linux-stable/drivers/tty/tty_buffer.c

600 lines
16 KiB
C
Raw Normal View History

/*
* Tty buffer allocation management
*/
#include <linux/types.h>
#include <linux/errno.h>
#include <linux/tty.h>
#include <linux/tty_driver.h>
#include <linux/tty_flip.h>
#include <linux/timer.h>
#include <linux/string.h>
#include <linux/slab.h>
#include <linux/sched.h>
#include <linux/wait.h>
#include <linux/bitops.h>
#include <linux/delay.h>
#include <linux/module.h>
#include <linux/ratelimit.h>
#define MIN_TTYB_SIZE 256
#define TTYB_ALIGN_MASK 255
/*
* Byte threshold to limit memory consumption for flip buffers.
* The actual memory limit is > 2x this amount.
*/
#define TTYB_DEFAULT_MEM_LIMIT 65536
/*
* We default to dicing tty buffer allocations to this many characters
* in order to avoid multiple page allocations. We know the size of
* tty_buffer itself but it must also be taken into account that the
* the buffer is 256 byte aligned. See tty_buffer_find for the allocation
* logic this must match
*/
#define TTY_BUFFER_PAGE (((PAGE_SIZE - sizeof(struct tty_buffer)) / 2) & ~0xFF)
pty: Fix input race when closing A read() from a pty master may mistakenly indicate EOF (errno == -EIO) after the pty slave has closed, even though input data remains to be read. For example, pty slave | input worker | pty master | | | | n_tty_read() pty_write() | | input avail? no add data | | sleep schedule worker --->| | . |---> flush_to_ldisc() | . pty_close() | fill read buffer | . wait for worker | wakeup reader --->| . | read buffer full? |---> input avail ? yes |<--- yes - exit worker | copy 4096 bytes to user TTY_OTHER_CLOSED <---| |<--- kick worker | | **** New read() before worker starts **** | | n_tty_read() | | input avail? no | | TTY_OTHER_CLOSED? yes | | return -EIO Several conditions are required to trigger this race: 1. the ldisc read buffer must become full so the input worker exits 2. the read() count parameter must be >= 4096 so the ldisc read buffer is empty 3. the subsequent read() occurs before the kicked worker has processed more input However, the underlying cause of the race is that data is pipelined, while tty state is not; ie., data already written by the pty slave end is not yet visible to the pty master end, but state changes by the pty slave end are visible to the pty master end immediately. Pipeline the TTY_OTHER_CLOSED state through input worker to the reader. 1. Introduce TTY_OTHER_DONE which is set by the input worker when TTY_OTHER_CLOSED is set and either the input buffers are flushed or input processing has completed. Readers/polls are woken when TTY_OTHER_DONE is set. 2. Reader/poll checks TTY_OTHER_DONE instead of TTY_OTHER_CLOSED. 3. A new input worker is started from pty_close() after setting TTY_OTHER_CLOSED, which ensures the TTY_OTHER_DONE state will be set if the last input worker is already finished (or just about to exit). Remove tty_flush_to_ldisc(); no in-tree callers. Fixes: 52bce7f8d4fc ("pty, n_tty: Simplify input processing on final close") Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96311 BugLink: http://bugs.launchpad.net/bugs/1429756 Cc: <stable@vger.kernel.org> # 3.19+ Reported-by: Andy Whitcroft <apw@canonical.com> Reported-by: H.J. Lu <hjl.tools@gmail.com> Signed-off-by: Peter Hurley <peter@hurleysoftware.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2015-04-13 17:24:34 +00:00
/*
* If all tty flip buffers have been processed by flush_to_ldisc() or
* dropped by tty_buffer_flush(), check if the linked pty has been closed.
* If so, wake the reader/poll to process
*/
static inline void check_other_closed(struct tty_struct *tty)
{
unsigned long flags, old;
/* transition from TTY_OTHER_CLOSED => TTY_OTHER_DONE must be atomic */
for (flags = ACCESS_ONCE(tty->flags);
test_bit(TTY_OTHER_CLOSED, &flags);
) {
old = flags;
__set_bit(TTY_OTHER_DONE, &flags);
flags = cmpxchg(&tty->flags, old, flags);
if (old == flags) {
wake_up_interruptible(&tty->read_wait);
break;
}
}
}
/**
* tty_buffer_lock_exclusive - gain exclusive access to buffer
* tty_buffer_unlock_exclusive - release exclusive access
*
* @port - tty_port owning the flip buffer
*
* Guarantees safe use of the line discipline's receive_buf() method by
* excluding the buffer work and any pending flush from using the flip
* buffer. Data can continue to be added concurrently to the flip buffer
* from the driver side.
*
* On release, the buffer work is restarted if there is data in the
* flip buffer
*/
void tty_buffer_lock_exclusive(struct tty_port *port)
{
struct tty_bufhead *buf = &port->buf;
atomic_inc(&buf->priority);
mutex_lock(&buf->lock);
}
EXPORT_SYMBOL_GPL(tty_buffer_lock_exclusive);
void tty_buffer_unlock_exclusive(struct tty_port *port)
{
struct tty_bufhead *buf = &port->buf;
int restart;
restart = buf->head->commit != buf->head->read;
atomic_dec(&buf->priority);
mutex_unlock(&buf->lock);
if (restart)
queue_work(system_unbound_wq, &buf->work);
}
EXPORT_SYMBOL_GPL(tty_buffer_unlock_exclusive);
/**
* tty_buffer_space_avail - return unused buffer space
* @port - tty_port owning the flip buffer
*
* Returns the # of bytes which can be written by the driver without
* reaching the buffer limit.
*
* Note: this does not guarantee that memory is available to write
* the returned # of bytes (use tty_prepare_flip_string_xxx() to
* pre-allocate if memory guarantee is required).
*/
int tty_buffer_space_avail(struct tty_port *port)
{
int space = port->buf.mem_limit - atomic_read(&port->buf.mem_used);
return max(space, 0);
}
EXPORT_SYMBOL_GPL(tty_buffer_space_avail);
static void tty_buffer_reset(struct tty_buffer *p, size_t size)
{
p->used = 0;
p->size = size;
p->next = NULL;
p->commit = 0;
p->read = 0;
p->flags = 0;
}
/**
* tty_buffer_free_all - free buffers used by a tty
* @tty: tty to free from
*
* Remove all the buffers pending on a tty whether queued with data
* or in the free ring. Must be called when the tty is no longer in use
*/
void tty_buffer_free_all(struct tty_port *port)
{
struct tty_bufhead *buf = &port->buf;
struct tty_buffer *p, *next;
struct llist_node *llist;
while ((p = buf->head) != NULL) {
buf->head = p->next;
if (p->size > 0)
kfree(p);
}
llist = llist_del_all(&buf->free);
llist_for_each_entry_safe(p, next, llist, free)
kfree(p);
tty_buffer_reset(&buf->sentinel, 0);
buf->head = &buf->sentinel;
buf->tail = &buf->sentinel;
atomic_set(&buf->mem_used, 0);
}
/**
* tty_buffer_alloc - allocate a tty buffer
* @tty: tty device
* @size: desired size (characters)
*
* Allocate a new tty buffer to hold the desired number of characters.
* We round our buffers off in 256 character chunks to get better
* allocation behaviour.
* Return NULL if out of memory or the allocation would exceed the
* per device queue
*/
static struct tty_buffer *tty_buffer_alloc(struct tty_port *port, size_t size)
{
struct llist_node *free;
struct tty_buffer *p;
/* Round the buffer size out */
size = __ALIGN_MASK(size, TTYB_ALIGN_MASK);
if (size <= MIN_TTYB_SIZE) {
free = llist_del_first(&port->buf.free);
if (free) {
p = llist_entry(free, struct tty_buffer, free);
goto found;
}
}
/* Should possibly check if this fails for the largest buffer we
have queued and recycle that ? */
if (atomic_read(&port->buf.mem_used) > port->buf.mem_limit)
return NULL;
p = kmalloc(sizeof(struct tty_buffer) + 2 * size, GFP_ATOMIC);
if (p == NULL)
return NULL;
found:
tty_buffer_reset(p, size);
atomic_add(size, &port->buf.mem_used);
return p;
}
/**
* tty_buffer_free - free a tty buffer
* @tty: tty owning the buffer
* @b: the buffer to free
*
* Free a tty buffer, or add it to the free list according to our
* internal strategy
*/
static void tty_buffer_free(struct tty_port *port, struct tty_buffer *b)
{
struct tty_bufhead *buf = &port->buf;
/* Dumb strategy for now - should keep some stats */
WARN_ON(atomic_sub_return(b->size, &buf->mem_used) < 0);
if (b->size > MIN_TTYB_SIZE)
kfree(b);
else if (b->size > 0)
llist_add(&b->free, &buf->free);
}
/**
* tty_buffer_flush - flush full tty buffers
* @tty: tty to flush
* @ld: optional ldisc ptr (must be referenced)
*
* flush all the buffers containing receive data. If ld != NULL,
* flush the ldisc input buffer.
*
* Locking: takes buffer lock to ensure single-threaded flip buffer
* 'consumer'
*/
void tty_buffer_flush(struct tty_struct *tty, struct tty_ldisc *ld)
{
struct tty_port *port = tty->port;
struct tty_bufhead *buf = &port->buf;
struct tty_buffer *next;
atomic_inc(&buf->priority);
mutex_lock(&buf->lock);
/* paired w/ release in __tty_buffer_request_room; ensures there are
* no pending memory accesses to the freed buffer
*/
while ((next = smp_load_acquire(&buf->head->next)) != NULL) {
tty_buffer_free(port, buf->head);
buf->head = next;
}
buf->head->read = buf->head->commit;
if (ld && ld->ops->flush_buffer)
ld->ops->flush_buffer(tty);
pty: Fix input race when closing A read() from a pty master may mistakenly indicate EOF (errno == -EIO) after the pty slave has closed, even though input data remains to be read. For example, pty slave | input worker | pty master | | | | n_tty_read() pty_write() | | input avail? no add data | | sleep schedule worker --->| | . |---> flush_to_ldisc() | . pty_close() | fill read buffer | . wait for worker | wakeup reader --->| . | read buffer full? |---> input avail ? yes |<--- yes - exit worker | copy 4096 bytes to user TTY_OTHER_CLOSED <---| |<--- kick worker | | **** New read() before worker starts **** | | n_tty_read() | | input avail? no | | TTY_OTHER_CLOSED? yes | | return -EIO Several conditions are required to trigger this race: 1. the ldisc read buffer must become full so the input worker exits 2. the read() count parameter must be >= 4096 so the ldisc read buffer is empty 3. the subsequent read() occurs before the kicked worker has processed more input However, the underlying cause of the race is that data is pipelined, while tty state is not; ie., data already written by the pty slave end is not yet visible to the pty master end, but state changes by the pty slave end are visible to the pty master end immediately. Pipeline the TTY_OTHER_CLOSED state through input worker to the reader. 1. Introduce TTY_OTHER_DONE which is set by the input worker when TTY_OTHER_CLOSED is set and either the input buffers are flushed or input processing has completed. Readers/polls are woken when TTY_OTHER_DONE is set. 2. Reader/poll checks TTY_OTHER_DONE instead of TTY_OTHER_CLOSED. 3. A new input worker is started from pty_close() after setting TTY_OTHER_CLOSED, which ensures the TTY_OTHER_DONE state will be set if the last input worker is already finished (or just about to exit). Remove tty_flush_to_ldisc(); no in-tree callers. Fixes: 52bce7f8d4fc ("pty, n_tty: Simplify input processing on final close") Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96311 BugLink: http://bugs.launchpad.net/bugs/1429756 Cc: <stable@vger.kernel.org> # 3.19+ Reported-by: Andy Whitcroft <apw@canonical.com> Reported-by: H.J. Lu <hjl.tools@gmail.com> Signed-off-by: Peter Hurley <peter@hurleysoftware.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2015-04-13 17:24:34 +00:00
check_other_closed(tty);
atomic_dec(&buf->priority);
mutex_unlock(&buf->lock);
}
/**
* tty_buffer_request_room - grow tty buffer if needed
* @tty: tty structure
* @size: size desired
* @flags: buffer flags if new buffer allocated (default = 0)
*
* Make at least size bytes of linear space available for the tty
* buffer. If we fail return the size we managed to find.
*
* Will change over to a new buffer if the current buffer is encoded as
* TTY_NORMAL (so has no flags buffer) and the new buffer requires
* a flags buffer.
*/
static int __tty_buffer_request_room(struct tty_port *port, size_t size,
int flags)
{
struct tty_bufhead *buf = &port->buf;
struct tty_buffer *b, *n;
int left, change;
b = buf->tail;
if (b->flags & TTYB_NORMAL)
left = 2 * b->size - b->used;
else
left = b->size - b->used;
change = (b->flags & TTYB_NORMAL) && (~flags & TTYB_NORMAL);
if (change || left < size) {
/* This is the slow path - looking for new buffers to use */
n = tty_buffer_alloc(port, size);
if (n != NULL) {
n->flags = flags;
buf->tail = n;
/* paired w/ acquire in flush_to_ldisc(); ensures
* flush_to_ldisc() sees buffer data.
*/
smp_store_release(&b->commit, b->used);
/* paired w/ acquire in flush_to_ldisc(); ensures the
* latest commit value can be read before the head is
* advanced to the next buffer
*/
smp_store_release(&b->next, n);
} else if (change)
size = 0;
else
size = left;
}
return size;
}
int tty_buffer_request_room(struct tty_port *port, size_t size)
{
return __tty_buffer_request_room(port, size, 0);
}
EXPORT_SYMBOL_GPL(tty_buffer_request_room);
/**
* tty_insert_flip_string_fixed_flag - Add characters to the tty buffer
* @port: tty port
* @chars: characters
* @flag: flag value for each character
* @size: size
*
* Queue a series of bytes to the tty buffering. All the characters
* passed are marked with the supplied flag. Returns the number added.
*/
int tty_insert_flip_string_fixed_flag(struct tty_port *port,
const unsigned char *chars, char flag, size_t size)
{
int copied = 0;
do {
int goal = min_t(size_t, size - copied, TTY_BUFFER_PAGE);
int flags = (flag == TTY_NORMAL) ? TTYB_NORMAL : 0;
int space = __tty_buffer_request_room(port, goal, flags);
struct tty_buffer *tb = port->buf.tail;
if (unlikely(space == 0))
break;
memcpy(char_buf_ptr(tb, tb->used), chars, space);
if (~tb->flags & TTYB_NORMAL)
memset(flag_buf_ptr(tb, tb->used), flag, space);
tb->used += space;
copied += space;
chars += space;
/* There is a small chance that we need to split the data over
several buffers. If this is the case we must loop */
} while (unlikely(size > copied));
return copied;
}
EXPORT_SYMBOL(tty_insert_flip_string_fixed_flag);
/**
* tty_insert_flip_string_flags - Add characters to the tty buffer
* @port: tty port
* @chars: characters
* @flags: flag bytes
* @size: size
*
* Queue a series of bytes to the tty buffering. For each character
* the flags array indicates the status of the character. Returns the
* number added.
*/
int tty_insert_flip_string_flags(struct tty_port *port,
const unsigned char *chars, const char *flags, size_t size)
{
int copied = 0;
do {
int goal = min_t(size_t, size - copied, TTY_BUFFER_PAGE);
int space = tty_buffer_request_room(port, goal);
struct tty_buffer *tb = port->buf.tail;
if (unlikely(space == 0))
break;
memcpy(char_buf_ptr(tb, tb->used), chars, space);
memcpy(flag_buf_ptr(tb, tb->used), flags, space);
tb->used += space;
copied += space;
chars += space;
flags += space;
/* There is a small chance that we need to split the data over
several buffers. If this is the case we must loop */
} while (unlikely(size > copied));
return copied;
}
EXPORT_SYMBOL(tty_insert_flip_string_flags);
/**
* tty_schedule_flip - push characters to ldisc
* @port: tty port to push from
*
* Takes any pending buffers and transfers their ownership to the
* ldisc side of the queue. It then schedules those characters for
* processing by the line discipline.
*/
void tty_schedule_flip(struct tty_port *port)
{
struct tty_bufhead *buf = &port->buf;
/* paired w/ acquire in flush_to_ldisc(); ensures
* flush_to_ldisc() sees buffer data.
*/
smp_store_release(&buf->tail->commit, buf->tail->used);
queue_work(system_unbound_wq, &buf->work);
}
EXPORT_SYMBOL(tty_schedule_flip);
/**
* tty_prepare_flip_string - make room for characters
* @port: tty port
* @chars: return pointer for character write area
* @size: desired size
*
* Prepare a block of space in the buffer for data. Returns the length
* available and buffer pointer to the space which is now allocated and
* accounted for as ready for normal characters. This is used for drivers
* that need their own block copy routines into the buffer. There is no
* guarantee the buffer is a DMA target!
*/
int tty_prepare_flip_string(struct tty_port *port, unsigned char **chars,
size_t size)
{
int space = __tty_buffer_request_room(port, size, TTYB_NORMAL);
if (likely(space)) {
struct tty_buffer *tb = port->buf.tail;
*chars = char_buf_ptr(tb, tb->used);
if (~tb->flags & TTYB_NORMAL)
memset(flag_buf_ptr(tb, tb->used), TTY_NORMAL, space);
tb->used += space;
}
return space;
}
EXPORT_SYMBOL_GPL(tty_prepare_flip_string);
static int
receive_buf(struct tty_struct *tty, struct tty_buffer *head, int count)
{
struct tty_ldisc *disc = tty->ldisc;
unsigned char *p = char_buf_ptr(head, head->read);
char *f = NULL;
if (~head->flags & TTYB_NORMAL)
f = flag_buf_ptr(head, head->read);
if (disc->ops->receive_buf2)
count = disc->ops->receive_buf2(tty, p, f, count);
else {
count = min_t(int, count, tty->receive_room);
tty: Fix GPF in flush_to_ldisc() A line discipline which does not define a receive_buf() method can can cause a GPF if data is ever received [1]. Oddly, this was known to the author of n_tracesink in 2011, but never fixed. [1] GPF report BUG: unable to handle kernel NULL pointer dereference at (null) IP: [< (null)>] (null) PGD 3752d067 PUD 37a7b067 PMD 0 Oops: 0010 [#1] SMP KASAN Modules linked in: CPU: 2 PID: 148 Comm: kworker/u10:2 Not tainted 4.4.0-rc2+ #51 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011 Workqueue: events_unbound flush_to_ldisc task: ffff88006da94440 ti: ffff88006db60000 task.ti: ffff88006db60000 RIP: 0010:[<0000000000000000>] [< (null)>] (null) RSP: 0018:ffff88006db67b50 EFLAGS: 00010246 RAX: 0000000000000102 RBX: ffff88003ab32f88 RCX: 0000000000000102 RDX: 0000000000000000 RSI: ffff88003ab330a6 RDI: ffff88003aabd388 RBP: ffff88006db67c48 R08: ffff88003ab32f9c R09: ffff88003ab31fb0 R10: ffff88003ab32fa8 R11: 0000000000000000 R12: dffffc0000000000 R13: ffff88006db67c20 R14: ffffffff863df820 R15: ffff88003ab31fb8 FS: 0000000000000000(0000) GS:ffff88006dc00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 0000000000000000 CR3: 0000000037938000 CR4: 00000000000006e0 Stack: ffffffff829f46f1 ffff88006da94bf8 ffff88006da94bf8 0000000000000000 ffff88003ab31fb0 ffff88003aabd438 ffff88003ab31ff8 ffff88006430fd90 ffff88003ab32f9c ffffed0007557a87 1ffff1000db6cf78 ffff88003ab32078 Call Trace: [<ffffffff8127cf91>] process_one_work+0x8f1/0x17a0 kernel/workqueue.c:2030 [<ffffffff8127df14>] worker_thread+0xd4/0x1180 kernel/workqueue.c:2162 [<ffffffff8128faaf>] kthread+0x1cf/0x270 drivers/block/aoe/aoecmd.c:1302 [<ffffffff852a7c2f>] ret_from_fork+0x3f/0x70 arch/x86/entry/entry_64.S:468 Code: Bad RIP value. RIP [< (null)>] (null) RSP <ffff88006db67b50> CR2: 0000000000000000 ---[ end trace a587f8947e54d6ea ]--- Reported-by: Dmitry Vyukov <dvyukov@google.com> Cc: <stable@vger.kernel.org> Signed-off-by: Peter Hurley <peter@hurleysoftware.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2015-11-27 19:25:08 +00:00
if (count && disc->ops->receive_buf)
disc->ops->receive_buf(tty, p, f, count);
}
return count;
}
/**
* flush_to_ldisc
* @work: tty structure passed from work queue.
*
* This routine is called out of the software interrupt to flush data
* from the buffer chain to the line discipline.
*
* The receive_buf method is single threaded for each tty instance.
*
* Locking: takes buffer lock to ensure single-threaded flip buffer
* 'consumer'
*/
static void flush_to_ldisc(struct work_struct *work)
{
struct tty_port *port = container_of(work, struct tty_port, buf.work);
struct tty_bufhead *buf = &port->buf;
struct tty_struct *tty;
struct tty_ldisc *disc;
tty = READ_ONCE(port->itty);
if (tty == NULL)
return;
disc = tty_ldisc_ref(tty);
if (disc == NULL)
return;
mutex_lock(&buf->lock);
while (1) {
struct tty_buffer *head = buf->head;
struct tty_buffer *next;
int count;
/* Ldisc or user is trying to gain exclusive access */
if (atomic_read(&buf->priority))
break;
/* paired w/ release in __tty_buffer_request_room();
* ensures commit value read is not stale if the head
* is advancing to the next buffer
*/
next = smp_load_acquire(&head->next);
/* paired w/ release in __tty_buffer_request_room() or in
* tty_buffer_flush(); ensures we see the committed buffer data
*/
count = smp_load_acquire(&head->commit) - head->read;
if (!count) {
pty: Fix input race when closing A read() from a pty master may mistakenly indicate EOF (errno == -EIO) after the pty slave has closed, even though input data remains to be read. For example, pty slave | input worker | pty master | | | | n_tty_read() pty_write() | | input avail? no add data | | sleep schedule worker --->| | . |---> flush_to_ldisc() | . pty_close() | fill read buffer | . wait for worker | wakeup reader --->| . | read buffer full? |---> input avail ? yes |<--- yes - exit worker | copy 4096 bytes to user TTY_OTHER_CLOSED <---| |<--- kick worker | | **** New read() before worker starts **** | | n_tty_read() | | input avail? no | | TTY_OTHER_CLOSED? yes | | return -EIO Several conditions are required to trigger this race: 1. the ldisc read buffer must become full so the input worker exits 2. the read() count parameter must be >= 4096 so the ldisc read buffer is empty 3. the subsequent read() occurs before the kicked worker has processed more input However, the underlying cause of the race is that data is pipelined, while tty state is not; ie., data already written by the pty slave end is not yet visible to the pty master end, but state changes by the pty slave end are visible to the pty master end immediately. Pipeline the TTY_OTHER_CLOSED state through input worker to the reader. 1. Introduce TTY_OTHER_DONE which is set by the input worker when TTY_OTHER_CLOSED is set and either the input buffers are flushed or input processing has completed. Readers/polls are woken when TTY_OTHER_DONE is set. 2. Reader/poll checks TTY_OTHER_DONE instead of TTY_OTHER_CLOSED. 3. A new input worker is started from pty_close() after setting TTY_OTHER_CLOSED, which ensures the TTY_OTHER_DONE state will be set if the last input worker is already finished (or just about to exit). Remove tty_flush_to_ldisc(); no in-tree callers. Fixes: 52bce7f8d4fc ("pty, n_tty: Simplify input processing on final close") Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96311 BugLink: http://bugs.launchpad.net/bugs/1429756 Cc: <stable@vger.kernel.org> # 3.19+ Reported-by: Andy Whitcroft <apw@canonical.com> Reported-by: H.J. Lu <hjl.tools@gmail.com> Signed-off-by: Peter Hurley <peter@hurleysoftware.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2015-04-13 17:24:34 +00:00
if (next == NULL) {
check_other_closed(tty);
break;
pty: Fix input race when closing A read() from a pty master may mistakenly indicate EOF (errno == -EIO) after the pty slave has closed, even though input data remains to be read. For example, pty slave | input worker | pty master | | | | n_tty_read() pty_write() | | input avail? no add data | | sleep schedule worker --->| | . |---> flush_to_ldisc() | . pty_close() | fill read buffer | . wait for worker | wakeup reader --->| . | read buffer full? |---> input avail ? yes |<--- yes - exit worker | copy 4096 bytes to user TTY_OTHER_CLOSED <---| |<--- kick worker | | **** New read() before worker starts **** | | n_tty_read() | | input avail? no | | TTY_OTHER_CLOSED? yes | | return -EIO Several conditions are required to trigger this race: 1. the ldisc read buffer must become full so the input worker exits 2. the read() count parameter must be >= 4096 so the ldisc read buffer is empty 3. the subsequent read() occurs before the kicked worker has processed more input However, the underlying cause of the race is that data is pipelined, while tty state is not; ie., data already written by the pty slave end is not yet visible to the pty master end, but state changes by the pty slave end are visible to the pty master end immediately. Pipeline the TTY_OTHER_CLOSED state through input worker to the reader. 1. Introduce TTY_OTHER_DONE which is set by the input worker when TTY_OTHER_CLOSED is set and either the input buffers are flushed or input processing has completed. Readers/polls are woken when TTY_OTHER_DONE is set. 2. Reader/poll checks TTY_OTHER_DONE instead of TTY_OTHER_CLOSED. 3. A new input worker is started from pty_close() after setting TTY_OTHER_CLOSED, which ensures the TTY_OTHER_DONE state will be set if the last input worker is already finished (or just about to exit). Remove tty_flush_to_ldisc(); no in-tree callers. Fixes: 52bce7f8d4fc ("pty, n_tty: Simplify input processing on final close") Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96311 BugLink: http://bugs.launchpad.net/bugs/1429756 Cc: <stable@vger.kernel.org> # 3.19+ Reported-by: Andy Whitcroft <apw@canonical.com> Reported-by: H.J. Lu <hjl.tools@gmail.com> Signed-off-by: Peter Hurley <peter@hurleysoftware.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2015-04-13 17:24:34 +00:00
}
buf->head = next;
tty_buffer_free(port, head);
continue;
}
count = receive_buf(tty, head, count);
if (!count)
break;
head->read += count;
}
mutex_unlock(&buf->lock);
tty_ldisc_deref(disc);
}
/**
* tty_flip_buffer_push - terminal
* @port: tty port to push
*
tty: Fix low_latency BUG The user-settable knob, low_latency, has been the source of several BUG reports which stem from flush_to_ldisc() running in interrupt context. Since 3.12, which added several sleeping locks (termios_rwsem and buf->lock) to the input processing path, the frequency of these BUG reports has increased. Note that changes in 3.12 did not introduce this regression; sleeping locks were first added to the input processing path with the removal of the BKL from N_TTY in commit a88a69c91256418c5907c2f1f8a0ec0a36f9e6cc, 'n_tty: Fix loss of echoed characters and remove bkl from n_tty' and later in commit 38db89799bdf11625a831c5af33938dcb11908b6, 'tty: throttling race fix'. Since those changes, executing flush_to_ldisc() in interrupt_context (ie, low_latency set), is unsafe. However, since most devices do not validate if the low_latency setting is appropriate for the context (process or interrupt) in which they receive data, some reports are due to misconfiguration. Further, serial dma devices for which dma fails, resort to interrupt receiving as a backup without resetting low_latency. Historically, low_latency was used to force wake-up the reading process rather than wait for the next scheduler tick. The effect was to trim multiple milliseconds of latency from when the process would receive new data. Recent tests [1] have shown that the reading process now receives data with only 10's of microseconds latency without low_latency set. Remove the low_latency rx steering from tty_flip_buffer_push(); however, leave the knob as an optional hint to drivers that can tune their rx fifos and such like. Cleanup stale code comments regarding low_latency. [1] https://lkml.org/lkml/2014/2/20/434 "Yay.. thats an annoying historical pain in the butt gone." -- Alan Cox Reported-by: Beat Bolli <bbolli@ewanet.ch> Reported-by: Pavel Roskin <proski@gnu.org> Acked-by: David Sterba <dsterba@suse.cz> Cc: Grant Edwards <grant.b.edwards@gmail.com> Cc: Stanislaw Gruszka <sgruszka@redhat.com> Cc: Hal Murray <murray+fedora@ip-64-139-1-69.sjc.megapath.net> Cc: <stable@vger.kernel.org> # 3.12.x+ Signed-off-by: Peter Hurley <peter@hurleysoftware.com> Signed-off-by: Alan Cox <alan@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2014-02-22 12:31:21 +00:00
* Queue a push of the terminal flip buffers to the line discipline.
* Can be called from IRQ/atomic context.
*
* In the event of the queue being busy for flipping the work will be
* held off and retried later.
*/
void tty_flip_buffer_push(struct tty_port *port)
{
tty: Fix low_latency BUG The user-settable knob, low_latency, has been the source of several BUG reports which stem from flush_to_ldisc() running in interrupt context. Since 3.12, which added several sleeping locks (termios_rwsem and buf->lock) to the input processing path, the frequency of these BUG reports has increased. Note that changes in 3.12 did not introduce this regression; sleeping locks were first added to the input processing path with the removal of the BKL from N_TTY in commit a88a69c91256418c5907c2f1f8a0ec0a36f9e6cc, 'n_tty: Fix loss of echoed characters and remove bkl from n_tty' and later in commit 38db89799bdf11625a831c5af33938dcb11908b6, 'tty: throttling race fix'. Since those changes, executing flush_to_ldisc() in interrupt_context (ie, low_latency set), is unsafe. However, since most devices do not validate if the low_latency setting is appropriate for the context (process or interrupt) in which they receive data, some reports are due to misconfiguration. Further, serial dma devices for which dma fails, resort to interrupt receiving as a backup without resetting low_latency. Historically, low_latency was used to force wake-up the reading process rather than wait for the next scheduler tick. The effect was to trim multiple milliseconds of latency from when the process would receive new data. Recent tests [1] have shown that the reading process now receives data with only 10's of microseconds latency without low_latency set. Remove the low_latency rx steering from tty_flip_buffer_push(); however, leave the knob as an optional hint to drivers that can tune their rx fifos and such like. Cleanup stale code comments regarding low_latency. [1] https://lkml.org/lkml/2014/2/20/434 "Yay.. thats an annoying historical pain in the butt gone." -- Alan Cox Reported-by: Beat Bolli <bbolli@ewanet.ch> Reported-by: Pavel Roskin <proski@gnu.org> Acked-by: David Sterba <dsterba@suse.cz> Cc: Grant Edwards <grant.b.edwards@gmail.com> Cc: Stanislaw Gruszka <sgruszka@redhat.com> Cc: Hal Murray <murray+fedora@ip-64-139-1-69.sjc.megapath.net> Cc: <stable@vger.kernel.org> # 3.12.x+ Signed-off-by: Peter Hurley <peter@hurleysoftware.com> Signed-off-by: Alan Cox <alan@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2014-02-22 12:31:21 +00:00
tty_schedule_flip(port);
}
EXPORT_SYMBOL(tty_flip_buffer_push);
/**
* tty_buffer_init - prepare a tty buffer structure
* @tty: tty to initialise
*
* Set up the initial state of the buffer management for a tty device.
* Must be called before the other tty buffer functions are used.
*/
void tty_buffer_init(struct tty_port *port)
{
struct tty_bufhead *buf = &port->buf;
mutex_init(&buf->lock);
tty_buffer_reset(&buf->sentinel, 0);
buf->head = &buf->sentinel;
buf->tail = &buf->sentinel;
init_llist_head(&buf->free);
atomic_set(&buf->mem_used, 0);
atomic_set(&buf->priority, 0);
INIT_WORK(&buf->work, flush_to_ldisc);
buf->mem_limit = TTYB_DEFAULT_MEM_LIMIT;
}
/**
* tty_buffer_set_limit - change the tty buffer memory limit
* @port: tty port to change
*
* Change the tty buffer memory limit.
* Must be called before the other tty buffer functions are used.
*/
int tty_buffer_set_limit(struct tty_port *port, int limit)
{
if (limit < MIN_TTYB_SIZE)
return -EINVAL;
port->buf.mem_limit = limit;
return 0;
}
EXPORT_SYMBOL_GPL(tty_buffer_set_limit);
/* slave ptys can claim nested buffer lock when handling BRK and INTR */
void tty_buffer_set_lock_subclass(struct tty_port *port)
{
lockdep_set_subclass(&port->buf.lock, TTY_LOCK_SLAVE);
}
bool tty_buffer_restart_work(struct tty_port *port)
{
return queue_work(system_unbound_wq, &port->buf.work);
}
bool tty_buffer_cancel_work(struct tty_port *port)
{
return cancel_work_sync(&port->buf.work);
}