push(port);
gsm_control_reply(gsm, CMD_RLS, data, clen);
}
static void gsm_dlci_begin_close(struct gsm_dlci *dlci);
/**
* gsm_control_message - DLCI 0 control processing
* @gsm: our GSM mux
* @command: the command EA
* @data: data beyond the command/length EAs
* @clen: length
*
* Input processor for control messages from the other end of the link.
* Processes the incoming request and queues a response frame or an
* NSC response if not supported
*/
static void gsm_control_message(struct gsm_mux *gsm, unsigned int command,
const u8 *data, int clen)
{
u8 buf[1];
switch (command) {
case CMD_CLD: {
tty: n_gsm: fix broken virtual tty handling Dynamic virtual tty registration was introduced to allow the user to handle these cases with uevent rules. The following commits relate to this: Commit 5b87686e3203 ("tty: n_gsm: Modify gsmtty driver register method when config requester") Commit 0b91b5332368 ("tty: n_gsm: Save dlci address open status when config requester") Commit 46292622ad73 ("tty: n_gsm: clean up indenting in gsm_queue()") However, the following behavior can be seen with this implementation: - n_gsm ldisc is activated via ioctl - all configuration parameters are set to their default value (initiator=0) - the mux gets activated and attached and gsmtty0 is being registered in in gsm_dlci_open() after DLCI 0 was established (DLCI 0 is the control channel) - the user configures n_gsm via ioctl GSMIOC_SETCONF as initiator - this re-attaches the n_gsm mux - no new gsmtty devices are registered in gsmld_attach_gsm() because the mux is already active - the initiator side registered only the control channel as gsmtty0 (which should never happen) and no user channel tty The commits above make it impossible to operate the initiator side as no user channel tty is or will be available. On the other hand, this behavior will make it also impossible to allow DLCI parameter negotiation on responder side in the future. The responder side first needs to provide a device for the application before the application can set its parameters of the associated DLCI via ioctl. Note that the user application is still able to detect a link establishment without relaying to uevent by waiting for DTR open on responder side. This is the same behavior as on a physical serial interface. And on initiator side a tty hangup can be detected if a link establishment request failed. Revert the commits above completely to always register all user channels and no control channel after mux attachment. No other changes are made. Fixes: 5b87686e3203 ("tty: n_gsm: Modify gsmtty driver register method when config requester") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:23 +00:00
struct gsm_dlci *dlci = gsm->dlci[0];
/* Modem wishes to close down */
if (dlci) {
dlci->dead = true;
gsm->dead = true;
tty: n_gsm: fix broken virtual tty handling Dynamic virtual tty registration was introduced to allow the user to handle these cases with uevent rules. The following commits relate to this: Commit 5b87686e3203 ("tty: n_gsm: Modify gsmtty driver register method when config requester") Commit 0b91b5332368 ("tty: n_gsm: Save dlci address open status when config requester") Commit 46292622ad73 ("tty: n_gsm: clean up indenting in gsm_queue()") However, the following behavior can be seen with this implementation: - n_gsm ldisc is activated via ioctl - all configuration parameters are set to their default value (initiator=0) - the mux gets activated and attached and gsmtty0 is being registered in in gsm_dlci_open() after DLCI 0 was established (DLCI 0 is the control channel) - the user configures n_gsm via ioctl GSMIOC_SETCONF as initiator - this re-attaches the n_gsm mux - no new gsmtty devices are registered in gsmld_attach_gsm() because the mux is already active - the initiator side registered only the control channel as gsmtty0 (which should never happen) and no user channel tty The commits above make it impossible to operate the initiator side as no user channel tty is or will be available. On the other hand, this behavior will make it also impossible to allow DLCI parameter negotiation on responder side in the future. The responder side first needs to provide a device for the application before the application can set its parameters of the associated DLCI via ioctl. Note that the user application is still able to detect a link establishment without relaying to uevent by waiting for DTR open on responder side. This is the same behavior as on a physical serial interface. And on initiator side a tty hangup can be detected if a link establishment request failed. Revert the commits above completely to always register all user channels and no control channel after mux attachment. No other changes are made. Fixes: 5b87686e3203 ("tty: n_gsm: Modify gsmtty driver register method when config requester") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:23 +00:00
gsm_dlci_begin_close(dlci);
}
}
break;
case CMD_TEST:
/* Modem wishes to test, reply with the data */
gsm_control_reply(gsm, CMD_TEST, data, clen);
break;
case CMD_FCON:
/* Modem can accept data again */
gsm->constipated = false;
gsm_control_reply(gsm, CMD_FCON, NULL, 0);
/* Kick the link in case it is idling */
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsmld_write_trigger(gsm);
break;
case CMD_FCOFF:
/* Modem wants us to STFU */
gsm->constipated = true;
gsm_control_reply(gsm, CMD_FCOFF, NULL, 0);
break;
case CMD_MSC:
/* Out of band modem line change indicator for a DLCI */
gsm_control_modem(gsm, data, clen);
break;
case CMD_RLS:
/* Out of band error reception for a DLCI */
gsm_control_rls(gsm, data, clen);
break;
case CMD_PSC:
/* Modem wishes to enter power saving state */
gsm_control_reply(gsm, CMD_PSC, NULL, 0);
break;
/* Optional commands */
case CMD_PN:
/* Modem sends a parameter negotiation command */
gsm_control_negotiation(gsm, 1, data, clen);
break;
/* Optional unsupported commands */
case CMD_RPN: /* Remote port negotiation */
case CMD_SNC: /* Service negotiation command */
default:
/* Reply to bad commands with an NSC */
buf[0] = command;
gsm_control_reply(gsm, CMD_NSC, buf, 1);
break;
}
}
/**
* gsm_control_response - process a response to our control
* @gsm: our GSM mux
* @command: the command (response) EA
* @data: data beyond the command/length EA
* @clen: length
*
* Process a response to an outstanding command. We only allow a single
* control message in flight so this is fairly easy. All the clean up
* is done by the caller, we just update the fields, flag it as done
* and return
*/
static void gsm_control_response(struct gsm_mux *gsm, unsigned int command,
const u8 *data, int clen)
{
struct gsm_control *ctrl;
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
struct gsm_dlci *dlci;
unsigned long flags;
spin_lock_irqsave(&gsm->control_lock, flags);
ctrl = gsm->pending_cmd;
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
dlci = gsm->dlci[0];
command |= 1;
/* Does the reply match our command */
if (ctrl != NULL && (command == ctrl->cmd || command == CMD_NSC)) {
/* Our command was replied to, kill the retry timer */
del_timer(&gsm->t2_timer);
gsm->pending_cmd = NULL;
/* Rejected by the other end */
if (command == CMD_NSC)
ctrl->error = -EOPNOTSUPP;
ctrl->done = 1;
wake_up(&gsm->event);
/* Or did we receive the PN response to our PN command */
} else if (command == CMD_PN) {
gsm_control_negotiation(gsm, 0, data, clen);
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
/* Or did we receive the TEST response to our TEST command */
} else if (command == CMD_TEST && clen == 1 && *data == gsm->ka_num) {
gsm->ka_retries = -1; /* trigger new keep-alive message */
if (dlci && !dlci->dead)
mod_timer(&gsm->ka_timer, jiffies + gsm->keep_alive * HZ / 100);
}
spin_unlock_irqrestore(&gsm->control_lock, flags);
}
/**
* gsm_control_keep_alive - check timeout or start keep-alive
* @t: timer contained in our gsm object
*
* Called off the keep-alive timer expiry signaling that our link
* partner is not responding anymore. Link will be closed.
* This is also called to startup our timer.
*/
static void gsm_control_keep_alive(struct timer_list *t)
{
struct gsm_mux *gsm = from_timer(gsm, t, ka_timer);
unsigned long flags;
spin_lock_irqsave(&gsm->control_lock, flags);
if (gsm->ka_num && gsm->ka_retries == 0) {
/* Keep-alive expired -> close the link */
if (debug & DBG_ERRORS)
pr_debug("%s keep-alive timed out\n", __func__);
spin_unlock_irqrestore(&gsm->control_lock, flags);
if (gsm->dlci[0])
gsm_dlci_begin_close(gsm->dlci[0]);
return;
} else if (gsm->keep_alive && gsm->dlci[0] && !gsm->dlci[0]->dead) {
if (gsm->ka_retries > 0) {
/* T2 expired for keep-alive -> resend */
gsm->ka_retries--;
} else {
/* Start keep-alive timer */
gsm->ka_num++;
if (!gsm->ka_num)
gsm->ka_num++;
gsm->ka_retries = (signed int)gsm->n2;
}
gsm_control_command(gsm, CMD_TEST, &gsm->ka_num,
sizeof(gsm->ka_num));
mod_timer(&gsm->ka_timer,
jiffies + gsm->t2 * HZ / 100);
}
spin_unlock_irqrestore(&gsm->control_lock, flags);
}
/**
* gsm_control_transmit - send control packet
* @gsm: gsm mux
* @ctrl: frame to send
*
* Send out a pending control command (called under control lock)
*/
static void gsm_control_transmit(struct gsm_mux *gsm, struct gsm_control *ctrl)
{
gsm_control_command(gsm, ctrl->cmd, ctrl->data, ctrl->len);
}
/**
* gsm_control_retransmit - retransmit a control frame
* @t: timer contained in our gsm object
*
* Called off the T2 timer expiry in order to retransmit control frames
* that have been lost in the system somewhere. The control_lock protects
* us from colliding with another sender or a receive completion event.
* In that situation the timer may still occur in a small window but
* gsm->pending_cmd will be NULL and we just let the timer expire.
*/
treewide: setup_timer() -> timer_setup() This converts all remaining cases of the old setup_timer() API into using timer_setup(), where the callback argument is the structure already holding the struct timer_list. These should have no behavioral changes, since they just change which pointer is passed into the callback with the same available pointers after conversion. It handles the following examples, in addition to some other variations. Casting from unsigned long: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... setup_timer(&ptr->my_timer, my_callback, ptr); and forced object casts: void my_callback(struct something *ptr) { ... } ... setup_timer(&ptr->my_timer, my_callback, (unsigned long)ptr); become: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... timer_setup(&ptr->my_timer, my_callback, 0); Direct function assignments: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... ptr->my_timer.function = my_callback; have a temporary cast added, along with converting the args: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... ptr->my_timer.function = (TIMER_FUNC_TYPE)my_callback; And finally, callbacks without a data assignment: void my_callback(unsigned long data) { ... } ... setup_timer(&ptr->my_timer, my_callback, 0); have their argument renamed to verify they're unused during conversion: void my_callback(struct timer_list *unused) { ... } ... timer_setup(&ptr->my_timer, my_callback, 0); The conversion is done with the following Coccinelle script: spatch --very-quiet --all-includes --include-headers \ -I ./arch/x86/include -I ./arch/x86/include/generated \ -I ./include -I ./arch/x86/include/uapi \ -I ./arch/x86/include/generated/uapi -I ./include/uapi \ -I ./include/generated/uapi --include ./include/linux/kconfig.h \ --dir . \ --cocci-file ~/src/data/timer_setup.cocci @fix_address_of@ expression e; @@ setup_timer( -&(e) +&e , ...) // Update any raw setup_timer() usages that have a NULL callback, but // would otherwise match change_timer_function_usage, since the latter // will update all function assignments done in the face of a NULL // function initialization in setup_timer(). @change_timer_function_usage_NULL@ expression _E; identifier _timer; type _cast_data; @@ ( -setup_timer(&_E->_timer, NULL, _E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E->_timer, NULL, (_cast_data)_E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E._timer, NULL, &_E); +timer_setup(&_E._timer, NULL, 0); | -setup_timer(&_E._timer, NULL, (_cast_data)&_E); +timer_setup(&_E._timer, NULL, 0); ) @change_timer_function_usage@ expression _E; identifier _timer; struct timer_list _stl; identifier _callback; type _cast_func, _cast_data; @@ ( -setup_timer(&_E->_timer, _callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | _E->_timer@_stl.function = _callback; | _E->_timer@_stl.function = &_callback; | _E->_timer@_stl.function = (_cast_func)_callback; | _E->_timer@_stl.function = (_cast_func)&_callback; | _E._timer@_stl.function = _callback; | _E._timer@_stl.function = &_callback; | _E._timer@_stl.function = (_cast_func)_callback; | _E._timer@_stl.function = (_cast_func)&_callback; ) // callback(unsigned long arg) @change_callback_handle_cast depends on change_timer_function_usage@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; identifier _handle; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { ( ... when != _origarg _handletype *_handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg ) } // callback(unsigned long arg) without existing variable @change_callback_handle_cast_no_arg depends on change_timer_function_usage && !change_callback_handle_cast@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { + _handletype *_origarg = from_timer(_origarg, t, _timer); + ... when != _origarg - (_handletype *)_origarg + _origarg ... when != _origarg } // Avoid already converted callbacks. @match_callback_converted depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier t; @@ void _callback(struct timer_list *t) { ... } // callback(struct something *handle) @change_callback_handle_arg depends on change_timer_function_usage && !match_callback_converted && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; @@ void _callback( -_handletype *_handle +struct timer_list *t ) { + _handletype *_handle = from_timer(_handle, t, _timer); ... } // If change_callback_handle_arg ran on an empty function, remove // the added handler. @unchange_callback_handle_arg depends on change_timer_function_usage && change_callback_handle_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; identifier t; @@ void _callback(struct timer_list *t) { - _handletype *_handle = from_timer(_handle, t, _timer); } // We only want to refactor the setup_timer() data argument if we've found // the matching callback. This undoes changes in change_timer_function_usage. @unchange_timer_function_usage depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg && !change_callback_handle_arg@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type change_timer_function_usage._cast_data; @@ ( -timer_setup(&_E->_timer, _callback, 0); +setup_timer(&_E->_timer, _callback, (_cast_data)_E); | -timer_setup(&_E._timer, _callback, 0); +setup_timer(&_E._timer, _callback, (_cast_data)&_E); ) // If we fixed a callback from a .function assignment, fix the // assignment cast now. @change_timer_function_assignment depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_func; typedef TIMER_FUNC_TYPE; @@ ( _E->_timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -&_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)_callback; +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -&_callback; +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; ) // Sometimes timer functions are called directly. Replace matched args. @change_timer_function_calls depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression _E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_data; @@ _callback( ( -(_cast_data)_E +&_E->_timer | -(_cast_data)&_E +&_E._timer | -_E +&_E->_timer ) ) // If a timer has been configured without a data argument, it can be // converted without regard to the callback argument, since it is unused. @match_timer_function_unused_data@ expression _E; identifier _timer; identifier _callback; @@ ( -setup_timer(&_E->_timer, _callback, 0); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0L); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0UL); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0L); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0UL); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_timer, _callback, 0); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0L); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0UL); +timer_setup(&_timer, _callback, 0); | -setup_timer(_timer, _callback, 0); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0L); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0UL); +timer_setup(_timer, _callback, 0); ) @change_callback_unused_data depends on match_timer_function_unused_data@ identifier match_timer_function_unused_data._callback; type _origtype; identifier _origarg; @@ void _callback( -_origtype _origarg +struct timer_list *unused ) { ... when != _origarg } Signed-off-by: Kees Cook <keescook@chromium.org>
2017-10-16 21:43:17 +00:00
static void gsm_control_retransmit(struct timer_list *t)
{
treewide: setup_timer() -> timer_setup() This converts all remaining cases of the old setup_timer() API into using timer_setup(), where the callback argument is the structure already holding the struct timer_list. These should have no behavioral changes, since they just change which pointer is passed into the callback with the same available pointers after conversion. It handles the following examples, in addition to some other variations. Casting from unsigned long: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... setup_timer(&ptr->my_timer, my_callback, ptr); and forced object casts: void my_callback(struct something *ptr) { ... } ... setup_timer(&ptr->my_timer, my_callback, (unsigned long)ptr); become: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... timer_setup(&ptr->my_timer, my_callback, 0); Direct function assignments: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... ptr->my_timer.function = my_callback; have a temporary cast added, along with converting the args: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... ptr->my_timer.function = (TIMER_FUNC_TYPE)my_callback; And finally, callbacks without a data assignment: void my_callback(unsigned long data) { ... } ... setup_timer(&ptr->my_timer, my_callback, 0); have their argument renamed to verify they're unused during conversion: void my_callback(struct timer_list *unused) { ... } ... timer_setup(&ptr->my_timer, my_callback, 0); The conversion is done with the following Coccinelle script: spatch --very-quiet --all-includes --include-headers \ -I ./arch/x86/include -I ./arch/x86/include/generated \ -I ./include -I ./arch/x86/include/uapi \ -I ./arch/x86/include/generated/uapi -I ./include/uapi \ -I ./include/generated/uapi --include ./include/linux/kconfig.h \ --dir . \ --cocci-file ~/src/data/timer_setup.cocci @fix_address_of@ expression e; @@ setup_timer( -&(e) +&e , ...) // Update any raw setup_timer() usages that have a NULL callback, but // would otherwise match change_timer_function_usage, since the latter // will update all function assignments done in the face of a NULL // function initialization in setup_timer(). @change_timer_function_usage_NULL@ expression _E; identifier _timer; type _cast_data; @@ ( -setup_timer(&_E->_timer, NULL, _E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E->_timer, NULL, (_cast_data)_E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E._timer, NULL, &_E); +timer_setup(&_E._timer, NULL, 0); | -setup_timer(&_E._timer, NULL, (_cast_data)&_E); +timer_setup(&_E._timer, NULL, 0); ) @change_timer_function_usage@ expression _E; identifier _timer; struct timer_list _stl; identifier _callback; type _cast_func, _cast_data; @@ ( -setup_timer(&_E->_timer, _callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | _E->_timer@_stl.function = _callback; | _E->_timer@_stl.function = &_callback; | _E->_timer@_stl.function = (_cast_func)_callback; | _E->_timer@_stl.function = (_cast_func)&_callback; | _E._timer@_stl.function = _callback; | _E._timer@_stl.function = &_callback; | _E._timer@_stl.function = (_cast_func)_callback; | _E._timer@_stl.function = (_cast_func)&_callback; ) // callback(unsigned long arg) @change_callback_handle_cast depends on change_timer_function_usage@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; identifier _handle; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { ( ... when != _origarg _handletype *_handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg ) } // callback(unsigned long arg) without existing variable @change_callback_handle_cast_no_arg depends on change_timer_function_usage && !change_callback_handle_cast@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { + _handletype *_origarg = from_timer(_origarg, t, _timer); + ... when != _origarg - (_handletype *)_origarg + _origarg ... when != _origarg } // Avoid already converted callbacks. @match_callback_converted depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier t; @@ void _callback(struct timer_list *t) { ... } // callback(struct something *handle) @change_callback_handle_arg depends on change_timer_function_usage && !match_callback_converted && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; @@ void _callback( -_handletype *_handle +struct timer_list *t ) { + _handletype *_handle = from_timer(_handle, t, _timer); ... } // If change_callback_handle_arg ran on an empty function, remove // the added handler. @unchange_callback_handle_arg depends on change_timer_function_usage && change_callback_handle_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; identifier t; @@ void _callback(struct timer_list *t) { - _handletype *_handle = from_timer(_handle, t, _timer); } // We only want to refactor the setup_timer() data argument if we've found // the matching callback. This undoes changes in change_timer_function_usage. @unchange_timer_function_usage depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg && !change_callback_handle_arg@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type change_timer_function_usage._cast_data; @@ ( -timer_setup(&_E->_timer, _callback, 0); +setup_timer(&_E->_timer, _callback, (_cast_data)_E); | -timer_setup(&_E._timer, _callback, 0); +setup_timer(&_E._timer, _callback, (_cast_data)&_E); ) // If we fixed a callback from a .function assignment, fix the // assignment cast now. @change_timer_function_assignment depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_func; typedef TIMER_FUNC_TYPE; @@ ( _E->_timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -&_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)_callback; +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -&_callback; +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; ) // Sometimes timer functions are called directly. Replace matched args. @change_timer_function_calls depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression _E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_data; @@ _callback( ( -(_cast_data)_E +&_E->_timer | -(_cast_data)&_E +&_E._timer | -_E +&_E->_timer ) ) // If a timer has been configured without a data argument, it can be // converted without regard to the callback argument, since it is unused. @match_timer_function_unused_data@ expression _E; identifier _timer; identifier _callback; @@ ( -setup_timer(&_E->_timer, _callback, 0); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0L); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0UL); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0L); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0UL); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_timer, _callback, 0); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0L); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0UL); +timer_setup(&_timer, _callback, 0); | -setup_timer(_timer, _callback, 0); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0L); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0UL); +timer_setup(_timer, _callback, 0); ) @change_callback_unused_data depends on match_timer_function_unused_data@ identifier match_timer_function_unused_data._callback; type _origtype; identifier _origarg; @@ void _callback( -_origtype _origarg +struct timer_list *unused ) { ... when != _origarg } Signed-off-by: Kees Cook <keescook@chromium.org>
2017-10-16 21:43:17 +00:00
struct gsm_mux *gsm = from_timer(gsm, t, t2_timer);
struct gsm_control *ctrl;
unsigned long flags;
spin_lock_irqsave(&gsm->control_lock, flags);
ctrl = gsm->pending_cmd;
if (ctrl) {
if (gsm->cretries == 0 || !gsm->dlci[0] || gsm->dlci[0]->dead) {
gsm->pending_cmd = NULL;
ctrl->error = -ETIMEDOUT;
ctrl->done = 1;
spin_unlock_irqrestore(&gsm->control_lock, flags);
wake_up(&gsm->event);
return;
}
gsm->cretries--;
gsm_control_transmit(gsm, ctrl);
mod_timer(&gsm->t2_timer, jiffies + gsm->t2 * HZ / 100);
}
spin_unlock_irqrestore(&gsm->control_lock, flags);
}
/**
* gsm_control_send - send a control frame on DLCI 0
* @gsm: the GSM channel
* @command: command to send including CR bit
* @data: bytes of data (must be kmalloced)
* @clen: length of the block to send
*
* Queue and dispatch a control command. Only one command can be
* active at a time. In theory more can be outstanding but the matching
* gets really complicated so for now stick to one outstanding.
*/
static struct gsm_control *gsm_control_send(struct gsm_mux *gsm,
unsigned int command, u8 *data, int clen)
{
struct gsm_control *ctrl = kzalloc(sizeof(struct gsm_control),
GFP_ATOMIC);
unsigned long flags;
if (ctrl == NULL)
return NULL;
retry:
wait_event(gsm->event, gsm->pending_cmd == NULL);
spin_lock_irqsave(&gsm->control_lock, flags);
if (gsm->pending_cmd != NULL) {
spin_unlock_irqrestore(&gsm->control_lock, flags);
goto retry;
}
ctrl->cmd = command;
ctrl->data = data;
ctrl->len = clen;
gsm->pending_cmd = ctrl;
tty: n_gsm: Fix long delays with control frame timeouts in ADM mode Commit ea3d8465ab9b ("tty: n_gsm: Allow ADM response in addition to UA for control dlci") added support for DLCI to stay in Asynchronous Disconnected Mode (ADM). But we still get long delays waiting for commands to other DLCI to complete: --> 5) C: SABM(P) Q> 0) C: UIH(F) Q> 0) C: UIH(F) Q> 0) C: UIH(F) ... This happens because gsm_control_send() sets cretries timer to T2 that is by default set to 34. This will cause resend for T2 times for the control frame. In ADM mode, we will never get a response so the control frame, so retries are just delaying all the commands. Let's fix the issue by setting DLCI_MODE_ADM flag after detecting the ADM mode for the control DLCI. Then we can use that in gsm_control_send() to set retries to 1. This means the control frame will be sent once allowing the other end at an opportunity to switch from ADM to ABM mode. Note that retries will be decremented in gsm_control_retransmit() so we don't want to set it to 0 here. Fixes: ea3d8465ab9b ("tty: n_gsm: Allow ADM response in addition to UA for control dlci") Cc: linux-serial@vger.kernel.org Cc: Alan Cox <alan@llwyncelyn.cymru> Cc: Dan Williams <dcbw@redhat.com> Cc: Jiri Prchal <jiri.prchal@aksignal.cz> Cc: Jiri Slaby <jslaby@suse.cz> Cc: Marcel Partap <mpartap@gmx.net> Cc: Merlijn Wajer <merlijn@wizzup.org> Cc: Michael Nazzareno Trimarchi <michael@amarulasolutions.com> Cc: Michael Scott <michael.scott@linaro.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Peter Hurley <peter@hurleysoftware.com> Cc: Russ Gorby <russ.gorby@intel.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sebastian Reichel <sre@kernel.org> Signed-off-by: Tony Lindgren <tony@atomide.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-04-07 17:19:50 +00:00
/* If DLCI0 is in ADM mode skip retries, it won't respond */
if (gsm->dlci[0]->mode == DLCI_MODE_ADM)
gsm->cretries = 0;
tty: n_gsm: Fix long delays with control frame timeouts in ADM mode Commit ea3d8465ab9b ("tty: n_gsm: Allow ADM response in addition to UA for control dlci") added support for DLCI to stay in Asynchronous Disconnected Mode (ADM). But we still get long delays waiting for commands to other DLCI to complete: --> 5) C: SABM(P) Q> 0) C: UIH(F) Q> 0) C: UIH(F) Q> 0) C: UIH(F) ... This happens because gsm_control_send() sets cretries timer to T2 that is by default set to 34. This will cause resend for T2 times for the control frame. In ADM mode, we will never get a response so the control frame, so retries are just delaying all the commands. Let's fix the issue by setting DLCI_MODE_ADM flag after detecting the ADM mode for the control DLCI. Then we can use that in gsm_control_send() to set retries to 1. This means the control frame will be sent once allowing the other end at an opportunity to switch from ADM to ABM mode. Note that retries will be decremented in gsm_control_retransmit() so we don't want to set it to 0 here. Fixes: ea3d8465ab9b ("tty: n_gsm: Allow ADM response in addition to UA for control dlci") Cc: linux-serial@vger.kernel.org Cc: Alan Cox <alan@llwyncelyn.cymru> Cc: Dan Williams <dcbw@redhat.com> Cc: Jiri Prchal <jiri.prchal@aksignal.cz> Cc: Jiri Slaby <jslaby@suse.cz> Cc: Marcel Partap <mpartap@gmx.net> Cc: Merlijn Wajer <merlijn@wizzup.org> Cc: Michael Nazzareno Trimarchi <michael@amarulasolutions.com> Cc: Michael Scott <michael.scott@linaro.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Peter Hurley <peter@hurleysoftware.com> Cc: Russ Gorby <russ.gorby@intel.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sebastian Reichel <sre@kernel.org> Signed-off-by: Tony Lindgren <tony@atomide.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-04-07 17:19:50 +00:00
else
gsm->cretries = gsm->n2;
mod_timer(&gsm->t2_timer, jiffies + gsm->t2 * HZ / 100);
gsm_control_transmit(gsm, ctrl);
spin_unlock_irqrestore(&gsm->control_lock, flags);
return ctrl;
}
/**
* gsm_control_wait - wait for a control to finish
* @gsm: GSM mux
* @control: control we are waiting on
*
* Waits for the control to complete or time out. Frees any used
* resources and returns 0 for success, or an error if the remote
* rejected or ignored the request.
*/
static int gsm_control_wait(struct gsm_mux *gsm, struct gsm_control *control)
{
int err;
wait_event(gsm->event, control->done == 1);
err = control->error;
kfree(control);
return err;
}
/*
* DLCI level handling: Needs krefs
*/
/*
* State transitions and timers
*/
/**
* gsm_dlci_close - a DLCI has closed
* @dlci: DLCI that closed
*
* Perform processing when moving a DLCI into closed state. If there
* is an attached tty this is hung up
*/
static void gsm_dlci_close(struct gsm_dlci *dlci)
{
del_timer(&dlci->t1);
if (debug & DBG_ERRORS)
pr_debug("DLCI %d goes closed.\n", dlci->addr);
dlci->state = DLCI_CLOSED;
/* Prevent us from sending data before the link is up again */
dlci->constipated = true;
if (dlci->addr != 0) {
tty_port_tty_hangup(&dlci->port, false);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsm_dlci_clear_queues(dlci->gsm, dlci);
/* Ensure that gsmtty_open() can return. */
tty_port_set_initialized(&dlci->port, false);
wake_up_interruptible(&dlci->port.open_wait);
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
} else {
del_timer(&dlci->gsm->ka_timer);
dlci->gsm->dead = true;
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
}
/* A DLCI 0 close is a MUX termination so we need to kick that
back to userspace somehow */
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsm_dlci_data_kick(dlci);
wake_up_all(&dlci->gsm->event);
}
/**
* gsm_dlci_open - a DLCI has opened
* @dlci: DLCI that opened
*
* Perform processing when moving a DLCI into open state.
*/
static void gsm_dlci_open(struct gsm_dlci *dlci)
{
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
struct gsm_mux *gsm = dlci->gsm;
/* Note that SABM UA .. SABM UA first UA lost can mean that we go
open -> open */
del_timer(&dlci->t1);
/* This will let a tty open continue */
dlci->state = DLCI_OPEN;
dlci->constipated = false;
if (debug & DBG_ERRORS)
pr_debug("DLCI %d goes open.\n", dlci->addr);
/* Send current modem state */
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
if (dlci->addr) {
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
gsm_modem_update(dlci, 0);
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
} else {
/* Start keep-alive control */
gsm->ka_num = 0;
gsm->ka_retries = -1;
mod_timer(&gsm->ka_timer,
jiffies + gsm->keep_alive * HZ / 100);
}
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsm_dlci_data_kick(dlci);
wake_up(&dlci->gsm->event);
}
/**
* gsm_dlci_negotiate - start parameter negotiation
* @dlci: DLCI to open
*
* Starts the parameter negotiation for the new DLCI. This needs to be done
* before the DLCI initialized the channel via SABM.
*/
static int gsm_dlci_negotiate(struct gsm_dlci *dlci)
{
struct gsm_mux *gsm = dlci->gsm;
struct gsm_dlci_param_bits params;
int ret;
ret = gsm_encode_params(dlci, &params);
if (ret != 0)
return ret;
/* We cannot asynchronous wait for the command response with
* gsm_command() and gsm_control_wait() at this point.
*/
ret = gsm_control_command(gsm, CMD_PN, (const u8 *)&params,
sizeof(params));
return ret;
}
/**
* gsm_dlci_t1 - T1 timer expiry
* @t: timer contained in the DLCI that opened
*
* The T1 timer handles retransmits of control frames (essentially of
* SABM and DISC). We resend the command until the retry count runs out
* in which case an opening port goes back to closed and a closing port
* is simply put into closed state (any further frames from the other
* end will get a DM response)
tty: n_gsm: Allow ADM response in addition to UA for control dlci Some devices have the control dlci stay in ADM mode instead of the UA mode. This can seen at least on droid 4 when trying to open the ts 27.010 mux port. Enabling n_gsm debug mode shows the control dlci always respond with DM to SABM instead of UA: # modprobe n_gsm debug=0xff # ldattach -d GSM0710 /dev/ttyS0 & gsmld_output: 00000000: f9 03 3f 01 1c f9 --> 0) C: SABM(P) gsmld_receive: 00000000: f9 03 1f 01 36 f9 <-- 0) C: DM(P) ... $ minicom -D /dev/gsmtty1 minicom: cannot open /dev/gsmtty1: No error information $ strace minicom -D /dev/gsmtty1 ... open("/dev/gsmtty1", O_RDWR|O_NOCTTY|O_NONBLOCK|O_LARGEFILE) = -1 EL2HLT Note that this is different issue from other n_gsm -EL2HLT issues such as timeouts when the control dlci does not respond at all. The ADM mode seems to be a quite common according to "RF Wireless World" article "GSM Issue-UE sends SABM and gets a DM response instead of UA response": This issue is most commonly observed in GSM networks where in UE sends SABM and expects network to send UA response but it ends up receiving DM response from the network. SABM stands for Set asynchronous balanced mode, UA stands for Unnumbered Acknowledge and DA stands for Disconnected Mode. An RLP entity can be in one of two modes: - Asynchronous Balanced Mode (ABM) - Asynchronous Disconnected Mode (ADM) Currently Linux kernel closes the control dlci after several retries in gsm_dlci_t1() on DM. This causes n_gsm /dev/gsmtty ports to produce error code -EL2HLT when trying to open them as the closing of control dlci has already set gsm->dead. Let's fix the issue by allowing control dlci stay in ADM mode after the retries so the /dev/gsmtty ports can be opened and used. It seems that it might take several attempts to get any response from the control dlci, so it's best to allow ADM mode only after the SABM retries are done. Note that for droid 4 additional patches are needed to mux the ttyS0 pins and to toggle RTS gpio_149 to wake up the mdm6600 modem are also needed to use n_gsm. And the mdm6600 modem needs to be powered on. Cc: linux-serial@vger.kernel.org Cc: Alan Cox <alan@llwyncelyn.cymru> Cc: Jiri Prchal <jiri.prchal@aksignal.cz> Cc: Jiri Slaby <jslaby@suse.cz> Cc: Marcel Partap <mpartap@gmx.net> Cc: Michael Scott <michael.scott@linaro.org> Cc: Peter Hurley <peter@hurleysoftware.com> Cc: Russ Gorby <russ.gorby@intel.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sebastian Reichel <sre@kernel.org> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-01-03 18:18:03 +00:00
*
* Some control dlci can stay in ADM mode with other dlci working just
* fine. In that case we can just keep the control dlci open after the
* DLCI_OPENING retries time out.
*/
treewide: setup_timer() -> timer_setup() This converts all remaining cases of the old setup_timer() API into using timer_setup(), where the callback argument is the structure already holding the struct timer_list. These should have no behavioral changes, since they just change which pointer is passed into the callback with the same available pointers after conversion. It handles the following examples, in addition to some other variations. Casting from unsigned long: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... setup_timer(&ptr->my_timer, my_callback, ptr); and forced object casts: void my_callback(struct something *ptr) { ... } ... setup_timer(&ptr->my_timer, my_callback, (unsigned long)ptr); become: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... timer_setup(&ptr->my_timer, my_callback, 0); Direct function assignments: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... ptr->my_timer.function = my_callback; have a temporary cast added, along with converting the args: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... ptr->my_timer.function = (TIMER_FUNC_TYPE)my_callback; And finally, callbacks without a data assignment: void my_callback(unsigned long data) { ... } ... setup_timer(&ptr->my_timer, my_callback, 0); have their argument renamed to verify they're unused during conversion: void my_callback(struct timer_list *unused) { ... } ... timer_setup(&ptr->my_timer, my_callback, 0); The conversion is done with the following Coccinelle script: spatch --very-quiet --all-includes --include-headers \ -I ./arch/x86/include -I ./arch/x86/include/generated \ -I ./include -I ./arch/x86/include/uapi \ -I ./arch/x86/include/generated/uapi -I ./include/uapi \ -I ./include/generated/uapi --include ./include/linux/kconfig.h \ --dir . \ --cocci-file ~/src/data/timer_setup.cocci @fix_address_of@ expression e; @@ setup_timer( -&(e) +&e , ...) // Update any raw setup_timer() usages that have a NULL callback, but // would otherwise match change_timer_function_usage, since the latter // will update all function assignments done in the face of a NULL // function initialization in setup_timer(). @change_timer_function_usage_NULL@ expression _E; identifier _timer; type _cast_data; @@ ( -setup_timer(&_E->_timer, NULL, _E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E->_timer, NULL, (_cast_data)_E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E._timer, NULL, &_E); +timer_setup(&_E._timer, NULL, 0); | -setup_timer(&_E._timer, NULL, (_cast_data)&_E); +timer_setup(&_E._timer, NULL, 0); ) @change_timer_function_usage@ expression _E; identifier _timer; struct timer_list _stl; identifier _callback; type _cast_func, _cast_data; @@ ( -setup_timer(&_E->_timer, _callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | _E->_timer@_stl.function = _callback; | _E->_timer@_stl.function = &_callback; | _E->_timer@_stl.function = (_cast_func)_callback; | _E->_timer@_stl.function = (_cast_func)&_callback; | _E._timer@_stl.function = _callback; | _E._timer@_stl.function = &_callback; | _E._timer@_stl.function = (_cast_func)_callback; | _E._timer@_stl.function = (_cast_func)&_callback; ) // callback(unsigned long arg) @change_callback_handle_cast depends on change_timer_function_usage@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; identifier _handle; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { ( ... when != _origarg _handletype *_handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg ) } // callback(unsigned long arg) without existing variable @change_callback_handle_cast_no_arg depends on change_timer_function_usage && !change_callback_handle_cast@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { + _handletype *_origarg = from_timer(_origarg, t, _timer); + ... when != _origarg - (_handletype *)_origarg + _origarg ... when != _origarg } // Avoid already converted callbacks. @match_callback_converted depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier t; @@ void _callback(struct timer_list *t) { ... } // callback(struct something *handle) @change_callback_handle_arg depends on change_timer_function_usage && !match_callback_converted && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; @@ void _callback( -_handletype *_handle +struct timer_list *t ) { + _handletype *_handle = from_timer(_handle, t, _timer); ... } // If change_callback_handle_arg ran on an empty function, remove // the added handler. @unchange_callback_handle_arg depends on change_timer_function_usage && change_callback_handle_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; identifier t; @@ void _callback(struct timer_list *t) { - _handletype *_handle = from_timer(_handle, t, _timer); } // We only want to refactor the setup_timer() data argument if we've found // the matching callback. This undoes changes in change_timer_function_usage. @unchange_timer_function_usage depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg && !change_callback_handle_arg@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type change_timer_function_usage._cast_data; @@ ( -timer_setup(&_E->_timer, _callback, 0); +setup_timer(&_E->_timer, _callback, (_cast_data)_E); | -timer_setup(&_E._timer, _callback, 0); +setup_timer(&_E._timer, _callback, (_cast_data)&_E); ) // If we fixed a callback from a .function assignment, fix the // assignment cast now. @change_timer_function_assignment depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_func; typedef TIMER_FUNC_TYPE; @@ ( _E->_timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -&_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)_callback; +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -&_callback; +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; ) // Sometimes timer functions are called directly. Replace matched args. @change_timer_function_calls depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression _E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_data; @@ _callback( ( -(_cast_data)_E +&_E->_timer | -(_cast_data)&_E +&_E._timer | -_E +&_E->_timer ) ) // If a timer has been configured without a data argument, it can be // converted without regard to the callback argument, since it is unused. @match_timer_function_unused_data@ expression _E; identifier _timer; identifier _callback; @@ ( -setup_timer(&_E->_timer, _callback, 0); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0L); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0UL); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0L); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0UL); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_timer, _callback, 0); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0L); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0UL); +timer_setup(&_timer, _callback, 0); | -setup_timer(_timer, _callback, 0); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0L); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0UL); +timer_setup(_timer, _callback, 0); ) @change_callback_unused_data depends on match_timer_function_unused_data@ identifier match_timer_function_unused_data._callback; type _origtype; identifier _origarg; @@ void _callback( -_origtype _origarg +struct timer_list *unused ) { ... when != _origarg } Signed-off-by: Kees Cook <keescook@chromium.org>
2017-10-16 21:43:17 +00:00
static void gsm_dlci_t1(struct timer_list *t)
{
treewide: setup_timer() -> timer_setup() This converts all remaining cases of the old setup_timer() API into using timer_setup(), where the callback argument is the structure already holding the struct timer_list. These should have no behavioral changes, since they just change which pointer is passed into the callback with the same available pointers after conversion. It handles the following examples, in addition to some other variations. Casting from unsigned long: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... setup_timer(&ptr->my_timer, my_callback, ptr); and forced object casts: void my_callback(struct something *ptr) { ... } ... setup_timer(&ptr->my_timer, my_callback, (unsigned long)ptr); become: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... timer_setup(&ptr->my_timer, my_callback, 0); Direct function assignments: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... ptr->my_timer.function = my_callback; have a temporary cast added, along with converting the args: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... ptr->my_timer.function = (TIMER_FUNC_TYPE)my_callback; And finally, callbacks without a data assignment: void my_callback(unsigned long data) { ... } ... setup_timer(&ptr->my_timer, my_callback, 0); have their argument renamed to verify they're unused during conversion: void my_callback(struct timer_list *unused) { ... } ... timer_setup(&ptr->my_timer, my_callback, 0); The conversion is done with the following Coccinelle script: spatch --very-quiet --all-includes --include-headers \ -I ./arch/x86/include -I ./arch/x86/include/generated \ -I ./include -I ./arch/x86/include/uapi \ -I ./arch/x86/include/generated/uapi -I ./include/uapi \ -I ./include/generated/uapi --include ./include/linux/kconfig.h \ --dir . \ --cocci-file ~/src/data/timer_setup.cocci @fix_address_of@ expression e; @@ setup_timer( -&(e) +&e , ...) // Update any raw setup_timer() usages that have a NULL callback, but // would otherwise match change_timer_function_usage, since the latter // will update all function assignments done in the face of a NULL // function initialization in setup_timer(). @change_timer_function_usage_NULL@ expression _E; identifier _timer; type _cast_data; @@ ( -setup_timer(&_E->_timer, NULL, _E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E->_timer, NULL, (_cast_data)_E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E._timer, NULL, &_E); +timer_setup(&_E._timer, NULL, 0); | -setup_timer(&_E._timer, NULL, (_cast_data)&_E); +timer_setup(&_E._timer, NULL, 0); ) @change_timer_function_usage@ expression _E; identifier _timer; struct timer_list _stl; identifier _callback; type _cast_func, _cast_data; @@ ( -setup_timer(&_E->_timer, _callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | _E->_timer@_stl.function = _callback; | _E->_timer@_stl.function = &_callback; | _E->_timer@_stl.function = (_cast_func)_callback; | _E->_timer@_stl.function = (_cast_func)&_callback; | _E._timer@_stl.function = _callback; | _E._timer@_stl.function = &_callback; | _E._timer@_stl.function = (_cast_func)_callback; | _E._timer@_stl.function = (_cast_func)&_callback; ) // callback(unsigned long arg) @change_callback_handle_cast depends on change_timer_function_usage@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; identifier _handle; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { ( ... when != _origarg _handletype *_handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg ) } // callback(unsigned long arg) without existing variable @change_callback_handle_cast_no_arg depends on change_timer_function_usage && !change_callback_handle_cast@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { + _handletype *_origarg = from_timer(_origarg, t, _timer); + ... when != _origarg - (_handletype *)_origarg + _origarg ... when != _origarg } // Avoid already converted callbacks. @match_callback_converted depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier t; @@ void _callback(struct timer_list *t) { ... } // callback(struct something *handle) @change_callback_handle_arg depends on change_timer_function_usage && !match_callback_converted && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; @@ void _callback( -_handletype *_handle +struct timer_list *t ) { + _handletype *_handle = from_timer(_handle, t, _timer); ... } // If change_callback_handle_arg ran on an empty function, remove // the added handler. @unchange_callback_handle_arg depends on change_timer_function_usage && change_callback_handle_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; identifier t; @@ void _callback(struct timer_list *t) { - _handletype *_handle = from_timer(_handle, t, _timer); } // We only want to refactor the setup_timer() data argument if we've found // the matching callback. This undoes changes in change_timer_function_usage. @unchange_timer_function_usage depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg && !change_callback_handle_arg@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type change_timer_function_usage._cast_data; @@ ( -timer_setup(&_E->_timer, _callback, 0); +setup_timer(&_E->_timer, _callback, (_cast_data)_E); | -timer_setup(&_E._timer, _callback, 0); +setup_timer(&_E._timer, _callback, (_cast_data)&_E); ) // If we fixed a callback from a .function assignment, fix the // assignment cast now. @change_timer_function_assignment depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_func; typedef TIMER_FUNC_TYPE; @@ ( _E->_timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -&_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)_callback; +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -&_callback; +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; ) // Sometimes timer functions are called directly. Replace matched args. @change_timer_function_calls depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression _E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_data; @@ _callback( ( -(_cast_data)_E +&_E->_timer | -(_cast_data)&_E +&_E._timer | -_E +&_E->_timer ) ) // If a timer has been configured without a data argument, it can be // converted without regard to the callback argument, since it is unused. @match_timer_function_unused_data@ expression _E; identifier _timer; identifier _callback; @@ ( -setup_timer(&_E->_timer, _callback, 0); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0L); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0UL); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0L); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0UL); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_timer, _callback, 0); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0L); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0UL); +timer_setup(&_timer, _callback, 0); | -setup_timer(_timer, _callback, 0); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0L); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0UL); +timer_setup(_timer, _callback, 0); ) @change_callback_unused_data depends on match_timer_function_unused_data@ identifier match_timer_function_unused_data._callback; type _origtype; identifier _origarg; @@ void _callback( -_origtype _origarg +struct timer_list *unused ) { ... when != _origarg } Signed-off-by: Kees Cook <keescook@chromium.org>
2017-10-16 21:43:17 +00:00
struct gsm_dlci *dlci = from_timer(dlci, t, t1);
struct gsm_mux *gsm = dlci->gsm;
switch (dlci->state) {
case DLCI_CONFIGURE:
if (dlci->retries && gsm_dlci_negotiate(dlci) == 0) {
dlci->retries--;
mod_timer(&dlci->t1, jiffies + gsm->t1 * HZ / 100);
} else {
gsm_dlci_begin_close(dlci); /* prevent half open link */
}
break;
case DLCI_OPENING:
if (dlci->retries) {
dlci->retries--;
gsm_command(dlci->gsm, dlci->addr, SABM|PF);
mod_timer(&dlci->t1, jiffies + gsm->t1 * HZ / 100);
tty: n_gsm: Allow ADM response in addition to UA for control dlci Some devices have the control dlci stay in ADM mode instead of the UA mode. This can seen at least on droid 4 when trying to open the ts 27.010 mux port. Enabling n_gsm debug mode shows the control dlci always respond with DM to SABM instead of UA: # modprobe n_gsm debug=0xff # ldattach -d GSM0710 /dev/ttyS0 & gsmld_output: 00000000: f9 03 3f 01 1c f9 --> 0) C: SABM(P) gsmld_receive: 00000000: f9 03 1f 01 36 f9 <-- 0) C: DM(P) ... $ minicom -D /dev/gsmtty1 minicom: cannot open /dev/gsmtty1: No error information $ strace minicom -D /dev/gsmtty1 ... open("/dev/gsmtty1", O_RDWR|O_NOCTTY|O_NONBLOCK|O_LARGEFILE) = -1 EL2HLT Note that this is different issue from other n_gsm -EL2HLT issues such as timeouts when the control dlci does not respond at all. The ADM mode seems to be a quite common according to "RF Wireless World" article "GSM Issue-UE sends SABM and gets a DM response instead of UA response": This issue is most commonly observed in GSM networks where in UE sends SABM and expects network to send UA response but it ends up receiving DM response from the network. SABM stands for Set asynchronous balanced mode, UA stands for Unnumbered Acknowledge and DA stands for Disconnected Mode. An RLP entity can be in one of two modes: - Asynchronous Balanced Mode (ABM) - Asynchronous Disconnected Mode (ADM) Currently Linux kernel closes the control dlci after several retries in gsm_dlci_t1() on DM. This causes n_gsm /dev/gsmtty ports to produce error code -EL2HLT when trying to open them as the closing of control dlci has already set gsm->dead. Let's fix the issue by allowing control dlci stay in ADM mode after the retries so the /dev/gsmtty ports can be opened and used. It seems that it might take several attempts to get any response from the control dlci, so it's best to allow ADM mode only after the SABM retries are done. Note that for droid 4 additional patches are needed to mux the ttyS0 pins and to toggle RTS gpio_149 to wake up the mdm6600 modem are also needed to use n_gsm. And the mdm6600 modem needs to be powered on. Cc: linux-serial@vger.kernel.org Cc: Alan Cox <alan@llwyncelyn.cymru> Cc: Jiri Prchal <jiri.prchal@aksignal.cz> Cc: Jiri Slaby <jslaby@suse.cz> Cc: Marcel Partap <mpartap@gmx.net> Cc: Michael Scott <michael.scott@linaro.org> Cc: Peter Hurley <peter@hurleysoftware.com> Cc: Russ Gorby <russ.gorby@intel.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sebastian Reichel <sre@kernel.org> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-01-03 18:18:03 +00:00
} else if (!dlci->addr && gsm->control == (DM | PF)) {
if (debug & DBG_ERRORS)
tty: n_gsm: Allow ADM response in addition to UA for control dlci Some devices have the control dlci stay in ADM mode instead of the UA mode. This can seen at least on droid 4 when trying to open the ts 27.010 mux port. Enabling n_gsm debug mode shows the control dlci always respond with DM to SABM instead of UA: # modprobe n_gsm debug=0xff # ldattach -d GSM0710 /dev/ttyS0 & gsmld_output: 00000000: f9 03 3f 01 1c f9 --> 0) C: SABM(P) gsmld_receive: 00000000: f9 03 1f 01 36 f9 <-- 0) C: DM(P) ... $ minicom -D /dev/gsmtty1 minicom: cannot open /dev/gsmtty1: No error information $ strace minicom -D /dev/gsmtty1 ... open("/dev/gsmtty1", O_RDWR|O_NOCTTY|O_NONBLOCK|O_LARGEFILE) = -1 EL2HLT Note that this is different issue from other n_gsm -EL2HLT issues such as timeouts when the control dlci does not respond at all. The ADM mode seems to be a quite common according to "RF Wireless World" article "GSM Issue-UE sends SABM and gets a DM response instead of UA response": This issue is most commonly observed in GSM networks where in UE sends SABM and expects network to send UA response but it ends up receiving DM response from the network. SABM stands for Set asynchronous balanced mode, UA stands for Unnumbered Acknowledge and DA stands for Disconnected Mode. An RLP entity can be in one of two modes: - Asynchronous Balanced Mode (ABM) - Asynchronous Disconnected Mode (ADM) Currently Linux kernel closes the control dlci after several retries in gsm_dlci_t1() on DM. This causes n_gsm /dev/gsmtty ports to produce error code -EL2HLT when trying to open them as the closing of control dlci has already set gsm->dead. Let's fix the issue by allowing control dlci stay in ADM mode after the retries so the /dev/gsmtty ports can be opened and used. It seems that it might take several attempts to get any response from the control dlci, so it's best to allow ADM mode only after the SABM retries are done. Note that for droid 4 additional patches are needed to mux the ttyS0 pins and to toggle RTS gpio_149 to wake up the mdm6600 modem are also needed to use n_gsm. And the mdm6600 modem needs to be powered on. Cc: linux-serial@vger.kernel.org Cc: Alan Cox <alan@llwyncelyn.cymru> Cc: Jiri Prchal <jiri.prchal@aksignal.cz> Cc: Jiri Slaby <jslaby@suse.cz> Cc: Marcel Partap <mpartap@gmx.net> Cc: Michael Scott <michael.scott@linaro.org> Cc: Peter Hurley <peter@hurleysoftware.com> Cc: Russ Gorby <russ.gorby@intel.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sebastian Reichel <sre@kernel.org> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-01-03 18:18:03 +00:00
pr_info("DLCI %d opening in ADM mode.\n",
dlci->addr);
tty: n_gsm: Fix long delays with control frame timeouts in ADM mode Commit ea3d8465ab9b ("tty: n_gsm: Allow ADM response in addition to UA for control dlci") added support for DLCI to stay in Asynchronous Disconnected Mode (ADM). But we still get long delays waiting for commands to other DLCI to complete: --> 5) C: SABM(P) Q> 0) C: UIH(F) Q> 0) C: UIH(F) Q> 0) C: UIH(F) ... This happens because gsm_control_send() sets cretries timer to T2 that is by default set to 34. This will cause resend for T2 times for the control frame. In ADM mode, we will never get a response so the control frame, so retries are just delaying all the commands. Let's fix the issue by setting DLCI_MODE_ADM flag after detecting the ADM mode for the control DLCI. Then we can use that in gsm_control_send() to set retries to 1. This means the control frame will be sent once allowing the other end at an opportunity to switch from ADM to ABM mode. Note that retries will be decremented in gsm_control_retransmit() so we don't want to set it to 0 here. Fixes: ea3d8465ab9b ("tty: n_gsm: Allow ADM response in addition to UA for control dlci") Cc: linux-serial@vger.kernel.org Cc: Alan Cox <alan@llwyncelyn.cymru> Cc: Dan Williams <dcbw@redhat.com> Cc: Jiri Prchal <jiri.prchal@aksignal.cz> Cc: Jiri Slaby <jslaby@suse.cz> Cc: Marcel Partap <mpartap@gmx.net> Cc: Merlijn Wajer <merlijn@wizzup.org> Cc: Michael Nazzareno Trimarchi <michael@amarulasolutions.com> Cc: Michael Scott <michael.scott@linaro.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Peter Hurley <peter@hurleysoftware.com> Cc: Russ Gorby <russ.gorby@intel.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sebastian Reichel <sre@kernel.org> Signed-off-by: Tony Lindgren <tony@atomide.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-04-07 17:19:50 +00:00
dlci->mode = DLCI_MODE_ADM;
tty: n_gsm: Allow ADM response in addition to UA for control dlci Some devices have the control dlci stay in ADM mode instead of the UA mode. This can seen at least on droid 4 when trying to open the ts 27.010 mux port. Enabling n_gsm debug mode shows the control dlci always respond with DM to SABM instead of UA: # modprobe n_gsm debug=0xff # ldattach -d GSM0710 /dev/ttyS0 & gsmld_output: 00000000: f9 03 3f 01 1c f9 --> 0) C: SABM(P) gsmld_receive: 00000000: f9 03 1f 01 36 f9 <-- 0) C: DM(P) ... $ minicom -D /dev/gsmtty1 minicom: cannot open /dev/gsmtty1: No error information $ strace minicom -D /dev/gsmtty1 ... open("/dev/gsmtty1", O_RDWR|O_NOCTTY|O_NONBLOCK|O_LARGEFILE) = -1 EL2HLT Note that this is different issue from other n_gsm -EL2HLT issues such as timeouts when the control dlci does not respond at all. The ADM mode seems to be a quite common according to "RF Wireless World" article "GSM Issue-UE sends SABM and gets a DM response instead of UA response": This issue is most commonly observed in GSM networks where in UE sends SABM and expects network to send UA response but it ends up receiving DM response from the network. SABM stands for Set asynchronous balanced mode, UA stands for Unnumbered Acknowledge and DA stands for Disconnected Mode. An RLP entity can be in one of two modes: - Asynchronous Balanced Mode (ABM) - Asynchronous Disconnected Mode (ADM) Currently Linux kernel closes the control dlci after several retries in gsm_dlci_t1() on DM. This causes n_gsm /dev/gsmtty ports to produce error code -EL2HLT when trying to open them as the closing of control dlci has already set gsm->dead. Let's fix the issue by allowing control dlci stay in ADM mode after the retries so the /dev/gsmtty ports can be opened and used. It seems that it might take several attempts to get any response from the control dlci, so it's best to allow ADM mode only after the SABM retries are done. Note that for droid 4 additional patches are needed to mux the ttyS0 pins and to toggle RTS gpio_149 to wake up the mdm6600 modem are also needed to use n_gsm. And the mdm6600 modem needs to be powered on. Cc: linux-serial@vger.kernel.org Cc: Alan Cox <alan@llwyncelyn.cymru> Cc: Jiri Prchal <jiri.prchal@aksignal.cz> Cc: Jiri Slaby <jslaby@suse.cz> Cc: Marcel Partap <mpartap@gmx.net> Cc: Michael Scott <michael.scott@linaro.org> Cc: Peter Hurley <peter@hurleysoftware.com> Cc: Russ Gorby <russ.gorby@intel.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sebastian Reichel <sre@kernel.org> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-01-03 18:18:03 +00:00
gsm_dlci_open(dlci);
} else {
gsm_dlci_begin_close(dlci); /* prevent half open link */
tty: n_gsm: Allow ADM response in addition to UA for control dlci Some devices have the control dlci stay in ADM mode instead of the UA mode. This can seen at least on droid 4 when trying to open the ts 27.010 mux port. Enabling n_gsm debug mode shows the control dlci always respond with DM to SABM instead of UA: # modprobe n_gsm debug=0xff # ldattach -d GSM0710 /dev/ttyS0 & gsmld_output: 00000000: f9 03 3f 01 1c f9 --> 0) C: SABM(P) gsmld_receive: 00000000: f9 03 1f 01 36 f9 <-- 0) C: DM(P) ... $ minicom -D /dev/gsmtty1 minicom: cannot open /dev/gsmtty1: No error information $ strace minicom -D /dev/gsmtty1 ... open("/dev/gsmtty1", O_RDWR|O_NOCTTY|O_NONBLOCK|O_LARGEFILE) = -1 EL2HLT Note that this is different issue from other n_gsm -EL2HLT issues such as timeouts when the control dlci does not respond at all. The ADM mode seems to be a quite common according to "RF Wireless World" article "GSM Issue-UE sends SABM and gets a DM response instead of UA response": This issue is most commonly observed in GSM networks where in UE sends SABM and expects network to send UA response but it ends up receiving DM response from the network. SABM stands for Set asynchronous balanced mode, UA stands for Unnumbered Acknowledge and DA stands for Disconnected Mode. An RLP entity can be in one of two modes: - Asynchronous Balanced Mode (ABM) - Asynchronous Disconnected Mode (ADM) Currently Linux kernel closes the control dlci after several retries in gsm_dlci_t1() on DM. This causes n_gsm /dev/gsmtty ports to produce error code -EL2HLT when trying to open them as the closing of control dlci has already set gsm->dead. Let's fix the issue by allowing control dlci stay in ADM mode after the retries so the /dev/gsmtty ports can be opened and used. It seems that it might take several attempts to get any response from the control dlci, so it's best to allow ADM mode only after the SABM retries are done. Note that for droid 4 additional patches are needed to mux the ttyS0 pins and to toggle RTS gpio_149 to wake up the mdm6600 modem are also needed to use n_gsm. And the mdm6600 modem needs to be powered on. Cc: linux-serial@vger.kernel.org Cc: Alan Cox <alan@llwyncelyn.cymru> Cc: Jiri Prchal <jiri.prchal@aksignal.cz> Cc: Jiri Slaby <jslaby@suse.cz> Cc: Marcel Partap <mpartap@gmx.net> Cc: Michael Scott <michael.scott@linaro.org> Cc: Peter Hurley <peter@hurleysoftware.com> Cc: Russ Gorby <russ.gorby@intel.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sebastian Reichel <sre@kernel.org> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-01-03 18:18:03 +00:00
}
break;
case DLCI_CLOSING:
if (dlci->retries) {
dlci->retries--;
gsm_command(dlci->gsm, dlci->addr, DISC|PF);
mod_timer(&dlci->t1, jiffies + gsm->t1 * HZ / 100);
} else
gsm_dlci_close(dlci);
break;
default:
pr_debug("%s: unhandled state: %d\n", __func__, dlci->state);
break;
}
}
/**
* gsm_dlci_begin_open - start channel open procedure
* @dlci: DLCI to open
*
* Commence opening a DLCI from the Linux side. We issue SABM messages
tty: n_gsm: Allow ADM response in addition to UA for control dlci Some devices have the control dlci stay in ADM mode instead of the UA mode. This can seen at least on droid 4 when trying to open the ts 27.010 mux port. Enabling n_gsm debug mode shows the control dlci always respond with DM to SABM instead of UA: # modprobe n_gsm debug=0xff # ldattach -d GSM0710 /dev/ttyS0 & gsmld_output: 00000000: f9 03 3f 01 1c f9 --> 0) C: SABM(P) gsmld_receive: 00000000: f9 03 1f 01 36 f9 <-- 0) C: DM(P) ... $ minicom -D /dev/gsmtty1 minicom: cannot open /dev/gsmtty1: No error information $ strace minicom -D /dev/gsmtty1 ... open("/dev/gsmtty1", O_RDWR|O_NOCTTY|O_NONBLOCK|O_LARGEFILE) = -1 EL2HLT Note that this is different issue from other n_gsm -EL2HLT issues such as timeouts when the control dlci does not respond at all. The ADM mode seems to be a quite common according to "RF Wireless World" article "GSM Issue-UE sends SABM and gets a DM response instead of UA response": This issue is most commonly observed in GSM networks where in UE sends SABM and expects network to send UA response but it ends up receiving DM response from the network. SABM stands for Set asynchronous balanced mode, UA stands for Unnumbered Acknowledge and DA stands for Disconnected Mode. An RLP entity can be in one of two modes: - Asynchronous Balanced Mode (ABM) - Asynchronous Disconnected Mode (ADM) Currently Linux kernel closes the control dlci after several retries in gsm_dlci_t1() on DM. This causes n_gsm /dev/gsmtty ports to produce error code -EL2HLT when trying to open them as the closing of control dlci has already set gsm->dead. Let's fix the issue by allowing control dlci stay in ADM mode after the retries so the /dev/gsmtty ports can be opened and used. It seems that it might take several attempts to get any response from the control dlci, so it's best to allow ADM mode only after the SABM retries are done. Note that for droid 4 additional patches are needed to mux the ttyS0 pins and to toggle RTS gpio_149 to wake up the mdm6600 modem are also needed to use n_gsm. And the mdm6600 modem needs to be powered on. Cc: linux-serial@vger.kernel.org Cc: Alan Cox <alan@llwyncelyn.cymru> Cc: Jiri Prchal <jiri.prchal@aksignal.cz> Cc: Jiri Slaby <jslaby@suse.cz> Cc: Marcel Partap <mpartap@gmx.net> Cc: Michael Scott <michael.scott@linaro.org> Cc: Peter Hurley <peter@hurleysoftware.com> Cc: Russ Gorby <russ.gorby@intel.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sebastian Reichel <sre@kernel.org> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-01-03 18:18:03 +00:00
* to the modem which should then reply with a UA or ADM, at which point
* we will move into open state. Opening is done asynchronously with retry
* running off timers and the responses.
* Parameter negotiation is performed before SABM if required.
*/
static void gsm_dlci_begin_open(struct gsm_dlci *dlci)
{
struct gsm_mux *gsm = dlci ? dlci->gsm : NULL;
bool need_pn = false;
if (!gsm)
return;
if (dlci->addr != 0) {
if (gsm->adaption != 1 || gsm->adaption != dlci->adaption)
need_pn = true;
if (dlci->prio != (roundup(dlci->addr + 1, 8) - 1))
need_pn = true;
if (gsm->ftype != dlci->ftype)
need_pn = true;
}
switch (dlci->state) {
case DLCI_CLOSED:
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
case DLCI_WAITING_CONFIG:
case DLCI_CLOSING:
dlci->retries = gsm->n2;
if (!need_pn) {
dlci->state = DLCI_OPENING;
gsm_command(gsm, dlci->addr, SABM|PF);
} else {
/* Configure DLCI before setup */
dlci->state = DLCI_CONFIGURE;
if (gsm_dlci_negotiate(dlci) != 0) {
gsm_dlci_close(dlci);
return;
}
}
mod_timer(&dlci->t1, jiffies + gsm->t1 * HZ / 100);
break;
default:
break;
}
}
/**
* gsm_dlci_set_opening - change state to opening
* @dlci: DLCI to open
*
* Change internal state to wait for DLCI open from initiator side.
* We set off timers and responses upon reception of an SABM.
*/
static void gsm_dlci_set_opening(struct gsm_dlci *dlci)
{
switch (dlci->state) {
case DLCI_CLOSED:
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
case DLCI_WAITING_CONFIG:
case DLCI_CLOSING:
dlci->state = DLCI_OPENING;
break;
default:
break;
}
}
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
/**
* gsm_dlci_set_wait_config - wait for channel configuration
* @dlci: DLCI to configure
*
* Wait for a DLCI configuration from the application.
*/
static void gsm_dlci_set_wait_config(struct gsm_dlci *dlci)
{
switch (dlci->state) {
case DLCI_CLOSED:
case DLCI_CLOSING:
dlci->state = DLCI_WAITING_CONFIG;
break;
default:
break;
}
}
/**
* gsm_dlci_begin_close - start channel open procedure
* @dlci: DLCI to open
*
* Commence closing a DLCI from the Linux side. We issue DISC messages
* to the modem which should then reply with a UA, at which point we
* will move into closed state. Closing is done asynchronously with retry
* off timers. We may also receive a DM reply from the other end which
* indicates the channel was already closed.
*/
static void gsm_dlci_begin_close(struct gsm_dlci *dlci)
{
struct gsm_mux *gsm = dlci->gsm;
if (dlci->state == DLCI_CLOSED || dlci->state == DLCI_CLOSING)
return;
dlci->retries = gsm->n2;
dlci->state = DLCI_CLOSING;
gsm_command(dlci->gsm, dlci->addr, DISC|PF);
mod_timer(&dlci->t1, jiffies + gsm->t1 * HZ / 100);
wake_up_interruptible(&gsm->event);
}
/**
* gsm_dlci_data - data arrived
* @dlci: channel
* @data: block of bytes received
* @clen: length of received block
*
* A UI or UIH frame has arrived which contains data for a channel
* other than the control channel. If the relevant virtual tty is
* open we shovel the bits down it, if not we drop them.
*/
static void gsm_dlci_data(struct gsm_dlci *dlci, const u8 *data, int clen)
{
/* krefs .. */
struct tty_port *port = &dlci->port;
struct tty_struct *tty;
unsigned int modem = 0;
int len;
if (debug & DBG_TTY)
pr_debug("%d bytes for tty\n", clen);
switch (dlci->adaption) {
/* Unsupported types */
case 4: /* Packetised interruptible data */
break;
case 3: /* Packetised uininterruptible voice/data */
break;
case 2: /* Asynchronous serial with line state in each frame */
len = gsm_read_ea_val(&modem, data, clen);
if (len < 1)
return;
tty = tty_port_tty_get(port);
if (tty) {
gsm_process_modem(tty, dlci, modem, len);
tty_wakeup(tty);
tty_kref_put(tty);
}
/* Skip processed modem data */
data += len;
clen -= len;
fallthrough;
case 1: /* Line state will go via DLCI 0 controls only */
default:
tty_insert_flip_string(port, data, clen);
tty_flip_buffer_push(port);
}
}
/**
* gsm_dlci_command - data arrived on control channel
* @dlci: channel
* @data: block of bytes received
* @len: length of received block
*
* A UI or UIH frame has arrived which contains data for DLCI 0 the
* control channel. This should contain a command EA followed by
* control data bytes. The command EA contains a command/response bit
* and we divide up the work accordingly.
*/
static void gsm_dlci_command(struct gsm_dlci *dlci, const u8 *data, int len)
{
/* See what command is involved */
unsigned int command = 0;
unsigned int clen = 0;
unsigned int dlen;
/* read the command */
dlen = gsm_read_ea_val(&command, data, len);
len -= dlen;
data += dlen;
/* read any control data */
dlen = gsm_read_ea_val(&clen, data, len);
len -= dlen;
data += dlen;
/* Malformed command? */
if (clen > len)
return;
if (command & 1)
gsm_control_message(dlci->gsm, command, data, clen);
else
gsm_control_response(dlci->gsm, command, data, clen);
}
/**
* gsm_kick_timer - transmit if possible
* @t: timer contained in our gsm object
*
* Transmit data from DLCIs if the queue is empty. We can't rely on
* a tty wakeup except when we filled the pipe so we need to fire off
* new data ourselves in other cases.
*/
static void gsm_kick_timer(struct timer_list *t)
{
struct gsm_mux *gsm = from_timer(gsm, t, kick_timer);
unsigned long flags;
int sent = 0;
spin_lock_irqsave(&gsm->tx_lock, flags);
/* If we have nothing running then we need to fire up */
if (gsm->tx_bytes < TX_THRESH_LO)
sent = gsm_dlci_data_sweep(gsm);
spin_unlock_irqrestore(&gsm->tx_lock, flags);
if (sent && debug & DBG_DATA)
pr_info("%s TX queue stalled\n", __func__);
}
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
/**
* gsm_dlci_copy_config_values - copy DLCI configuration
* @dlci: source DLCI
* @dc: configuration structure to fill
*/
static void gsm_dlci_copy_config_values(struct gsm_dlci *dlci, struct gsm_dlci_config *dc)
{
memset(dc, 0, sizeof(*dc));
dc->channel = (u32)dlci->addr;
dc->adaption = (u32)dlci->adaption;
dc->mtu = (u32)dlci->mtu;
dc->priority = (u32)dlci->prio;
if (dlci->ftype == UIH)
dc->i = 1;
else
dc->i = 2;
dc->k = (u32)dlci->k;
}
/**
* gsm_dlci_config - configure DLCI from configuration
* @dlci: DLCI to configure
* @dc: DLCI configuration
* @open: open DLCI after configuration?
*/
static int gsm_dlci_config(struct gsm_dlci *dlci, struct gsm_dlci_config *dc, int open)
{
struct gsm_mux *gsm;
bool need_restart = false;
bool need_open = false;
unsigned int i;
/*
* Check that userspace doesn't put stuff in here to prevent breakages
* in the future.
*/
for (i = 0; i < ARRAY_SIZE(dc->reserved); i++)
if (dc->reserved[i])
return -EINVAL;
if (!dlci)
return -EINVAL;
gsm = dlci->gsm;
/* Stuff we don't support yet - I frame transport */
if (dc->adaption != 1 && dc->adaption != 2)
return -EOPNOTSUPP;
if (dc->mtu > MAX_MTU || dc->mtu < MIN_MTU || dc->mtu > gsm->mru)
return -EINVAL;
if (dc->priority >= 64)
return -EINVAL;
if (dc->i == 0 || dc->i > 2) /* UIH and UI only */
return -EINVAL;
if (dc->k > 7)
return -EINVAL;
/*
* See what is needed for reconfiguration
*/
/* Framing fields */
if (dc->adaption != dlci->adaption)
need_restart = true;
if (dc->mtu != dlci->mtu)
need_restart = true;
if (dc->i != dlci->ftype)
need_restart = true;
/* Requires care */
if (dc->priority != dlci->prio)
need_restart = true;
if ((open && gsm->wait_config) || need_restart)
need_open = true;
if (dlci->state == DLCI_WAITING_CONFIG) {
need_restart = false;
need_open = true;
}
/*
* Close down what is needed, restart and initiate the new
* configuration.
*/
if (need_restart) {
gsm_dlci_begin_close(dlci);
wait_event_interruptible(gsm->event, dlci->state == DLCI_CLOSED);
if (signal_pending(current))
return -EINTR;
}
/*
* Setup the new configuration values
*/
dlci->adaption = (int)dc->adaption;
if (dc->mtu)
dlci->mtu = (unsigned int)dc->mtu;
else
dlci->mtu = gsm->mtu;
if (dc->priority)
dlci->prio = (u8)dc->priority;
else
dlci->prio = roundup(dlci->addr + 1, 8) - 1;
if (dc->i == 1)
dlci->ftype = UIH;
else if (dc->i == 2)
dlci->ftype = UI;
if (dc->k)
dlci->k = (u8)dc->k;
else
dlci->k = gsm->k;
if (need_open) {
if (gsm->initiator)
gsm_dlci_begin_open(dlci);
else
gsm_dlci_set_opening(dlci);
}
return 0;
}
/*
* Allocate/Free DLCI channels
*/
/**
* gsm_dlci_alloc - allocate a DLCI
* @gsm: GSM mux
* @addr: address of the DLCI
*
* Allocate and install a new DLCI object into the GSM mux.
*
* FIXME: review locking races
*/
static struct gsm_dlci *gsm_dlci_alloc(struct gsm_mux *gsm, int addr)
{
struct gsm_dlci *dlci = kzalloc(sizeof(struct gsm_dlci), GFP_ATOMIC);
if (dlci == NULL)
return NULL;
spin_lock_init(&dlci->lock);
mutex_init(&dlci->mutex);
if (kfifo_alloc(&dlci->fifo, TX_SIZE, GFP_KERNEL) < 0) {
kfree(dlci);
return NULL;
}
skb_queue_head_init(&dlci->skb_list);
treewide: setup_timer() -> timer_setup() This converts all remaining cases of the old setup_timer() API into using timer_setup(), where the callback argument is the structure already holding the struct timer_list. These should have no behavioral changes, since they just change which pointer is passed into the callback with the same available pointers after conversion. It handles the following examples, in addition to some other variations. Casting from unsigned long: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... setup_timer(&ptr->my_timer, my_callback, ptr); and forced object casts: void my_callback(struct something *ptr) { ... } ... setup_timer(&ptr->my_timer, my_callback, (unsigned long)ptr); become: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... timer_setup(&ptr->my_timer, my_callback, 0); Direct function assignments: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... ptr->my_timer.function = my_callback; have a temporary cast added, along with converting the args: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... ptr->my_timer.function = (TIMER_FUNC_TYPE)my_callback; And finally, callbacks without a data assignment: void my_callback(unsigned long data) { ... } ... setup_timer(&ptr->my_timer, my_callback, 0); have their argument renamed to verify they're unused during conversion: void my_callback(struct timer_list *unused) { ... } ... timer_setup(&ptr->my_timer, my_callback, 0); The conversion is done with the following Coccinelle script: spatch --very-quiet --all-includes --include-headers \ -I ./arch/x86/include -I ./arch/x86/include/generated \ -I ./include -I ./arch/x86/include/uapi \ -I ./arch/x86/include/generated/uapi -I ./include/uapi \ -I ./include/generated/uapi --include ./include/linux/kconfig.h \ --dir . \ --cocci-file ~/src/data/timer_setup.cocci @fix_address_of@ expression e; @@ setup_timer( -&(e) +&e , ...) // Update any raw setup_timer() usages that have a NULL callback, but // would otherwise match change_timer_function_usage, since the latter // will update all function assignments done in the face of a NULL // function initialization in setup_timer(). @change_timer_function_usage_NULL@ expression _E; identifier _timer; type _cast_data; @@ ( -setup_timer(&_E->_timer, NULL, _E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E->_timer, NULL, (_cast_data)_E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E._timer, NULL, &_E); +timer_setup(&_E._timer, NULL, 0); | -setup_timer(&_E._timer, NULL, (_cast_data)&_E); +timer_setup(&_E._timer, NULL, 0); ) @change_timer_function_usage@ expression _E; identifier _timer; struct timer_list _stl; identifier _callback; type _cast_func, _cast_data; @@ ( -setup_timer(&_E->_timer, _callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | _E->_timer@_stl.function = _callback; | _E->_timer@_stl.function = &_callback; | _E->_timer@_stl.function = (_cast_func)_callback; | _E->_timer@_stl.function = (_cast_func)&_callback; | _E._timer@_stl.function = _callback; | _E._timer@_stl.function = &_callback; | _E._timer@_stl.function = (_cast_func)_callback; | _E._timer@_stl.function = (_cast_func)&_callback; ) // callback(unsigned long arg) @change_callback_handle_cast depends on change_timer_function_usage@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; identifier _handle; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { ( ... when != _origarg _handletype *_handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg ) } // callback(unsigned long arg) without existing variable @change_callback_handle_cast_no_arg depends on change_timer_function_usage && !change_callback_handle_cast@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { + _handletype *_origarg = from_timer(_origarg, t, _timer); + ... when != _origarg - (_handletype *)_origarg + _origarg ... when != _origarg } // Avoid already converted callbacks. @match_callback_converted depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier t; @@ void _callback(struct timer_list *t) { ... } // callback(struct something *handle) @change_callback_handle_arg depends on change_timer_function_usage && !match_callback_converted && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; @@ void _callback( -_handletype *_handle +struct timer_list *t ) { + _handletype *_handle = from_timer(_handle, t, _timer); ... } // If change_callback_handle_arg ran on an empty function, remove // the added handler. @unchange_callback_handle_arg depends on change_timer_function_usage && change_callback_handle_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; identifier t; @@ void _callback(struct timer_list *t) { - _handletype *_handle = from_timer(_handle, t, _timer); } // We only want to refactor the setup_timer() data argument if we've found // the matching callback. This undoes changes in change_timer_function_usage. @unchange_timer_function_usage depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg && !change_callback_handle_arg@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type change_timer_function_usage._cast_data; @@ ( -timer_setup(&_E->_timer, _callback, 0); +setup_timer(&_E->_timer, _callback, (_cast_data)_E); | -timer_setup(&_E._timer, _callback, 0); +setup_timer(&_E._timer, _callback, (_cast_data)&_E); ) // If we fixed a callback from a .function assignment, fix the // assignment cast now. @change_timer_function_assignment depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_func; typedef TIMER_FUNC_TYPE; @@ ( _E->_timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -&_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)_callback; +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -&_callback; +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; ) // Sometimes timer functions are called directly. Replace matched args. @change_timer_function_calls depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression _E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_data; @@ _callback( ( -(_cast_data)_E +&_E->_timer | -(_cast_data)&_E +&_E._timer | -_E +&_E->_timer ) ) // If a timer has been configured without a data argument, it can be // converted without regard to the callback argument, since it is unused. @match_timer_function_unused_data@ expression _E; identifier _timer; identifier _callback; @@ ( -setup_timer(&_E->_timer, _callback, 0); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0L); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0UL); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0L); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0UL); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_timer, _callback, 0); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0L); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0UL); +timer_setup(&_timer, _callback, 0); | -setup_timer(_timer, _callback, 0); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0L); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0UL); +timer_setup(_timer, _callback, 0); ) @change_callback_unused_data depends on match_timer_function_unused_data@ identifier match_timer_function_unused_data._callback; type _origtype; identifier _origarg; @@ void _callback( -_origtype _origarg +struct timer_list *unused ) { ... when != _origarg } Signed-off-by: Kees Cook <keescook@chromium.org>
2017-10-16 21:43:17 +00:00
timer_setup(&dlci->t1, gsm_dlci_t1, 0);
tty_port_init(&dlci->port);
dlci->port.ops = &gsm_port_ops;
dlci->gsm = gsm;
dlci->addr = addr;
dlci->adaption = gsm->adaption;
dlci->mtu = gsm->mtu;
if (addr == 0)
dlci->prio = 0;
else
dlci->prio = roundup(addr + 1, 8) - 1;
dlci->ftype = gsm->ftype;
dlci->k = gsm->k;
dlci->state = DLCI_CLOSED;
if (addr) {
dlci->data = gsm_dlci_data;
/* Prevent us from sending data before the link is up */
dlci->constipated = true;
} else {
dlci->data = gsm_dlci_command;
}
gsm->dlci[addr] = dlci;
return dlci;
}
/**
* gsm_dlci_free - free DLCI
* @port: tty port for DLCI to free
*
* Free up a DLCI.
*
* Can sleep.
*/
static void gsm_dlci_free(struct tty_port *port)
{
struct gsm_dlci *dlci = container_of(port, struct gsm_dlci, port);
timer_shutdown_sync(&dlci->t1);
dlci->gsm->dlci[dlci->addr] = NULL;
kfifo_free(&dlci->fifo);
while ((dlci->skb = skb_dequeue(&dlci->skb_list)))
dev_kfree_skb(dlci->skb);
kfree(dlci);
}
static inline void dlci_get(struct gsm_dlci *dlci)
{
tty_port_get(&dlci->port);
}
static inline void dlci_put(struct gsm_dlci *dlci)
{
tty_port_put(&dlci->port);
}
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
static void gsm_destroy_network(struct gsm_dlci *dlci);
/**
* gsm_dlci_release - release DLCI
* @dlci: DLCI to destroy
*
* Release a DLCI. Actual free is deferred until either
* mux is closed or tty is closed - whichever is last.
*
* Can sleep.
*/
static void gsm_dlci_release(struct gsm_dlci *dlci)
{
struct tty_struct *tty = tty_port_tty_get(&dlci->port);
if (tty) {
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
mutex_lock(&dlci->mutex);
gsm_destroy_network(dlci);
mutex_unlock(&dlci->mutex);
/* We cannot use tty_hangup() because in tty_kref_put() the tty
* driver assumes that the hangup queue is free and reuses it to
* queue release_one_tty() -> NULL pointer panic in
* process_one_work().
*/
tty_vhangup(tty);
TTY/n_gsm: Removing the wrong tty_unlock/lock() in gsm_dlci_release() Commit 4d9b109060f690f5c835(tty: Prevent deadlock in n_gsm driver) tried to close all the virtual ports synchronously before closing the phycial ports, so the tty_vhangup() is used. But the tty_unlock/lock() is wrong: tty_release tty_ldisc_release tty_lock_pair(tty, o_tty) < == Here the tty is for physical port tty_ldisc_kill gsmld_close gsm_cleanup_mux gsm_dlci_release tty = tty_port_tty_get(&dlci->port) < == Here the tty(s) are for virtual port They are different ttys, so before tty_vhangup(virtual tty), do not need to call the tty_unlock(virtual tty) at all which causes unbalanced unlock warning. When enabling mutex debugging option, we will hit the below warning also: [ 99.276903] ===================================== [ 99.282172] [ BUG: bad unlock balance detected! ] [ 99.287442] 3.10.20-261976-gaec5ba0 #44 Tainted: G O [ 99.293972] ------------------------------------- [ 99.299240] mmgr/152 is trying to release lock (&tty->legacy_mutex) at: [ 99.306693] [<c1b2dcad>] mutex_unlock+0xd/0x10 [ 99.311669] but there are no more locks to release! [ 99.317131] [ 99.317131] other info that might help us debug this: [ 99.324440] 3 locks held by mmgr/152: [ 99.328542] #0: (&tty->legacy_mutex/1){......}, at: [<c1b30ab0>] tty_lock_nested+0x40/0x90 [ 99.338116] #1: (&tty->ldisc_mutex){......}, at: [<c15dbd02>] tty_ldisc_kill+0x22/0xd0 [ 99.347284] #2: (&gsm->mutex){......}, at: [<c15e3d83>] gsm_cleanup_mux+0x73/0x170 [ 99.356060] [ 99.356060] stack backtrace: [ 99.360932] CPU: 0 PID: 152 Comm: mmgr Tainted: G O 3.10.20-261976-gaec5ba0 #44 [ 99.370086] ef4a4de0 ef4a4de0 ef4c1d98 c1b27b91 ef4c1db8 c1292655 c1dd10f5 c1b2dcad [ 99.378921] c1b2dcad ef4a4de0 ef4a528c ffffffff ef4c1dfc c12930dd 00000246 00000000 [ 99.387754] 00000000 00000000 c15e1926 00000000 00000001 ddfa7530 00000003 c1b2dcad [ 99.396588] Call Trace: [ 99.399326] [<c1b27b91>] dump_stack+0x16/0x18 [ 99.404307] [<c1292655>] print_unlock_imbalance_bug+0xe5/0xf0 [ 99.410840] [<c1b2dcad>] ? mutex_unlock+0xd/0x10 [ 99.416110] [<c1b2dcad>] ? mutex_unlock+0xd/0x10 [ 99.421382] [<c12930dd>] lock_release_non_nested+0x1cd/0x210 [ 99.427818] [<c15e1926>] ? gsm_destroy_network+0x36/0x130 [ 99.433964] [<c1b2dcad>] ? mutex_unlock+0xd/0x10 [ 99.439235] [<c12931a2>] lock_release+0x82/0x1c0 [ 99.444505] [<c1b2dcad>] ? mutex_unlock+0xd/0x10 [ 99.449776] [<c1b2dcad>] ? mutex_unlock+0xd/0x10 [ 99.455047] [<c1b2dc2f>] __mutex_unlock_slowpath+0x5f/0xd0 [ 99.461288] [<c1b2dcad>] mutex_unlock+0xd/0x10 [ 99.466365] [<c1b30bb1>] tty_unlock+0x21/0x50 [ 99.471345] [<c15e3dd1>] gsm_cleanup_mux+0xc1/0x170 [ 99.476906] [<c15e44d2>] gsmld_close+0x52/0x90 [ 99.481983] [<c15db905>] tty_ldisc_close.isra.1+0x35/0x50 [ 99.488127] [<c15dbd0c>] tty_ldisc_kill+0x2c/0xd0 [ 99.493494] [<c15dc7af>] tty_ldisc_release+0x2f/0x50 [ 99.499152] [<c15d572c>] tty_release+0x37c/0x4b0 [ 99.504424] [<c1b2dcad>] ? mutex_unlock+0xd/0x10 [ 99.509695] [<c1b2dcad>] ? mutex_unlock+0xd/0x10 [ 99.514967] [<c1372f6e>] ? eventpoll_release_file+0x7e/0x90 [ 99.521307] [<c1335849>] __fput+0xd9/0x200 [ 99.525996] [<c133597d>] ____fput+0xd/0x10 [ 99.530685] [<c125c731>] task_work_run+0x81/0xb0 [ 99.535957] [<c12019e9>] do_notify_resume+0x49/0x70 [ 99.541520] [<c1b30dc4>] work_notifysig+0x29/0x31 [ 99.546897] ------------[ cut here ]------------ So here we can call tty_vhangup() directly which is for virtual port. Reviewed-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Liu, Chuansheng <chuansheng.liu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-12-18 05:30:11 +00:00
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
tty_port_tty_set(&dlci->port, NULL);
tty_kref_put(tty);
}
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
dlci->state = DLCI_CLOSED;
dlci_put(dlci);
}
/*
* LAPBish link layer logic
*/
/**
* gsm_queue - a GSM frame is ready to process
* @gsm: pointer to our gsm mux
*
* At this point in time a frame has arrived and been demangled from
* the line encoding. All the differences between the encodings have
* been handled below us and the frame is unpacked into the structures.
* The fcs holds the header FCS but any data FCS must be added here.
*/
static void gsm_queue(struct gsm_mux *gsm)
{
struct gsm_dlci *dlci;
u8 cr;
int address;
if (gsm->fcs != GOOD_FCS) {
gsm->bad_fcs++;
if (debug & DBG_DATA)
pr_debug("BAD FCS %02x\n", gsm->fcs);
return;
}
address = gsm->address >> 1;
if (address >= NUM_DLCI)
goto invalid;
cr = gsm->address & 1; /* C/R bit */
cr ^= gsm->initiator ? 0 : 1; /* Flip so 1 always means command */
gsm_print_packet("<--", address, cr, gsm->control, gsm->buf, gsm->len);
dlci = gsm->dlci[address];
switch (gsm->control) {
case SABM|PF:
if (cr == 1)
goto invalid;
if (dlci == NULL)
dlci = gsm_dlci_alloc(gsm, address);
if (dlci == NULL)
return;
if (dlci->dead)
gsm_response(gsm, address, DM|PF);
else {
gsm_response(gsm, address, UA|PF);
gsm_dlci_open(dlci);
}
break;
case DISC|PF:
if (cr == 1)
goto invalid;
if (dlci == NULL || dlci->state == DLCI_CLOSED) {
gsm_response(gsm, address, DM|PF);
return;
}
/* Real close complete */
tty: n_gsm: fix broken virtual tty handling Dynamic virtual tty registration was introduced to allow the user to handle these cases with uevent rules. The following commits relate to this: Commit 5b87686e3203 ("tty: n_gsm: Modify gsmtty driver register method when config requester") Commit 0b91b5332368 ("tty: n_gsm: Save dlci address open status when config requester") Commit 46292622ad73 ("tty: n_gsm: clean up indenting in gsm_queue()") However, the following behavior can be seen with this implementation: - n_gsm ldisc is activated via ioctl - all configuration parameters are set to their default value (initiator=0) - the mux gets activated and attached and gsmtty0 is being registered in in gsm_dlci_open() after DLCI 0 was established (DLCI 0 is the control channel) - the user configures n_gsm via ioctl GSMIOC_SETCONF as initiator - this re-attaches the n_gsm mux - no new gsmtty devices are registered in gsmld_attach_gsm() because the mux is already active - the initiator side registered only the control channel as gsmtty0 (which should never happen) and no user channel tty The commits above make it impossible to operate the initiator side as no user channel tty is or will be available. On the other hand, this behavior will make it also impossible to allow DLCI parameter negotiation on responder side in the future. The responder side first needs to provide a device for the application before the application can set its parameters of the associated DLCI via ioctl. Note that the user application is still able to detect a link establishment without relaying to uevent by waiting for DTR open on responder side. This is the same behavior as on a physical serial interface. And on initiator side a tty hangup can be detected if a link establishment request failed. Revert the commits above completely to always register all user channels and no control channel after mux attachment. No other changes are made. Fixes: 5b87686e3203 ("tty: n_gsm: Modify gsmtty driver register method when config requester") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:23 +00:00
gsm_response(gsm, address, UA|PF);
gsm_dlci_close(dlci);
break;
case UA|PF:
if (cr == 0 || dlci == NULL)
break;
switch (dlci->state) {
case DLCI_CLOSING:
gsm_dlci_close(dlci);
break;
case DLCI_OPENING:
gsm_dlci_open(dlci);
break;
default:
pr_debug("%s: unhandled state: %d\n", __func__,
dlci->state);
break;
}
break;
case DM: /* DM can be valid unsolicited */
case DM|PF:
if (cr)
goto invalid;
if (dlci == NULL)
return;
gsm_dlci_close(dlci);
break;
case UI:
case UI|PF:
case UIH:
case UIH|PF:
if (dlci == NULL || dlci->state != DLCI_OPEN) {
gsm_response(gsm, address, DM|PF);
return;
}
dlci->data(dlci, gsm->buf, gsm->len);
break;
default:
goto invalid;
}
return;
invalid:
gsm->malformed++;
return;
}
/**
* gsm0_receive - perform processing for non-transparency
* @gsm: gsm data for this ldisc instance
* @c: character
*
* Receive bytes in gsm mode 0
*/
static void gsm0_receive(struct gsm_mux *gsm, unsigned char c)
{
unsigned int len;
switch (gsm->state) {
case GSM_SEARCH: /* SOF marker */
if (c == GSM0_SOF) {
gsm->state = GSM_ADDRESS;
gsm->address = 0;
gsm->len = 0;
gsm->fcs = INIT_FCS;
}
break;
case GSM_ADDRESS: /* Address EA */
gsm->fcs = gsm_fcs_add(gsm->fcs, c);
if (gsm_read_ea(&gsm->address, c))
gsm->state = GSM_CONTROL;
break;
case GSM_CONTROL: /* Control Byte */
gsm->fcs = gsm_fcs_add(gsm->fcs, c);
gsm->control = c;
gsm->state = GSM_LEN0;
break;
case GSM_LEN0: /* Length EA */
gsm->fcs = gsm_fcs_add(gsm->fcs, c);
if (gsm_read_ea(&gsm->len, c)) {
if (gsm->len > gsm->mru) {
gsm->bad_size++;
gsm->state = GSM_SEARCH;
break;
}
gsm->count = 0;
if (!gsm->len)
gsm->state = GSM_FCS;
else
gsm->state = GSM_DATA;
break;
}
gsm->state = GSM_LEN1;
break;
case GSM_LEN1:
gsm->fcs = gsm_fcs_add(gsm->fcs, c);
len = c;
gsm->len |= len << 7;
if (gsm->len > gsm->mru) {
gsm->bad_size++;
gsm->state = GSM_SEARCH;
break;
}
gsm->count = 0;
if (!gsm->len)
gsm->state = GSM_FCS;
else
gsm->state = GSM_DATA;
break;
case GSM_DATA: /* Data */
gsm->buf[gsm->count++] = c;
if (gsm->count == gsm->len) {
/* Calculate final FCS for UI frames over all data */
if ((gsm->control & ~PF) != UIH) {
gsm->fcs = gsm_fcs_add_block(gsm->fcs, gsm->buf,
gsm->count);
}
gsm->state = GSM_FCS;
}
break;
case GSM_FCS: /* FCS follows the packet */
gsm->fcs = gsm_fcs_add(gsm->fcs, c);
gsm->state = GSM_SSOF;
break;
case GSM_SSOF:
gsm->state = GSM_SEARCH;
if (c == GSM0_SOF)
gsm_queue(gsm);
else
gsm->bad_size++;
break;
default:
pr_debug("%s: unhandled state: %d\n", __func__, gsm->state);
break;
}
}
/**
* gsm1_receive - perform processing for non-transparency
* @gsm: gsm data for this ldisc instance
* @c: character
*
* Receive bytes in mode 1 (Advanced option)
*/
static void gsm1_receive(struct gsm_mux *gsm, unsigned char c)
{
tty: n_gsm: fix software flow control handling n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.8.1 states that XON/XOFF characters shall be used instead of Fcon/Fcoff command in advanced option mode to handle flow control. Chapter 5.4.8.2 describes how XON/XOFF characters shall be handled. Basic option mode only used Fcon/Fcoff commands and no XON/XOFF characters. These are treated as data bytes here. The current implementation uses the gsm_mux field 'constipated' to handle flow control from the remote peer and the gsm_dlci field 'constipated' to handle flow control from each DLCI. The later is unrelated to this patch. The gsm_mux field is correctly set for Fcon/Fcoff commands in gsm_control_message(). However, the same is not true for XON/XOFF characters in gsm1_receive(). Disable software flow control handling in the tty to allow explicit handling by n_gsm. Add the missing handling in advanced option mode for gsm_mux in gsm1_receive() to comply with the standard. This patch depends on the following commit: Commit 8838b2af23ca ("tty: n_gsm: fix SW flow control encoding/handling") Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-3-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:25 +00:00
/* handle XON/XOFF */
if ((c & ISO_IEC_646_MASK) == XON) {
gsm->constipated = true;
return;
} else if ((c & ISO_IEC_646_MASK) == XOFF) {
gsm->constipated = false;
/* Kick the link in case it is idling */
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsmld_write_trigger(gsm);
tty: n_gsm: fix software flow control handling n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.8.1 states that XON/XOFF characters shall be used instead of Fcon/Fcoff command in advanced option mode to handle flow control. Chapter 5.4.8.2 describes how XON/XOFF characters shall be handled. Basic option mode only used Fcon/Fcoff commands and no XON/XOFF characters. These are treated as data bytes here. The current implementation uses the gsm_mux field 'constipated' to handle flow control from the remote peer and the gsm_dlci field 'constipated' to handle flow control from each DLCI. The later is unrelated to this patch. The gsm_mux field is correctly set for Fcon/Fcoff commands in gsm_control_message(). However, the same is not true for XON/XOFF characters in gsm1_receive(). Disable software flow control handling in the tty to allow explicit handling by n_gsm. Add the missing handling in advanced option mode for gsm_mux in gsm1_receive() to comply with the standard. This patch depends on the following commit: Commit 8838b2af23ca ("tty: n_gsm: fix SW flow control encoding/handling") Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-3-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:25 +00:00
return;
}
if (c == GSM1_SOF) {
/* EOF is only valid in frame if we have got to the data state */
if (gsm->state == GSM_DATA) {
if (gsm->count < 1) {
/* Missing FSC */
gsm->malformed++;
gsm->state = GSM_START;
return;
}
/* Remove the FCS from data */
gsm->count--;
if ((gsm->control & ~PF) != UIH) {
/* Calculate final FCS for UI frames over all
* data but FCS
*/
gsm->fcs = gsm_fcs_add_block(gsm->fcs, gsm->buf,
gsm->count);
}
/* Add the FCS itself to test against GOOD_FCS */
gsm->fcs = gsm_fcs_add(gsm->fcs, gsm->buf[gsm->count]);
gsm->len = gsm->count;
gsm_queue(gsm);
gsm->state = GSM_START;
return;
}
/* Any partial frame was a runt so go back to start */
if (gsm->state != GSM_START) {
if (gsm->state != GSM_SEARCH)
gsm->malformed++;
gsm->state = GSM_START;
}
/* A SOF in GSM_START means we are still reading idling or
framing bytes */
return;
}
if (c == GSM1_ESCAPE) {
gsm->escape = true;
return;
}
/* Only an unescaped SOF gets us out of GSM search */
if (gsm->state == GSM_SEARCH)
return;
if (gsm->escape) {
c ^= GSM1_ESCAPE_BITS;
gsm->escape = false;
}
switch (gsm->state) {
case GSM_START: /* First byte after SOF */
gsm->address = 0;
gsm->state = GSM_ADDRESS;
gsm->fcs = INIT_FCS;
fallthrough;
case GSM_ADDRESS: /* Address continuation */
gsm->fcs = gsm_fcs_add(gsm->fcs, c);
if (gsm_read_ea(&gsm->address, c))
gsm->state = GSM_CONTROL;
break;
case GSM_CONTROL: /* Control Byte */
gsm->fcs = gsm_fcs_add(gsm->fcs, c);
gsm->control = c;
gsm->count = 0;
gsm->state = GSM_DATA;
break;
case GSM_DATA: /* Data */
if (gsm->count > gsm->mru) { /* Allow one for the FCS */
gsm->state = GSM_OVERRUN;
gsm->bad_size++;
} else
gsm->buf[gsm->count++] = c;
break;
case GSM_OVERRUN: /* Over-long - eg a dropped SOF */
break;
default:
pr_debug("%s: unhandled state: %d\n", __func__, gsm->state);
break;
}
}
/**
* gsm_error - handle tty error
* @gsm: ldisc data
*
* Handle an error in the receipt of data for a frame. Currently we just
* go back to hunting for a SOF.
*
* FIXME: better diagnostics ?
*/
static void gsm_error(struct gsm_mux *gsm)
{
gsm->state = GSM_SEARCH;
gsm->io_error++;
}
/**
* gsm_cleanup_mux - generic GSM protocol cleanup
* @gsm: our mux
tty: n_gsm: fix restart handling via CLD command n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.8.2 states that both sides will revert to the non-multiplexed mode via a close-down message (CLD). The usual program flow is as following: - start multiplex mode by sending AT+CMUX to the mobile - establish the control channel (DLCI 0) - establish user channels (DLCI >0) - terminate user channels - send close-down message (CLD) - revert to AT protocol (i.e. leave multiplexed mode) The AT protocol is out of scope of the n_gsm driver. However, gsm_disconnect() sends CLD if gsm_config() detects that the requested parameters require the mux protocol to restart. The next immediate action is to start the mux protocol by opening DLCI 0 again. Any responder side which handles CLD commands correctly forces us to fail at this point because AT+CMUX needs to be sent to the mobile to start the mux again. Therefore, remove the CLD command in this phase and keep both sides in multiplexed mode. Remove the gsm_disconnect() function as it become unnecessary and merge the remaining parts into gsm_cleanup_mux() to handle the termination order and locking correctly. Fixes: 71e077915396 ("tty: n_gsm: do not send/receive in ldisc close path") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220414094225.4527-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-14 09:42:07 +00:00
* @disc: disconnect link?
*
* Clean up the bits of the mux which are the same for all framing
* protocols. Remove the mux from the mux table, stop all the timers
* and then shut down each device hanging up the channels as we go.
*/
tty: n_gsm: fix restart handling via CLD command n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.8.2 states that both sides will revert to the non-multiplexed mode via a close-down message (CLD). The usual program flow is as following: - start multiplex mode by sending AT+CMUX to the mobile - establish the control channel (DLCI 0) - establish user channels (DLCI >0) - terminate user channels - send close-down message (CLD) - revert to AT protocol (i.e. leave multiplexed mode) The AT protocol is out of scope of the n_gsm driver. However, gsm_disconnect() sends CLD if gsm_config() detects that the requested parameters require the mux protocol to restart. The next immediate action is to start the mux protocol by opening DLCI 0 again. Any responder side which handles CLD commands correctly forces us to fail at this point because AT+CMUX needs to be sent to the mobile to start the mux again. Therefore, remove the CLD command in this phase and keep both sides in multiplexed mode. Remove the gsm_disconnect() function as it become unnecessary and merge the remaining parts into gsm_cleanup_mux() to handle the termination order and locking correctly. Fixes: 71e077915396 ("tty: n_gsm: do not send/receive in ldisc close path") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220414094225.4527-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-14 09:42:07 +00:00
static void gsm_cleanup_mux(struct gsm_mux *gsm, bool disc)
{
int i;
struct gsm_dlci *dlci = gsm->dlci[0];
struct gsm_msg *txq, *ntxq;
gsm->dead = true;
tty: n_gsm: fix restart handling via CLD command n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.8.2 states that both sides will revert to the non-multiplexed mode via a close-down message (CLD). The usual program flow is as following: - start multiplex mode by sending AT+CMUX to the mobile - establish the control channel (DLCI 0) - establish user channels (DLCI >0) - terminate user channels - send close-down message (CLD) - revert to AT protocol (i.e. leave multiplexed mode) The AT protocol is out of scope of the n_gsm driver. However, gsm_disconnect() sends CLD if gsm_config() detects that the requested parameters require the mux protocol to restart. The next immediate action is to start the mux protocol by opening DLCI 0 again. Any responder side which handles CLD commands correctly forces us to fail at this point because AT+CMUX needs to be sent to the mobile to start the mux again. Therefore, remove the CLD command in this phase and keep both sides in multiplexed mode. Remove the gsm_disconnect() function as it become unnecessary and merge the remaining parts into gsm_cleanup_mux() to handle the termination order and locking correctly. Fixes: 71e077915396 ("tty: n_gsm: do not send/receive in ldisc close path") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220414094225.4527-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-14 09:42:07 +00:00
mutex_lock(&gsm->mutex);
tty: n_gsm: fix restart handling via CLD command n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.8.2 states that both sides will revert to the non-multiplexed mode via a close-down message (CLD). The usual program flow is as following: - start multiplex mode by sending AT+CMUX to the mobile - establish the control channel (DLCI 0) - establish user channels (DLCI >0) - terminate user channels - send close-down message (CLD) - revert to AT protocol (i.e. leave multiplexed mode) The AT protocol is out of scope of the n_gsm driver. However, gsm_disconnect() sends CLD if gsm_config() detects that the requested parameters require the mux protocol to restart. The next immediate action is to start the mux protocol by opening DLCI 0 again. Any responder side which handles CLD commands correctly forces us to fail at this point because AT+CMUX needs to be sent to the mobile to start the mux again. Therefore, remove the CLD command in this phase and keep both sides in multiplexed mode. Remove the gsm_disconnect() function as it become unnecessary and merge the remaining parts into gsm_cleanup_mux() to handle the termination order and locking correctly. Fixes: 71e077915396 ("tty: n_gsm: do not send/receive in ldisc close path") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220414094225.4527-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-14 09:42:07 +00:00
if (dlci) {
if (disc && dlci->state != DLCI_CLOSED) {
gsm_dlci_begin_close(dlci);
wait_event(gsm->event, dlci->state == DLCI_CLOSED);
}
tty: n_gsm: fix restart handling via CLD command n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.8.2 states that both sides will revert to the non-multiplexed mode via a close-down message (CLD). The usual program flow is as following: - start multiplex mode by sending AT+CMUX to the mobile - establish the control channel (DLCI 0) - establish user channels (DLCI >0) - terminate user channels - send close-down message (CLD) - revert to AT protocol (i.e. leave multiplexed mode) The AT protocol is out of scope of the n_gsm driver. However, gsm_disconnect() sends CLD if gsm_config() detects that the requested parameters require the mux protocol to restart. The next immediate action is to start the mux protocol by opening DLCI 0 again. Any responder side which handles CLD commands correctly forces us to fail at this point because AT+CMUX needs to be sent to the mobile to start the mux again. Therefore, remove the CLD command in this phase and keep both sides in multiplexed mode. Remove the gsm_disconnect() function as it become unnecessary and merge the remaining parts into gsm_cleanup_mux() to handle the termination order and locking correctly. Fixes: 71e077915396 ("tty: n_gsm: do not send/receive in ldisc close path") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220414094225.4527-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-14 09:42:07 +00:00
dlci->dead = true;
}
tty: n_gsm: fix restart handling via CLD command n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.8.2 states that both sides will revert to the non-multiplexed mode via a close-down message (CLD). The usual program flow is as following: - start multiplex mode by sending AT+CMUX to the mobile - establish the control channel (DLCI 0) - establish user channels (DLCI >0) - terminate user channels - send close-down message (CLD) - revert to AT protocol (i.e. leave multiplexed mode) The AT protocol is out of scope of the n_gsm driver. However, gsm_disconnect() sends CLD if gsm_config() detects that the requested parameters require the mux protocol to restart. The next immediate action is to start the mux protocol by opening DLCI 0 again. Any responder side which handles CLD commands correctly forces us to fail at this point because AT+CMUX needs to be sent to the mobile to start the mux again. Therefore, remove the CLD command in this phase and keep both sides in multiplexed mode. Remove the gsm_disconnect() function as it become unnecessary and merge the remaining parts into gsm_cleanup_mux() to handle the termination order and locking correctly. Fixes: 71e077915396 ("tty: n_gsm: do not send/receive in ldisc close path") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220414094225.4527-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-14 09:42:07 +00:00
/* Finish outstanding timers, making sure they are done */
del_timer_sync(&gsm->kick_timer);
del_timer_sync(&gsm->t2_timer);
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
del_timer_sync(&gsm->ka_timer);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
/* Finish writing to ldisc */
flush_work(&gsm->tx_work);
/* Free up any link layer users and finally the control channel */
if (gsm->has_devices) {
gsm_unregister_devices(gsm_tty_driver, gsm->num);
gsm->has_devices = false;
}
for (i = NUM_DLCI - 1; i >= 0; i--)
if (gsm->dlci[i])
gsm_dlci_release(gsm->dlci[i]);
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
mutex_unlock(&gsm->mutex);
/* Now wipe the queues */
tty_ldisc_flush(gsm->tty);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
list_for_each_entry_safe(txq, ntxq, &gsm->tx_ctrl_list, list)
kfree(txq);
INIT_LIST_HEAD(&gsm->tx_ctrl_list);
list_for_each_entry_safe(txq, ntxq, &gsm->tx_data_list, list)
kfree(txq);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
INIT_LIST_HEAD(&gsm->tx_data_list);
}
/**
* gsm_activate_mux - generic GSM setup
* @gsm: our mux
*
* Set up the bits of the mux which are the same for all framing
* protocols. Add the mux to the mux table so it can be opened and
* finally kick off connecting to DLCI 0 on the modem.
*/
static int gsm_activate_mux(struct gsm_mux *gsm)
{
struct gsm_dlci *dlci;
int ret;
dlci = gsm_dlci_alloc(gsm, 0);
if (dlci == NULL)
return -ENOMEM;
if (gsm->encoding == GSM_BASIC_OPT)
gsm->receive = gsm0_receive;
else
gsm->receive = gsm1_receive;
ret = gsm_register_devices(gsm_tty_driver, gsm->num);
if (ret)
return ret;
gsm->has_devices = true;
gsm->dead = false; /* Tty opens are now permissible */
return 0;
}
/**
* gsm_free_mux - free up a mux
* @gsm: mux to free
*
* Dispose of allocated resources for a dead mux
*/
static void gsm_free_mux(struct gsm_mux *gsm)
{
int i;
for (i = 0; i < MAX_MUX; i++) {
if (gsm == gsm_mux[i]) {
gsm_mux[i] = NULL;
break;
}
}
mutex_destroy(&gsm->mutex);
kfree(gsm->txframe);
kfree(gsm->buf);
kfree(gsm);
}
/**
* gsm_free_muxr - free up a mux
* @ref: kreference to the mux to free
*
* Dispose of allocated resources for a dead mux
*/
static void gsm_free_muxr(struct kref *ref)
{
struct gsm_mux *gsm = container_of(ref, struct gsm_mux, ref);
gsm_free_mux(gsm);
}
static inline void mux_get(struct gsm_mux *gsm)
{
unsigned long flags;
spin_lock_irqsave(&gsm_mux_lock, flags);
kref_get(&gsm->ref);
spin_unlock_irqrestore(&gsm_mux_lock, flags);
}
static inline void mux_put(struct gsm_mux *gsm)
{
unsigned long flags;
spin_lock_irqsave(&gsm_mux_lock, flags);
kref_put(&gsm->ref, gsm_free_muxr);
spin_unlock_irqrestore(&gsm_mux_lock, flags);
}
static inline unsigned int mux_num_to_base(struct gsm_mux *gsm)
{
return gsm->num * NUM_DLCI;
}
static inline unsigned int mux_line_to_num(unsigned int line)
{
return line / NUM_DLCI;
}
/**
* gsm_alloc_mux - allocate a mux
*
* Creates a new mux ready for activation.
*/
static struct gsm_mux *gsm_alloc_mux(void)
{
int i;
struct gsm_mux *gsm = kzalloc(sizeof(struct gsm_mux), GFP_KERNEL);
if (gsm == NULL)
return NULL;
gsm->buf = kmalloc(MAX_MRU + 1, GFP_KERNEL);
if (gsm->buf == NULL) {
kfree(gsm);
return NULL;
}
gsm->txframe = kmalloc(2 * (MAX_MTU + PROT_OVERHEAD - 1), GFP_KERNEL);
if (gsm->txframe == NULL) {
kfree(gsm->buf);
kfree(gsm);
return NULL;
}
spin_lock_init(&gsm->lock);
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
mutex_init(&gsm->mutex);
kref_init(&gsm->ref);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
INIT_LIST_HEAD(&gsm->tx_ctrl_list);
INIT_LIST_HEAD(&gsm->tx_data_list);
timer_setup(&gsm->kick_timer, gsm_kick_timer, 0);
timer_setup(&gsm->t2_timer, gsm_control_retransmit, 0);
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
timer_setup(&gsm->ka_timer, gsm_control_keep_alive, 0);
INIT_WORK(&gsm->tx_work, gsmld_write_task);
init_waitqueue_head(&gsm->event);
spin_lock_init(&gsm->control_lock);
spin_lock_init(&gsm->tx_lock);
gsm->t1 = T1;
gsm->t2 = T2;
gsm->t3 = T3;
gsm->n2 = N2;
gsm->k = K;
gsm->ftype = UIH;
gsm->adaption = 1;
gsm->encoding = GSM_ADV_OPT;
gsm->mru = 64; /* Default to encoding 1 so these should be 64 */
gsm->mtu = 64;
gsm->dead = true; /* Avoid early tty opens */
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
gsm->wait_config = false; /* Disabled */
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
gsm->keep_alive = 0; /* Disabled */
/* Store the instance to the mux array or abort if no space is
* available.
*/
spin_lock(&gsm_mux_lock);
for (i = 0; i < MAX_MUX; i++) {
if (!gsm_mux[i]) {
gsm_mux[i] = gsm;
gsm->num = i;
break;
}
}
spin_unlock(&gsm_mux_lock);
if (i == MAX_MUX) {
mutex_destroy(&gsm->mutex);
kfree(gsm->txframe);
kfree(gsm->buf);
kfree(gsm);
return NULL;
}
return gsm;
}
static void gsm_copy_config_values(struct gsm_mux *gsm,
struct gsm_config *c)
{
memset(c, 0, sizeof(*c));
c->adaption = gsm->adaption;
c->encapsulation = gsm->encoding;
c->initiator = gsm->initiator;
c->t1 = gsm->t1;
c->t2 = gsm->t2;
c->t3 = gsm->t3;
c->n2 = gsm->n2;
if (gsm->ftype == UIH)
c->i = 1;
else
c->i = 2;
pr_debug("Ftype %d i %d\n", gsm->ftype, c->i);
c->mru = gsm->mru;
c->mtu = gsm->mtu;
c->k = gsm->k;
}
static int gsm_config(struct gsm_mux *gsm, struct gsm_config *c)
{
int ret = 0;
int need_close = 0;
int need_restart = 0;
/* Stuff we don't support yet - UI or I frame transport */
if (c->adaption != 1 && c->adaption != 2)
return -EOPNOTSUPP;
/* Check the MRU/MTU range looks sane */
if (c->mru < MIN_MTU || c->mtu < MIN_MTU)
return -EINVAL;
if (c->mru > MAX_MRU || c->mtu > MAX_MTU)
return -EINVAL;
if (c->t3 > MAX_T3)
return -EINVAL;
if (c->n2 > 255)
return -EINVAL;
if (c->encapsulation > 1) /* Basic, advanced, no I */
return -EINVAL;
if (c->initiator > 1)
return -EINVAL;
if (c->k > MAX_WINDOW_SIZE)
return -EINVAL;
if (c->i == 0 || c->i > 2) /* UIH and UI only */
return -EINVAL;
/*
* See what is needed for reconfiguration
*/
/* Timing fields */
if (c->t1 != 0 && c->t1 != gsm->t1)
need_restart = 1;
if (c->t2 != 0 && c->t2 != gsm->t2)
need_restart = 1;
if (c->encapsulation != gsm->encoding)
need_restart = 1;
if (c->adaption != gsm->adaption)
need_restart = 1;
/* Requires care */
if (c->initiator != gsm->initiator)
need_close = 1;
if (c->mru != gsm->mru)
need_restart = 1;
if (c->mtu != gsm->mtu)
need_restart = 1;
/*
* Close down what is needed, restart and initiate the new
tty: n_gsm: fix restart handling via CLD command n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.8.2 states that both sides will revert to the non-multiplexed mode via a close-down message (CLD). The usual program flow is as following: - start multiplex mode by sending AT+CMUX to the mobile - establish the control channel (DLCI 0) - establish user channels (DLCI >0) - terminate user channels - send close-down message (CLD) - revert to AT protocol (i.e. leave multiplexed mode) The AT protocol is out of scope of the n_gsm driver. However, gsm_disconnect() sends CLD if gsm_config() detects that the requested parameters require the mux protocol to restart. The next immediate action is to start the mux protocol by opening DLCI 0 again. Any responder side which handles CLD commands correctly forces us to fail at this point because AT+CMUX needs to be sent to the mobile to start the mux again. Therefore, remove the CLD command in this phase and keep both sides in multiplexed mode. Remove the gsm_disconnect() function as it become unnecessary and merge the remaining parts into gsm_cleanup_mux() to handle the termination order and locking correctly. Fixes: 71e077915396 ("tty: n_gsm: do not send/receive in ldisc close path") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220414094225.4527-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-14 09:42:07 +00:00
* configuration. On the first time there is no DLCI[0]
* and closing or cleaning up is not necessary.
*/
tty: n_gsm: fix restart handling via CLD command n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.8.2 states that both sides will revert to the non-multiplexed mode via a close-down message (CLD). The usual program flow is as following: - start multiplex mode by sending AT+CMUX to the mobile - establish the control channel (DLCI 0) - establish user channels (DLCI >0) - terminate user channels - send close-down message (CLD) - revert to AT protocol (i.e. leave multiplexed mode) The AT protocol is out of scope of the n_gsm driver. However, gsm_disconnect() sends CLD if gsm_config() detects that the requested parameters require the mux protocol to restart. The next immediate action is to start the mux protocol by opening DLCI 0 again. Any responder side which handles CLD commands correctly forces us to fail at this point because AT+CMUX needs to be sent to the mobile to start the mux again. Therefore, remove the CLD command in this phase and keep both sides in multiplexed mode. Remove the gsm_disconnect() function as it become unnecessary and merge the remaining parts into gsm_cleanup_mux() to handle the termination order and locking correctly. Fixes: 71e077915396 ("tty: n_gsm: do not send/receive in ldisc close path") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220414094225.4527-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-14 09:42:07 +00:00
if (need_close || need_restart)
gsm_cleanup_mux(gsm, true);
gsm->initiator = c->initiator;
gsm->mru = c->mru;
gsm->mtu = c->mtu;
gsm->encoding = c->encapsulation ? GSM_ADV_OPT : GSM_BASIC_OPT;
gsm->adaption = c->adaption;
gsm->n2 = c->n2;
if (c->i == 1)
gsm->ftype = UIH;
else if (c->i == 2)
gsm->ftype = UI;
if (c->t1)
gsm->t1 = c->t1;
if (c->t2)
gsm->t2 = c->t2;
if (c->t3)
gsm->t3 = c->t3;
if (c->k)
gsm->k = c->k;
/*
* FIXME: We need to separate activation/deactivation from adding
* and removing from the mux array
*/
if (gsm->dead) {
ret = gsm_activate_mux(gsm);
if (ret)
return ret;
if (gsm->initiator)
gsm_dlci_begin_open(gsm->dlci[0]);
}
return 0;
}
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
static void gsm_copy_config_ext_values(struct gsm_mux *gsm,
struct gsm_config_ext *ce)
{
memset(ce, 0, sizeof(*ce));
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
ce->wait_config = gsm->wait_config ? 1 : 0;
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
ce->keep_alive = gsm->keep_alive;
}
static int gsm_config_ext(struct gsm_mux *gsm, struct gsm_config_ext *ce)
{
unsigned int i;
/*
* Check that userspace doesn't put stuff in here to prevent breakages
* in the future.
*/
for (i = 0; i < ARRAY_SIZE(ce->reserved); i++)
if (ce->reserved[i])
return -EINVAL;
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
/*
* Setup the new configuration values
*/
gsm->wait_config = ce->wait_config ? true : false;
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
gsm->keep_alive = ce->keep_alive;
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
return 0;
}
/**
* gsmld_output - write to link
* @gsm: our mux
* @data: bytes to output
* @len: size
*
* Write a block of data from the GSM mux to the data channel. This
* will eventually be serialized from above but at the moment isn't.
*/
static int gsmld_output(struct gsm_mux *gsm, u8 *data, int len)
{
if (tty_write_room(gsm->tty) < len) {
set_bit(TTY_DO_WRITE_WAKEUP, &gsm->tty->flags);
return -ENOSPC;
}
if (debug & DBG_DATA)
gsm_hex_dump_bytes(__func__, data, len);
return gsm->tty->ops->write(gsm->tty, data, len);
}
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
/**
* gsmld_write_trigger - schedule ldisc write task
* @gsm: our mux
*/
static void gsmld_write_trigger(struct gsm_mux *gsm)
{
if (!gsm || !gsm->dlci[0] || gsm->dlci[0]->dead)
return;
schedule_work(&gsm->tx_work);
}
/**
* gsmld_write_task - ldisc write task
* @work: our tx write work
*
* Writes out data to the ldisc if possible. We are doing this here to
* avoid dead-locking. This returns if no space or data is left for output.
*/
static void gsmld_write_task(struct work_struct *work)
{
struct gsm_mux *gsm = container_of(work, struct gsm_mux, tx_work);
unsigned long flags;
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
int i, ret;
/* All outstanding control channel and control messages and one data
* frame is sent.
*/
ret = -ENODEV;
spin_lock_irqsave(&gsm->tx_lock, flags);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
if (gsm->tty)
ret = gsm_data_kick(gsm);
spin_unlock_irqrestore(&gsm->tx_lock, flags);
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
if (ret >= 0)
for (i = 0; i < NUM_DLCI; i++)
if (gsm->dlci[i])
tty_port_tty_wakeup(&gsm->dlci[i]->port);
}
/**
* gsmld_attach_gsm - mode set up
* @tty: our tty structure
* @gsm: our mux
*
* Set up the MUX for basic mode and commence connecting to the
* modem. Currently called from the line discipline set up but
* will need moving to an ioctl path.
*/
static void gsmld_attach_gsm(struct tty_struct *tty, struct gsm_mux *gsm)
{
gsm->tty = tty_kref_get(tty);
tty: n_gsm: fix software flow control handling n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.8.1 states that XON/XOFF characters shall be used instead of Fcon/Fcoff command in advanced option mode to handle flow control. Chapter 5.4.8.2 describes how XON/XOFF characters shall be handled. Basic option mode only used Fcon/Fcoff commands and no XON/XOFF characters. These are treated as data bytes here. The current implementation uses the gsm_mux field 'constipated' to handle flow control from the remote peer and the gsm_dlci field 'constipated' to handle flow control from each DLCI. The later is unrelated to this patch. The gsm_mux field is correctly set for Fcon/Fcoff commands in gsm_control_message(). However, the same is not true for XON/XOFF characters in gsm1_receive(). Disable software flow control handling in the tty to allow explicit handling by n_gsm. Add the missing handling in advanced option mode for gsm_mux in gsm1_receive() to comply with the standard. This patch depends on the following commit: Commit 8838b2af23ca ("tty: n_gsm: fix SW flow control encoding/handling") Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-3-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:25 +00:00
/* Turn off tty XON/XOFF handling to handle it explicitly. */
gsm->old_c_iflag = tty->termios.c_iflag;
tty->termios.c_iflag &= (IXON | IXOFF);
}
/**
* gsmld_detach_gsm - stop doing 0710 mux
* @tty: tty attached to the mux
* @gsm: mux
*
* Shutdown and then clean up the resources used by the line discipline
*/
static void gsmld_detach_gsm(struct tty_struct *tty, struct gsm_mux *gsm)
{
WARN_ON(tty != gsm->tty);
tty: n_gsm: fix software flow control handling n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.8.1 states that XON/XOFF characters shall be used instead of Fcon/Fcoff command in advanced option mode to handle flow control. Chapter 5.4.8.2 describes how XON/XOFF characters shall be handled. Basic option mode only used Fcon/Fcoff commands and no XON/XOFF characters. These are treated as data bytes here. The current implementation uses the gsm_mux field 'constipated' to handle flow control from the remote peer and the gsm_dlci field 'constipated' to handle flow control from each DLCI. The later is unrelated to this patch. The gsm_mux field is correctly set for Fcon/Fcoff commands in gsm_control_message(). However, the same is not true for XON/XOFF characters in gsm1_receive(). Disable software flow control handling in the tty to allow explicit handling by n_gsm. Add the missing handling in advanced option mode for gsm_mux in gsm1_receive() to comply with the standard. This patch depends on the following commit: Commit 8838b2af23ca ("tty: n_gsm: fix SW flow control encoding/handling") Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-3-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:25 +00:00
/* Restore tty XON/XOFF handling. */
gsm->tty->termios.c_iflag = gsm->old_c_iflag;
tty_kref_put(gsm->tty);
gsm->tty = NULL;
}
Revert "tty: make receive_buf() return the amout of bytes received" This reverts commit b1c43f82c5aa265442f82dba31ce985ebb7aa71c. It was broken in so many ways, and results in random odd pty issues. It re-introduced the buggy schedule_work() in flush_to_ldisc() that can cause endless work-loops (see commit a5660b41af6a: "tty: fix endless work loop when the buffer fills up"). It also used an "unsigned int" return value fo the ->receive_buf() function, but then made multiple functions return a negative error code, and didn't actually check for the error in the caller. And it didn't actually work at all. BenH bisected down odd tty behavior to it: "It looks like the patch is causing some major malfunctions of the X server for me, possibly related to PTYs. For example, cat'ing a large file in a gnome terminal hangs the kernel for -minutes- in a loop of what looks like flush_to_ldisc/workqueue code, (some ftrace data in the quoted bits further down). ... Some more data: It -looks- like what happens is that the flush_to_ldisc work queue entry constantly re-queues itself (because the PTY is full ?) and the workqueue thread will basically loop forver calling it without ever scheduling, thus starving the consumer process that could have emptied the PTY." which is pretty much exactly the problem we fixed in a5660b41af6a. Milton Miller pointed out the 'unsigned int' issue. Reported-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Reported-by: Milton Miller <miltonm@bga.com> Cc: Stefan Bigler <stefan.bigler@keymile.com> Cc: Toby Gray <toby.gray@realvnc.com> Cc: Felipe Balbi <balbi@ti.com> Cc: Greg Kroah-Hartman <gregkh@suse.de> Cc: Alan Cox <alan@lxorguk.ukuu.org.uk> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-06-03 21:33:24 +00:00
static void gsmld_receive_buf(struct tty_struct *tty, const unsigned char *cp,
const char *fp, int count)
{
struct gsm_mux *gsm = tty->disc_data;
char flags = TTY_NORMAL;
if (debug & DBG_DATA)
gsm_hex_dump_bytes(__func__, cp, count);
for (; count; count--, cp++) {
if (fp)
flags = *fp++;
switch (flags) {
case TTY_NORMAL:
tty: n_gsm: add sanity check for gsm->receive in gsm_receive_buf() A null pointer dereference can happen when attempting to access the "gsm->receive()" function in gsmld_receive_buf(). Currently, the code assumes that gsm->recieve is only called after MUX activation. Since the gsmld_receive_buf() function can be accessed without the need to initialize the MUX, the gsm->receive() function will not be set and a NULL pointer dereference will occur. Fix this by avoiding the call to "gsm->receive()" in case the function is not initialized by adding a sanity check. Call Trace: <TASK> gsmld_receive_buf+0x1c2/0x2f0 drivers/tty/n_gsm.c:2861 tiocsti drivers/tty/tty_io.c:2293 [inline] tty_ioctl+0xa75/0x15d0 drivers/tty/tty_io.c:2692 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:870 [inline] __se_sys_ioctl fs/ioctl.c:856 [inline] __x64_sys_ioctl+0x193/0x200 fs/ioctl.c:856 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd Link: https://syzkaller.appspot.com/bug?id=bdf035c61447f8c6e0e6920315d577cb5cc35ac5 Fixes: 01aecd917114 ("tty: n_gsm: fix tty registration before control channel open") Cc: stable <stable@kernel.org> Reported-and-tested-by: syzbot+e3563f0c94e188366dbb@syzkaller.appspotmail.com Signed-off-by: Mazin Al Haddad <mazinalhaddad05@gmail.com> Link: https://lore.kernel.org/r/20220814015211.84180-1-mazinalhaddad05@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-08-14 01:52:12 +00:00
if (gsm->receive)
gsm->receive(gsm, *cp);
break;
case TTY_OVERRUN:
case TTY_BREAK:
case TTY_PARITY:
case TTY_FRAME:
gsm_error(gsm);
break;
default:
WARN_ONCE(1, "%s: unknown flag %d\n",
tty_name(tty), flags);
break;
}
}
/* FASYNC if needed ? */
/* If clogged call tty_throttle(tty); */
}
/**
* gsmld_flush_buffer - clean input queue
* @tty: terminal device
*
* Flush the input buffer. Called when the line discipline is
* being closed, when the tty layer wants the buffer flushed (eg
* at hangup).
*/
static void gsmld_flush_buffer(struct tty_struct *tty)
{
}
/**
* gsmld_close - close the ldisc for this tty
* @tty: device
*
* Called from the terminal layer when this line discipline is
* being shut down, either because of a close or becsuse of a
* discipline change. The function will not be called while other
* ldisc methods are in progress.
*/
static void gsmld_close(struct tty_struct *tty)
{
struct gsm_mux *gsm = tty->disc_data;
/* The ldisc locks and closes the port before calling our close. This
* means we have no way to do a proper disconnect. We will not bother
* to do one.
*/
gsm_cleanup_mux(gsm, false);
gsmld_detach_gsm(tty, gsm);
gsmld_flush_buffer(tty);
/* Do other clean up here */
mux_put(gsm);
}
/**
* gsmld_open - open an ldisc
* @tty: terminal to open
*
* Called when this line discipline is being attached to the
* terminal device. Can sleep. Called serialized so that no
* other events will occur in parallel. No further open will occur
* until a close.
*/
static int gsmld_open(struct tty_struct *tty)
{
struct gsm_mux *gsm;
if (tty->ops->write == NULL)
return -EINVAL;
/* Attach our ldisc data */
gsm = gsm_alloc_mux();
if (gsm == NULL)
return -ENOMEM;
tty->disc_data = gsm;
tty->receive_room = 65536;
/* Attach the initial passive connection */
gsmld_attach_gsm(tty, gsm);
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
/* The mux will not be activated yet, we wait for correct
* configuration first.
*/
if (gsm->encoding == GSM_BASIC_OPT)
gsm->receive = gsm0_receive;
else
gsm->receive = gsm1_receive;
return 0;
}
/**
* gsmld_write_wakeup - asynchronous I/O notifier
* @tty: tty device
*
* Required for the ptys, serial driver etc. since processes
* that attach themselves to the master and rely on ASYNC
* IO must be woken up
*/
static void gsmld_write_wakeup(struct tty_struct *tty)
{
struct gsm_mux *gsm = tty->disc_data;
/* Queue poll */
tty: n_gsm: fix deadlock and link starvation in outgoing data path The current implementation queues up new control and user packets as needed and processes this queue down to the ldisc in the same code path. That means that the upper and the lower layer are hard coupled in the code. Due to this deadlocks can happen as seen below while transmitting data, especially during ldisc congestion. Furthermore, the data channels starve the control channel on high transmission load on the ldisc. Introduce an additional control channel data queue to prevent timeouts and link hangups during ldisc congestion. This is being processed before the user channel data queue in gsm_data_kick(), i.e. with the highest priority. Put the queue to ldisc data path into a workqueue and trigger it whenever new data has been put into the transmission queue. Change gsm_dlci_data_sweep() accordingly to fill up the transmission queue until TX_THRESH_HI. This solves the locking issue, keeps latency low and provides good performance on high data load. Note that now all packets from a DLCI are removed from the internal queue if the associated DLCI was closed. This ensures that no data is sent by the introduced write task to an already closed DLCI. BUG: spinlock recursion on CPU#0, test_v24_loop/124 lock: serial8250_ports+0x3a8/0x7500, .magic: dead4ead, .owner: test_v24_loop/124, .owner_cpu: 0 CPU: 0 PID: 124 Comm: test_v24_loop Tainted: G O 5.18.0-rc2 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 Call Trace: <IRQ> dump_stack_lvl+0x34/0x44 do_raw_spin_lock+0x76/0xa0 _raw_spin_lock_irqsave+0x72/0x80 uart_write_room+0x3b/0xc0 gsm_data_kick+0x14b/0x240 [n_gsm] gsmld_write_wakeup+0x35/0x70 [n_gsm] tty_wakeup+0x53/0x60 tty_port_default_wakeup+0x1b/0x30 serial8250_tx_chars+0x12f/0x220 serial8250_handle_irq.part.0+0xfe/0x150 serial8250_default_handle_irq+0x48/0x80 serial8250_interrupt+0x56/0xa0 __handle_irq_event_percpu+0x78/0x1f0 handle_irq_event+0x34/0x70 handle_fasteoi_irq+0x90/0x1e0 __common_interrupt+0x69/0x100 common_interrupt+0x48/0xc0 asm_common_interrupt+0x1e/0x40 RIP: 0010:__do_softirq+0x83/0x34e Code: 2a 0a ff 0f b7 ed c7 44 24 10 0a 00 00 00 48 c7 c7 51 2a 64 82 e8 2d e2 d5 ff 65 66 c7 05 83 af 1e 7e 00 00 fb b8 ff ff ff ff <49> c7 c2 40 61 80 82 0f bc c5 41 89 c4 41 83 c4 01 0f 84 e6 00 00 RSP: 0018:ffffc90000003f98 EFLAGS: 00000286 RAX: 00000000ffffffff RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82642a51 RDI: ffffffff825bb5e7 RBP: 0000000000000200 R08: 00000008de3271a8 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000030 R14: 0000000000000000 R15: 0000000000000000 ? __do_softirq+0x73/0x34e irq_exit_rcu+0xb5/0x100 common_interrupt+0xa4/0xc0 </IRQ> <TASK> asm_common_interrupt+0x1e/0x40 RIP: 0010:_raw_spin_unlock_irqrestore+0x2e/0x50 Code: 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 85 28 36 ff 48 89 ef e8 cd 58 36 ff 80 e7 02 74 01 fb bf 01 00 00 00 <e8> 3d 97 33 ff 65 8b 05 96 23 2b 7e 85 c0 74 03 5b 5d c3 0f 1f 44 RSP: 0018:ffffc9000020fd08 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000246 RCX: 0000000000000000 RDX: 0000000000000004 RSI: ffffffff8257fd74 RDI: 0000000000000001 RBP: ffff8880057de3a0 R08: 00000008de233000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000100 R14: 0000000000000202 R15: ffff8880057df0b8 ? _raw_spin_unlock_irqrestore+0x23/0x50 gsmtty_write+0x65/0x80 [n_gsm] n_tty_write+0x33f/0x530 ? swake_up_all+0xe0/0xe0 file_tty_write.constprop.0+0x1b1/0x320 ? n_tty_flush_buffer+0xb0/0xb0 new_sync_write+0x10c/0x190 vfs_write+0x282/0x310 ksys_write+0x68/0xe0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f3e5e35c15c Code: 8b 7c 24 08 89 c5 e8 c5 ff ff ff 89 ef 89 44 24 08 e8 58 bc 02 00 8b 44 24 08 48 83 c4 10 5d c3 48 63 ff b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 10 48 8b 15 fd fc 05 00 f7 d8 64 89 02 48 83 RSP: 002b:00007ffcee77cd18 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00007ffcee77cd70 RCX: 00007f3e5e35c15c RDX: 0000000000000100 RSI: 00007ffcee77cd90 RDI: 0000000000000003 RBP: 0000000000000100 R08: 0000000000000000 R09: 7efefefefefefeff R10: 00007f3e5e3bddeb R11: 0000000000000246 R12: 00007ffcee77ce8f R13: 0000000000000001 R14: 000056214404e010 R15: 00007ffcee77cd90 </TASK> Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220701122332.2039-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-07-01 12:23:31 +00:00
gsmld_write_trigger(gsm);
}
/**
* gsmld_read - read function for tty
* @tty: tty device
* @file: file object
* @buf: userspace buffer pointer
* @nr: size of I/O
* @cookie: unused
* @offset: unused
*
* Perform reads for the line discipline. We are guaranteed that the
* line discipline will not be closed under us but we may get multiple
* parallel readers and must handle this ourselves. We may also get
* a hangup. Always called in user context, may sleep.
*
* This code must be sure never to sleep through a hangup.
*/
static ssize_t gsmld_read(struct tty_struct *tty, struct file *file,
unsigned char *buf, size_t nr,
void **cookie, unsigned long offset)
{
return -EOPNOTSUPP;
}
/**
* gsmld_write - write function for tty
* @tty: tty device
* @file: file object
* @buf: userspace buffer pointer
* @nr: size of I/O
*
* Called when the owner of the device wants to send a frame
* itself (or some other control data). The data is transferred
* as-is and must be properly framed and checksummed as appropriate
* by userspace. Frames are either sent whole or not at all as this
* avoids pain user side.
*/
static ssize_t gsmld_write(struct tty_struct *tty, struct file *file,
const unsigned char *buf, size_t nr)
{
struct gsm_mux *gsm = tty->disc_data;
unsigned long flags;
int space;
int ret;
if (!gsm)
return -ENODEV;
ret = -ENOBUFS;
spin_lock_irqsave(&gsm->tx_lock, flags);
space = tty_write_room(tty);
if (space >= nr)
ret = tty->ops->write(tty, buf, nr);
else
set_bit(TTY_DO_WRITE_WAKEUP, &tty->flags);
spin_unlock_irqrestore(&gsm->tx_lock, flags);
return ret;
}
/**
* gsmld_poll - poll method for N_GSM0710
* @tty: terminal device
* @file: file accessing it
* @wait: poll table
*
* Called when the line discipline is asked to poll() for data or
* for special events. This code is not serialized with respect to
* other events save open/close.
*
* This code must be sure never to sleep through a hangup.
* Called without the kernel lock held - fine
*/
static __poll_t gsmld_poll(struct tty_struct *tty, struct file *file,
poll_table *wait)
{
__poll_t mask = 0;
struct gsm_mux *gsm = tty->disc_data;
poll_wait(file, &tty->read_wait, wait);
poll_wait(file, &tty->write_wait, wait);
if (gsm->dead)
mask |= EPOLLHUP;
if (tty_hung_up_p(file))
mask |= EPOLLHUP;
if (test_bit(TTY_OTHER_CLOSED, &tty->flags))
mask |= EPOLLHUP;
if (!tty_is_writelocked(tty) && tty_write_room(tty) > 0)
mask |= EPOLLOUT | EPOLLWRNORM;
return mask;
}
static int gsmld_ioctl(struct tty_struct *tty, unsigned int cmd,
unsigned long arg)
{
struct gsm_config c;
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
struct gsm_config_ext ce;
struct gsm_dlci_config dc;
struct gsm_mux *gsm = tty->disc_data;
unsigned int base, addr;
struct gsm_dlci *dlci;
switch (cmd) {
case GSMIOC_GETCONF:
gsm_copy_config_values(gsm, &c);
if (copy_to_user((void __user *)arg, &c, sizeof(c)))
return -EFAULT;
return 0;
case GSMIOC_SETCONF:
if (copy_from_user(&c, (void __user *)arg, sizeof(c)))
return -EFAULT;
return gsm_config(gsm, &c);
case GSMIOC_GETFIRST:
base = mux_num_to_base(gsm);
return put_user(base + 1, (__u32 __user *)arg);
tty: n_gsm: add keep alive support n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapters 5.4.6.3.4 and 5.1.8.1.3 describe the test command which can be used to test the mux connection between both sides. Currently, no algorithm is implemented to make use of this command. This requires that each multiplexed upper layer protocol supervises the underlying muxer connection to handle possible connection losses. Introduce ioctl commands and functions to optionally enable keep alive handling via the test command as described in chapter 5.4.6.3.4. A single incrementing octet "ka_num" is being used for unique identification of each single keep alive packet. Retries will use the same "ka_num" value as the original packet. Retry count and interval are taken from the general parameters N2 and T2. Add usage description and basic example for the new ioctl to the n_gsm documentation. Note that support for the test command is mandatory and already present in the muxer implementation since the very first version. Also note that the previous ioctl structure gsm_config cannot be extended due to missing checks against zero of the field "unused". Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230214122737.1976-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-02-14 12:27:37 +00:00
case GSMIOC_GETCONF_EXT:
gsm_copy_config_ext_values(gsm, &ce);
if (copy_to_user((void __user *)arg, &ce, sizeof(ce)))
return -EFAULT;
return 0;
case GSMIOC_SETCONF_EXT:
if (copy_from_user(&ce, (void __user *)arg, sizeof(ce)))
return -EFAULT;
return gsm_config_ext(gsm, &ce);
case GSMIOC_GETCONF_DLCI:
if (copy_from_user(&dc, (void __user *)arg, sizeof(dc)))
return -EFAULT;
if (dc.channel == 0 || dc.channel >= NUM_DLCI)
return -EINVAL;
addr = array_index_nospec(dc.channel, NUM_DLCI);
dlci = gsm->dlci[addr];
if (!dlci) {
dlci = gsm_dlci_alloc(gsm, addr);
if (!dlci)
return -ENOMEM;
}
gsm_dlci_copy_config_values(dlci, &dc);
if (copy_to_user((void __user *)arg, &dc, sizeof(dc)))
return -EFAULT;
return 0;
case GSMIOC_SETCONF_DLCI:
if (copy_from_user(&dc, (void __user *)arg, sizeof(dc)))
return -EFAULT;
if (dc.channel == 0 || dc.channel >= NUM_DLCI)
return -EINVAL;
addr = array_index_nospec(dc.channel, NUM_DLCI);
dlci = gsm->dlci[addr];
if (!dlci) {
dlci = gsm_dlci_alloc(gsm, addr);
if (!dlci)
return -ENOMEM;
}
return gsm_dlci_config(dlci, &dc, 0);
default:
return n_tty_ioctl_helper(tty, cmd, arg);
}
}
/*
* Network interface
*
*/
static int gsm_mux_net_open(struct net_device *net)
{
pr_debug("%s called\n", __func__);
netif_start_queue(net);
return 0;
}
static int gsm_mux_net_close(struct net_device *net)
{
netif_stop_queue(net);
return 0;
}
static void dlci_net_free(struct gsm_dlci *dlci)
{
if (!dlci->net) {
WARN_ON(1);
return;
}
dlci->adaption = dlci->prev_adaption;
dlci->data = dlci->prev_data;
free_netdev(dlci->net);
dlci->net = NULL;
}
static void net_free(struct kref *ref)
{
struct gsm_mux_net *mux_net;
struct gsm_dlci *dlci;
mux_net = container_of(ref, struct gsm_mux_net, ref);
dlci = mux_net->dlci;
if (dlci->net) {
unregister_netdev(dlci->net);
dlci_net_free(dlci);
}
}
static inline void muxnet_get(struct gsm_mux_net *mux_net)
{
kref_get(&mux_net->ref);
}
static inline void muxnet_put(struct gsm_mux_net *mux_net)
{
kref_put(&mux_net->ref, net_free);
}
static netdev_tx_t gsm_mux_net_start_xmit(struct sk_buff *skb,
struct net_device *net)
{
struct gsm_mux_net *mux_net = netdev_priv(net);
struct gsm_dlci *dlci = mux_net->dlci;
muxnet_get(mux_net);
skb_queue_head(&dlci->skb_list, skb);
net->stats.tx_packets++;
net->stats.tx_bytes += skb->len;
gsm_dlci_data_kick(dlci);
/* And tell the kernel when the last transmit started. */
netif_trans_update(net);
muxnet_put(mux_net);
return NETDEV_TX_OK;
}
/* called when a packet did not ack after watchdogtimeout */
netdev: pass the stuck queue to the timeout handler This allows incrementing the correct timeout statistic without any mess. Down the road, devices can learn to reset just the specific queue. The patch was generated with the following script: use strict; use warnings; our $^I = '.bak'; my @work = ( ["arch/m68k/emu/nfeth.c", "nfeth_tx_timeout"], ["arch/um/drivers/net_kern.c", "uml_net_tx_timeout"], ["arch/um/drivers/vector_kern.c", "vector_net_tx_timeout"], ["arch/xtensa/platforms/iss/network.c", "iss_net_tx_timeout"], ["drivers/char/pcmcia/synclink_cs.c", "hdlcdev_tx_timeout"], ["drivers/infiniband/ulp/ipoib/ipoib_main.c", "ipoib_timeout"], ["drivers/infiniband/ulp/ipoib/ipoib_main.c", "ipoib_timeout"], ["drivers/message/fusion/mptlan.c", "mpt_lan_tx_timeout"], ["drivers/misc/sgi-xp/xpnet.c", "xpnet_dev_tx_timeout"], ["drivers/net/appletalk/cops.c", "cops_timeout"], ["drivers/net/arcnet/arcdevice.h", "arcnet_timeout"], ["drivers/net/arcnet/arcnet.c", "arcnet_timeout"], ["drivers/net/arcnet/com20020.c", "arcnet_timeout"], ["drivers/net/ethernet/3com/3c509.c", "el3_tx_timeout"], ["drivers/net/ethernet/3com/3c515.c", "corkscrew_timeout"], ["drivers/net/ethernet/3com/3c574_cs.c", "el3_tx_timeout"], ["drivers/net/ethernet/3com/3c589_cs.c", "el3_tx_timeout"], ["drivers/net/ethernet/3com/3c59x.c", "vortex_tx_timeout"], ["drivers/net/ethernet/3com/3c59x.c", "vortex_tx_timeout"], ["drivers/net/ethernet/3com/typhoon.c", "typhoon_tx_timeout"], ["drivers/net/ethernet/8390/8390.h", "ei_tx_timeout"], ["drivers/net/ethernet/8390/8390.h", "eip_tx_timeout"], ["drivers/net/ethernet/8390/8390.c", "ei_tx_timeout"], ["drivers/net/ethernet/8390/8390p.c", "eip_tx_timeout"], ["drivers/net/ethernet/8390/ax88796.c", "ax_ei_tx_timeout"], ["drivers/net/ethernet/8390/axnet_cs.c", "axnet_tx_timeout"], ["drivers/net/ethernet/8390/etherh.c", "__ei_tx_timeout"], ["drivers/net/ethernet/8390/hydra.c", "__ei_tx_timeout"], ["drivers/net/ethernet/8390/mac8390.c", "__ei_tx_timeout"], ["drivers/net/ethernet/8390/mcf8390.c", "__ei_tx_timeout"], ["drivers/net/ethernet/8390/lib8390.c", "__ei_tx_timeout"], ["drivers/net/ethernet/8390/ne2k-pci.c", "ei_tx_timeout"], ["drivers/net/ethernet/8390/pcnet_cs.c", "ei_tx_timeout"], ["drivers/net/ethernet/8390/smc-ultra.c", "ei_tx_timeout"], ["drivers/net/ethernet/8390/wd.c", "ei_tx_timeout"], ["drivers/net/ethernet/8390/zorro8390.c", "__ei_tx_timeout"], ["drivers/net/ethernet/adaptec/starfire.c", "tx_timeout"], ["drivers/net/ethernet/agere/et131x.c", "et131x_tx_timeout"], ["drivers/net/ethernet/allwinner/sun4i-emac.c", "emac_timeout"], ["drivers/net/ethernet/alteon/acenic.c", "ace_watchdog"], ["drivers/net/ethernet/amazon/ena/ena_netdev.c", "ena_tx_timeout"], ["drivers/net/ethernet/amd/7990.h", "lance_tx_timeout"], ["drivers/net/ethernet/amd/7990.c", "lance_tx_timeout"], ["drivers/net/ethernet/amd/a2065.c", "lance_tx_timeout"], ["drivers/net/ethernet/amd/am79c961a.c", "am79c961_timeout"], ["drivers/net/ethernet/amd/amd8111e.c", "amd8111e_tx_timeout"], ["drivers/net/ethernet/amd/ariadne.c", "ariadne_tx_timeout"], ["drivers/net/ethernet/amd/atarilance.c", "lance_tx_timeout"], ["drivers/net/ethernet/amd/au1000_eth.c", "au1000_tx_timeout"], ["drivers/net/ethernet/amd/declance.c", "lance_tx_timeout"], ["drivers/net/ethernet/amd/lance.c", "lance_tx_timeout"], ["drivers/net/ethernet/amd/mvme147.c", "lance_tx_timeout"], ["drivers/net/ethernet/amd/ni65.c", "ni65_timeout"], ["drivers/net/ethernet/amd/nmclan_cs.c", "mace_tx_timeout"], ["drivers/net/ethernet/amd/pcnet32.c", "pcnet32_tx_timeout"], ["drivers/net/ethernet/amd/sunlance.c", "lance_tx_timeout"], ["drivers/net/ethernet/amd/xgbe/xgbe-drv.c", "xgbe_tx_timeout"], ["drivers/net/ethernet/apm/xgene-v2/main.c", "xge_timeout"], ["drivers/net/ethernet/apm/xgene/xgene_enet_main.c", "xgene_enet_timeout"], ["drivers/net/ethernet/apple/macmace.c", "mace_tx_timeout"], ["drivers/net/ethernet/atheros/ag71xx.c", "ag71xx_tx_timeout"], ["drivers/net/ethernet/atheros/alx/main.c", "alx_tx_timeout"], ["drivers/net/ethernet/atheros/atl1c/atl1c_main.c", "atl1c_tx_timeout"], ["drivers/net/ethernet/atheros/atl1e/atl1e_main.c", "atl1e_tx_timeout"], ["drivers/net/ethernet/atheros/atlx/atl.c", "atlx_tx_timeout"], ["drivers/net/ethernet/atheros/atlx/atl1.c", "atlx_tx_timeout"], ["drivers/net/ethernet/atheros/atlx/atl2.c", "atl2_tx_timeout"], ["drivers/net/ethernet/broadcom/b44.c", "b44_tx_timeout"], ["drivers/net/ethernet/broadcom/bcmsysport.c", "bcm_sysport_tx_timeout"], ["drivers/net/ethernet/broadcom/bnx2.c", "bnx2_tx_timeout"], ["drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h", "bnx2x_tx_timeout"], ["drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c", "bnx2x_tx_timeout"], ["drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c", "bnx2x_tx_timeout"], ["drivers/net/ethernet/broadcom/bnxt/bnxt.c", "bnxt_tx_timeout"], ["drivers/net/ethernet/broadcom/genet/bcmgenet.c", "bcmgenet_timeout"], ["drivers/net/ethernet/broadcom/sb1250-mac.c", "sbmac_tx_timeout"], ["drivers/net/ethernet/broadcom/tg3.c", "tg3_tx_timeout"], ["drivers/net/ethernet/calxeda/xgmac.c", "xgmac_tx_timeout"], ["drivers/net/ethernet/cavium/liquidio/lio_main.c", "liquidio_tx_timeout"], ["drivers/net/ethernet/cavium/liquidio/lio_vf_main.c", "liquidio_tx_timeout"], ["drivers/net/ethernet/cavium/liquidio/lio_vf_rep.c", "lio_vf_rep_tx_timeout"], ["drivers/net/ethernet/cavium/thunder/nicvf_main.c", "nicvf_tx_timeout"], ["drivers/net/ethernet/cirrus/cs89x0.c", "net_timeout"], ["drivers/net/ethernet/cisco/enic/enic_main.c", "enic_tx_timeout"], ["drivers/net/ethernet/cisco/enic/enic_main.c", "enic_tx_timeout"], ["drivers/net/ethernet/cortina/gemini.c", "gmac_tx_timeout"], ["drivers/net/ethernet/davicom/dm9000.c", "dm9000_timeout"], ["drivers/net/ethernet/dec/tulip/de2104x.c", "de_tx_timeout"], ["drivers/net/ethernet/dec/tulip/tulip_core.c", "tulip_tx_timeout"], ["drivers/net/ethernet/dec/tulip/winbond-840.c", "tx_timeout"], ["drivers/net/ethernet/dlink/dl2k.c", "rio_tx_timeout"], ["drivers/net/ethernet/dlink/sundance.c", "tx_timeout"], ["drivers/net/ethernet/emulex/benet/be_main.c", "be_tx_timeout"], ["drivers/net/ethernet/ethoc.c", "ethoc_tx_timeout"], ["drivers/net/ethernet/faraday/ftgmac100.c", "ftgmac100_tx_timeout"], ["drivers/net/ethernet/fealnx.c", "fealnx_tx_timeout"], ["drivers/net/ethernet/freescale/dpaa/dpaa_eth.c", "dpaa_tx_timeout"], ["drivers/net/ethernet/freescale/fec_main.c", "fec_timeout"], ["drivers/net/ethernet/freescale/fec_mpc52xx.c", "mpc52xx_fec_tx_timeout"], ["drivers/net/ethernet/freescale/fs_enet/fs_enet-main.c", "fs_timeout"], ["drivers/net/ethernet/freescale/gianfar.c", "gfar_timeout"], ["drivers/net/ethernet/freescale/ucc_geth.c", "ucc_geth_timeout"], ["drivers/net/ethernet/fujitsu/fmvj18x_cs.c", "fjn_tx_timeout"], ["drivers/net/ethernet/google/gve/gve_main.c", "gve_tx_timeout"], ["drivers/net/ethernet/hisilicon/hip04_eth.c", "hip04_timeout"], ["drivers/net/ethernet/hisilicon/hix5hd2_gmac.c", "hix5hd2_net_timeout"], ["drivers/net/ethernet/hisilicon/hns/hns_enet.c", "hns_nic_net_timeout"], ["drivers/net/ethernet/hisilicon/hns3/hns3_enet.c", "hns3_nic_net_timeout"], ["drivers/net/ethernet/huawei/hinic/hinic_main.c", "hinic_tx_timeout"], ["drivers/net/ethernet/i825xx/82596.c", "i596_tx_timeout"], ["drivers/net/ethernet/i825xx/ether1.c", "ether1_timeout"], ["drivers/net/ethernet/i825xx/lib82596.c", "i596_tx_timeout"], ["drivers/net/ethernet/i825xx/sun3_82586.c", "sun3_82586_timeout"], ["drivers/net/ethernet/ibm/ehea/ehea_main.c", "ehea_tx_watchdog"], ["drivers/net/ethernet/ibm/emac/core.c", "emac_tx_timeout"], ["drivers/net/ethernet/ibm/emac/core.c", "emac_tx_timeout"], ["drivers/net/ethernet/ibm/ibmvnic.c", "ibmvnic_tx_timeout"], ["drivers/net/ethernet/intel/e100.c", "e100_tx_timeout"], ["drivers/net/ethernet/intel/e1000/e1000_main.c", "e1000_tx_timeout"], ["drivers/net/ethernet/intel/e1000e/netdev.c", "e1000_tx_timeout"], ["drivers/net/ethernet/intel/fm10k/fm10k_netdev.c", "fm10k_tx_timeout"], ["drivers/net/ethernet/intel/i40e/i40e_main.c", "i40e_tx_timeout"], ["drivers/net/ethernet/intel/iavf/iavf_main.c", "iavf_tx_timeout"], ["drivers/net/ethernet/intel/ice/ice_main.c", "ice_tx_timeout"], ["drivers/net/ethernet/intel/ice/ice_main.c", "ice_tx_timeout"], ["drivers/net/ethernet/intel/igb/igb_main.c", "igb_tx_timeout"], ["drivers/net/ethernet/intel/igbvf/netdev.c", "igbvf_tx_timeout"], ["drivers/net/ethernet/intel/ixgb/ixgb_main.c", "ixgb_tx_timeout"], ["drivers/net/ethernet/intel/ixgbe/ixgbe_debugfs.c", "adapter->netdev->netdev_ops->ndo_tx_timeout(adapter->netdev);"], ["drivers/net/ethernet/intel/ixgbe/ixgbe_main.c", "ixgbe_tx_timeout"], ["drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c", "ixgbevf_tx_timeout"], ["drivers/net/ethernet/jme.c", "jme_tx_timeout"], ["drivers/net/ethernet/korina.c", "korina_tx_timeout"], ["drivers/net/ethernet/lantiq_etop.c", "ltq_etop_tx_timeout"], ["drivers/net/ethernet/marvell/mv643xx_eth.c", "mv643xx_eth_tx_timeout"], ["drivers/net/ethernet/marvell/pxa168_eth.c", "pxa168_eth_tx_timeout"], ["drivers/net/ethernet/marvell/skge.c", "skge_tx_timeout"], ["drivers/net/ethernet/marvell/sky2.c", "sky2_tx_timeout"], ["drivers/net/ethernet/marvell/sky2.c", "sky2_tx_timeout"], ["drivers/net/ethernet/mediatek/mtk_eth_soc.c", "mtk_tx_timeout"], ["drivers/net/ethernet/mellanox/mlx4/en_netdev.c", "mlx4_en_tx_timeout"], ["drivers/net/ethernet/mellanox/mlx4/en_netdev.c", "mlx4_en_tx_timeout"], ["drivers/net/ethernet/mellanox/mlx5/core/en_main.c", "mlx5e_tx_timeout"], ["drivers/net/ethernet/micrel/ks8842.c", "ks8842_tx_timeout"], ["drivers/net/ethernet/micrel/ksz884x.c", "netdev_tx_timeout"], ["drivers/net/ethernet/microchip/enc28j60.c", "enc28j60_tx_timeout"], ["drivers/net/ethernet/microchip/encx24j600.c", "encx24j600_tx_timeout"], ["drivers/net/ethernet/natsemi/sonic.h", "sonic_tx_timeout"], ["drivers/net/ethernet/natsemi/sonic.c", "sonic_tx_timeout"], ["drivers/net/ethernet/natsemi/jazzsonic.c", "sonic_tx_timeout"], ["drivers/net/ethernet/natsemi/macsonic.c", "sonic_tx_timeout"], ["drivers/net/ethernet/natsemi/natsemi.c", "ns_tx_timeout"], ["drivers/net/ethernet/natsemi/ns83820.c", "ns83820_tx_timeout"], ["drivers/net/ethernet/natsemi/xtsonic.c", "sonic_tx_timeout"], ["drivers/net/ethernet/neterion/s2io.h", "s2io_tx_watchdog"], ["drivers/net/ethernet/neterion/s2io.c", "s2io_tx_watchdog"], ["drivers/net/ethernet/neterion/vxge/vxge-main.c", "vxge_tx_watchdog"], ["drivers/net/ethernet/netronome/nfp/nfp_net_common.c", "nfp_net_tx_timeout"], ["drivers/net/ethernet/nvidia/forcedeth.c", "nv_tx_timeout"], ["drivers/net/ethernet/nvidia/forcedeth.c", "nv_tx_timeout"], ["drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c", "pch_gbe_tx_timeout"], ["drivers/net/ethernet/packetengines/hamachi.c", "hamachi_tx_timeout"], ["drivers/net/ethernet/packetengines/yellowfin.c", "yellowfin_tx_timeout"], ["drivers/net/ethernet/pensando/ionic/ionic_lif.c", "ionic_tx_timeout"], ["drivers/net/ethernet/qlogic/netxen/netxen_nic_main.c", "netxen_tx_timeout"], ["drivers/net/ethernet/qlogic/qla3xxx.c", "ql3xxx_tx_timeout"], ["drivers/net/ethernet/qlogic/qlcnic/qlcnic_main.c", "qlcnic_tx_timeout"], ["drivers/net/ethernet/qualcomm/emac/emac.c", "emac_tx_timeout"], ["drivers/net/ethernet/qualcomm/qca_spi.c", "qcaspi_netdev_tx_timeout"], ["drivers/net/ethernet/qualcomm/qca_uart.c", "qcauart_netdev_tx_timeout"], ["drivers/net/ethernet/rdc/r6040.c", "r6040_tx_timeout"], ["drivers/net/ethernet/realtek/8139cp.c", "cp_tx_timeout"], ["drivers/net/ethernet/realtek/8139too.c", "rtl8139_tx_timeout"], ["drivers/net/ethernet/realtek/atp.c", "tx_timeout"], ["drivers/net/ethernet/realtek/r8169_main.c", "rtl8169_tx_timeout"], ["drivers/net/ethernet/renesas/ravb_main.c", "ravb_tx_timeout"], ["drivers/net/ethernet/renesas/sh_eth.c", "sh_eth_tx_timeout"], ["drivers/net/ethernet/renesas/sh_eth.c", "sh_eth_tx_timeout"], ["drivers/net/ethernet/samsung/sxgbe/sxgbe_main.c", "sxgbe_tx_timeout"], ["drivers/net/ethernet/seeq/ether3.c", "ether3_timeout"], ["drivers/net/ethernet/seeq/sgiseeq.c", "timeout"], ["drivers/net/ethernet/sfc/efx.c", "efx_watchdog"], ["drivers/net/ethernet/sfc/falcon/efx.c", "ef4_watchdog"], ["drivers/net/ethernet/sgi/ioc3-eth.c", "ioc3_timeout"], ["drivers/net/ethernet/sgi/meth.c", "meth_tx_timeout"], ["drivers/net/ethernet/silan/sc92031.c", "sc92031_tx_timeout"], ["drivers/net/ethernet/sis/sis190.c", "sis190_tx_timeout"], ["drivers/net/ethernet/sis/sis900.c", "sis900_tx_timeout"], ["drivers/net/ethernet/smsc/epic100.c", "epic_tx_timeout"], ["drivers/net/ethernet/smsc/smc911x.c", "smc911x_timeout"], ["drivers/net/ethernet/smsc/smc9194.c", "smc_timeout"], ["drivers/net/ethernet/smsc/smc91c92_cs.c", "smc_tx_timeout"], ["drivers/net/ethernet/smsc/smc91x.c", "smc_timeout"], ["drivers/net/ethernet/stmicro/stmmac/stmmac_main.c", "stmmac_tx_timeout"], ["drivers/net/ethernet/sun/cassini.c", "cas_tx_timeout"], ["drivers/net/ethernet/sun/ldmvsw.c", "sunvnet_tx_timeout_common"], ["drivers/net/ethernet/sun/niu.c", "niu_tx_timeout"], ["drivers/net/ethernet/sun/sunbmac.c", "bigmac_tx_timeout"], ["drivers/net/ethernet/sun/sungem.c", "gem_tx_timeout"], ["drivers/net/ethernet/sun/sunhme.c", "happy_meal_tx_timeout"], ["drivers/net/ethernet/sun/sunqe.c", "qe_tx_timeout"], ["drivers/net/ethernet/sun/sunvnet.c", "sunvnet_tx_timeout_common"], ["drivers/net/ethernet/sun/sunvnet_common.c", "sunvnet_tx_timeout_common"], ["drivers/net/ethernet/sun/sunvnet_common.h", "sunvnet_tx_timeout_common"], ["drivers/net/ethernet/synopsys/dwc-xlgmac-net.c", "xlgmac_tx_timeout"], ["drivers/net/ethernet/ti/cpmac.c", "cpmac_tx_timeout"], ["drivers/net/ethernet/ti/cpsw.c", "cpsw_ndo_tx_timeout"], ["drivers/net/ethernet/ti/cpsw_priv.c", "cpsw_ndo_tx_timeout"], ["drivers/net/ethernet/ti/cpsw_priv.h", "cpsw_ndo_tx_timeout"], ["drivers/net/ethernet/ti/davinci_emac.c", "emac_dev_tx_timeout"], ["drivers/net/ethernet/ti/netcp_core.c", "netcp_ndo_tx_timeout"], ["drivers/net/ethernet/ti/tlan.c", "tlan_tx_timeout"], ["drivers/net/ethernet/toshiba/ps3_gelic_net.h", "gelic_net_tx_timeout"], ["drivers/net/ethernet/toshiba/ps3_gelic_net.c", "gelic_net_tx_timeout"], ["drivers/net/ethernet/toshiba/ps3_gelic_wireless.c", "gelic_net_tx_timeout"], ["drivers/net/ethernet/toshiba/spider_net.c", "spider_net_tx_timeout"], ["drivers/net/ethernet/toshiba/tc35815.c", "tc35815_tx_timeout"], ["drivers/net/ethernet/via/via-rhine.c", "rhine_tx_timeout"], ["drivers/net/ethernet/wiznet/w5100.c", "w5100_tx_timeout"], ["drivers/net/ethernet/wiznet/w5300.c", "w5300_tx_timeout"], ["drivers/net/ethernet/xilinx/xilinx_emaclite.c", "xemaclite_tx_timeout"], ["drivers/net/ethernet/xircom/xirc2ps_cs.c", "xirc_tx_timeout"], ["drivers/net/fjes/fjes_main.c", "fjes_tx_retry"], ["drivers/net/slip/slip.c", "sl_tx_timeout"], ["include/linux/usb/usbnet.h", "usbnet_tx_timeout"], ["drivers/net/usb/aqc111.c", "usbnet_tx_timeout"], ["drivers/net/usb/asix_devices.c", "usbnet_tx_timeout"], ["drivers/net/usb/asix_devices.c", "usbnet_tx_timeout"], ["drivers/net/usb/asix_devices.c", "usbnet_tx_timeout"], ["drivers/net/usb/ax88172a.c", "usbnet_tx_timeout"], ["drivers/net/usb/ax88179_178a.c", "usbnet_tx_timeout"], ["drivers/net/usb/catc.c", "catc_tx_timeout"], ["drivers/net/usb/cdc_mbim.c", "usbnet_tx_timeout"], ["drivers/net/usb/cdc_ncm.c", "usbnet_tx_timeout"], ["drivers/net/usb/dm9601.c", "usbnet_tx_timeout"], ["drivers/net/usb/hso.c", "hso_net_tx_timeout"], ["drivers/net/usb/int51x1.c", "usbnet_tx_timeout"], ["drivers/net/usb/ipheth.c", "ipheth_tx_timeout"], ["drivers/net/usb/kaweth.c", "kaweth_tx_timeout"], ["drivers/net/usb/lan78xx.c", "lan78xx_tx_timeout"], ["drivers/net/usb/mcs7830.c", "usbnet_tx_timeout"], ["drivers/net/usb/pegasus.c", "pegasus_tx_timeout"], ["drivers/net/usb/qmi_wwan.c", "usbnet_tx_timeout"], ["drivers/net/usb/r8152.c", "rtl8152_tx_timeout"], ["drivers/net/usb/rndis_host.c", "usbnet_tx_timeout"], ["drivers/net/usb/rtl8150.c", "rtl8150_tx_timeout"], ["drivers/net/usb/sierra_net.c", "usbnet_tx_timeout"], ["drivers/net/usb/smsc75xx.c", "usbnet_tx_timeout"], ["drivers/net/usb/smsc95xx.c", "usbnet_tx_timeout"], ["drivers/net/usb/sr9700.c", "usbnet_tx_timeout"], ["drivers/net/usb/sr9800.c", "usbnet_tx_timeout"], ["drivers/net/usb/usbnet.c", "usbnet_tx_timeout"], ["drivers/net/vmxnet3/vmxnet3_drv.c", "vmxnet3_tx_timeout"], ["drivers/net/wan/cosa.c", "cosa_net_timeout"], ["drivers/net/wan/farsync.c", "fst_tx_timeout"], ["drivers/net/wan/fsl_ucc_hdlc.c", "uhdlc_tx_timeout"], ["drivers/net/wan/lmc/lmc_main.c", "lmc_driver_timeout"], ["drivers/net/wan/x25_asy.c", "x25_asy_timeout"], ["drivers/net/wimax/i2400m/netdev.c", "i2400m_tx_timeout"], ["drivers/net/wireless/intel/ipw2x00/ipw2100.c", "ipw2100_tx_timeout"], ["drivers/net/wireless/intersil/hostap/hostap_main.c", "prism2_tx_timeout"], ["drivers/net/wireless/intersil/hostap/hostap_main.c", "prism2_tx_timeout"], ["drivers/net/wireless/intersil/hostap/hostap_main.c", "prism2_tx_timeout"], ["drivers/net/wireless/intersil/orinoco/main.c", "orinoco_tx_timeout"], ["drivers/net/wireless/intersil/orinoco/orinoco_usb.c", "orinoco_tx_timeout"], ["drivers/net/wireless/intersil/orinoco/orinoco.h", "orinoco_tx_timeout"], ["drivers/net/wireless/intersil/prism54/islpci_dev.c", "islpci_eth_tx_timeout"], ["drivers/net/wireless/intersil/prism54/islpci_eth.c", "islpci_eth_tx_timeout"], ["drivers/net/wireless/intersil/prism54/islpci_eth.h", "islpci_eth_tx_timeout"], ["drivers/net/wireless/marvell/mwifiex/main.c", "mwifiex_tx_timeout"], ["drivers/net/wireless/quantenna/qtnfmac/core.c", "qtnf_netdev_tx_timeout"], ["drivers/net/wireless/quantenna/qtnfmac/core.h", "qtnf_netdev_tx_timeout"], ["drivers/net/wireless/rndis_wlan.c", "usbnet_tx_timeout"], ["drivers/net/wireless/wl3501_cs.c", "wl3501_tx_timeout"], ["drivers/net/wireless/zydas/zd1201.c", "zd1201_tx_timeout"], ["drivers/s390/net/qeth_core.h", "qeth_tx_timeout"], ["drivers/s390/net/qeth_core_main.c", "qeth_tx_timeout"], ["drivers/s390/net/qeth_l2_main.c", "qeth_tx_timeout"], ["drivers/s390/net/qeth_l2_main.c", "qeth_tx_timeout"], ["drivers/s390/net/qeth_l3_main.c", "qeth_tx_timeout"], ["drivers/s390/net/qeth_l3_main.c", "qeth_tx_timeout"], ["drivers/staging/ks7010/ks_wlan_net.c", "ks_wlan_tx_timeout"], ["drivers/staging/qlge/qlge_main.c", "qlge_tx_timeout"], ["drivers/staging/rtl8192e/rtl8192e/rtl_core.c", "_rtl92e_tx_timeout"], ["drivers/staging/rtl8192u/r8192U_core.c", "tx_timeout"], ["drivers/staging/unisys/visornic/visornic_main.c", "visornic_xmit_timeout"], ["drivers/staging/wlan-ng/p80211netdev.c", "p80211knetdev_tx_timeout"], ["drivers/tty/n_gsm.c", "gsm_mux_net_tx_timeout"], ["drivers/tty/synclink.c", "hdlcdev_tx_timeout"], ["drivers/tty/synclink_gt.c", "hdlcdev_tx_timeout"], ["drivers/tty/synclinkmp.c", "hdlcdev_tx_timeout"], ["net/atm/lec.c", "lec_tx_timeout"], ["net/bluetooth/bnep/netdev.c", "bnep_net_timeout"] ); for my $p (@work) { my @pair = @$p; my $file = $pair[0]; my $func = $pair[1]; print STDERR $file , ": ", $func,"\n"; our @ARGV = ($file); while (<ARGV>) { if (m/($func\s*\(struct\s+net_device\s+\*[A-Za-z_]?[A-Za-z-0-9_]*)(\))/) { print STDERR "found $1+$2 in $file\n"; } if (s/($func\s*\(struct\s+net_device\s+\*[A-Za-z_]?[A-Za-z-0-9_]*)(\))/$1, unsigned int txqueue$2/) { print STDERR "$func found in $file\n"; } print; } } where the list of files and functions is simply from: git grep ndo_tx_timeout, with manual addition of headers in the rare cases where the function is from a header, then manually changing the few places which actually call ndo_tx_timeout. Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Acked-by: Heiner Kallweit <hkallweit1@gmail.com> Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com> Acked-by: Shannon Nelson <snelson@pensando.io> Reviewed-by: Martin Habets <mhabets@solarflare.com> changes from v9: fixup a forward declaration changes from v9: more leftovers from v3 change changes from v8: fix up a missing direct call to timeout rebased on net-next changes from v7: fixup leftovers from v3 change changes from v6: fix typo in rtl driver changes from v5: add missing files (allow any net device argument name) changes from v4: add a missing driver header changes from v3: change queue # to unsigned Changes from v2: added headers Changes from v1: Fix errors found by kbuild: generalize the pattern a bit, to pick up a couple of instances missed by the previous version. Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-10 14:23:51 +00:00
static void gsm_mux_net_tx_timeout(struct net_device *net, unsigned int txqueue)
{
/* Tell syslog we are hosed. */
dev_dbg(&net->dev, "Tx timed out.\n");
/* Update statistics */
net->stats.tx_errors++;
}
static void gsm_mux_rx_netchar(struct gsm_dlci *dlci,
const unsigned char *in_buf, int size)
{
struct net_device *net = dlci->net;
struct sk_buff *skb;
struct gsm_mux_net *mux_net = netdev_priv(net);
muxnet_get(mux_net);
/* Allocate an sk_buff */
skb = dev_alloc_skb(size + NET_IP_ALIGN);
if (!skb) {
/* We got no receive buffer. */
net->stats.rx_dropped++;
muxnet_put(mux_net);
return;
}
skb_reserve(skb, NET_IP_ALIGN);
skb_put_data(skb, in_buf, size);
skb->dev = net;
skb->protocol = htons(ETH_P_IP);
/* Ship it off to the kernel */
netif_rx(skb);
/* update out statistics */
net->stats.rx_packets++;
net->stats.rx_bytes += size;
muxnet_put(mux_net);
return;
}
static void gsm_mux_net_init(struct net_device *net)
{
static const struct net_device_ops gsm_netdev_ops = {
.ndo_open = gsm_mux_net_open,
.ndo_stop = gsm_mux_net_close,
.ndo_start_xmit = gsm_mux_net_start_xmit,
.ndo_tx_timeout = gsm_mux_net_tx_timeout,
};
net->netdev_ops = &gsm_netdev_ops;
/* fill in the other fields */
net->watchdog_timeo = GSM_NET_TX_TIMEOUT;
net->flags = IFF_POINTOPOINT | IFF_NOARP | IFF_MULTICAST;
net->type = ARPHRD_NONE;
net->tx_queue_len = 10;
}
/* caller holds the dlci mutex */
static void gsm_destroy_network(struct gsm_dlci *dlci)
{
struct gsm_mux_net *mux_net;
pr_debug("destroy network interface\n");
if (!dlci->net)
return;
mux_net = netdev_priv(dlci->net);
muxnet_put(mux_net);
}
/* caller holds the dlci mutex */
static int gsm_create_network(struct gsm_dlci *dlci, struct gsm_netconfig *nc)
{
char *netname;
int retval = 0;
struct net_device *net;
struct gsm_mux_net *mux_net;
if (!capable(CAP_NET_ADMIN))
return -EPERM;
/* Already in a non tty mode */
if (dlci->adaption > 2)
return -EBUSY;
if (nc->protocol != htons(ETH_P_IP))
return -EPROTONOSUPPORT;
if (nc->adaption != 3 && nc->adaption != 4)
return -EPROTONOSUPPORT;
pr_debug("create network interface\n");
netname = "gsm%d";
if (nc->if_name[0] != '\0')
netname = nc->if_name;
net = alloc_netdev(sizeof(struct gsm_mux_net), netname,
NET_NAME_UNKNOWN, gsm_mux_net_init);
if (!net) {
pr_err("alloc_netdev failed\n");
return -ENOMEM;
}
net->mtu = dlci->mtu;
net->min_mtu = MIN_MTU;
net->max_mtu = dlci->mtu;
mux_net = netdev_priv(net);
mux_net->dlci = dlci;
kref_init(&mux_net->ref);
strncpy(nc->if_name, net->name, IFNAMSIZ); /* return net name */
/* reconfigure dlci for network */
dlci->prev_adaption = dlci->adaption;
dlci->prev_data = dlci->data;
dlci->adaption = nc->adaption;
dlci->data = gsm_mux_rx_netchar;
dlci->net = net;
pr_debug("register netdev\n");
retval = register_netdev(net);
if (retval) {
pr_err("network register fail %d\n", retval);
dlci_net_free(dlci);
return retval;
}
return net->ifindex; /* return network index */
}
/* Line discipline for real tty */
static struct tty_ldisc_ops tty_ldisc_packet = {
.owner = THIS_MODULE,
.num = N_GSM0710,
.name = "n_gsm",
.open = gsmld_open,
.close = gsmld_close,
.flush_buffer = gsmld_flush_buffer,
.read = gsmld_read,
.write = gsmld_write,
.ioctl = gsmld_ioctl,
.poll = gsmld_poll,
.receive_buf = gsmld_receive_buf,
.write_wakeup = gsmld_write_wakeup
};
/*
* Virtual tty side
*/
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
/**
* gsm_modem_upd_via_data - send modem bits via convergence layer
* @dlci: channel
* @brk: break signal
*
* Send an empty frame to signal mobile state changes and to transmit the
* break signal for adaption 2.
*/
static void gsm_modem_upd_via_data(struct gsm_dlci *dlci, u8 brk)
{
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
struct gsm_mux *gsm = dlci->gsm;
unsigned long flags;
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
if (dlci->state != DLCI_OPEN || dlci->adaption != 2)
return;
spin_lock_irqsave(&gsm->tx_lock, flags);
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
gsm_dlci_modem_output(gsm, dlci, brk);
spin_unlock_irqrestore(&gsm->tx_lock, flags);
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
}
/**
* gsm_modem_upd_via_msc - send modem bits via control frame
* @dlci: channel
* @brk: break signal
*/
static int gsm_modem_upd_via_msc(struct gsm_dlci *dlci, u8 brk)
{
u8 modembits[3];
struct gsm_control *ctrl;
int len = 2;
if (dlci->gsm->encoding != GSM_BASIC_OPT)
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
return 0;
modembits[0] = (dlci->addr << 2) | 2 | EA; /* DLCI, Valid, EA */
tty: n_gsm: fix wrong signal octets encoding in MSC n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. The value of the modem status command (MSC) frame contains an address field, control signal and optional break signal octet. The address field is encoded as described in chapter 5.2.1.2 with only one octet (may be extended to more in future versions of the standard). Whereas the control signal and break signal octet are always one byte each. This is strange at first glance as it makes the EA bit redundant. However, the same two octets are also encoded as header in convergence layer type 2 as described in chapter 5.5.2. No header length field is given and the only way to test if there is an optional break signal octet is via the EA flag which extends the control signal octet with a break signal octet. Now it becomes obvious how the EA bit for those two octets shall be encoded in the MSC frame. The current implementation treats the signal octet different for MSC frame and convergence layer type 2 header even though the standard describes it for both in the same way. Use the EA bit to encode the signal octets not only in the convergence layer type 2 header but also in the MSC frame in the same way with either 1 or 2 bytes in case of an optional break signal. Adjust the receiving path accordingly in gsm_control_modem(). Fixes: 3ac06b905655 ("tty: n_gsm: Fix for modems with brk in modem status control") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220414094225.4527-13-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-14 09:42:18 +00:00
if (!brk) {
modembits[1] = (gsm_encode_modem(dlci) << 1) | EA;
} else {
modembits[1] = gsm_encode_modem(dlci) << 1;
modembits[2] = (brk << 4) | 2 | EA; /* Length, Break, EA */
len++;
}
ctrl = gsm_control_send(dlci->gsm, CMD_MSC, modembits, len);
if (ctrl == NULL)
return -ENOMEM;
return gsm_control_wait(dlci->gsm, ctrl);
}
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
/**
* gsm_modem_update - send modem status line state
* @dlci: channel
* @brk: break signal
*/
static int gsm_modem_update(struct gsm_dlci *dlci, u8 brk)
{
if (dlci->adaption == 2) {
/* Send convergence layer type 2 empty data frame. */
gsm_modem_upd_via_data(dlci, brk);
return 0;
} else if (dlci->gsm->encoding == GSM_BASIC_OPT) {
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
/* Send as MSC control message. */
return gsm_modem_upd_via_msc(dlci, brk);
}
/* Modem status lines are not supported. */
return -EPROTONOSUPPORT;
}
/**
* gsm_wait_modem_change - wait for modem status line change
* @dlci: channel
* @mask: modem status line bits
*
* The function returns if:
* - any given modem status line bit changed
* - the wait event function got interrupted (e.g. by a signal)
* - the underlying DLCI was closed
* - the underlying ldisc device was removed
*/
static int gsm_wait_modem_change(struct gsm_dlci *dlci, u32 mask)
{
struct gsm_mux *gsm = dlci->gsm;
u32 old = dlci->modem_rx;
int ret;
ret = wait_event_interruptible(gsm->event, gsm->dead ||
dlci->state != DLCI_OPEN ||
(old ^ dlci->modem_rx) & mask);
if (gsm->dead)
return -ENODEV;
if (dlci->state != DLCI_OPEN)
return -EL2NSYNC;
return ret;
}
static bool gsm_carrier_raised(struct tty_port *port)
{
struct gsm_dlci *dlci = container_of(port, struct gsm_dlci, port);
struct gsm_mux *gsm = dlci->gsm;
/* Not yet open so no carrier info */
if (dlci->state != DLCI_OPEN)
return false;
if (debug & DBG_CD_ON)
return true;
/*
* Basic mode with control channel in ADM mode may not respond
* to CMD_MSC at all and modem_rx is empty.
*/
if (gsm->encoding == GSM_BASIC_OPT &&
gsm->dlci[0]->mode == DLCI_MODE_ADM && !dlci->modem_rx)
return true;
return dlci->modem_rx & TIOCM_CD;
}
static void gsm_dtr_rts(struct tty_port *port, bool active)
{
struct gsm_dlci *dlci = container_of(port, struct gsm_dlci, port);
unsigned int modem_tx = dlci->modem_tx;
if (active)
modem_tx |= TIOCM_DTR | TIOCM_RTS;
else
modem_tx &= ~(TIOCM_DTR | TIOCM_RTS);
if (modem_tx != dlci->modem_tx) {
dlci->modem_tx = modem_tx;
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
gsm_modem_update(dlci, 0);
}
}
static const struct tty_port_operations gsm_port_ops = {
.carrier_raised = gsm_carrier_raised,
.dtr_rts = gsm_dtr_rts,
.destruct = gsm_dlci_free,
};
static int gsmtty_install(struct tty_driver *driver, struct tty_struct *tty)
{
struct gsm_mux *gsm;
struct gsm_dlci *dlci;
unsigned int line = tty->index;
unsigned int mux = mux_line_to_num(line);
bool alloc = false;
int ret;
line = line & 0x3F;
if (mux >= MAX_MUX)
return -ENXIO;
/* FIXME: we need to lock gsm_mux for lifetimes of ttys eventually */
if (gsm_mux[mux] == NULL)
return -EUNATCH;
if (line == 0 || line > 61) /* 62/63 reserved */
return -ECHRNG;
gsm = gsm_mux[mux];
if (gsm->dead)
return -EL2HLT;
/* If DLCI 0 is not yet fully open return an error.
This is ok from a locking
perspective as we don't have to worry about this
if DLCI0 is lost */
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
mutex_lock(&gsm->mutex);
if (gsm->dlci[0] && gsm->dlci[0]->state != DLCI_OPEN) {
mutex_unlock(&gsm->mutex);
return -EL2NSYNC;
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
}
dlci = gsm->dlci[line];
if (dlci == NULL) {
alloc = true;
dlci = gsm_dlci_alloc(gsm, line);
}
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
if (dlci == NULL) {
mutex_unlock(&gsm->mutex);
return -ENOMEM;
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
}
ret = tty_port_install(&dlci->port, driver, tty);
if (ret) {
if (alloc)
dlci_put(dlci);
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
mutex_unlock(&gsm->mutex);
return ret;
}
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
dlci_get(dlci);
dlci_get(gsm->dlci[0]);
mux_get(gsm);
tty->driver_data = dlci;
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
mutex_unlock(&gsm->mutex);
return 0;
}
static int gsmtty_open(struct tty_struct *tty, struct file *filp)
{
struct gsm_dlci *dlci = tty->driver_data;
struct tty_port *port = &dlci->port;
port->count++;
tty_port_tty_set(port, tty);
dlci->modem_rx = 0;
/* We could in theory open and close before we wait - eg if we get
a DM straight back. This is ok as that will have caused a hangup */
tty_port_set_initialized(port, true);
/* Start sending off SABM messages */
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
if (!dlci->gsm->wait_config) {
/* Start sending off SABM messages */
if (dlci->gsm->initiator)
gsm_dlci_begin_open(dlci);
else
gsm_dlci_set_opening(dlci);
} else {
gsm_dlci_set_wait_config(dlci);
}
/* And wait for virtual carrier */
return tty_port_block_til_ready(port, tty, filp);
}
static void gsmtty_close(struct tty_struct *tty, struct file *filp)
{
struct gsm_dlci *dlci = tty->driver_data;
if (dlci == NULL)
return;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return;
mutex_lock(&dlci->mutex);
gsm_destroy_network(dlci);
mutex_unlock(&dlci->mutex);
if (tty_port_close_start(&dlci->port, tty, filp) == 0)
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
return;
gsm_dlci_begin_close(dlci);
if (tty_port_initialized(&dlci->port) && C_HUPCL(tty))
tty_port_lower_dtr_rts(&dlci->port);
tty_port_close_end(&dlci->port, tty);
tty_port_tty_set(&dlci->port, NULL);
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
return;
}
static void gsmtty_hangup(struct tty_struct *tty)
{
struct gsm_dlci *dlci = tty->driver_data;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return;
tty_port_hangup(&dlci->port);
gsm_dlci_begin_close(dlci);
}
static int gsmtty_write(struct tty_struct *tty, const unsigned char *buf,
int len)
{
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
int sent;
struct gsm_dlci *dlci = tty->driver_data;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return -EINVAL;
/* Stuff the bytes into the fifo queue */
sent = kfifo_in_locked(&dlci->fifo, buf, len, &dlci->lock);
/* Need to kick the channel */
gsm_dlci_data_kick(dlci);
return sent;
}
tty: make tty_operations::write_room return uint Line disciplines expect a positive value or zero returned from tty->ops->write_room (invoked by tty_write_room). So make this assumption explicit by using unsigned int as a return value. Both of tty->ops->write_room and tty_write_room. Signed-off-by: Jiri Slaby <jslaby@suse.cz> Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com> Acked-by: Alex Elder <elder@linaro.org> Acked-by: Max Filippov <jcmvbkbc@gmail.com> # xtensa Acked-by: David Sterba <dsterba@suse.com> Acked-By: Anton Ivanov <anton.ivanov@cambridgegreys.com> Acked-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Jeff Dike <jdike@addtoit.com> Cc: Richard Weinberger <richard@nod.at> Cc: Chris Zankel <chris@zankel.net> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Samuel Iglesias Gonsalvez <siglesias@igalia.com> Cc: Jens Taprogge <jens.taprogge@taprogge.org> Cc: Karsten Keil <isdn@linux-pingi.de> Cc: Scott Branden <scott.branden@broadcom.com> Cc: Ulf Hansson <ulf.hansson@linaro.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Jakub Kicinski <kuba@kernel.org> Cc: Heiko Carstens <hca@linux.ibm.com> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: David Lin <dtwlin@gmail.com> Cc: Johan Hovold <johan@kernel.org> Cc: Jiri Kosina <jikos@kernel.org> Cc: Shawn Guo <shawnguo@kernel.org> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Oliver Neukum <oneukum@suse.com> Cc: Felipe Balbi <balbi@kernel.org> Cc: Mathias Nyman <mathias.nyman@intel.com> Cc: Marcel Holtmann <marcel@holtmann.org> Cc: Johan Hedberg <johan.hedberg@gmail.com> Cc: Luiz Augusto von Dentz <luiz.dentz@gmail.com> Link: https://lore.kernel.org/r/20210505091928.22010-23-jslaby@suse.cz Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-05-05 09:19:15 +00:00
static unsigned int gsmtty_write_room(struct tty_struct *tty)
{
struct gsm_dlci *dlci = tty->driver_data;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return 0;
return kfifo_avail(&dlci->fifo);
}
static unsigned int gsmtty_chars_in_buffer(struct tty_struct *tty)
{
struct gsm_dlci *dlci = tty->driver_data;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return 0;
return kfifo_len(&dlci->fifo);
}
static void gsmtty_flush_buffer(struct tty_struct *tty)
{
struct gsm_dlci *dlci = tty->driver_data;
unsigned long flags;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return;
/* Caution needed: If we implement reliable transport classes
then the data being transmitted can't simply be junked once
it has first hit the stack. Until then we can just blow it
away */
spin_lock_irqsave(&dlci->lock, flags);
kfifo_reset(&dlci->fifo);
spin_unlock_irqrestore(&dlci->lock, flags);
/* Need to unhook this DLCI from the transmit queue logic */
}
static void gsmtty_wait_until_sent(struct tty_struct *tty, int timeout)
{
/* The FIFO handles the queue so the kernel will do the right
thing waiting on chars_in_buffer before calling us. No work
to do here */
}
static int gsmtty_tiocmget(struct tty_struct *tty)
{
struct gsm_dlci *dlci = tty->driver_data;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return -EINVAL;
return dlci->modem_rx;
}
static int gsmtty_tiocmset(struct tty_struct *tty,
unsigned int set, unsigned int clear)
{
struct gsm_dlci *dlci = tty->driver_data;
unsigned int modem_tx = dlci->modem_tx;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return -EINVAL;
modem_tx &= ~clear;
modem_tx |= set;
if (modem_tx != dlci->modem_tx) {
dlci->modem_tx = modem_tx;
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
return gsm_modem_update(dlci, 0);
}
return 0;
}
static int gsmtty_ioctl(struct tty_struct *tty,
unsigned int cmd, unsigned long arg)
{
struct gsm_dlci *dlci = tty->driver_data;
struct gsm_netconfig nc;
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
struct gsm_dlci_config dc;
int index;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return -EINVAL;
switch (cmd) {
case GSMIOC_ENABLE_NET:
if (copy_from_user(&nc, (void __user *)arg, sizeof(nc)))
return -EFAULT;
nc.if_name[IFNAMSIZ-1] = '\0';
/* return net interface index or error code */
mutex_lock(&dlci->mutex);
index = gsm_create_network(dlci, &nc);
mutex_unlock(&dlci->mutex);
if (copy_to_user((void __user *)arg, &nc, sizeof(nc)))
return -EFAULT;
return index;
case GSMIOC_DISABLE_NET:
if (!capable(CAP_NET_ADMIN))
return -EPERM;
mutex_lock(&dlci->mutex);
gsm_destroy_network(dlci);
mutex_unlock(&dlci->mutex);
return 0;
tty: n_gsm: add ioctl for DLC specific parameter configuration Parameter negotiation has been introduced with commit 92f1f0c3290d ("tty: n_gsm: add parameter negotiation support") However, means to set individual parameters per DLCI are not yet implemented. Furthermore, it is currently not possible to keep a DLCI half open until the user application sets the right parameters for it. This is required to allow a user application to set its specific parameters before the underlying link is established. Otherwise, the link is opened and re-established right afterwards if the user application sets incompatible parameters. This may be an unexpected behavior for the peer. Add parameter 'wait_config' to 'gsm_config' to support setups where the DLCI specific user application sets its specific parameters after open() and before the link gets fully established. Setting this to zero disables the user application specific DLCI configuration option. Add the ioctls 'GSMIOC_GETCONF_DLCI' and 'GSMIOC_SETCONF_DLCI' for the ldisc and virtual ttys. This gets/sets the DLCI specific parameters and may trigger a reconnect of the DLCI if incompatible values have been set. Only the parameters for the DLCI associated with the virtual tty can be set or retrieved if called on these. Add remark within the documentation to introduce the new ioctls. Link: https://lore.kernel.org/oe-kbuild-all/202302281856.S9Lz4gHB-lkp@intel.com/ Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20230315105354.6234-1-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2023-03-15 10:53:52 +00:00
case GSMIOC_GETCONF_DLCI:
if (copy_from_user(&dc, (void __user *)arg, sizeof(dc)))
return -EFAULT;
if (dc.channel != dlci->addr)
return -EPERM;
gsm_dlci_copy_config_values(dlci, &dc);
if (copy_to_user((void __user *)arg, &dc, sizeof(dc)))
return -EFAULT;
return 0;
case GSMIOC_SETCONF_DLCI:
if (copy_from_user(&dc, (void __user *)arg, sizeof(dc)))
return -EFAULT;
if (dc.channel >= NUM_DLCI)
return -EINVAL;
if (dc.channel != 0 && dc.channel != dlci->addr)
return -EPERM;
return gsm_dlci_config(dlci, &dc, 1);
case TIOCMIWAIT:
return gsm_wait_modem_change(dlci, (u32)arg);
default:
return -ENOIOCTLCMD;
}
}
static void gsmtty_set_termios(struct tty_struct *tty,
const struct ktermios *old)
{
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
struct gsm_dlci *dlci = tty->driver_data;
if (dlci->state == DLCI_CLOSED)
return;
/* For the moment its fixed. In actual fact the speed information
for the virtual channel can be propogated in both directions by
the RPN control message. This however rapidly gets nasty as we
then have to remap modem signals each way according to whether
our virtual cable is null modem etc .. */
tty_termios_copy_hw(&tty->termios, old);
}
static void gsmtty_throttle(struct tty_struct *tty)
{
struct gsm_dlci *dlci = tty->driver_data;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return;
if (C_CRTSCTS(tty))
dlci->modem_tx &= ~TIOCM_RTS;
dlci->throttled = true;
/* Send an MSC with RTS cleared */
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
gsm_modem_update(dlci, 0);
}
static void gsmtty_unthrottle(struct tty_struct *tty)
{
struct gsm_dlci *dlci = tty->driver_data;
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return;
if (C_CRTSCTS(tty))
dlci->modem_tx |= TIOCM_RTS;
dlci->throttled = false;
/* Send an MSC with RTS set */
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
gsm_modem_update(dlci, 0);
}
static int gsmtty_break_ctl(struct tty_struct *tty, int state)
{
struct gsm_dlci *dlci = tty->driver_data;
int encode = 0; /* Off */
tty: Prevent deadlock in n_gsm driver This change fixes a deadlock when the multiplexer is closed while there are still client side ports open. When the multiplexer is closed and there are active tty's it tries to close them with tty_vhangup. This has a problem though, because tty_vhangup needs the tty_lock. This patch changes it to unlock the tty_lock before attempting the hangup and relocks afterwards. The additional call to tty_port_tty_set is needed because otherwise the port stays active because of the reference counter. This change also exposed another problem that other code paths don't expect that the multiplexer could have been closed. This patch also adds checks for these cases in the gsmtty_ class of function that could be called. The documentation explicitly states that "first close all virtual ports before closing the physical port" but we've found this to not always reality in our field situations. The GPRS / UTMS modem sometimes crashes and needs a power cycle in that case which means cleanly shutting down everything is not always possible. This change makes it much more robust for our situation where at least the system is recoverable with this patch and doesn't hang in a deadlock situation inside the kernel. The patch is against the long term support kernel (3.4.27) and should apply cleanly to more recent branches. Tested with a Telit GE864-QUADV2 and Telit HE910 modem. Signed-off-by: Dirkjan Bussink <dirkjan.bussink@nedap.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-30 10:44:50 +00:00
if (dlci->state == DLCI_CLOSED)
return -EINVAL;
if (state == -1) /* "On indefinitely" - we can't encode this
properly */
encode = 0x0F;
else if (state > 0) {
encode = state / 200; /* mS to encoding */
if (encode > 0x0F)
encode = 0x0F; /* Best effort */
}
tty: n_gsm: fix invalid use of MSC in advanced option n_gsm is based on the 3GPP 07.010 and its newer version is the 3GPP 27.010. See https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=1516 The changes from 07.010 to 27.010 are non-functional. Therefore, I refer to the newer 27.010 here. Chapter 5.4.6.3.7 states that the Modem Status Command (MSC) shall only be used if the basic option was chosen. The current implementation uses MSC frames even if advanced option was chosen to inform the peer about modem line state updates. A standard conform peer may choose to discard these frames in advanced option mode. Furthermore, gsmtty_modem_update() is not part of the 'tty_operations' functions despite its name. Rename gsmtty_modem_update() to gsm_modem_update() to clarify this. Split its function into gsm_modem_upd_via_data() and gsm_modem_upd_via_msc() depending on the encoding and adaption. Introduce gsm_dlci_modem_output() as adaption of gsm_dlci_data_output() to encode and queue empty frames in advanced option mode. Use it in gsm_modem_upd_via_data(). gsm_modem_upd_via_msc() is based on the initial gsmtty_modem_update() function which used only MSC frames to update modem states. Fixes: e1eaea46bb40 ("tty: n_gsm line discipline") Cc: stable@vger.kernel.org Signed-off-by: Daniel Starke <daniel.starke@siemens.com> Link: https://lore.kernel.org/r/20220422071025.5490-2-daniel.starke@siemens.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-04-22 07:10:24 +00:00
return gsm_modem_update(dlci, encode);
}
static void gsmtty_cleanup(struct tty_struct *tty)
n_gsm: race between ld close and gsmtty open ttyA has ld associated to n_gsm, when ttyA is closing, it triggers to release gsmttyB's ld data dlci[B], then race would happen if gsmttyB is opening in parallel. (Note: This patch set differs from previous set in that it uses mutex instead of spin lock to avoid race, so that it avoids sleeping in automic context) Here are race cases we found recently in test: CASE #1 ==================================================================== releasing dlci[B] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(gsmttyB), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[B]) ----- | | gsm_dlci_free(dlci[B]) ----- | | ----- gsmtty_open(gsmttyB) gsmtty_open() { struct gsm_dlci *dlci = tty->driver_data; => here it uses dlci[B] ... } In gsmtty_open(gsmttyA), it uses dlci[B] which was release, so hit a panic. ===================================================================== CASE #2 ===================================================================== releasing dlci[0] race with gsmtty_install(gsmttyB), then panic in gsmtty_open(), as below: tty_release(ttyA) tty_open(gsmttyB) | | ----- gsmtty_install(gsmttyB) | | ----- gsm_dlci_alloc(gsmttyB) => alloc dlci[B] | | ----- gsmtty_open(gsmttyB) fail | | ----- tty_release(gsmttyB) | | ----- gsmtty_close(gsmttyB) | | ----- gsmtty_detach_dlci(dlci[B]) | | ----- dlci_put(dlci[B]) | | tty_ldisc_release(ttyA) ----- | | gsm_dlci_release(dlci[0]) ----- | | gsm_dlci_free(dlci[0]) ----- | | ----- dlci_put(dlci[0]) In gsmtty_detach_dlci(dlci[B]), it tries to use dlci[0] which was released, then hit panic. ===================================================================== IMHO, n_gsm tty operations would refer released ldisc, as long as gsm_dlci_release() has chance to release ldisc data when some gsmtty operations are ongoing.. This patch is try to avoid it by: 1) in n_gsm driver, use a global gsm mutex lock to avoid gsm_dlci_release() run in parallel with gsmtty_install(); 2) Increase dlci's ref count in gsmtty_install() instead of in gsmtty_open(), the purpose is to prevent gsm_dlci_release() releasing dlci after gsmtty_install() allocats dlci but before gsmtty_open increases dlci's ref count; 3) Decrease dlci's ref count in gsmtty_remove(), a tty framework API, this is the opposite process of step 2). Signed-off-by: Chao Bi <chao.bi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-26 04:09:39 +00:00
{
struct gsm_dlci *dlci = tty->driver_data;
struct gsm_mux *gsm = dlci->gsm;
dlci_put(dlci);
dlci_put(gsm->dlci[0]);
mux_put(gsm);
}
/* Virtual ttys for the demux */
static const struct tty_operations gsmtty_ops = {
.install = gsmtty_install,
.open = gsmtty_open,
.close = gsmtty_close,
.write = gsmtty_write,
.write_room = gsmtty_write_room,
.chars_in_buffer = gsmtty_chars_in_buffer,
.flush_buffer = gsmtty_flush_buffer,
.ioctl = gsmtty_ioctl,
.throttle = gsmtty_throttle,
.unthrottle = gsmtty_unthrottle,
.set_termios = gsmtty_set_termios,
.hangup = gsmtty_hangup,
.wait_until_sent = gsmtty_wait_until_sent,
.tiocmget = gsmtty_tiocmget,
.tiocmset = gsmtty_tiocmset,
.break_ctl = gsmtty_break_ctl,
.cleanup = gsmtty_cleanup,
};
static int __init gsm_init(void)
{
/* Fill in our line protocol discipline, and register it */
int status = tty_register_ldisc(&tty_ldisc_packet);
if (status != 0) {
pr_err("n_gsm: can't register line discipline (err = %d)\n",
status);
return status;
}
gsm_tty_driver = tty_alloc_driver(GSM_TTY_MINORS, TTY_DRIVER_REAL_RAW |
tty: stop using alloc_tty_driver alloc_tty_driver was deprecated by tty_alloc_driver in commit 7f0bc6a68ed9 (TTY: pass flags to alloc_tty_driver) in 2012. I never got into eliminating alloc_tty_driver until now. So we still have two functions for allocating drivers which might be confusing. So get rid of alloc_tty_driver uses to eliminate it for good in the next patch. Note we need to switch return value checking as tty_alloc_driver uses ERR_PTR. And flags are now a parameter of tty_alloc_driver. Cc: Richard Henderson <rth@twiddle.net>(odd fixer:ALPHA PORT) Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Jeff Dike <jdike@addtoit.com> Cc: Richard Weinberger <richard@nod.at> Cc: Anton Ivanov <anton.ivanov@cambridgegreys.com> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Samuel Iglesias Gonsalvez <siglesias@igalia.com> Cc: Jens Taprogge <jens.taprogge@taprogge.org> Cc: Karsten Keil <isdn@linux-pingi.de> Cc: Ulf Hansson <ulf.hansson@linaro.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Jakub Kicinski <kuba@kernel.org> Cc: Heiko Carstens <hca@linux.ibm.com> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Laurentiu Tudor <laurentiu.tudor@nxp.com> Cc: Jiri Kosina <jikos@kernel.org> Cc: David Sterba <dsterba@suse.com> Cc: Shawn Guo <shawnguo@kernel.org> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Oliver Neukum <oneukum@suse.com> Cc: Felipe Balbi <balbi@kernel.org> Cc: Johan Hovold <johan@kernel.org> Cc: Marcel Holtmann <marcel@holtmann.org> Cc: Johan Hedberg <johan.hedberg@gmail.com> Cc: Luiz Augusto von Dentz <luiz.dentz@gmail.com> Acked-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com> Acked-by: Max Filippov <jcmvbkbc@gmail.com> Acked-by: David Sterba <dsterba@suse.com> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Signed-off-by: Jiri Slaby <jslaby@suse.cz> Link: https://lore.kernel.org/r/20210723074317.32690-5-jslaby@suse.cz Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-07-23 07:43:13 +00:00
TTY_DRIVER_DYNAMIC_DEV | TTY_DRIVER_HARDWARE_BREAK);
if (IS_ERR(gsm_tty_driver)) {
pr_err("gsm_init: tty allocation failed.\n");
tty: stop using alloc_tty_driver alloc_tty_driver was deprecated by tty_alloc_driver in commit 7f0bc6a68ed9 (TTY: pass flags to alloc_tty_driver) in 2012. I never got into eliminating alloc_tty_driver until now. So we still have two functions for allocating drivers which might be confusing. So get rid of alloc_tty_driver uses to eliminate it for good in the next patch. Note we need to switch return value checking as tty_alloc_driver uses ERR_PTR. And flags are now a parameter of tty_alloc_driver. Cc: Richard Henderson <rth@twiddle.net>(odd fixer:ALPHA PORT) Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Jeff Dike <jdike@addtoit.com> Cc: Richard Weinberger <richard@nod.at> Cc: Anton Ivanov <anton.ivanov@cambridgegreys.com> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Samuel Iglesias Gonsalvez <siglesias@igalia.com> Cc: Jens Taprogge <jens.taprogge@taprogge.org> Cc: Karsten Keil <isdn@linux-pingi.de> Cc: Ulf Hansson <ulf.hansson@linaro.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Jakub Kicinski <kuba@kernel.org> Cc: Heiko Carstens <hca@linux.ibm.com> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Laurentiu Tudor <laurentiu.tudor@nxp.com> Cc: Jiri Kosina <jikos@kernel.org> Cc: David Sterba <dsterba@suse.com> Cc: Shawn Guo <shawnguo@kernel.org> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Oliver Neukum <oneukum@suse.com> Cc: Felipe Balbi <balbi@kernel.org> Cc: Johan Hovold <johan@kernel.org> Cc: Marcel Holtmann <marcel@holtmann.org> Cc: Johan Hedberg <johan.hedberg@gmail.com> Cc: Luiz Augusto von Dentz <luiz.dentz@gmail.com> Acked-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com> Acked-by: Max Filippov <jcmvbkbc@gmail.com> Acked-by: David Sterba <dsterba@suse.com> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Signed-off-by: Jiri Slaby <jslaby@suse.cz> Link: https://lore.kernel.org/r/20210723074317.32690-5-jslaby@suse.cz Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-07-23 07:43:13 +00:00
status = PTR_ERR(gsm_tty_driver);
goto err_unreg_ldisc;
}
gsm_tty_driver->driver_name = "gsmtty";
gsm_tty_driver->name = "gsmtty";
gsm_tty_driver->major = 0; /* Dynamic */
gsm_tty_driver->minor_start = 0;
gsm_tty_driver->type = TTY_DRIVER_TYPE_SERIAL;
gsm_tty_driver->subtype = SERIAL_TYPE_NORMAL;
gsm_tty_driver->init_termios = tty_std_termios;
/* Fixme */
gsm_tty_driver->init_termios.c_lflag &= ~ECHO;
tty_set_operations(gsm_tty_driver, &gsmtty_ops);
if (tty_register_driver(gsm_tty_driver)) {
pr_err("gsm_init: tty registration failed.\n");
status = -EBUSY;
goto err_put_driver;
}
pr_debug("gsm_init: loaded as %d,%d.\n",
gsm_tty_driver->major, gsm_tty_driver->minor_start);
return 0;
err_put_driver:
tty: drop put_tty_driver put_tty_driver() is an alias for tty_driver_kref_put(). There is no need for two exported identical functions, therefore switch all users of old put_tty_driver() to new tty_driver_kref_put() and remove the former for good. Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Jeff Dike <jdike@addtoit.com> Cc: Richard Weinberger <richard@nod.at> Cc: Anton Ivanov <anton.ivanov@cambridgegreys.com> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Samuel Iglesias Gonsalvez <siglesias@igalia.com> Cc: Jens Taprogge <jens.taprogge@taprogge.org> Cc: Karsten Keil <isdn@linux-pingi.de> Cc: Scott Branden <scott.branden@broadcom.com> Cc: Ulf Hansson <ulf.hansson@linaro.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Jakub Kicinski <kuba@kernel.org> Cc: Heiko Carstens <hca@linux.ibm.com> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: David Lin <dtwlin@gmail.com> Cc: Johan Hovold <johan@kernel.org> Cc: Alex Elder <elder@kernel.org> Cc: Jiri Slaby <jirislaby@kernel.org> Cc: Laurentiu Tudor <laurentiu.tudor@nxp.com> Cc: Jiri Kosina <jikos@kernel.org> Cc: David Sterba <dsterba@suse.com> Cc: Shawn Guo <shawnguo@kernel.org> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Pengutronix Kernel Team <kernel@pengutronix.de> Cc: Fabio Estevam <festevam@gmail.com> Cc: NXP Linux Team <linux-imx@nxp.com> Cc: Oliver Neukum <oneukum@suse.com> Cc: Felipe Balbi <balbi@kernel.org> Cc: Mathias Nyman <mathias.nyman@intel.com> Cc: Marcel Holtmann <marcel@holtmann.org> Cc: Johan Hedberg <johan.hedberg@gmail.com> Cc: Luiz Augusto von Dentz <luiz.dentz@gmail.com> Acked-by: Alex Elder <elder@linaro.org> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Acked-by: Max Filippov <jcmvbkbc@gmail.com> Acked-by: David Sterba <dsterba@suse.com> Acked-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com> Signed-off-by: Jiri Slaby <jslaby@suse.cz> Link: https://lore.kernel.org/r/20210723074317.32690-8-jslaby@suse.cz Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-07-23 07:43:16 +00:00
tty_driver_kref_put(gsm_tty_driver);
err_unreg_ldisc:
tty_unregister_ldisc(&tty_ldisc_packet);
return status;
}
static void __exit gsm_exit(void)
{
tty_unregister_ldisc(&tty_ldisc_packet);
tty_unregister_driver(gsm_tty_driver);
tty: drop put_tty_driver put_tty_driver() is an alias for tty_driver_kref_put(). There is no need for two exported identical functions, therefore switch all users of old put_tty_driver() to new tty_driver_kref_put() and remove the former for good. Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Jeff Dike <jdike@addtoit.com> Cc: Richard Weinberger <richard@nod.at> Cc: Anton Ivanov <anton.ivanov@cambridgegreys.com> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Samuel Iglesias Gonsalvez <siglesias@igalia.com> Cc: Jens Taprogge <jens.taprogge@taprogge.org> Cc: Karsten Keil <isdn@linux-pingi.de> Cc: Scott Branden <scott.branden@broadcom.com> Cc: Ulf Hansson <ulf.hansson@linaro.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Jakub Kicinski <kuba@kernel.org> Cc: Heiko Carstens <hca@linux.ibm.com> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: David Lin <dtwlin@gmail.com> Cc: Johan Hovold <johan@kernel.org> Cc: Alex Elder <elder@kernel.org> Cc: Jiri Slaby <jirislaby@kernel.org> Cc: Laurentiu Tudor <laurentiu.tudor@nxp.com> Cc: Jiri Kosina <jikos@kernel.org> Cc: David Sterba <dsterba@suse.com> Cc: Shawn Guo <shawnguo@kernel.org> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Pengutronix Kernel Team <kernel@pengutronix.de> Cc: Fabio Estevam <festevam@gmail.com> Cc: NXP Linux Team <linux-imx@nxp.com> Cc: Oliver Neukum <oneukum@suse.com> Cc: Felipe Balbi <balbi@kernel.org> Cc: Mathias Nyman <mathias.nyman@intel.com> Cc: Marcel Holtmann <marcel@holtmann.org> Cc: Johan Hedberg <johan.hedberg@gmail.com> Cc: Luiz Augusto von Dentz <luiz.dentz@gmail.com> Acked-by: Alex Elder <elder@linaro.org> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Acked-by: Max Filippov <jcmvbkbc@gmail.com> Acked-by: David Sterba <dsterba@suse.com> Acked-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com> Signed-off-by: Jiri Slaby <jslaby@suse.cz> Link: https://lore.kernel.org/r/20210723074317.32690-8-jslaby@suse.cz Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-07-23 07:43:16 +00:00
tty_driver_kref_put(gsm_tty_driver);
}
module_init(gsm_init);
module_exit(gsm_exit);
MODULE_LICENSE("GPL");
MODULE_ALIAS_LDISC(N_GSM0710);