mirrors/linux-stable
1
0
Fork 0
mirror of https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git synced 2024-11-01 08:58:07 +00:00
Code Issues Releases Wiki Activity
linux-stable/drivers/scsi/osst.c

6102 lines
183 KiB
C
Raw Normal View History

Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
SCSI Tape Driver for Linux version 1.1 and newer. See the accompanying
file Documentation/scsi/st.txt for more information.
History:
OnStream SCSI Tape support (osst) cloned from st.c by
Willem Riede (osst@riede.org) Feb 2000
Fixes ... Kurt Garloff <garloff@suse.de> Mar 2000
Rewritten from Dwayne Forsyth's SCSI tape driver by Kai Makisara.
Contribution and ideas from several people including (in alphabetical
order) Klaus Ehrenfried, Wolfgang Denk, Steve Hirsch, Andreas Koppenh"ofer,
Michael Leodolter, Eyal Lebedinsky, J"org Weule, and Eric Youngdale.
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
Copyright 1992 - 2002 Kai Makisara / 2000 - 2006 Willem Riede
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
email osst@riede.org
$Header: /cvsroot/osst/Driver/osst.c,v 1.73 2005/01/01 21:13:34 wriede Exp $
Microscopic alterations - Rik Ling, 2000/12/21
Last st.c sync: Tue Oct 15 22:01:04 2002 by makisara
Some small formal changes - aeb, 950809
*/
static const char * cvsid = "$Id: osst.c,v 1.73 2005/01/01 21:13:34 wriede Exp $";
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static const char * osst_version = "0.99.4";
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/* The "failure to reconnect" firmware bug */
#define OSST_FW_NEED_POLL_MIN 10601 /*(107A)*/
#define OSST_FW_NEED_POLL_MAX 10704 /*(108D)*/
#define OSST_FW_NEED_POLL(x,d) ((x) >= OSST_FW_NEED_POLL_MIN && (x) <= OSST_FW_NEED_POLL_MAX && d->host->this_id != 7)
#include <linux/module.h>
#include <linux/fs.h>
#include <linux/kernel.h>
sched/headers: Prepare to move signal wakeup & sigpending methods from <linux/sched.h> into <linux/sched/signal.h> Fix up affected files that include this signal functionality via sched.h. Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-02-02 18:15:33 +00:00
#include <linux/sched/signal.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#include <linux/proc_fs.h>
#include <linux/mm.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#include <linux/init.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/mtio.h>
#include <linux/ioctl.h>
#include <linux/fcntl.h>
#include <linux/spinlock.h>
#include <linux/vmalloc.h>
#include <linux/blkdev.h>
#include <linux/moduleparam.h>
#include <linux/delay.h>
[PATCH] drivers/scsi/*: use time_after() and friends They deal with wrapping correctly and are nicer to read. Signed-off-by: Marcelo Feitoza Parisi <marcelo@feitoza.com.br> Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Cc: James Bottomley <James.Bottomley@steeleye.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-28 09:56:47 +00:00
#include <linux/jiffies.h>
scsi: autoconvert trivial BKL users to private mutex All these files use the big kernel lock in a trivial way to serialize their private file operations, typically resulting from an earlier semi-automatic pushdown from VFS. None of these drivers appears to want to lock against other code, and they all use the BKL as the top-level lock in their file operations, meaning that there is no lock-order inversion problem. Consequently, we can remove the BKL completely, replacing it with a per-file mutex in every case. Using a scripted approach means we can avoid typos. file=$1 name=$2 if grep -q lock_kernel ${file} ; then if grep -q 'include.*linux.mutex.h' ${file} ; then sed -i '/include.*<linux\/smp_lock.h>/d' ${file} else sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file} fi sed -i ${file} \ -e "/^#include.*linux.mutex.h/,$ { 1,/^\(static\|int\|long\)/ { /^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex); } }" \ -e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \ -e '/[ ]*cycle_kernel_lock();/d' else sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \ -e '/cycle_kernel_lock()/d' fi Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: linux-scsi@vger.kernel.org Cc: "James E.J. Bottomley" <James.Bottomley@suse.de>
2010-06-02 12:28:52 +00:00
#include <linux/mutex.h>
Replace <asm/uaccess.h> with <linux/uaccess.h> globally This was entirely automated, using the script by Al: PATT='^[[:blank:]]*#[[:blank:]]*include[[:blank:]]*<asm/uaccess.h>' sed -i -e "s!$PATT!#include <linux/uaccess.h>!" \ $(git grep -l "$PATT"|grep -v ^include/linux/uaccess.h) to do the replacement at the end of the merge window. Requested-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-24 19:46:01 +00:00
#include <linux/uaccess.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#include <asm/dma.h>
/* The driver prints some debugging information on the console if DEBUG
is defined and non-zero. */
#define DEBUG 0
/* The message level for the debug messages is currently set to KERN_NOTICE
so that people can easily see the messages. Later when the debugging messages
in the drivers are more widely classified, this may be changed to KERN_DEBUG. */
#define OSST_DEB_MSG KERN_NOTICE
#include <scsi/scsi.h>
#include <scsi/scsi_dbg.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_driver.h>
#include <scsi/scsi_eh.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_ioctl.h>
#define ST_KILOBYTE 1024
#include "st.h"
#include "osst.h"
#include "osst_options.h"
#include "osst_detect.h"
scsi: autoconvert trivial BKL users to private mutex All these files use the big kernel lock in a trivial way to serialize their private file operations, typically resulting from an earlier semi-automatic pushdown from VFS. None of these drivers appears to want to lock against other code, and they all use the BKL as the top-level lock in their file operations, meaning that there is no lock-order inversion problem. Consequently, we can remove the BKL completely, replacing it with a per-file mutex in every case. Using a scripted approach means we can avoid typos. file=$1 name=$2 if grep -q lock_kernel ${file} ; then if grep -q 'include.*linux.mutex.h' ${file} ; then sed -i '/include.*<linux\/smp_lock.h>/d' ${file} else sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file} fi sed -i ${file} \ -e "/^#include.*linux.mutex.h/,$ { 1,/^\(static\|int\|long\)/ { /^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex); } }" \ -e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \ -e '/[ ]*cycle_kernel_lock();/d' else sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \ -e '/cycle_kernel_lock()/d' fi Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: linux-scsi@vger.kernel.org Cc: "James E.J. Bottomley" <James.Bottomley@suse.de>
2010-06-02 12:28:52 +00:00
static DEFINE_MUTEX(osst_int_mutex);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
static int max_dev = 0;
static int write_threshold_kbs = 0;
static int max_sg_segs = 0;
#ifdef MODULE
MODULE_AUTHOR("Willem Riede");
MODULE_DESCRIPTION("OnStream {DI-|FW-|SC-|USB}{30|50} Tape Driver");
MODULE_LICENSE("GPL");
[SCSI] MODULE_ALIAS_{BLOCK,CHAR}DEV_MAJOR for drivers/scsi Add device-major aliases in drivers/scsi, allowing kmod autoload: MODULE_ALIAS_CHARDEV_MAJOR(SCSI_CHANGER_MAJOR) MODULE_ALIAS_CHARDEV_MAJOR(OSST_MAJOR) MODULE_ALIAS_CHARDEV_MAJOR(SCSI_TAPE_MAJOR) MODULE_ALIAS_BLOCKDEV_MAJOR(SCSI_CDROM_MAJOR) MODULE_ALIAS_BLOCKDEV_MAJOR(SCSI_DISKN_MAJOR) Signed-off-by: Andrew Morton <akpm@osdl.org> Acked-by: Kai Makisara <kai.makisara@kolumbus.fi> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-03-08 08:14:20 +00:00
MODULE_ALIAS_CHARDEV_MAJOR(OSST_MAJOR);
[SCSI] modalias for scsi devices The following patch adds support for sysfs/uevent modalias attribute for scsi devices (like disks, tapes, cdroms etc), based on whatever current sd.c, sr.c, st.c and osst.c drivers supports. The modalias format is like this: scsi:type-0x04 (for TYPE_WORM, handled by sr.c now). Several comments. o This hexadecimal type value is because all TYPE_XXX constants in include/scsi/scsi.h are given in hex, but __stringify() will not convert them to decimal (so it will NOT be scsi:type-4). Since it does not really matter in which format it is, while both modalias in module and modalias attribute match each other, I descided to go for that 0x%02x format (and added a comment in include/scsi/scsi.h to keep them that way), instead of changing them all to decimal. o There was no .uevent routine for SCSI bus. It might be a good idea to add some more ueven environment variables in there. o osst.c driver handles tapes too, like st.c, but only SOME tapes. With this setup, hotplug scripts (or whatever is used by the user) will try to load both st and osst modules for all SCSI tapes found, because both modules have scsi:type-0x01 alias). It is not harmful, but one extra module is no good either. It is possible to solve this, by exporting more info in modalias attribute, including vendor and device identification strings, so that modalias becomes something like scsi:type-0x12:vendor-Adaptec LTD:device-OnStream Tape Drive and having that, match for all 3 attributes, not only device type. But oh well, vendor and device strings may be large, and they do contain spaces and whatnot. So I left them for now, awaiting for comments first. Signed-off-by: Michael Tokarev <mjt@tls.msk.ru> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-27 12:02:37 +00:00
MODULE_ALIAS_SCSI_DEVICE(TYPE_TAPE);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
module_param(max_dev, int, 0444);
MODULE_PARM_DESC(max_dev, "Maximum number of OnStream Tape Drives to attach (4)");
module_param(write_threshold_kbs, int, 0644);
MODULE_PARM_DESC(write_threshold_kbs, "Asynchronous write threshold (KB; 32)");
module_param(max_sg_segs, int, 0644);
MODULE_PARM_DESC(max_sg_segs, "Maximum number of scatter/gather segments to use (9)");
#else
static struct osst_dev_parm {
char *name;
int *val;
} parms[] __initdata = {
{ "max_dev", &max_dev },
{ "write_threshold_kbs", &write_threshold_kbs },
{ "max_sg_segs", &max_sg_segs }
};
#endif
/* Some default definitions have been moved to osst_options.h */
#define OSST_BUFFER_SIZE (OSST_BUFFER_BLOCKS * ST_KILOBYTE)
#define OSST_WRITE_THRESHOLD (OSST_WRITE_THRESHOLD_BLOCKS * ST_KILOBYTE)
/* The buffer size should fit into the 24 bits for length in the
6-byte SCSI read and write commands. */
#if OSST_BUFFER_SIZE >= (2 << 24 - 1)
#error "Buffer size should not exceed (2 << 24 - 1) bytes!"
#endif
#if DEBUG
static int debugging = 1;
/* uncomment define below to test error recovery */
// #define OSST_INJECT_ERRORS 1
#endif
/* Do not retry! The drive firmware already retries when appropriate,
and when it tries to tell us something, we had better listen... */
#define MAX_RETRIES 0
#define NO_TAPE NOT_READY
#define OSST_WAIT_POSITION_COMPLETE (HZ > 200 ? HZ / 200 : 1)
#define OSST_WAIT_WRITE_COMPLETE (HZ / 12)
#define OSST_WAIT_LONG_WRITE_COMPLETE (HZ / 2)
#define OSST_TIMEOUT (200 * HZ)
#define OSST_LONG_TIMEOUT (1800 * HZ)
#define TAPE_NR(x) (iminor(x) & ~(-1 << ST_MODE_SHIFT))
#define TAPE_MODE(x) ((iminor(x) & ST_MODE_MASK) >> ST_MODE_SHIFT)
#define TAPE_REWIND(x) ((iminor(x) & 0x80) == 0)
#define TAPE_IS_RAW(x) (TAPE_MODE(x) & (ST_NBR_MODES >> 1))
/* Internal ioctl to set both density (uppermost 8 bits) and blocksize (lower
24 bits) */
#define SET_DENS_AND_BLK 0x10001
static int osst_buffer_size = OSST_BUFFER_SIZE;
static int osst_write_threshold = OSST_WRITE_THRESHOLD;
static int osst_max_sg_segs = OSST_MAX_SG;
static int osst_max_dev = OSST_MAX_TAPES;
static int osst_nr_dev;
static struct osst_tape **os_scsi_tapes = NULL;
static DEFINE_RWLOCK(os_scsi_tapes_lock);
static int modes_defined = 0;
static struct osst_buffer *new_tape_buffer(int, int, int);
static int enlarge_buffer(struct osst_buffer *, int);
static void normalize_buffer(struct osst_buffer *);
static int append_to_buffer(const char __user *, struct osst_buffer *, int);
static int from_buffer(struct osst_buffer *, char __user *, int);
static int osst_zero_buffer_tail(struct osst_buffer *);
static int osst_copy_to_buffer(struct osst_buffer *, unsigned char *);
static int osst_copy_from_buffer(struct osst_buffer *, unsigned char *);
static int osst_probe(struct device *);
static int osst_remove(struct device *);
static struct scsi_driver osst_template = {
.gendrv = {
.name = "osst",
scsi: remove scsi_driver owner field The driver core driver structure has grown an owner field and now requires it to be set for all modular drivers. Set it up for all scsi_driver instances and get rid of the now superflous scsi_driver owner field. Signed-off-by: Christoph Hellwig <hch@lst.de> Reported-by: Shane M Seymour <shane.seymour@hp.com> Reviewed-by: Ewan D. Milne <emilne@redhat.com Reviewed-by: Hannes Reinecke <hare@suse.de>
2014-11-12 17:34:51 +00:00
.owner = THIS_MODULE,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
.probe = osst_probe,
.remove = osst_remove,
}
};
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_int_ioctl(struct osst_tape *STp, struct osst_request ** aSRpnt,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned int cmd_in, unsigned long arg);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_set_frame_position(struct osst_tape *STp, struct osst_request ** aSRpnt, int frame, int skip);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_get_frame_position(struct osst_tape *STp, struct osst_request ** aSRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_flush_write_buffer(struct osst_tape *STp, struct osst_request ** aSRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_write_error_recovery(struct osst_tape * STp, struct osst_request ** aSRpnt, int pending);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
static inline char *tape_name(struct osst_tape *tape)
{
return tape->drive->disk_name;
}
/* Routines that handle the interaction with mid-layer SCSI routines */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
/* Normalize Sense */
static void osst_analyze_sense(struct osst_request *SRpnt, struct st_cmdstatus *s)
{
const u8 *ucp;
const u8 *sense = SRpnt->sense;
s->have_sense = scsi_normalize_sense(SRpnt->sense,
SCSI_SENSE_BUFFERSIZE, &s->sense_hdr);
s->flags = 0;
if (s->have_sense) {
s->deferred = 0;
s->remainder_valid =
scsi_get_sense_info_fld(sense, SCSI_SENSE_BUFFERSIZE, &s->uremainder64);
switch (sense[0] & 0x7f) {
case 0x71:
s->deferred = 1;
case 0x70:
s->fixed_format = 1;
s->flags = sense[2] & 0xe0;
break;
case 0x73:
s->deferred = 1;
case 0x72:
s->fixed_format = 0;
ucp = scsi_sense_desc_find(sense, SCSI_SENSE_BUFFERSIZE, 4);
s->flags = ucp ? (ucp[3] & 0xe0) : 0;
break;
}
}
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/* Convert the result to success code */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_chk_result(struct osst_tape * STp, struct osst_request * SRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
char *name = tape_name(STp);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
int result = SRpnt->result;
u8 * sense = SRpnt->sense, scode;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
const char *stp;
#endif
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct st_cmdstatus *cmdstatp;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (!result)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return 0;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
cmdstatp = &STp->buffer->cmdstat;
osst_analyze_sense(SRpnt, cmdstatp);
if (cmdstatp->have_sense)
scode = STp->buffer->cmdstat.sense_hdr.sense_key;
else
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
scode = 0;
#if DEBUG
if (debugging) {
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
printk(OSST_DEB_MSG "%s:D: Error: %x, cmd: %x %x %x %x %x %x\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
name, result,
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
SRpnt->cmd[0], SRpnt->cmd[1], SRpnt->cmd[2],
SRpnt->cmd[3], SRpnt->cmd[4], SRpnt->cmd[5]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (scode) printk(OSST_DEB_MSG "%s:D: Sense: %02x, ASC: %02x, ASCQ: %02x\n",
scsi: use sdev as argument for sense code printing We should be using the standard dev_printk() variants for sense code printing. [hch: remove __scsi_print_sense call in xen-scsiback, Acked by Juergen] [hch: folded bracing fix from Dan Carpenter] Signed-off-by: Hannes Reinecke <hare@suse.de> Reviewed-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-10-24 12:26:45 +00:00
name, scode, sense[12], sense[13]);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (cmdstatp->have_sense)
scsi: use sdev as argument for sense code printing We should be using the standard dev_printk() variants for sense code printing. [hch: remove __scsi_print_sense call in xen-scsiback, Acked by Juergen] [hch: folded bracing fix from Dan Carpenter] Signed-off-by: Hannes Reinecke <hare@suse.de> Reviewed-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-10-24 12:26:45 +00:00
__scsi_print_sense(STp->device, name,
SRpnt->sense, SCSI_SENSE_BUFFERSIZE);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
else
#endif
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (cmdstatp->have_sense && (
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
scode != NO_SENSE &&
scode != RECOVERED_ERROR &&
/* scode != UNIT_ATTENTION && */
scode != BLANK_CHECK &&
scode != VOLUME_OVERFLOW &&
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
SRpnt->cmd[0] != MODE_SENSE &&
SRpnt->cmd[0] != TEST_UNIT_READY)) { /* Abnormal conditions for tape */
if (cmdstatp->have_sense) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_WARNING "%s:W: Command with sense data:\n", name);
scsi: use sdev as argument for sense code printing We should be using the standard dev_printk() variants for sense code printing. [hch: remove __scsi_print_sense call in xen-scsiback, Acked by Juergen] [hch: folded bracing fix from Dan Carpenter] Signed-off-by: Hannes Reinecke <hare@suse.de> Reviewed-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-10-24 12:26:45 +00:00
__scsi_print_sense(STp->device, name,
SRpnt->sense, SCSI_SENSE_BUFFERSIZE);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
else {
static int notyetprinted = 1;
printk(KERN_WARNING
[SCSI] osst: Remove SUGGEST flags Fix up remaining bit of SUGGEST flag removal done by patch commit 0f10274300857d98ea5ea4c800c561a9ad9ac89f Author: Martin K. Petersen <martin.petersen@oracle.com> Date: Sun Jan 4 03:14:11 2009 -0500 [SCSI] Remove SUGGEST flags Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2009-01-25 21:09:42 +00:00
"%s:W: Warning %x (driver bt 0x%x, host bt 0x%x).\n",
name, result, driver_byte(result),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
host_byte(result));
if (notyetprinted) {
notyetprinted = 0;
printk(KERN_INFO
"%s:I: This warning may be caused by your scsi controller,\n", name);
printk(KERN_INFO
"%s:I: it has been reported with some Buslogic cards.\n", name);
}
}
}
STp->pos_unknown |= STp->device->was_reset;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (cmdstatp->have_sense && scode == RECOVERED_ERROR) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->recover_count++;
STp->recover_erreg++;
#if DEBUG
if (debugging) {
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt->cmd[0] == READ_6)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
stp = "read";
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
else if (SRpnt->cmd[0] == WRITE_6)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
stp = "write";
else
stp = "ioctl";
printk(OSST_DEB_MSG "%s:D: Recovered %s error (%d).\n", name, stp,
STp->recover_count);
}
#endif
if ((sense[2] & 0xe0) == 0)
return 0;
}
return (-EIO);
}
/* Wakeup from interrupt */
block: introduce new block status code type Currently we use nornal Linux errno values in the block layer, and while we accept any error a few have overloaded magic meanings. This patch instead introduces a new blk_status_t value that holds block layer specific status codes and explicitly explains their meaning. Helpers to convert from and to the previous special meanings are provided for now, but I suspect we want to get rid of them in the long run - those drivers that have a errno input (e.g. networking) usually get errnos that don't know about the special block layer overloads, and similarly returning them to userspace will usually return somethings that strictly speaking isn't correct for file system operations, but that's left as an exercise for later. For now the set of errors is a very limited set that closely corresponds to the previous overloaded errno values, but there is some low hanging fruite to improve it. blk_status_t (ab)uses the sparse __bitwise annotations to allow for sparse typechecking, so that we can easily catch places passing the wrong values. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2017-06-03 07:38:04 +00:00
static void osst_end_async(struct request *req, blk_status_t status)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
block: split scsi_request out of struct request And require all drivers that want to support BLOCK_PC to allocate it as the first thing of their private data. To support this the legacy IDE and BSG code is switched to set cmd_size on their queues to let the block layer allocate the additional space. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2017-01-27 08:46:29 +00:00
struct scsi_request *rq = scsi_req(req);
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
struct osst_request *SRpnt = req->end_io_data;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_tape *STp = SRpnt->stp;
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
struct rq_map_data *mdata = &SRpnt->stp->buffer->map_data;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
scsi: introduce a result field in struct scsi_request This passes on the scsi_cmnd result field to users of passthrough requests. Currently we abuse req->errors for this purpose, but that field will go away in its current form. Note that the old IDE code abuses the errors field in very creative ways and stores all kinds of different values in it. I didn't dare to touch this magic, so the abuses are brought forward 1:1. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com> Reviewed-by: Bart Van Assche <Bart.VanAssche@sandisk.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-20 14:03:01 +00:00
STp->buffer->cmdstat.midlevel_result = SRpnt->result = rq->result;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
STp->write_pending = 0;
#endif
block: split scsi_request out of struct request And require all drivers that want to support BLOCK_PC to allocate it as the first thing of their private data. To support this the legacy IDE and BSG code is switched to set cmd_size on their queues to let the block layer allocate the additional space. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2017-01-27 08:46:29 +00:00
if (rq->sense_len)
memcpy(SRpnt->sense, rq->sense, SCSI_SENSE_BUFFERSIZE);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt->waiting)
complete(SRpnt->waiting);
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
if (SRpnt->bio) {
kfree(mdata->pages);
blk_rq_unmap_user(SRpnt->bio);
}
block: remove __blk_put_request() Now there's no difference between blk_put_request() and __blk_put_request() anymore, get rid of the underscore version and convert the few callers. Reviewed-by: Hannes Reinecke <hare@suse.com> Tested-by: Ming Lei <ming.lei@redhat.com> Reviewed-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-10-24 19:52:28 +00:00
blk_put_request(req);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
}
/* osst_request memory management */
static struct osst_request *osst_allocate_request(void)
{
return kzalloc(sizeof(struct osst_request), GFP_KERNEL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static void osst_release_request(struct osst_request *streq)
{
kfree(streq);
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
static int osst_execute(struct osst_request *SRpnt, const unsigned char *cmd,
int cmd_len, int data_direction, void *buffer, unsigned bufflen,
int use_sg, int timeout, int retries)
{
struct request *req;
block: split scsi_request out of struct request And require all drivers that want to support BLOCK_PC to allocate it as the first thing of their private data. To support this the legacy IDE and BSG code is switched to set cmd_size on their queues to let the block layer allocate the additional space. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2017-01-27 08:46:29 +00:00
struct scsi_request *rq;
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
struct page **pages = NULL;
struct rq_map_data *mdata = &SRpnt->stp->buffer->map_data;
int err = 0;
int write = (data_direction == DMA_TO_DEVICE);
block: fold cmd_type into the REQ_OP_ space Instead of keeping two levels of indirection for requests types, fold it all into the operations. The little caveat here is that previously cmd_type only applied to struct request, while the request and bio op fields were set to plain REQ_OP_READ/WRITE even for passthrough operations. Instead this patch adds new REQ_OP_* for SCSI passthrough and driver private requests, althought it has to add two for each so that we can communicate the data in/out nature of the request. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2017-01-31 15:57:31 +00:00
req = blk_get_request(SRpnt->stp->device->request_queue,
block: sanitize blk_get_request calling conventions Switch everyone to blk_get_request_flags, and then rename blk_get_request_flags to blk_get_request. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-05-09 07:54:05 +00:00
write ? REQ_OP_SCSI_OUT : REQ_OP_SCSI_IN, 0);
block,scsi: fixup blk_get_request dead queue scenarios The blk_get_request function may fail in low-memory conditions or during device removal (even if __GFP_WAIT is set). To distinguish between these errors, modify the blk_get_request call stack to return the appropriate ERR_PTR. Verify that all callers check the return status and consider IS_ERR instead of a simple NULL pointer check. For consistency, make a similar change to the blk_mq_alloc_request leg of blk_get_request. It may fail if the queue is dead, or the caller was unwilling to wait. Signed-off-by: Joe Lawrence <joe.lawrence@stratus.com> Acked-by: Jiri Kosina <jkosina@suse.cz> [for pktdvd] Acked-by: Boaz Harrosh <bharrosh@panasas.com> [for osd] Reviewed-by: Jeff Moyer <jmoyer@redhat.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2014-08-28 14:15:21 +00:00
if (IS_ERR(req))
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
return DRIVER_ERROR << 24;
block: split scsi_request out of struct request And require all drivers that want to support BLOCK_PC to allocate it as the first thing of their private data. To support this the legacy IDE and BSG code is switched to set cmd_size on their queues to let the block layer allocate the additional space. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2017-01-27 08:46:29 +00:00
rq = scsi_req(req);
block: split out request-only flags into a new namespace A lot of the REQ_* flags are only used on struct requests, and only of use to the block layer and a few drivers that dig into struct request internals. This patch adds a new req_flags_t rq_flags field to struct request for them, and thus dramatically shrinks the number of common requests. It also removes the unfortunate situation where we have to fit the fields from the same enum into 32 bits for struct bio and 64 bits for struct request. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Shaun Tancheff <shaun.tancheff@seagate.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2016-10-20 13:12:13 +00:00
req->rq_flags |= RQF_QUIET;
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
SRpnt->bio = NULL;
if (use_sg) {
struct scatterlist *sg, *sgl = (struct scatterlist *)buffer;
int i;
treewide: kzalloc() -> kcalloc() The kzalloc() function has a 2-factor argument form, kcalloc(). This patch replaces cases of: kzalloc(a * b, gfp) with: kcalloc(a * b, gfp) as well as handling cases of: kzalloc(a * b * c, gfp) with: kzalloc(array3_size(a, b, c), gfp) as it's slightly less ugly than: kzalloc_array(array_size(a, b), c, gfp) This does, however, attempt to ignore constant size factors like: kzalloc(4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ type TYPE; expression THING, E; @@ ( kzalloc( - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | kzalloc( - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression COUNT; typedef u8; typedef __u8; @@ ( kzalloc( - sizeof(u8) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(__u8) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(char) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(unsigned char) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(u8) * COUNT + COUNT , ...) | kzalloc( - sizeof(__u8) * COUNT + COUNT , ...) | kzalloc( - sizeof(char) * COUNT + COUNT , ...) | kzalloc( - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( - kzalloc + kcalloc ( - sizeof(TYPE) * (COUNT_ID) + COUNT_ID, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * COUNT_ID + COUNT_ID, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * (COUNT_CONST) + COUNT_CONST, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * COUNT_CONST + COUNT_CONST, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * (COUNT_ID) + COUNT_ID, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * COUNT_ID + COUNT_ID, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * (COUNT_CONST) + COUNT_CONST, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * COUNT_CONST + COUNT_CONST, sizeof(THING) , ...) ) // 2-factor product, only identifiers. @@ identifier SIZE, COUNT; @@ - kzalloc + kcalloc ( - SIZE * COUNT + COUNT, SIZE , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( kzalloc( - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kzalloc( - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kzalloc( - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kzalloc( - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( kzalloc( - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kzalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kzalloc( - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kzalloc( - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kzalloc( - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | kzalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ identifier STRIDE, SIZE, COUNT; @@ ( kzalloc( - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products, // when they're not all constants... @@ expression E1, E2, E3; constant C1, C2, C3; @@ ( kzalloc(C1 * C2 * C3, ...) | kzalloc( - (E1) * E2 * E3 + array3_size(E1, E2, E3) , ...) | kzalloc( - (E1) * (E2) * E3 + array3_size(E1, E2, E3) , ...) | kzalloc( - (E1) * (E2) * (E3) + array3_size(E1, E2, E3) , ...) | kzalloc( - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants, // keeping sizeof() as the second factor argument. @@ expression THING, E1, E2; type TYPE; constant C1, C2, C3; @@ ( kzalloc(sizeof(THING) * C2, ...) | kzalloc(sizeof(TYPE) * C2, ...) | kzalloc(C1 * C2 * C3, ...) | kzalloc(C1 * C2, ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * (E2) + E2, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * E2 + E2, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * (E2) + E2, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * E2 + E2, sizeof(THING) , ...) | - kzalloc + kcalloc ( - (E1) * E2 + E1, E2 , ...) | - kzalloc + kcalloc ( - (E1) * (E2) + E1, E2 , ...) | - kzalloc + kcalloc ( - E1 * E2 + E1, E2 , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-12 21:03:40 +00:00
pages = kcalloc(use_sg, sizeof(struct page *), GFP_KERNEL);
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
if (!pages)
goto free_req;
for_each_sg(sgl, sg, use_sg, i)
pages[i] = sg_page(sg);
mdata->null_mapped = 1;
mdata->page_order = get_order(sgl[0].length);
mdata->nr_entries =
DIV_ROUND_UP(bufflen, PAGE_SIZE << mdata->page_order);
mdata->offset = 0;
err = blk_rq_map_user(req->q, req, mdata, NULL, bufflen, GFP_KERNEL);
if (err) {
kfree(pages);
goto free_req;
}
SRpnt->bio = req->bio;
mdata->pages = pages;
} else if (bufflen) {
err = blk_rq_map_kern(req->q, req, buffer, bufflen, GFP_KERNEL);
if (err)
goto free_req;
}
block: split scsi_request out of struct request And require all drivers that want to support BLOCK_PC to allocate it as the first thing of their private data. To support this the legacy IDE and BSG code is switched to set cmd_size on their queues to let the block layer allocate the additional space. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2017-01-27 08:46:29 +00:00
rq->cmd_len = cmd_len;
memset(rq->cmd, 0, BLK_MAX_CDB); /* ATAPI hates garbage after CDB */
memcpy(rq->cmd, cmd, rq->cmd_len);
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
req->timeout = timeout;
block, scsi: move the retries field to struct scsi_request Instead of bloating the generic struct request with it. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-05 17:18:12 +00:00
rq->retries = retries;
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
req->end_io_data = SRpnt;
blk_execute_rq_nowait(req->q, NULL, req, 1, osst_end_async);
return 0;
free_req:
blk_put_request(req);
return DRIVER_ERROR << 24;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/* Do the scsi command. Waits until command performed if do_wait is true.
Otherwise osst_write_behind_check() is used to check that the command
has finished. */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static struct osst_request * osst_do_scsi(struct osst_request *SRpnt, struct osst_tape *STp,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned char *cmd, int bytes, int direction, int timeout, int retries, int do_wait)
{
unsigned char *bp;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
unsigned short use_sg;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#ifdef OSST_INJECT_ERRORS
static int inject = 0;
static int repeat = 0;
#endif
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct completion *waiting;
/* if async, make sure there's no command outstanding */
if (!do_wait && ((STp->buffer)->last_SRpnt)) {
printk(KERN_ERR "%s: Async command already active.\n",
tape_name(STp));
if (signal_pending(current))
(STp->buffer)->syscall_result = (-EINTR);
else
(STp->buffer)->syscall_result = (-EBUSY);
return NULL;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (SRpnt == NULL) {
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
SRpnt = osst_allocate_request();
if (SRpnt == NULL) {
printk(KERN_ERR "%s: Can't allocate SCSI request.\n",
tape_name(STp));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (signal_pending(current))
(STp->buffer)->syscall_result = (-EINTR);
else
(STp->buffer)->syscall_result = (-EBUSY);
return NULL;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
SRpnt->stp = STp;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
/* If async IO, set last_SRpnt. This ptr tells write_behind_check
which IO is outstanding. It's nulled out when the IO completes. */
if (!do_wait)
(STp->buffer)->last_SRpnt = SRpnt;
waiting = &STp->wait;
init_completion(waiting);
SRpnt->waiting = waiting;
use_sg = (bytes > STp->buffer->sg[0].length) ? STp->buffer->use_sg : 0;
if (use_sg) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
bp = (char *)&(STp->buffer->sg[0]);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (STp->buffer->sg_segs < use_sg)
use_sg = STp->buffer->sg_segs;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
else
bp = (STp->buffer)->b_data;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
memcpy(SRpnt->cmd, cmd, sizeof(SRpnt->cmd));
STp->buffer->cmdstat.have_sense = 0;
STp->buffer->syscall_result = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: replace scsi_execute_async with the block layer API This replaces scsi_execute_async with the block layer API. st does the same thing so it might make sense to have something like libst (there are other things that os and osst can share). Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-12-13 15:55:18 +00:00
if (osst_execute(SRpnt, cmd, COMMAND_SIZE(cmd[0]), direction, bp, bytes,
use_sg, timeout, retries))
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
/* could not allocate the buffer or request was too large */
(STp->buffer)->syscall_result = (-EBUSY);
else if (do_wait) {
wait_for_completion(waiting);
SRpnt->waiting = NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->buffer->syscall_result = osst_chk_result(STp, SRpnt);
#ifdef OSST_INJECT_ERRORS
if (STp->buffer->syscall_result == 0 &&
cmd[0] == READ_6 &&
cmd[4] &&
( (++ inject % 83) == 29 ||
(STp->first_frame_position == 240
/* or STp->read_error_frame to fail again on the block calculated above */ &&
++repeat < 3))) {
printk(OSST_DEB_MSG "%s:D: Injecting read error\n", tape_name(STp));
STp->buffer->last_result_fatal = 1;
}
#endif
}
return SRpnt;
}
/* Handle the write-behind checking (downs the semaphore) */
static void osst_write_behind_check(struct osst_tape *STp)
{
struct osst_buffer * STbuffer;
STbuffer = STp->buffer;
#if DEBUG
if (STp->write_pending)
STp->nbr_waits++;
else
STp->nbr_finished++;
#endif
wait_for_completion(&(STp->wait));
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
STp->buffer->last_SRpnt->waiting = NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->buffer->syscall_result = osst_chk_result(STp, STp->buffer->last_SRpnt);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (STp->buffer->syscall_result)
STp->buffer->syscall_result =
osst_write_error_recovery(STp, &(STp->buffer->last_SRpnt), 1);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
else
STp->first_frame_position++;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
osst_release_request(STp->buffer->last_SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (STbuffer->writing < STbuffer->buffer_bytes)
printk(KERN_WARNING "osst :A: write_behind_check: something left in buffer!\n");
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
STbuffer->last_SRpnt = NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STbuffer->buffer_bytes -= STbuffer->writing;
STbuffer->writing = 0;
return;
}
/* Onstream specific Routines */
/*
* Initialize the OnStream AUX
*/
static void osst_init_aux(struct osst_tape * STp, int frame_type, int frame_seq_number,
int logical_blk_num, int blk_sz, int blk_cnt)
{
os_aux_t *aux = STp->buffer->aux;
os_partition_t *par = &aux->partition;
os_dat_t *dat = &aux->dat;
if (STp->raw) return;
memset(aux, 0, sizeof(*aux));
aux->format_id = htonl(0);
memcpy(aux->application_sig, "LIN4", 4);
aux->hdwr = htonl(0);
aux->frame_type = frame_type;
switch (frame_type) {
case OS_FRAME_TYPE_HEADER:
aux->update_frame_cntr = htonl(STp->update_frame_cntr);
par->partition_num = OS_CONFIG_PARTITION;
par->par_desc_ver = OS_PARTITION_VERSION;
par->wrt_pass_cntr = htons(0xffff);
/* 0-4 = reserved, 5-9 = header, 2990-2994 = header, 2995-2999 = reserved */
par->first_frame_ppos = htonl(0);
par->last_frame_ppos = htonl(0xbb7);
aux->frame_seq_num = htonl(0);
aux->logical_blk_num_high = htonl(0);
aux->logical_blk_num = htonl(0);
aux->next_mark_ppos = htonl(STp->first_mark_ppos);
break;
case OS_FRAME_TYPE_DATA:
case OS_FRAME_TYPE_MARKER:
dat->dat_sz = 8;
dat->reserved1 = 0;
dat->entry_cnt = 1;
dat->reserved3 = 0;
dat->dat_list[0].blk_sz = htonl(blk_sz);
dat->dat_list[0].blk_cnt = htons(blk_cnt);
dat->dat_list[0].flags = frame_type==OS_FRAME_TYPE_MARKER?
OS_DAT_FLAGS_MARK:OS_DAT_FLAGS_DATA;
dat->dat_list[0].reserved = 0;
case OS_FRAME_TYPE_EOD:
aux->update_frame_cntr = htonl(0);
par->partition_num = OS_DATA_PARTITION;
par->par_desc_ver = OS_PARTITION_VERSION;
par->wrt_pass_cntr = htons(STp->wrt_pass_cntr);
par->first_frame_ppos = htonl(STp->first_data_ppos);
par->last_frame_ppos = htonl(STp->capacity);
aux->frame_seq_num = htonl(frame_seq_number);
aux->logical_blk_num_high = htonl(0);
aux->logical_blk_num = htonl(logical_blk_num);
break;
default: ; /* probably FILL */
}
[PATCH] osst endianness annotations Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-09 16:39:45 +00:00
aux->filemark_cnt = htonl(STp->filemark_cnt);
aux->phys_fm = htonl(0xffffffff);
aux->last_mark_ppos = htonl(STp->last_mark_ppos);
aux->last_mark_lbn = htonl(STp->last_mark_lbn);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
/*
* Verify that we have the correct tape frame
*/
static int osst_verify_frame(struct osst_tape * STp, int frame_seq_number, int quiet)
{
char * name = tape_name(STp);
os_aux_t * aux = STp->buffer->aux;
os_partition_t * par = &(aux->partition);
struct st_partstat * STps = &(STp->ps[STp->partition]);
scsi: osst: silence underflow warning in osst_verify_frame() The code looks like this: i = ntohl(aux->filemark_cnt); if (STp->header_cache != NULL && i < OS_FM_TAB_MAX && (i > STp->filemark_cnt || STp->first_frame_position - 1 != ntohl(STp->header_cache->dat_fm_tab.fm_tab_ent[i]))) { If i is negative then it's less than OS_FM_TAB_MAX so we read before the start of the STp->header_cache->dat_fm_tab.fm_tab_ent[] array. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2017-08-04 08:17:33 +00:00
unsigned int blk_cnt, blk_sz, i;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (STp->raw) {
if (STp->buffer->syscall_result) {
for (i=0; i < STp->buffer->sg_segs; i++)
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
memset(page_address(sg_page(&STp->buffer->sg[i])),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
0, STp->buffer->sg[i].length);
strcpy(STp->buffer->b_data, "READ ERROR ON FRAME");
} else
STp->buffer->buffer_bytes = OS_FRAME_SIZE;
return 1;
}
if (STp->buffer->syscall_result) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame, read error\n", name);
#endif
return 0;
}
if (ntohl(aux->format_id) != 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame, format_id %u\n", name, ntohl(aux->format_id));
#endif
goto err_out;
}
if (memcmp(aux->application_sig, STp->application_sig, 4) != 0 &&
(memcmp(aux->application_sig, "LIN3", 4) != 0 || STp->linux_media_version != 4)) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame, incorrect application signature\n", name);
#endif
goto err_out;
}
if (par->partition_num != OS_DATA_PARTITION) {
if (!STp->linux_media || STp->linux_media_version != 2) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame, partition num %d\n",
name, par->partition_num);
#endif
goto err_out;
}
}
if (par->par_desc_ver != OS_PARTITION_VERSION) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame, partition version %d\n", name, par->par_desc_ver);
#endif
goto err_out;
}
if (ntohs(par->wrt_pass_cntr) != STp->wrt_pass_cntr) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame, wrt_pass_cntr %d (expected %d)\n",
name, ntohs(par->wrt_pass_cntr), STp->wrt_pass_cntr);
#endif
goto err_out;
}
if (aux->frame_type != OS_FRAME_TYPE_DATA &&
aux->frame_type != OS_FRAME_TYPE_EOD &&
aux->frame_type != OS_FRAME_TYPE_MARKER) {
[SCSI] osst: fix if (...) \n #if... cases missing semicolons when false Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi> Acked-by: Willem Riede <osst@riede.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-10-23 20:42:04 +00:00
if (!quiet) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame, frame type %x\n", name, aux->frame_type);
#endif
[SCSI] osst: fix if (...) \n #if... cases missing semicolons when false Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi> Acked-by: Willem Riede <osst@riede.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-10-23 20:42:04 +00:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
goto err_out;
}
if (aux->frame_type == OS_FRAME_TYPE_EOD &&
STp->first_frame_position < STp->eod_frame_ppos) {
printk(KERN_INFO "%s:I: Skipping premature EOD frame %d\n", name,
STp->first_frame_position);
goto err_out;
}
if (frame_seq_number != -1 && ntohl(aux->frame_seq_num) != frame_seq_number) {
[SCSI] osst: fix if (...) \n #if... cases missing semicolons when false Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi> Acked-by: Willem Riede <osst@riede.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-10-23 20:42:04 +00:00
if (!quiet) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame, sequence number %u (expected %d)\n",
name, ntohl(aux->frame_seq_num), frame_seq_number);
#endif
[SCSI] osst: fix if (...) \n #if... cases missing semicolons when false Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi> Acked-by: Willem Riede <osst@riede.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-10-23 20:42:04 +00:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
goto err_out;
}
if (aux->frame_type == OS_FRAME_TYPE_MARKER) {
STps->eof = ST_FM_HIT;
i = ntohl(aux->filemark_cnt);
if (STp->header_cache != NULL && i < OS_FM_TAB_MAX && (i > STp->filemark_cnt ||
STp->first_frame_position - 1 != ntohl(STp->header_cache->dat_fm_tab.fm_tab_ent[i]))) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: %s filemark %d at frame pos %d\n", name,
STp->header_cache->dat_fm_tab.fm_tab_ent[i] == 0?"Learned":"Corrected",
i, STp->first_frame_position - 1);
#endif
STp->header_cache->dat_fm_tab.fm_tab_ent[i] = htonl(STp->first_frame_position - 1);
if (i >= STp->filemark_cnt)
STp->filemark_cnt = i+1;
}
}
if (aux->frame_type == OS_FRAME_TYPE_EOD) {
STps->eof = ST_EOD_1;
STp->frame_in_buffer = 1;
}
if (aux->frame_type == OS_FRAME_TYPE_DATA) {
blk_cnt = ntohs(aux->dat.dat_list[0].blk_cnt);
blk_sz = ntohl(aux->dat.dat_list[0].blk_sz);
STp->buffer->buffer_bytes = blk_cnt * blk_sz;
STp->buffer->read_pointer = 0;
STp->frame_in_buffer = 1;
/* See what block size was used to write file */
if (STp->block_size != blk_sz && blk_sz > 0) {
printk(KERN_INFO
"%s:I: File was written with block size %d%c, currently %d%c, adjusted to match.\n",
name, blk_sz<1024?blk_sz:blk_sz/1024,blk_sz<1024?'b':'k',
STp->block_size<1024?STp->block_size:STp->block_size/1024,
STp->block_size<1024?'b':'k');
STp->block_size = blk_sz;
STp->buffer->buffer_blocks = OS_DATA_SIZE / blk_sz;
}
STps->eof = ST_NOEOF;
}
STp->frame_seq_number = ntohl(aux->frame_seq_num);
STp->logical_blk_num = ntohl(aux->logical_blk_num);
return 1;
err_out:
if (STp->read_error_frame == 0)
STp->read_error_frame = STp->first_frame_position - 1;
return 0;
}
/*
* Wait for the unit to become Ready
*/
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_wait_ready(struct osst_tape * STp, struct osst_request ** aSRpnt,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned timeout, int initial_delay)
{
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned long startwait = jiffies;
#if DEBUG
int dbg = debugging;
char * name = tape_name(STp);
printk(OSST_DEB_MSG "%s:D: Reached onstream wait ready\n", name);
#endif
if (initial_delay > 0)
msleep(jiffies_to_msecs(initial_delay));
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = TEST_UNIT_READY;
SRpnt = osst_do_scsi(*aSRpnt, STp, cmd, 0, DMA_NONE, STp->timeout, MAX_RETRIES, 1);
*aSRpnt = SRpnt;
if (!SRpnt) return (-EBUSY);
while ( STp->buffer->syscall_result && time_before(jiffies, startwait + timeout*HZ) &&
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
(( SRpnt->sense[2] == 2 && SRpnt->sense[12] == 4 &&
(SRpnt->sense[13] == 1 || SRpnt->sense[13] == 8) ) ||
( SRpnt->sense[2] == 6 && SRpnt->sense[12] == 0x28 &&
SRpnt->sense[13] == 0 ) )) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
if (debugging) {
printk(OSST_DEB_MSG "%s:D: Sleeping in onstream wait ready\n", name);
printk(OSST_DEB_MSG "%s:D: Turning off debugging for a while\n", name);
debugging = 0;
}
#endif
msleep(100);
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = TEST_UNIT_READY;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, 0, DMA_NONE, STp->timeout, MAX_RETRIES, 1);
}
*aSRpnt = SRpnt;
#if DEBUG
debugging = dbg;
#endif
if ( STp->buffer->syscall_result &&
osst_write_error_recovery(STp, aSRpnt, 0) ) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Abnormal exit from onstream wait ready\n", name);
printk(OSST_DEB_MSG "%s:D: Result = %d, Sense: 0=%02x, 2=%02x, 12=%02x, 13=%02x\n", name,
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
STp->buffer->syscall_result, SRpnt->sense[0], SRpnt->sense[2],
SRpnt->sense[12], SRpnt->sense[13]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
return (-EIO);
}
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Normal exit from onstream wait ready\n", name);
#endif
return 0;
}
/*
* Wait for a tape to be inserted in the unit
*/
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_wait_for_medium(struct osst_tape * STp, struct osst_request ** aSRpnt, unsigned timeout)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned long startwait = jiffies;
#if DEBUG
int dbg = debugging;
char * name = tape_name(STp);
printk(OSST_DEB_MSG "%s:D: Reached onstream wait for medium\n", name);
#endif
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = TEST_UNIT_READY;
SRpnt = osst_do_scsi(*aSRpnt, STp, cmd, 0, DMA_NONE, STp->timeout, MAX_RETRIES, 1);
*aSRpnt = SRpnt;
if (!SRpnt) return (-EBUSY);
while ( STp->buffer->syscall_result && time_before(jiffies, startwait + timeout*HZ) &&
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
SRpnt->sense[2] == 2 && SRpnt->sense[12] == 0x3a && SRpnt->sense[13] == 0 ) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
if (debugging) {
printk(OSST_DEB_MSG "%s:D: Sleeping in onstream wait medium\n", name);
printk(OSST_DEB_MSG "%s:D: Turning off debugging for a while\n", name);
debugging = 0;
}
#endif
msleep(100);
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = TEST_UNIT_READY;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, 0, DMA_NONE, STp->timeout, MAX_RETRIES, 1);
}
*aSRpnt = SRpnt;
#if DEBUG
debugging = dbg;
#endif
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ( STp->buffer->syscall_result && SRpnt->sense[2] != 2 &&
SRpnt->sense[12] != 4 && SRpnt->sense[13] == 1) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Abnormal exit from onstream wait medium\n", name);
printk(OSST_DEB_MSG "%s:D: Result = %d, Sense: 0=%02x, 2=%02x, 12=%02x, 13=%02x\n", name,
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
STp->buffer->syscall_result, SRpnt->sense[0], SRpnt->sense[2],
SRpnt->sense[12], SRpnt->sense[13]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
return 0;
}
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Normal exit from onstream wait medium\n", name);
#endif
return 1;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_position_tape_and_confirm(struct osst_tape * STp, struct osst_request ** aSRpnt, int frame)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
int retval;
osst_wait_ready(STp, aSRpnt, 15 * 60, 0); /* TODO - can this catch a write error? */
retval = osst_set_frame_position(STp, aSRpnt, frame, 0);
if (retval) return (retval);
osst_wait_ready(STp, aSRpnt, 15 * 60, OSST_WAIT_POSITION_COMPLETE);
return (osst_get_frame_position(STp, aSRpnt));
}
/*
* Wait for write(s) to complete
*/
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_flush_drive_buffer(struct osst_tape * STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
int result = 0;
int delay = OSST_WAIT_WRITE_COMPLETE;
#if DEBUG
char * name = tape_name(STp);
printk(OSST_DEB_MSG "%s:D: Reached onstream flush drive buffer (write filemark)\n", name);
#endif
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = WRITE_FILEMARKS;
cmd[1] = 1;
SRpnt = osst_do_scsi(*aSRpnt, STp, cmd, 0, DMA_NONE, STp->timeout, MAX_RETRIES, 1);
*aSRpnt = SRpnt;
if (!SRpnt) return (-EBUSY);
if (STp->buffer->syscall_result) {
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[2] & 0x0f) == 2 && SRpnt->sense[12] == 4) {
if (SRpnt->sense[13] == 8) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
delay = OSST_WAIT_LONG_WRITE_COMPLETE;
}
} else
result = osst_write_error_recovery(STp, aSRpnt, 0);
}
result |= osst_wait_ready(STp, aSRpnt, 5 * 60, delay);
STp->ps[STp->partition].rw = OS_WRITING_COMPLETE;
return (result);
}
#define OSST_POLL_PER_SEC 10
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_wait_frame(struct osst_tape * STp, struct osst_request ** aSRpnt, int curr, int minlast, int to)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned long startwait = jiffies;
char * name = tape_name(STp);
#if DEBUG
char notyetprinted = 1;
#endif
if (minlast >= 0 && STp->ps[STp->partition].rw != ST_READING)
printk(KERN_ERR "%s:A: Waiting for frame without having initialized read!\n", name);
while (time_before (jiffies, startwait + to*HZ))
{
int result;
result = osst_get_frame_position(STp, aSRpnt);
if (result == -EIO)
if ((result = osst_write_error_recovery(STp, aSRpnt, 0)) == 0)
return 0; /* successful recovery leaves drive ready for frame */
if (result < 0) break;
if (STp->first_frame_position == curr &&
((minlast < 0 &&
(signed)STp->last_frame_position > (signed)curr + minlast) ||
(minlast >= 0 && STp->cur_frames > minlast)
) && result >= 0)
{
#if DEBUG
[PATCH] drivers/scsi/*: use time_after() and friends They deal with wrapping correctly and are nicer to read. Signed-off-by: Marcelo Feitoza Parisi <marcelo@feitoza.com.br> Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Cc: James Bottomley <James.Bottomley@steeleye.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-28 09:56:47 +00:00
if (debugging || time_after_eq(jiffies, startwait + 2*HZ/OSST_POLL_PER_SEC))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk (OSST_DEB_MSG
"%s:D: Succ wait f fr %i (>%i): %i-%i %i (%i): %3li.%li s\n",
name, curr, curr+minlast, STp->first_frame_position,
STp->last_frame_position, STp->cur_frames,
result, (jiffies-startwait)/HZ,
(((jiffies-startwait)%HZ)*10)/HZ);
#endif
return 0;
}
#if DEBUG
[PATCH] drivers/scsi/*: use time_after() and friends They deal with wrapping correctly and are nicer to read. Signed-off-by: Marcelo Feitoza Parisi <marcelo@feitoza.com.br> Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Cc: James Bottomley <James.Bottomley@steeleye.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-28 09:56:47 +00:00
if (time_after_eq(jiffies, startwait + 2*HZ/OSST_POLL_PER_SEC) && notyetprinted)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
printk (OSST_DEB_MSG "%s:D: Wait for frame %i (>%i): %i-%i %i (%i)\n",
name, curr, curr+minlast, STp->first_frame_position,
STp->last_frame_position, STp->cur_frames, result);
notyetprinted--;
}
#endif
msleep(1000 / OSST_POLL_PER_SEC);
}
#if DEBUG
printk (OSST_DEB_MSG "%s:D: Fail wait f fr %i (>%i): %i-%i %i: %3li.%li s\n",
name, curr, curr+minlast, STp->first_frame_position,
STp->last_frame_position, STp->cur_frames,
(jiffies-startwait)/HZ, (((jiffies-startwait)%HZ)*10)/HZ);
#endif
return -EBUSY;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_recover_wait_frame(struct osst_tape * STp, struct osst_request ** aSRpnt, int writing)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned char cmd[MAX_COMMAND_SIZE];
unsigned long startwait = jiffies;
int retval = 1;
char * name = tape_name(STp);
if (writing) {
char mybuf[24];
char * olddata = STp->buffer->b_data;
int oldsize = STp->buffer->buffer_size;
/* write zero fm then read pos - if shows write error, try to recover - if no progress, wait */
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = WRITE_FILEMARKS;
cmd[1] = 1;
SRpnt = osst_do_scsi(*aSRpnt, STp, cmd, 0, DMA_NONE, STp->timeout,
MAX_RETRIES, 1);
while (retval && time_before (jiffies, startwait + 5*60*HZ)) {
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (STp->buffer->syscall_result && (SRpnt->sense[2] & 0x0f) != 2) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/* some failure - not just not-ready */
retval = osst_write_error_recovery(STp, aSRpnt, 0);
break;
}
[PATCH] drivers/scsi: fix-up schedule_timeout() usage Use schedule_timeout_uninterruptible() instead of set_current_state()/schedule_timeout() to reduce kernel size. Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com> Cc: James Bottomley <James.Bottomley@steeleye.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-07 09:01:20 +00:00
schedule_timeout_interruptible(HZ / OSST_POLL_PER_SEC);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->buffer->b_data = mybuf; STp->buffer->buffer_size = 24;
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = READ_POSITION;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, 20, DMA_FROM_DEVICE, STp->timeout,
MAX_RETRIES, 1);
retval = ( STp->buffer->syscall_result || (STp->buffer)->b_data[15] > 25 );
STp->buffer->b_data = olddata; STp->buffer->buffer_size = oldsize;
}
if (retval)
printk(KERN_ERR "%s:E: Device did not succeed to write buffered data\n", name);
} else
/* TODO - figure out which error conditions can be handled */
if (STp->buffer->syscall_result)
printk(KERN_WARNING
"%s:W: Recover_wait_frame(read) cannot handle %02x:%02x:%02x\n", name,
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
(*aSRpnt)->sense[ 2] & 0x0f,
(*aSRpnt)->sense[12],
(*aSRpnt)->sense[13]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return retval;
}
/*
* Read the next OnStream tape frame at the current location
*/
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_read_frame(struct osst_tape * STp, struct osst_request ** aSRpnt, int timeout)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
int retval = 0;
#if DEBUG
os_aux_t * aux = STp->buffer->aux;
char * name = tape_name(STp);
#endif
if (STp->poll)
if (osst_wait_frame (STp, aSRpnt, STp->first_frame_position, 0, timeout))
retval = osst_recover_wait_frame(STp, aSRpnt, 0);
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = READ_6;
cmd[1] = 1;
cmd[4] = 1;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Reading frame from OnStream tape\n", name);
#endif
SRpnt = osst_do_scsi(*aSRpnt, STp, cmd, OS_FRAME_SIZE, DMA_FROM_DEVICE,
STp->timeout, MAX_RETRIES, 1);
*aSRpnt = SRpnt;
if (!SRpnt)
return (-EBUSY);
if ((STp->buffer)->syscall_result) {
retval = 1;
if (STp->read_error_frame == 0) {
STp->read_error_frame = STp->first_frame_position;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Recording read error at %d\n", name, STp->read_error_frame);
#endif
}
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Sense: %2x %2x %2x %2x %2x %2x %2x %2x\n",
name,
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
SRpnt->sense[0], SRpnt->sense[1],
SRpnt->sense[2], SRpnt->sense[3],
SRpnt->sense[4], SRpnt->sense[5],
SRpnt->sense[6], SRpnt->sense[7]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
}
else
STp->first_frame_position++;
#if DEBUG
if (debugging) {
char sig[8]; int i;
for (i=0;i<4;i++)
sig[i] = aux->application_sig[i]<32?'^':aux->application_sig[i];
sig[4] = '\0';
printk(OSST_DEB_MSG
"%s:D: AUX: %s UpdFrCt#%d Wpass#%d %s FrSeq#%d LogBlk#%d Qty=%d Sz=%d\n", name, sig,
ntohl(aux->update_frame_cntr), ntohs(aux->partition.wrt_pass_cntr),
aux->frame_type==1?"EOD":aux->frame_type==2?"MARK":
aux->frame_type==8?"HEADR":aux->frame_type==0x80?"DATA":"FILL",
ntohl(aux->frame_seq_num), ntohl(aux->logical_blk_num),
ntohs(aux->dat.dat_list[0].blk_cnt), ntohl(aux->dat.dat_list[0].blk_sz) );
if (aux->frame_type==2)
printk(OSST_DEB_MSG "%s:D: mark_cnt=%d, last_mark_ppos=%d, last_mark_lbn=%d\n", name,
ntohl(aux->filemark_cnt), ntohl(aux->last_mark_ppos), ntohl(aux->last_mark_lbn));
printk(OSST_DEB_MSG "%s:D: Exit read frame from OnStream tape with code %d\n", name, retval);
}
#endif
return (retval);
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_initiate_read(struct osst_tape * STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct st_partstat * STps = &(STp->ps[STp->partition]);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt ;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned char cmd[MAX_COMMAND_SIZE];
int retval = 0;
char * name = tape_name(STp);
if (STps->rw != ST_READING) { /* Initialize read operation */
if (STps->rw == ST_WRITING || STp->dirty) {
STp->write_type = OS_WRITE_DATA;
osst_flush_write_buffer(STp, aSRpnt);
osst_flush_drive_buffer(STp, aSRpnt);
}
STps->rw = ST_READING;
STp->frame_in_buffer = 0;
/*
* Issue a read 0 command to get the OnStream drive
* read frames into its buffer.
*/
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = READ_6;
cmd[1] = 1;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Start Read Ahead on OnStream tape\n", name);
#endif
SRpnt = osst_do_scsi(*aSRpnt, STp, cmd, 0, DMA_NONE, STp->timeout, MAX_RETRIES, 1);
*aSRpnt = SRpnt;
if ((retval = STp->buffer->syscall_result))
printk(KERN_WARNING "%s:W: Error starting read ahead\n", name);
}
return retval;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_get_logical_frame(struct osst_tape * STp, struct osst_request ** aSRpnt,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
int frame_seq_number, int quiet)
{
struct st_partstat * STps = &(STp->ps[STp->partition]);
char * name = tape_name(STp);
int cnt = 0,
bad = 0,
past = 0,
x,
position;
/*
* If we want just any frame (-1) and there is a frame in the buffer, return it
*/
if (frame_seq_number == -1 && STp->frame_in_buffer) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Frame %d still in buffer\n", name, STp->frame_seq_number);
#endif
return (STps->eof);
}
/*
* Search and wait for the next logical tape frame
*/
while (1) {
if (cnt++ > 400) {
printk(KERN_ERR "%s:E: Couldn't find logical frame %d, aborting\n",
name, frame_seq_number);
if (STp->read_error_frame) {
osst_set_frame_position(STp, aSRpnt, STp->read_error_frame, 0);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Repositioning tape to bad frame %d\n",
name, STp->read_error_frame);
#endif
STp->read_error_frame = 0;
STp->abort_count++;
}
return (-EIO);
}
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Looking for frame %d, attempt %d\n",
name, frame_seq_number, cnt);
#endif
if ( osst_initiate_read(STp, aSRpnt)
|| ( (!STp->frame_in_buffer) && osst_read_frame(STp, aSRpnt, 30) ) ) {
if (STp->raw)
return (-EIO);
position = osst_get_frame_position(STp, aSRpnt);
if (position >= 0xbae && position < 0xbb8)
position = 0xbb8;
else if (position > STp->eod_frame_ppos || ++bad == 10) {
position = STp->read_error_frame - 1;
bad = 0;
}
else {
position += 29;
cnt += 19;
}
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Bad frame detected, positioning tape to block %d\n",
name, position);
#endif
osst_set_frame_position(STp, aSRpnt, position, 0);
continue;
}
if (osst_verify_frame(STp, frame_seq_number, quiet))
break;
if (osst_verify_frame(STp, -1, quiet)) {
x = ntohl(STp->buffer->aux->frame_seq_num);
if (STp->fast_open) {
printk(KERN_WARNING
"%s:W: Found logical frame %d instead of %d after fast open\n",
name, x, frame_seq_number);
STp->header_ok = 0;
STp->read_error_frame = 0;
return (-EIO);
}
if (x > frame_seq_number) {
if (++past > 3) {
/* positioning backwards did not bring us to the desired frame */
position = STp->read_error_frame - 1;
}
else {
position = osst_get_frame_position(STp, aSRpnt)
+ frame_seq_number - x - 1;
if (STp->first_frame_position >= 3000 && position < 3000)
position -= 10;
}
#if DEBUG
printk(OSST_DEB_MSG
"%s:D: Found logical frame %d while looking for %d: back up %d\n",
name, x, frame_seq_number,
STp->first_frame_position - position);
#endif
osst_set_frame_position(STp, aSRpnt, position, 0);
cnt += 10;
}
else
past = 0;
}
if (osst_get_frame_position(STp, aSRpnt) == 0xbaf) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping config partition\n", name);
#endif
osst_set_frame_position(STp, aSRpnt, 0xbb8, 0);
cnt--;
}
STp->frame_in_buffer = 0;
}
if (cnt > 1) {
STp->recover_count++;
STp->recover_erreg++;
printk(KERN_WARNING "%s:I: Don't worry, Read error at position %d recovered\n",
name, STp->read_error_frame);
}
STp->read_count++;
#if DEBUG
if (debugging || STps->eof)
printk(OSST_DEB_MSG
"%s:D: Exit get logical frame (%d=>%d) from OnStream tape with code %d\n",
name, frame_seq_number, STp->frame_seq_number, STps->eof);
#endif
STp->fast_open = 0;
STp->read_error_frame = 0;
return (STps->eof);
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_seek_logical_blk(struct osst_tape * STp, struct osst_request ** aSRpnt, int logical_blk_num)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct st_partstat * STps = &(STp->ps[STp->partition]);
char * name = tape_name(STp);
int retries = 0;
int frame_seq_estimate, ppos_estimate, move;
if (logical_blk_num < 0) logical_blk_num = 0;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Seeking logical block %d (now at %d, size %d%c)\n",
name, logical_blk_num, STp->logical_blk_num,
STp->block_size<1024?STp->block_size:STp->block_size/1024,
STp->block_size<1024?'b':'k');
#endif
/* Do we know where we are? */
if (STps->drv_block >= 0) {
move = logical_blk_num - STp->logical_blk_num;
if (move < 0) move -= (OS_DATA_SIZE / STp->block_size) - 1;
move /= (OS_DATA_SIZE / STp->block_size);
frame_seq_estimate = STp->frame_seq_number + move;
} else
frame_seq_estimate = logical_blk_num * STp->block_size / OS_DATA_SIZE;
if (frame_seq_estimate < 2980) ppos_estimate = frame_seq_estimate + 10;
else ppos_estimate = frame_seq_estimate + 20;
while (++retries < 10) {
if (ppos_estimate > STp->eod_frame_ppos-2) {
frame_seq_estimate += STp->eod_frame_ppos - 2 - ppos_estimate;
ppos_estimate = STp->eod_frame_ppos - 2;
}
if (frame_seq_estimate < 0) {
frame_seq_estimate = 0;
ppos_estimate = 10;
}
osst_set_frame_position(STp, aSRpnt, ppos_estimate, 0);
if (osst_get_logical_frame(STp, aSRpnt, frame_seq_estimate, 1) >= 0) {
/* we've located the estimated frame, now does it have our block? */
if (logical_blk_num < STp->logical_blk_num ||
logical_blk_num >= STp->logical_blk_num + ntohs(STp->buffer->aux->dat.dat_list[0].blk_cnt)) {
if (STps->eof == ST_FM_HIT)
move = logical_blk_num < STp->logical_blk_num? -2 : 1;
else {
move = logical_blk_num - STp->logical_blk_num;
if (move < 0) move -= (OS_DATA_SIZE / STp->block_size) - 1;
move /= (OS_DATA_SIZE / STp->block_size);
}
if (!move) move = logical_blk_num > STp->logical_blk_num ? 1 : -1;
#if DEBUG
printk(OSST_DEB_MSG
"%s:D: Seek retry %d at ppos %d fsq %d (est %d) lbn %d (need %d) move %d\n",
name, retries, ppos_estimate, STp->frame_seq_number, frame_seq_estimate,
STp->logical_blk_num, logical_blk_num, move);
#endif
frame_seq_estimate += move;
ppos_estimate += move;
continue;
} else {
STp->buffer->read_pointer = (logical_blk_num - STp->logical_blk_num) * STp->block_size;
STp->buffer->buffer_bytes -= STp->buffer->read_pointer;
STp->logical_blk_num = logical_blk_num;
#if DEBUG
printk(OSST_DEB_MSG
"%s:D: Seek success at ppos %d fsq %d in_buf %d, bytes %d, ptr %d*%d\n",
name, ppos_estimate, STp->frame_seq_number, STp->frame_in_buffer,
STp->buffer->buffer_bytes, STp->buffer->read_pointer / STp->block_size,
STp->block_size);
#endif
STps->drv_file = ntohl(STp->buffer->aux->filemark_cnt);
if (STps->eof == ST_FM_HIT) {
STps->drv_file++;
STps->drv_block = 0;
} else {
STps->drv_block = ntohl(STp->buffer->aux->last_mark_lbn)?
STp->logical_blk_num -
(STps->drv_file ? ntohl(STp->buffer->aux->last_mark_lbn) + 1 : 0):
-1;
}
STps->eof = (STp->first_frame_position >= STp->eod_frame_ppos)?ST_EOD:ST_NOEOF;
return 0;
}
}
if (osst_get_logical_frame(STp, aSRpnt, -1, 1) < 0)
goto error;
/* we are not yet at the estimated frame, adjust our estimate of its physical position */
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Seek retry %d at ppos %d fsq %d (est %d) lbn %d (need %d)\n",
name, retries, ppos_estimate, STp->frame_seq_number, frame_seq_estimate,
STp->logical_blk_num, logical_blk_num);
#endif
if (frame_seq_estimate != STp->frame_seq_number)
ppos_estimate += frame_seq_estimate - STp->frame_seq_number;
else
break;
}
error:
printk(KERN_ERR "%s:E: Couldn't seek to logical block %d (at %d), %d retries\n",
name, logical_blk_num, STp->logical_blk_num, retries);
return (-EIO);
}
/* The values below are based on the OnStream frame payload size of 32K == 2**15,
* that is, OSST_FRAME_SHIFT + OSST_SECTOR_SHIFT must be 15. With a minimum block
* size of 512 bytes, we need to be able to resolve 32K/512 == 64 == 2**6 positions
Fix common misspellings Fixes generated by 'codespell' and manually reviewed. Signed-off-by: Lucas De Marchi <lucas.demarchi@profusion.mobi>
2011-03-31 01:57:33 +00:00
* inside each frame. Finally, OSST_SECTOR_MASK == 2**OSST_FRAME_SHIFT - 1.
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
*/
#define OSST_FRAME_SHIFT 6
#define OSST_SECTOR_SHIFT 9
#define OSST_SECTOR_MASK 0x03F
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_get_sector(struct osst_tape * STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
int sector;
#if DEBUG
char * name = tape_name(STp);
printk(OSST_DEB_MSG
"%s:D: Positioned at ppos %d, frame %d, lbn %d, file %d, blk %d, %cptr %d, eof %d\n",
name, STp->first_frame_position, STp->frame_seq_number, STp->logical_blk_num,
STp->ps[STp->partition].drv_file, STp->ps[STp->partition].drv_block,
STp->ps[STp->partition].rw == ST_WRITING?'w':'r',
STp->ps[STp->partition].rw == ST_WRITING?STp->buffer->buffer_bytes:
STp->buffer->read_pointer, STp->ps[STp->partition].eof);
#endif
/* do we know where we are inside a file? */
if (STp->ps[STp->partition].drv_block >= 0) {
sector = (STp->frame_in_buffer ? STp->first_frame_position-1 :
STp->first_frame_position) << OSST_FRAME_SHIFT;
if (STp->ps[STp->partition].rw == ST_WRITING)
sector |= (STp->buffer->buffer_bytes >> OSST_SECTOR_SHIFT) & OSST_SECTOR_MASK;
else
sector |= (STp->buffer->read_pointer >> OSST_SECTOR_SHIFT) & OSST_SECTOR_MASK;
} else {
sector = osst_get_frame_position(STp, aSRpnt);
if (sector > 0)
sector <<= OSST_FRAME_SHIFT;
}
return sector;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_seek_sector(struct osst_tape * STp, struct osst_request ** aSRpnt, int sector)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct st_partstat * STps = &(STp->ps[STp->partition]);
int frame = sector >> OSST_FRAME_SHIFT,
offset = (sector & OSST_SECTOR_MASK) << OSST_SECTOR_SHIFT,
r;
#if DEBUG
char * name = tape_name(STp);
printk(OSST_DEB_MSG "%s:D: Seeking sector %d in frame %d at offset %d\n",
name, sector, frame, offset);
#endif
if (frame < 0 || frame >= STp->capacity) return (-ENXIO);
if (frame <= STp->first_data_ppos) {
STp->frame_seq_number = STp->logical_blk_num = STps->drv_file = STps->drv_block = 0;
return (osst_set_frame_position(STp, aSRpnt, frame, 0));
}
r = osst_set_frame_position(STp, aSRpnt, offset?frame:frame-1, 0);
if (r < 0) return r;
r = osst_get_logical_frame(STp, aSRpnt, -1, 1);
if (r < 0) return r;
if (osst_get_frame_position(STp, aSRpnt) != (offset?frame+1:frame)) return (-EIO);
if (offset) {
STp->logical_blk_num += offset / STp->block_size;
STp->buffer->read_pointer = offset;
STp->buffer->buffer_bytes -= offset;
} else {
STp->frame_seq_number++;
STp->frame_in_buffer = 0;
STp->logical_blk_num += ntohs(STp->buffer->aux->dat.dat_list[0].blk_cnt);
STp->buffer->buffer_bytes = STp->buffer->read_pointer = 0;
}
STps->drv_file = ntohl(STp->buffer->aux->filemark_cnt);
if (STps->eof == ST_FM_HIT) {
STps->drv_file++;
STps->drv_block = 0;
} else {
STps->drv_block = ntohl(STp->buffer->aux->last_mark_lbn)?
STp->logical_blk_num -
(STps->drv_file ? ntohl(STp->buffer->aux->last_mark_lbn) + 1 : 0):
-1;
}
STps->eof = (STp->first_frame_position >= STp->eod_frame_ppos)?ST_EOD:ST_NOEOF;
#if DEBUG
printk(OSST_DEB_MSG
"%s:D: Now positioned at ppos %d, frame %d, lbn %d, file %d, blk %d, rptr %d, eof %d\n",
name, STp->first_frame_position, STp->frame_seq_number, STp->logical_blk_num,
STps->drv_file, STps->drv_block, STp->buffer->read_pointer, STps->eof);
#endif
return 0;
}
/*
* Read back the drive's internal buffer contents, as a part
* of the write error recovery mechanism for old OnStream
* firmware revisions.
* Precondition for this function to work: all frames in the
* drive's buffer must be of one type (DATA, MARK or EOD)!
*/
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_read_back_buffer_and_rewrite(struct osst_tape * STp, struct osst_request ** aSRpnt,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned int frame, unsigned int skip, int pending)
{
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = * aSRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned char * buffer, * p;
unsigned char cmd[MAX_COMMAND_SIZE];
int flag, new_frame, i;
int nframes = STp->cur_frames;
int blks_per_frame = ntohs(STp->buffer->aux->dat.dat_list[0].blk_cnt);
int frame_seq_number = ntohl(STp->buffer->aux->frame_seq_num)
- (nframes + pending - 1);
int logical_blk_num = ntohl(STp->buffer->aux->logical_blk_num)
- (nframes + pending - 1) * blks_per_frame;
char * name = tape_name(STp);
unsigned long startwait = jiffies;
#if DEBUG
int dbg = debugging;
#endif
treewide: Use array_size() in vmalloc() The vmalloc() function has no 2-factor argument form, so multiplication factors need to be wrapped in array_size(). This patch replaces cases of: vmalloc(a * b) with: vmalloc(array_size(a, b)) as well as handling cases of: vmalloc(a * b * c) with: vmalloc(array3_size(a, b, c)) This does, however, attempt to ignore constant size factors like: vmalloc(4 * 1024) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ type TYPE; expression THING, E; @@ ( vmalloc( - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | vmalloc( - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression COUNT; typedef u8; typedef __u8; @@ ( vmalloc( - sizeof(u8) * (COUNT) + COUNT , ...) | vmalloc( - sizeof(__u8) * (COUNT) + COUNT , ...) | vmalloc( - sizeof(char) * (COUNT) + COUNT , ...) | vmalloc( - sizeof(unsigned char) * (COUNT) + COUNT , ...) | vmalloc( - sizeof(u8) * COUNT + COUNT , ...) | vmalloc( - sizeof(__u8) * COUNT + COUNT , ...) | vmalloc( - sizeof(char) * COUNT + COUNT , ...) | vmalloc( - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( vmalloc( - sizeof(TYPE) * (COUNT_ID) + array_size(COUNT_ID, sizeof(TYPE)) , ...) | vmalloc( - sizeof(TYPE) * COUNT_ID + array_size(COUNT_ID, sizeof(TYPE)) , ...) | vmalloc( - sizeof(TYPE) * (COUNT_CONST) + array_size(COUNT_CONST, sizeof(TYPE)) , ...) | vmalloc( - sizeof(TYPE) * COUNT_CONST + array_size(COUNT_CONST, sizeof(TYPE)) , ...) | vmalloc( - sizeof(THING) * (COUNT_ID) + array_size(COUNT_ID, sizeof(THING)) , ...) | vmalloc( - sizeof(THING) * COUNT_ID + array_size(COUNT_ID, sizeof(THING)) , ...) | vmalloc( - sizeof(THING) * (COUNT_CONST) + array_size(COUNT_CONST, sizeof(THING)) , ...) | vmalloc( - sizeof(THING) * COUNT_CONST + array_size(COUNT_CONST, sizeof(THING)) , ...) ) // 2-factor product, only identifiers. @@ identifier SIZE, COUNT; @@ vmalloc( - SIZE * COUNT + array_size(COUNT, SIZE) , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( vmalloc( - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | vmalloc( - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | vmalloc( - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | vmalloc( - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | vmalloc( - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | vmalloc( - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | vmalloc( - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | vmalloc( - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( vmalloc( - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | vmalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | vmalloc( - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | vmalloc( - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | vmalloc( - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | vmalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ identifier STRIDE, SIZE, COUNT; @@ ( vmalloc( - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | vmalloc( - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | vmalloc( - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | vmalloc( - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | vmalloc( - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | vmalloc( - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | vmalloc( - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | vmalloc( - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products // when they're not all constants... @@ expression E1, E2, E3; constant C1, C2, C3; @@ ( vmalloc(C1 * C2 * C3, ...) | vmalloc( - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants. @@ expression E1, E2; constant C1, C2; @@ ( vmalloc(C1 * C2, ...) | vmalloc( - E1 * E2 + array_size(E1, E2) , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-12 21:27:11 +00:00
if ((buffer = vmalloc(array_size((nframes + 1), OS_DATA_SIZE))) == NULL)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return (-EIO);
printk(KERN_INFO "%s:I: Reading back %d frames from drive buffer%s\n",
name, nframes, pending?" and one that was pending":"");
osst_copy_from_buffer(STp->buffer, (p = &buffer[nframes * OS_DATA_SIZE]));
#if DEBUG
if (pending && debugging)
printk(OSST_DEB_MSG "%s:D: Pending frame %d (lblk %d), data %02x %02x %02x %02x\n",
name, frame_seq_number + nframes,
logical_blk_num + nframes * blks_per_frame,
p[0], p[1], p[2], p[3]);
#endif
for (i = 0, p = buffer; i < nframes; i++, p += OS_DATA_SIZE) {
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = 0x3C; /* Buffer Read */
cmd[1] = 6; /* Retrieve Faulty Block */
cmd[7] = 32768 >> 8;
cmd[8] = 32768 & 0xff;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, OS_FRAME_SIZE, DMA_FROM_DEVICE,
STp->timeout, MAX_RETRIES, 1);
if ((STp->buffer)->syscall_result || !SRpnt) {
printk(KERN_ERR "%s:E: Failed to read frame back from OnStream buffer\n", name);
[PATCH] vfree and kfree cleanup in drivers/ This patch does a full cleanup of 'NULL checks before vfree', and a partial cleanup of calls to kfree for all of drivers/ - the kfree bit is partial in that I only did the files that also had vfree calls in them. The patch also gets rid of some redundant (void *) casts of pointers being passed to [vk]free, and a some tiny whitespace corrections also crept in. Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 07:26:54 +00:00
vfree(buffer);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
*aSRpnt = SRpnt;
return (-EIO);
}
osst_copy_from_buffer(STp->buffer, p);
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Read back logical frame %d, data %02x %02x %02x %02x\n",
name, frame_seq_number + i, p[0], p[1], p[2], p[3]);
#endif
}
*aSRpnt = SRpnt;
osst_get_frame_position(STp, aSRpnt);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Frames left in buffer: %d\n", name, STp->cur_frames);
#endif
/* Write synchronously so we can be sure we're OK again and don't have to recover recursively */
/* In the header we don't actually re-write the frames that fail, just the ones after them */
for (flag=1, new_frame=frame, p=buffer, i=0; i < nframes + pending; ) {
if (flag) {
if (STp->write_type == OS_WRITE_HEADER) {
i += skip;
p += skip * OS_DATA_SIZE;
}
else if (new_frame < 2990 && new_frame+skip+nframes+pending >= 2990)
new_frame = 3000-i;
else
new_frame += skip;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Position to frame %d, write fseq %d\n",
name, new_frame+i, frame_seq_number+i);
#endif
osst_set_frame_position(STp, aSRpnt, new_frame + i, 0);
osst_wait_ready(STp, aSRpnt, 60, OSST_WAIT_POSITION_COMPLETE);
osst_get_frame_position(STp, aSRpnt);
SRpnt = * aSRpnt;
if (new_frame > frame + 1000) {
printk(KERN_ERR "%s:E: Failed to find writable tape media\n", name);
[PATCH] vfree and kfree cleanup in drivers/ This patch does a full cleanup of 'NULL checks before vfree', and a partial cleanup of calls to kfree for all of drivers/ - the kfree bit is partial in that I only did the files that also had vfree calls in them. The patch also gets rid of some redundant (void *) casts of pointers being passed to [vk]free, and a some tiny whitespace corrections also crept in. Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 07:26:54 +00:00
vfree(buffer);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return (-EIO);
}
if ( i >= nframes + pending ) break;
flag = 0;
}
osst_copy_to_buffer(STp->buffer, p);
/*
* IMPORTANT: for error recovery to work, _never_ queue frames with mixed frame type!
*/
osst_init_aux(STp, STp->buffer->aux->frame_type, frame_seq_number+i,
logical_blk_num + i*blks_per_frame,
ntohl(STp->buffer->aux->dat.dat_list[0].blk_sz), blks_per_frame);
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = WRITE_6;
cmd[1] = 1;
cmd[4] = 1;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG
"%s:D: About to write frame %d, seq %d, lbn %d, data %02x %02x %02x %02x\n",
name, new_frame+i, frame_seq_number+i, logical_blk_num + i*blks_per_frame,
p[0], p[1], p[2], p[3]);
#endif
SRpnt = osst_do_scsi(SRpnt, STp, cmd, OS_FRAME_SIZE, DMA_TO_DEVICE,
STp->timeout, MAX_RETRIES, 1);
if (STp->buffer->syscall_result)
flag = 1;
else {
p += OS_DATA_SIZE; i++;
/* if we just sent the last frame, wait till all successfully written */
if ( i == nframes + pending ) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Check re-write successful\n", name);
#endif
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = WRITE_FILEMARKS;
cmd[1] = 1;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, 0, DMA_NONE,
STp->timeout, MAX_RETRIES, 1);
#if DEBUG
if (debugging) {
printk(OSST_DEB_MSG "%s:D: Sleeping in re-write wait ready\n", name);
printk(OSST_DEB_MSG "%s:D: Turning off debugging for a while\n", name);
debugging = 0;
}
#endif
flag = STp->buffer->syscall_result;
while ( !flag && time_before(jiffies, startwait + 60*HZ) ) {
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = TEST_UNIT_READY;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, 0, DMA_NONE, STp->timeout,
MAX_RETRIES, 1);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt->sense[2] == 2 && SRpnt->sense[12] == 4 &&
(SRpnt->sense[13] == 1 || SRpnt->sense[13] == 8)) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/* in the process of becoming ready */
msleep(100);
continue;
}
if (STp->buffer->syscall_result)
flag = 1;
break;
}
#if DEBUG
debugging = dbg;
printk(OSST_DEB_MSG "%s:D: Wait re-write finished\n", name);
#endif
}
}
*aSRpnt = SRpnt;
if (flag) {
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[ 2] & 0x0f) == 13 &&
SRpnt->sense[12] == 0 &&
SRpnt->sense[13] == 2) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_ERR "%s:E: Volume overflow in write error recovery\n", name);
[PATCH] vfree and kfree cleanup in drivers/ This patch does a full cleanup of 'NULL checks before vfree', and a partial cleanup of calls to kfree for all of drivers/ - the kfree bit is partial in that I only did the files that also had vfree calls in them. The patch also gets rid of some redundant (void *) casts of pointers being passed to [vk]free, and a some tiny whitespace corrections also crept in. Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 07:26:54 +00:00
vfree(buffer);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return (-EIO); /* hit end of tape = fail */
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
i = ((SRpnt->sense[3] << 24) |
(SRpnt->sense[4] << 16) |
(SRpnt->sense[5] << 8) |
SRpnt->sense[6] ) - new_frame;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
p = &buffer[i * OS_DATA_SIZE];
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Additional write error at %d\n", name, new_frame+i);
#endif
osst_get_frame_position(STp, aSRpnt);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: reported frame positions: host = %d, tape = %d, buffer = %d\n",
name, STp->first_frame_position, STp->last_frame_position, STp->cur_frames);
#endif
}
}
if (flag) {
/* error recovery did not successfully complete */
printk(KERN_ERR "%s:D: Write error recovery failed in %s\n", name,
STp->write_type == OS_WRITE_HEADER?"header":"body");
}
if (!pending)
osst_copy_to_buffer(STp->buffer, p); /* so buffer content == at entry in all cases */
[PATCH] vfree and kfree cleanup in drivers/ This patch does a full cleanup of 'NULL checks before vfree', and a partial cleanup of calls to kfree for all of drivers/ - the kfree bit is partial in that I only did the files that also had vfree calls in them. The patch also gets rid of some redundant (void *) casts of pointers being passed to [vk]free, and a some tiny whitespace corrections also crept in. Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 07:26:54 +00:00
vfree(buffer);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return 0;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_reposition_and_retry(struct osst_tape * STp, struct osst_request ** aSRpnt,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned int frame, unsigned int skip, int pending)
{
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
char * name = tape_name(STp);
int expected = 0;
int attempts = 1000 / skip;
int flag = 1;
unsigned long startwait = jiffies;
#if DEBUG
int dbg = debugging;
#endif
while (attempts && time_before(jiffies, startwait + 60*HZ)) {
if (flag) {
#if DEBUG
debugging = dbg;
#endif
if (frame < 2990 && frame+skip+STp->cur_frames+pending >= 2990)
frame = 3000-skip;
expected = frame+skip+STp->cur_frames+pending;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Position to fppos %d, re-write from fseq %d\n",
name, frame+skip, STp->frame_seq_number-STp->cur_frames-pending);
#endif
osst_set_frame_position(STp, aSRpnt, frame + skip, 1);
flag = 0;
attempts--;
[PATCH] drivers/scsi: fix-up schedule_timeout() usage Use schedule_timeout_uninterruptible() instead of set_current_state()/schedule_timeout() to reduce kernel size. Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com> Cc: James Bottomley <James.Bottomley@steeleye.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-07 09:01:20 +00:00
schedule_timeout_interruptible(msecs_to_jiffies(100));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
if (osst_get_frame_position(STp, aSRpnt) < 0) { /* additional write error */
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Addl error, host %d, tape %d, buffer %d\n",
name, STp->first_frame_position,
STp->last_frame_position, STp->cur_frames);
#endif
frame = STp->last_frame_position;
flag = 1;
continue;
}
if (pending && STp->cur_frames < 50) {
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = WRITE_6;
cmd[1] = 1;
cmd[4] = 1;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: About to write pending fseq %d at fppos %d\n",
name, STp->frame_seq_number-1, STp->first_frame_position);
#endif
SRpnt = osst_do_scsi(*aSRpnt, STp, cmd, OS_FRAME_SIZE, DMA_TO_DEVICE,
STp->timeout, MAX_RETRIES, 1);
*aSRpnt = SRpnt;
if (STp->buffer->syscall_result) { /* additional write error */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[ 2] & 0x0f) == 13 &&
SRpnt->sense[12] == 0 &&
SRpnt->sense[13] == 2) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_ERR
"%s:E: Volume overflow in write error recovery\n",
name);
break; /* hit end of tape = fail */
}
flag = 1;
}
else
pending = 0;
continue;
}
if (STp->cur_frames == 0) {
#if DEBUG
debugging = dbg;
printk(OSST_DEB_MSG "%s:D: Wait re-write finished\n", name);
#endif
if (STp->first_frame_position != expected) {
printk(KERN_ERR "%s:A: Actual position %d - expected %d\n",
name, STp->first_frame_position, expected);
return (-EIO);
}
return 0;
}
#if DEBUG
if (debugging) {
printk(OSST_DEB_MSG "%s:D: Sleeping in re-write wait ready\n", name);
printk(OSST_DEB_MSG "%s:D: Turning off debugging for a while\n", name);
debugging = 0;
}
#endif
[PATCH] drivers/scsi: fix-up schedule_timeout() usage Use schedule_timeout_uninterruptible() instead of set_current_state()/schedule_timeout() to reduce kernel size. Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com> Cc: James Bottomley <James.Bottomley@steeleye.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-07 09:01:20 +00:00
schedule_timeout_interruptible(msecs_to_jiffies(100));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
printk(KERN_ERR "%s:E: Failed to find valid tape media\n", name);
#if DEBUG
debugging = dbg;
#endif
return (-EIO);
}
/*
* Error recovery algorithm for the OnStream tape.
*/
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_write_error_recovery(struct osst_tape * STp, struct osst_request ** aSRpnt, int pending)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = * aSRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct st_partstat * STps = & STp->ps[STp->partition];
char * name = tape_name(STp);
int retval = 0;
int rw_state;
unsigned int frame, skip;
rw_state = STps->rw;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[ 2] & 0x0f) != 3
|| SRpnt->sense[12] != 12
|| SRpnt->sense[13] != 0) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Write error recovery cannot handle %02x:%02x:%02x\n", name,
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
SRpnt->sense[2], SRpnt->sense[12], SRpnt->sense[13]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
return (-EIO);
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
frame = (SRpnt->sense[3] << 24) |
(SRpnt->sense[4] << 16) |
(SRpnt->sense[5] << 8) |
SRpnt->sense[6];
skip = SRpnt->sense[9];
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Detected physical bad frame at %u, advised to skip %d\n", name, frame, skip);
#endif
osst_get_frame_position(STp, aSRpnt);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: reported frame positions: host = %d, tape = %d\n",
name, STp->first_frame_position, STp->last_frame_position);
#endif
switch (STp->write_type) {
case OS_WRITE_DATA:
case OS_WRITE_EOD:
case OS_WRITE_NEW_MARK:
printk(KERN_WARNING
"%s:I: Relocating %d buffered logical frames from position %u to %u\n",
name, STp->cur_frames, frame, (frame + skip > 3000 && frame < 3000)?3000:frame + skip);
if (STp->os_fw_rev >= 10600)
retval = osst_reposition_and_retry(STp, aSRpnt, frame, skip, pending);
else
retval = osst_read_back_buffer_and_rewrite(STp, aSRpnt, frame, skip, pending);
printk(KERN_WARNING "%s:%s: %sWrite error%srecovered\n", name,
retval?"E" :"I",
retval?"" :"Don't worry, ",
retval?" not ":" ");
break;
case OS_WRITE_LAST_MARK:
printk(KERN_ERR "%s:E: Bad frame in update last marker, fatal\n", name);
osst_set_frame_position(STp, aSRpnt, frame + STp->cur_frames + pending, 0);
retval = -EIO;
break;
case OS_WRITE_HEADER:
printk(KERN_WARNING "%s:I: Bad frame in header partition, skipped\n", name);
retval = osst_read_back_buffer_and_rewrite(STp, aSRpnt, frame, 1, pending);
break;
default:
printk(KERN_INFO "%s:I: Bad frame in filler, ignored\n", name);
osst_set_frame_position(STp, aSRpnt, frame + STp->cur_frames + pending, 0);
}
osst_get_frame_position(STp, aSRpnt);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Positioning complete, cur_frames %d, pos %d, tape pos %d\n",
name, STp->cur_frames, STp->first_frame_position, STp->last_frame_position);
printk(OSST_DEB_MSG "%s:D: next logical frame to write: %d\n", name, STp->logical_blk_num);
#endif
if (retval == 0) {
STp->recover_count++;
STp->recover_erreg++;
} else
STp->abort_count++;
STps->rw = rw_state;
return retval;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_space_over_filemarks_backward(struct osst_tape * STp, struct osst_request ** aSRpnt,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
int mt_op, int mt_count)
{
char * name = tape_name(STp);
int cnt;
int last_mark_ppos = -1;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reached space_over_filemarks_backwards %d %d\n", name, mt_op, mt_count);
#endif
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't get logical blk num in space_filemarks_bwd\n", name);
#endif
return -EIO;
}
if (STp->linux_media_version >= 4) {
/*
* direct lookup in header filemark list
*/
cnt = ntohl(STp->buffer->aux->filemark_cnt);
if (STp->header_ok &&
STp->header_cache != NULL &&
(cnt - mt_count) >= 0 &&
(cnt - mt_count) < OS_FM_TAB_MAX &&
(cnt - mt_count) < STp->filemark_cnt &&
STp->header_cache->dat_fm_tab.fm_tab_ent[cnt-1] == STp->buffer->aux->last_mark_ppos)
last_mark_ppos = ntohl(STp->header_cache->dat_fm_tab.fm_tab_ent[cnt - mt_count]);
#if DEBUG
if (STp->header_cache == NULL || (cnt - mt_count) < 0 || (cnt - mt_count) >= OS_FM_TAB_MAX)
printk(OSST_DEB_MSG "%s:D: Filemark lookup fail due to %s\n", name,
STp->header_cache == NULL?"lack of header cache":"count out of range");
else
printk(OSST_DEB_MSG "%s:D: Filemark lookup: prev mark %d (%s), skip %d to %d\n",
name, cnt,
((cnt == -1 && ntohl(STp->buffer->aux->last_mark_ppos) == -1) ||
(STp->header_cache->dat_fm_tab.fm_tab_ent[cnt-1] ==
STp->buffer->aux->last_mark_ppos))?"match":"error",
mt_count, last_mark_ppos);
#endif
if (last_mark_ppos > 10 && last_mark_ppos < STp->eod_frame_ppos) {
osst_position_tape_and_confirm(STp, aSRpnt, last_mark_ppos);
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG
"%s:D: Couldn't get logical blk num in space_filemarks\n", name);
#endif
return (-EIO);
}
if (STp->buffer->aux->frame_type != OS_FRAME_TYPE_MARKER) {
printk(KERN_WARNING "%s:W: Expected to find marker at ppos %d, not found\n",
name, last_mark_ppos);
return (-EIO);
}
goto found;
}
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reverting to scan filemark backwards\n", name);
#endif
}
cnt = 0;
while (cnt != mt_count) {
last_mark_ppos = ntohl(STp->buffer->aux->last_mark_ppos);
if (last_mark_ppos == -1)
return (-EIO);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Positioning to last mark at %d\n", name, last_mark_ppos);
#endif
osst_position_tape_and_confirm(STp, aSRpnt, last_mark_ppos);
cnt++;
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't get logical blk num in space_filemarks\n", name);
#endif
return (-EIO);
}
if (STp->buffer->aux->frame_type != OS_FRAME_TYPE_MARKER) {
printk(KERN_WARNING "%s:W: Expected to find marker at ppos %d, not found\n",
name, last_mark_ppos);
return (-EIO);
}
}
found:
if (mt_op == MTBSFM) {
STp->frame_seq_number++;
STp->frame_in_buffer = 0;
STp->buffer->buffer_bytes = 0;
STp->buffer->read_pointer = 0;
STp->logical_blk_num += ntohs(STp->buffer->aux->dat.dat_list[0].blk_cnt);
}
return 0;
}
/*
* ADRL 1.1 compatible "slow" space filemarks fwd version
*
* Just scans for the filemark sequentially.
*/
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_space_over_filemarks_forward_slow(struct osst_tape * STp, struct osst_request ** aSRpnt,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
int mt_op, int mt_count)
{
int cnt = 0;
#if DEBUG
char * name = tape_name(STp);
printk(OSST_DEB_MSG "%s:D: Reached space_over_filemarks_forward_slow %d %d\n", name, mt_op, mt_count);
#endif
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't get logical blk num in space_filemarks_fwd\n", name);
#endif
return (-EIO);
}
while (1) {
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't get logical blk num in space_filemarks\n", name);
#endif
return (-EIO);
}
if (STp->buffer->aux->frame_type == OS_FRAME_TYPE_MARKER)
cnt++;
if (STp->buffer->aux->frame_type == OS_FRAME_TYPE_EOD) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: space_fwd: EOD reached\n", name);
#endif
if (STp->first_frame_position > STp->eod_frame_ppos+1) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: EOD position corrected (%d=>%d)\n",
name, STp->eod_frame_ppos, STp->first_frame_position-1);
#endif
STp->eod_frame_ppos = STp->first_frame_position-1;
}
return (-EIO);
}
if (cnt == mt_count)
break;
STp->frame_in_buffer = 0;
}
if (mt_op == MTFSF) {
STp->frame_seq_number++;
STp->frame_in_buffer = 0;
STp->buffer->buffer_bytes = 0;
STp->buffer->read_pointer = 0;
STp->logical_blk_num += ntohs(STp->buffer->aux->dat.dat_list[0].blk_cnt);
}
return 0;
}
/*
* Fast linux specific version of OnStream FSF
*/
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_space_over_filemarks_forward_fast(struct osst_tape * STp, struct osst_request ** aSRpnt,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
int mt_op, int mt_count)
{
char * name = tape_name(STp);
int cnt = 0,
next_mark_ppos = -1;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reached space_over_filemarks_forward_fast %d %d\n", name, mt_op, mt_count);
#endif
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't get logical blk num in space_filemarks_fwd\n", name);
#endif
return (-EIO);
}
if (STp->linux_media_version >= 4) {
/*
* direct lookup in header filemark list
*/
cnt = ntohl(STp->buffer->aux->filemark_cnt) - 1;
if (STp->header_ok &&
STp->header_cache != NULL &&
(cnt + mt_count) < OS_FM_TAB_MAX &&
(cnt + mt_count) < STp->filemark_cnt &&
((cnt == -1 && ntohl(STp->buffer->aux->last_mark_ppos) == -1) ||
(STp->header_cache->dat_fm_tab.fm_tab_ent[cnt] == STp->buffer->aux->last_mark_ppos)))
next_mark_ppos = ntohl(STp->header_cache->dat_fm_tab.fm_tab_ent[cnt + mt_count]);
#if DEBUG
if (STp->header_cache == NULL || (cnt + mt_count) >= OS_FM_TAB_MAX)
printk(OSST_DEB_MSG "%s:D: Filemark lookup fail due to %s\n", name,
STp->header_cache == NULL?"lack of header cache":"count out of range");
else
printk(OSST_DEB_MSG "%s:D: Filemark lookup: prev mark %d (%s), skip %d to %d\n",
name, cnt,
((cnt == -1 && ntohl(STp->buffer->aux->last_mark_ppos) == -1) ||
(STp->header_cache->dat_fm_tab.fm_tab_ent[cnt] ==
STp->buffer->aux->last_mark_ppos))?"match":"error",
mt_count, next_mark_ppos);
#endif
if (next_mark_ppos <= 10 || next_mark_ppos > STp->eod_frame_ppos) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reverting to slow filemark space\n", name);
#endif
return osst_space_over_filemarks_forward_slow(STp, aSRpnt, mt_op, mt_count);
} else {
osst_position_tape_and_confirm(STp, aSRpnt, next_mark_ppos);
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't get logical blk num in space_filemarks\n",
name);
#endif
return (-EIO);
}
if (STp->buffer->aux->frame_type != OS_FRAME_TYPE_MARKER) {
printk(KERN_WARNING "%s:W: Expected to find marker at ppos %d, not found\n",
name, next_mark_ppos);
return (-EIO);
}
if (ntohl(STp->buffer->aux->filemark_cnt) != cnt + mt_count) {
printk(KERN_WARNING "%s:W: Expected to find marker %d at ppos %d, not %d\n",
name, cnt+mt_count, next_mark_ppos,
ntohl(STp->buffer->aux->filemark_cnt));
return (-EIO);
}
}
} else {
/*
* Find nearest (usually previous) marker, then jump from marker to marker
*/
while (1) {
if (STp->buffer->aux->frame_type == OS_FRAME_TYPE_MARKER)
break;
if (STp->buffer->aux->frame_type == OS_FRAME_TYPE_EOD) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: space_fwd: EOD reached\n", name);
#endif
return (-EIO);
}
if (ntohl(STp->buffer->aux->filemark_cnt) == 0) {
if (STp->first_mark_ppos == -1) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reverting to slow filemark space\n", name);
#endif
return osst_space_over_filemarks_forward_slow(STp, aSRpnt, mt_op, mt_count);
}
osst_position_tape_and_confirm(STp, aSRpnt, STp->first_mark_ppos);
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG
"%s:D: Couldn't get logical blk num in space_filemarks_fwd_fast\n",
name);
#endif
return (-EIO);
}
if (STp->buffer->aux->frame_type != OS_FRAME_TYPE_MARKER) {
printk(KERN_WARNING "%s:W: Expected to find filemark at %d\n",
name, STp->first_mark_ppos);
return (-EIO);
}
} else {
if (osst_space_over_filemarks_backward(STp, aSRpnt, MTBSF, 1) < 0)
return (-EIO);
mt_count++;
}
}
cnt++;
while (cnt != mt_count) {
next_mark_ppos = ntohl(STp->buffer->aux->next_mark_ppos);
if (!next_mark_ppos || next_mark_ppos > STp->eod_frame_ppos) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reverting to slow filemark space\n", name);
#endif
return osst_space_over_filemarks_forward_slow(STp, aSRpnt, mt_op, mt_count - cnt);
}
#if DEBUG
else printk(OSST_DEB_MSG "%s:D: Positioning to next mark at %d\n", name, next_mark_ppos);
#endif
osst_position_tape_and_confirm(STp, aSRpnt, next_mark_ppos);
cnt++;
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't get logical blk num in space_filemarks\n",
name);
#endif
return (-EIO);
}
if (STp->buffer->aux->frame_type != OS_FRAME_TYPE_MARKER) {
printk(KERN_WARNING "%s:W: Expected to find marker at ppos %d, not found\n",
name, next_mark_ppos);
return (-EIO);
}
}
}
if (mt_op == MTFSF) {
STp->frame_seq_number++;
STp->frame_in_buffer = 0;
STp->buffer->buffer_bytes = 0;
STp->buffer->read_pointer = 0;
STp->logical_blk_num += ntohs(STp->buffer->aux->dat.dat_list[0].blk_cnt);
}
return 0;
}
/*
* In debug mode, we want to see as many errors as possible
* to test the error recovery mechanism.
*/
#if DEBUG
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static void osst_set_retries(struct osst_tape * STp, struct osst_request ** aSRpnt, int retries)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = * aSRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
char * name = tape_name(STp);
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = MODE_SELECT;
cmd[1] = 0x10;
cmd[4] = NUMBER_RETRIES_PAGE_LENGTH + MODE_HEADER_LENGTH;
(STp->buffer)->b_data[0] = cmd[4] - 1;
(STp->buffer)->b_data[1] = 0; /* Medium Type - ignoring */
(STp->buffer)->b_data[2] = 0; /* Reserved */
(STp->buffer)->b_data[3] = 0; /* Block Descriptor Length */
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 0] = NUMBER_RETRIES_PAGE | (1 << 7);
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 1] = 2;
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 2] = 4;
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 3] = retries;
if (debugging)
printk(OSST_DEB_MSG "%s:D: Setting number of retries on OnStream tape to %d\n", name, retries);
SRpnt = osst_do_scsi(SRpnt, STp, cmd, cmd[4], DMA_TO_DEVICE, STp->timeout, 0, 1);
*aSRpnt = SRpnt;
if ((STp->buffer)->syscall_result)
printk (KERN_ERR "%s:D: Couldn't set retries to %d\n", name, retries);
}
#endif
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_write_filemark(struct osst_tape * STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
int result;
int this_mark_ppos = STp->first_frame_position;
int this_mark_lbn = STp->logical_blk_num;
#if DEBUG
char * name = tape_name(STp);
#endif
if (STp->raw) return 0;
STp->write_type = OS_WRITE_NEW_MARK;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Writing Filemark %i at fppos %d (fseq %d, lblk %d)\n",
name, STp->filemark_cnt, this_mark_ppos, STp->frame_seq_number, this_mark_lbn);
#endif
STp->dirty = 1;
result = osst_flush_write_buffer(STp, aSRpnt);
result |= osst_flush_drive_buffer(STp, aSRpnt);
STp->last_mark_ppos = this_mark_ppos;
STp->last_mark_lbn = this_mark_lbn;
if (STp->header_cache != NULL && STp->filemark_cnt < OS_FM_TAB_MAX)
STp->header_cache->dat_fm_tab.fm_tab_ent[STp->filemark_cnt] = htonl(this_mark_ppos);
if (STp->filemark_cnt++ == 0)
STp->first_mark_ppos = this_mark_ppos;
return result;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_write_eod(struct osst_tape * STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
int result;
#if DEBUG
char * name = tape_name(STp);
#endif
if (STp->raw) return 0;
STp->write_type = OS_WRITE_EOD;
STp->eod_frame_ppos = STp->first_frame_position;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Writing EOD at fppos %d (fseq %d, lblk %d)\n", name,
STp->eod_frame_ppos, STp->frame_seq_number, STp->logical_blk_num);
#endif
STp->dirty = 1;
result = osst_flush_write_buffer(STp, aSRpnt);
result |= osst_flush_drive_buffer(STp, aSRpnt);
STp->eod_frame_lfa = --(STp->frame_seq_number);
return result;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_write_filler(struct osst_tape * STp, struct osst_request ** aSRpnt, int where, int count)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
char * name = tape_name(STp);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reached onstream write filler group %d\n", name, where);
#endif
osst_wait_ready(STp, aSRpnt, 60 * 5, 0);
osst_set_frame_position(STp, aSRpnt, where, 0);
STp->write_type = OS_WRITE_FILLER;
while (count--) {
memcpy(STp->buffer->b_data, "Filler", 6);
STp->buffer->buffer_bytes = 6;
STp->dirty = 1;
if (osst_flush_write_buffer(STp, aSRpnt)) {
printk(KERN_INFO "%s:I: Couldn't write filler frame\n", name);
return (-EIO);
}
}
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Exiting onstream write filler group\n", name);
#endif
return osst_flush_drive_buffer(STp, aSRpnt);
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int __osst_write_header(struct osst_tape * STp, struct osst_request ** aSRpnt, int where, int count)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
char * name = tape_name(STp);
int result;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reached onstream write header group %d\n", name, where);
#endif
osst_wait_ready(STp, aSRpnt, 60 * 5, 0);
osst_set_frame_position(STp, aSRpnt, where, 0);
STp->write_type = OS_WRITE_HEADER;
while (count--) {
osst_copy_to_buffer(STp->buffer, (unsigned char *)STp->header_cache);
STp->buffer->buffer_bytes = sizeof(os_header_t);
STp->dirty = 1;
if (osst_flush_write_buffer(STp, aSRpnt)) {
printk(KERN_INFO "%s:I: Couldn't write header frame\n", name);
return (-EIO);
}
}
result = osst_flush_drive_buffer(STp, aSRpnt);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Write onstream header group %s\n", name, result?"failed":"done");
#endif
return result;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_write_header(struct osst_tape * STp, struct osst_request ** aSRpnt, int locate_eod)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
os_header_t * header;
int result;
char * name = tape_name(STp);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Writing tape header\n", name);
#endif
if (STp->raw) return 0;
if (STp->header_cache == NULL) {
scsi: Remove unnecessary casts of void ptr returning alloc function return values The [vk][cmz]alloc(_node) family of functions return void pointers which it's completely unnecessary/pointless to cast to other pointer types since that happens implicitly. This patch removes such casts from drivers/scsi/ Signed-off-by: Jesper Juhl <jj@chaosbits.net> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
2010-11-08 23:09:25 +00:00
if ((STp->header_cache = vmalloc(sizeof(os_header_t))) == NULL) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_ERR "%s:E: Failed to allocate header cache\n", name);
return (-ENOMEM);
}
memset(STp->header_cache, 0, sizeof(os_header_t));
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Allocated and cleared memory for header cache\n", name);
#endif
}
if (STp->header_ok) STp->update_frame_cntr++;
else STp->update_frame_cntr = 0;
header = STp->header_cache;
strcpy(header->ident_str, "ADR_SEQ");
header->major_rev = 1;
header->minor_rev = 4;
header->ext_trk_tb_off = htons(17192);
header->pt_par_num = 1;
header->partition[0].partition_num = OS_DATA_PARTITION;
header->partition[0].par_desc_ver = OS_PARTITION_VERSION;
header->partition[0].wrt_pass_cntr = htons(STp->wrt_pass_cntr);
header->partition[0].first_frame_ppos = htonl(STp->first_data_ppos);
header->partition[0].last_frame_ppos = htonl(STp->capacity);
header->partition[0].eod_frame_ppos = htonl(STp->eod_frame_ppos);
header->cfg_col_width = htonl(20);
header->dat_col_width = htonl(1500);
header->qfa_col_width = htonl(0);
header->ext_track_tb.nr_stream_part = 1;
header->ext_track_tb.et_ent_sz = 32;
header->ext_track_tb.dat_ext_trk_ey.et_part_num = 0;
header->ext_track_tb.dat_ext_trk_ey.fmt = 1;
header->ext_track_tb.dat_ext_trk_ey.fm_tab_off = htons(17736);
header->ext_track_tb.dat_ext_trk_ey.last_hlb_hi = 0;
header->ext_track_tb.dat_ext_trk_ey.last_hlb = htonl(STp->eod_frame_lfa);
header->ext_track_tb.dat_ext_trk_ey.last_pp = htonl(STp->eod_frame_ppos);
header->dat_fm_tab.fm_part_num = 0;
header->dat_fm_tab.fm_tab_ent_sz = 4;
header->dat_fm_tab.fm_tab_ent_cnt = htons(STp->filemark_cnt<OS_FM_TAB_MAX?
STp->filemark_cnt:OS_FM_TAB_MAX);
result = __osst_write_header(STp, aSRpnt, 0xbae, 5);
if (STp->update_frame_cntr == 0)
osst_write_filler(STp, aSRpnt, 0xbb3, 5);
result &= __osst_write_header(STp, aSRpnt, 5, 5);
if (locate_eod) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Locating back to eod frame addr %d\n", name, STp->eod_frame_ppos);
#endif
osst_set_frame_position(STp, aSRpnt, STp->eod_frame_ppos, 0);
}
if (result)
printk(KERN_ERR "%s:E: Write header failed\n", name);
else {
memcpy(STp->application_sig, "LIN4", 4);
STp->linux_media = 1;
STp->linux_media_version = 4;
STp->header_ok = 1;
}
return result;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_reset_header(struct osst_tape * STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
if (STp->header_cache != NULL)
memset(STp->header_cache, 0, sizeof(os_header_t));
STp->logical_blk_num = STp->frame_seq_number = 0;
STp->frame_in_buffer = 0;
STp->eod_frame_ppos = STp->first_data_ppos = 0x0000000A;
STp->filemark_cnt = 0;
STp->first_mark_ppos = STp->last_mark_ppos = STp->last_mark_lbn = -1;
return osst_write_header(STp, aSRpnt, 1);
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int __osst_analyze_headers(struct osst_tape * STp, struct osst_request ** aSRpnt, int ppos)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
char * name = tape_name(STp);
os_header_t * header;
os_aux_t * aux;
char id_string[8];
int linux_media_version,
update_frame_cntr;
if (STp->raw)
return 1;
if (ppos == 5 || ppos == 0xbae || STp->buffer->syscall_result) {
if (osst_set_frame_position(STp, aSRpnt, ppos, 0))
printk(KERN_WARNING "%s:W: Couldn't position tape\n", name);
osst_wait_ready(STp, aSRpnt, 60 * 15, 0);
if (osst_initiate_read (STp, aSRpnt)) {
printk(KERN_WARNING "%s:W: Couldn't initiate read\n", name);
return 0;
}
}
if (osst_read_frame(STp, aSRpnt, 180)) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't read header frame\n", name);
#endif
return 0;
}
header = (os_header_t *) STp->buffer->b_data; /* warning: only first segment addressable */
aux = STp->buffer->aux;
if (aux->frame_type != OS_FRAME_TYPE_HEADER) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping non-header frame (%d)\n", name, ppos);
#endif
return 0;
}
if (ntohl(aux->frame_seq_num) != 0 ||
ntohl(aux->logical_blk_num) != 0 ||
aux->partition.partition_num != OS_CONFIG_PARTITION ||
ntohl(aux->partition.first_frame_ppos) != 0 ||
ntohl(aux->partition.last_frame_ppos) != 0xbb7 ) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Invalid header frame (%d,%d,%d,%d,%d)\n", name,
ntohl(aux->frame_seq_num), ntohl(aux->logical_blk_num),
aux->partition.partition_num, ntohl(aux->partition.first_frame_ppos),
ntohl(aux->partition.last_frame_ppos));
#endif
return 0;
}
if (strncmp(header->ident_str, "ADR_SEQ", 7) != 0 &&
strncmp(header->ident_str, "ADR-SEQ", 7) != 0) {
strlcpy(id_string, header->ident_str, 8);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Invalid header identification string %s\n", name, id_string);
#endif
return 0;
}
update_frame_cntr = ntohl(aux->update_frame_cntr);
if (update_frame_cntr < STp->update_frame_cntr) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame %d with update_frame_counter %d<%d\n",
name, ppos, update_frame_cntr, STp->update_frame_cntr);
#endif
return 0;
}
if (header->major_rev != 1 || header->minor_rev != 4 ) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: %s revision %d.%d detected (1.4 supported)\n",
name, (header->major_rev != 1 || header->minor_rev < 2 ||
header->minor_rev > 4 )? "Invalid" : "Warning:",
header->major_rev, header->minor_rev);
#endif
if (header->major_rev != 1 || header->minor_rev < 2 || header->minor_rev > 4)
return 0;
}
#if DEBUG
if (header->pt_par_num != 1)
printk(KERN_INFO "%s:W: %d partitions defined, only one supported\n",
name, header->pt_par_num);
#endif
memcpy(id_string, aux->application_sig, 4);
id_string[4] = 0;
if (memcmp(id_string, "LIN", 3) == 0) {
STp->linux_media = 1;
linux_media_version = id_string[3] - '0';
if (linux_media_version != 4)
printk(KERN_INFO "%s:I: Linux media version %d detected (current 4)\n",
name, linux_media_version);
} else {
printk(KERN_WARNING "%s:W: Non Linux media detected (%s)\n", name, id_string);
return 0;
}
if (linux_media_version < STp->linux_media_version) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping frame %d with linux_media_version %d\n",
name, ppos, linux_media_version);
#endif
return 0;
}
if (linux_media_version > STp->linux_media_version) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Frame %d sets linux_media_version to %d\n",
name, ppos, linux_media_version);
#endif
memcpy(STp->application_sig, id_string, 5);
STp->linux_media_version = linux_media_version;
STp->update_frame_cntr = -1;
}
if (update_frame_cntr > STp->update_frame_cntr) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Frame %d sets update_frame_counter to %d\n",
name, ppos, update_frame_cntr);
#endif
if (STp->header_cache == NULL) {
scsi: Remove unnecessary casts of void ptr returning alloc function return values The [vk][cmz]alloc(_node) family of functions return void pointers which it's completely unnecessary/pointless to cast to other pointer types since that happens implicitly. This patch removes such casts from drivers/scsi/ Signed-off-by: Jesper Juhl <jj@chaosbits.net> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
2010-11-08 23:09:25 +00:00
if ((STp->header_cache = vmalloc(sizeof(os_header_t))) == NULL) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_ERR "%s:E: Failed to allocate header cache\n", name);
return 0;
}
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Allocated memory for header cache\n", name);
#endif
}
osst_copy_from_buffer(STp->buffer, (unsigned char *)STp->header_cache);
header = STp->header_cache; /* further accesses from cached (full) copy */
STp->wrt_pass_cntr = ntohs(header->partition[0].wrt_pass_cntr);
STp->first_data_ppos = ntohl(header->partition[0].first_frame_ppos);
STp->eod_frame_ppos = ntohl(header->partition[0].eod_frame_ppos);
STp->eod_frame_lfa = ntohl(header->ext_track_tb.dat_ext_trk_ey.last_hlb);
STp->filemark_cnt = ntohl(aux->filemark_cnt);
STp->first_mark_ppos = ntohl(aux->next_mark_ppos);
STp->last_mark_ppos = ntohl(aux->last_mark_ppos);
STp->last_mark_lbn = ntohl(aux->last_mark_lbn);
STp->update_frame_cntr = update_frame_cntr;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Detected write pass %d, update frame counter %d, filemark counter %d\n",
name, STp->wrt_pass_cntr, STp->update_frame_cntr, STp->filemark_cnt);
printk(OSST_DEB_MSG "%s:D: first data frame on tape = %d, last = %d, eod frame = %d\n", name,
STp->first_data_ppos,
ntohl(header->partition[0].last_frame_ppos),
ntohl(header->partition[0].eod_frame_ppos));
printk(OSST_DEB_MSG "%s:D: first mark on tape = %d, last = %d, eod frame = %d\n",
name, STp->first_mark_ppos, STp->last_mark_ppos, STp->eod_frame_ppos);
#endif
if (header->minor_rev < 4 && STp->linux_media_version == 4) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Moving filemark list to ADR 1.4 location\n", name);
#endif
memcpy((void *)header->dat_fm_tab.fm_tab_ent,
(void *)header->old_filemark_list, sizeof(header->dat_fm_tab.fm_tab_ent));
memset((void *)header->old_filemark_list, 0, sizeof(header->old_filemark_list));
}
if (header->minor_rev == 4 &&
(header->ext_trk_tb_off != htons(17192) ||
header->partition[0].partition_num != OS_DATA_PARTITION ||
header->partition[0].par_desc_ver != OS_PARTITION_VERSION ||
header->partition[0].last_frame_ppos != htonl(STp->capacity) ||
header->cfg_col_width != htonl(20) ||
header->dat_col_width != htonl(1500) ||
header->qfa_col_width != htonl(0) ||
header->ext_track_tb.nr_stream_part != 1 ||
header->ext_track_tb.et_ent_sz != 32 ||
header->ext_track_tb.dat_ext_trk_ey.et_part_num != OS_DATA_PARTITION ||
header->ext_track_tb.dat_ext_trk_ey.fmt != 1 ||
header->ext_track_tb.dat_ext_trk_ey.fm_tab_off != htons(17736) ||
header->ext_track_tb.dat_ext_trk_ey.last_hlb_hi != 0 ||
header->ext_track_tb.dat_ext_trk_ey.last_pp != htonl(STp->eod_frame_ppos) ||
header->dat_fm_tab.fm_part_num != OS_DATA_PARTITION ||
header->dat_fm_tab.fm_tab_ent_sz != 4 ||
header->dat_fm_tab.fm_tab_ent_cnt !=
htons(STp->filemark_cnt<OS_FM_TAB_MAX?STp->filemark_cnt:OS_FM_TAB_MAX)))
printk(KERN_WARNING "%s:W: Failed consistency check ADR 1.4 format\n", name);
}
return 1;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_analyze_headers(struct osst_tape * STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
int position, ppos;
int first, last;
int valid = 0;
char * name = tape_name(STp);
position = osst_get_frame_position(STp, aSRpnt);
if (STp->raw) {
STp->header_ok = STp->linux_media = 1;
STp->linux_media_version = 0;
return 1;
}
STp->header_ok = STp->linux_media = STp->linux_media_version = 0;
STp->wrt_pass_cntr = STp->update_frame_cntr = -1;
STp->eod_frame_ppos = STp->first_data_ppos = -1;
STp->first_mark_ppos = STp->last_mark_ppos = STp->last_mark_lbn = -1;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reading header\n", name);
#endif
/* optimization for speed - if we are positioned at ppos 10, read second group first */
/* TODO try the ADR 1.1 locations for the second group if we have no valid one yet... */
first = position==10?0xbae: 5;
last = position==10?0xbb3:10;
for (ppos = first; ppos < last; ppos++)
if (__osst_analyze_headers(STp, aSRpnt, ppos))
valid = 1;
first = position==10? 5:0xbae;
last = position==10?10:0xbb3;
for (ppos = first; ppos < last; ppos++)
if (__osst_analyze_headers(STp, aSRpnt, ppos))
valid = 1;
if (!valid) {
printk(KERN_ERR "%s:E: Failed to find valid ADRL header, new media?\n", name);
STp->eod_frame_ppos = STp->first_data_ppos = 0;
osst_set_frame_position(STp, aSRpnt, 10, 0);
return 0;
}
if (position <= STp->first_data_ppos) {
position = STp->first_data_ppos;
STp->ps[0].drv_file = STp->ps[0].drv_block = STp->frame_seq_number = STp->logical_blk_num = 0;
}
osst_set_frame_position(STp, aSRpnt, position, 0);
STp->header_ok = 1;
return 1;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_verify_position(struct osst_tape * STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
int frame_position = STp->first_frame_position;
int frame_seq_numbr = STp->frame_seq_number;
int logical_blk_num = STp->logical_blk_num;
int halfway_frame = STp->frame_in_buffer;
int read_pointer = STp->buffer->read_pointer;
int prev_mark_ppos = -1;
int actual_mark_ppos, i, n;
#if DEBUG
char * name = tape_name(STp);
printk(OSST_DEB_MSG "%s:D: Verify that the tape is really the one we think before writing\n", name);
#endif
osst_set_frame_position(STp, aSRpnt, frame_position - 1, 0);
if (osst_get_logical_frame(STp, aSRpnt, -1, 0) < 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't get logical blk num in verify_position\n", name);
#endif
return (-EIO);
}
if (STp->linux_media_version >= 4) {
for (i=0; i<STp->filemark_cnt; i++)
if ((n=ntohl(STp->header_cache->dat_fm_tab.fm_tab_ent[i])) < frame_position)
prev_mark_ppos = n;
} else
prev_mark_ppos = frame_position - 1; /* usually - we don't really know */
actual_mark_ppos = STp->buffer->aux->frame_type == OS_FRAME_TYPE_MARKER ?
frame_position - 1 : ntohl(STp->buffer->aux->last_mark_ppos);
if (frame_position != STp->first_frame_position ||
frame_seq_numbr != STp->frame_seq_number + (halfway_frame?0:1) ||
prev_mark_ppos != actual_mark_ppos ) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Block mismatch: fppos %d-%d, fseq %d-%d, mark %d-%d\n", name,
STp->first_frame_position, frame_position,
STp->frame_seq_number + (halfway_frame?0:1),
frame_seq_numbr, actual_mark_ppos, prev_mark_ppos);
#endif
return (-EIO);
}
if (halfway_frame) {
/* prepare buffer for append and rewrite on top of original */
osst_set_frame_position(STp, aSRpnt, frame_position - 1, 0);
STp->buffer->buffer_bytes = read_pointer;
STp->ps[STp->partition].rw = ST_WRITING;
STp->dirty = 1;
}
STp->frame_in_buffer = halfway_frame;
STp->frame_seq_number = frame_seq_numbr;
STp->logical_blk_num = logical_blk_num;
return 0;
}
/* Acc. to OnStream, the vers. numbering is the following:
* X.XX for released versions (X=digit),
* XXXY for unreleased versions (Y=letter)
* Ordering 1.05 < 106A < 106B < ... < 106a < ... < 1.06
* This fn makes monoton numbers out of this scheme ...
*/
static unsigned int osst_parse_firmware_rev (const char * str)
{
if (str[1] == '.') {
return (str[0]-'0')*10000
+(str[2]-'0')*1000
+(str[3]-'0')*100;
} else {
return (str[0]-'0')*10000
+(str[1]-'0')*1000
+(str[2]-'0')*100 - 100
+(str[3]-'@');
}
}
/*
* Configure the OnStream SCII tape drive for default operation
*/
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_configure_onstream(struct osst_tape *STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned char cmd[MAX_COMMAND_SIZE];
char * name = tape_name(STp);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = * aSRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
osst_mode_parameter_header_t * header;
osst_block_size_page_t * bs;
osst_capabilities_page_t * cp;
osst_tape_paramtr_page_t * prm;
int drive_buffer_size;
if (STp->ready != ST_READY) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Not Ready\n", name);
#endif
return (-EIO);
}
if (STp->os_fw_rev < 10600) {
printk(KERN_INFO "%s:I: Old OnStream firmware revision detected (%s),\n", name, STp->device->rev);
printk(KERN_INFO "%s:I: an upgrade to version 1.06 or above is recommended\n", name);
}
/*
* Configure 32.5KB (data+aux) frame size.
* Get the current frame size from the block size mode page
*/
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = MODE_SENSE;
cmd[1] = 8;
cmd[2] = BLOCK_SIZE_PAGE;
cmd[4] = BLOCK_SIZE_PAGE_LENGTH + MODE_HEADER_LENGTH;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, cmd[4], DMA_FROM_DEVICE, STp->timeout, 0, 1);
if (SRpnt == NULL) {
#if DEBUG
printk(OSST_DEB_MSG "osst :D: Busy\n");
#endif
return (-EBUSY);
}
*aSRpnt = SRpnt;
if ((STp->buffer)->syscall_result != 0) {
printk (KERN_ERR "%s:E: Can't get tape block size mode page\n", name);
return (-EIO);
}
header = (osst_mode_parameter_header_t *) (STp->buffer)->b_data;
bs = (osst_block_size_page_t *) ((STp->buffer)->b_data + sizeof(osst_mode_parameter_header_t) + header->bdl);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: 32KB play back: %s\n", name, bs->play32 ? "Yes" : "No");
printk(OSST_DEB_MSG "%s:D: 32.5KB play back: %s\n", name, bs->play32_5 ? "Yes" : "No");
printk(OSST_DEB_MSG "%s:D: 32KB record: %s\n", name, bs->record32 ? "Yes" : "No");
printk(OSST_DEB_MSG "%s:D: 32.5KB record: %s\n", name, bs->record32_5 ? "Yes" : "No");
#endif
/*
* Configure default auto columns mode, 32.5KB transfer mode
*/
bs->one = 1;
bs->play32 = 0;
bs->play32_5 = 1;
bs->record32 = 0;
bs->record32_5 = 1;
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = MODE_SELECT;
cmd[1] = 0x10;
cmd[4] = BLOCK_SIZE_PAGE_LENGTH + MODE_HEADER_LENGTH;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, cmd[4], DMA_TO_DEVICE, STp->timeout, 0, 1);
*aSRpnt = SRpnt;
if ((STp->buffer)->syscall_result != 0) {
printk (KERN_ERR "%s:E: Couldn't set tape block size mode page\n", name);
return (-EIO);
}
#if DEBUG
printk(KERN_INFO "%s:D: Drive Block Size changed to 32.5K\n", name);
/*
* In debug mode, we want to see as many errors as possible
* to test the error recovery mechanism.
*/
osst_set_retries(STp, aSRpnt, 0);
SRpnt = * aSRpnt;
#endif
/*
* Set vendor name to 'LIN4' for "Linux support version 4".
*/
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = MODE_SELECT;
cmd[1] = 0x10;
cmd[4] = VENDOR_IDENT_PAGE_LENGTH + MODE_HEADER_LENGTH;
header->mode_data_length = VENDOR_IDENT_PAGE_LENGTH + MODE_HEADER_LENGTH - 1;
header->medium_type = 0; /* Medium Type - ignoring */
header->dsp = 0; /* Reserved */
header->bdl = 0; /* Block Descriptor Length */
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 0] = VENDOR_IDENT_PAGE | (1 << 7);
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 1] = 6;
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 2] = 'L';
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 3] = 'I';
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 4] = 'N';
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 5] = '4';
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 6] = 0;
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 7] = 0;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, cmd[4], DMA_TO_DEVICE, STp->timeout, 0, 1);
*aSRpnt = SRpnt;
if ((STp->buffer)->syscall_result != 0) {
printk (KERN_ERR "%s:E: Couldn't set vendor name to %s\n", name,
(char *) ((STp->buffer)->b_data + MODE_HEADER_LENGTH + 2));
return (-EIO);
}
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = MODE_SENSE;
cmd[1] = 8;
cmd[2] = CAPABILITIES_PAGE;
cmd[4] = CAPABILITIES_PAGE_LENGTH + MODE_HEADER_LENGTH;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, cmd[4], DMA_FROM_DEVICE, STp->timeout, 0, 1);
*aSRpnt = SRpnt;
if ((STp->buffer)->syscall_result != 0) {
printk (KERN_ERR "%s:E: Can't get capabilities page\n", name);
return (-EIO);
}
header = (osst_mode_parameter_header_t *) (STp->buffer)->b_data;
cp = (osst_capabilities_page_t *) ((STp->buffer)->b_data +
sizeof(osst_mode_parameter_header_t) + header->bdl);
drive_buffer_size = ntohs(cp->buffer_size) / 2;
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = MODE_SENSE;
cmd[1] = 8;
cmd[2] = TAPE_PARAMTR_PAGE;
cmd[4] = TAPE_PARAMTR_PAGE_LENGTH + MODE_HEADER_LENGTH;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, cmd[4], DMA_FROM_DEVICE, STp->timeout, 0, 1);
*aSRpnt = SRpnt;
if ((STp->buffer)->syscall_result != 0) {
printk (KERN_ERR "%s:E: Can't get tape parameter page\n", name);
return (-EIO);
}
header = (osst_mode_parameter_header_t *) (STp->buffer)->b_data;
prm = (osst_tape_paramtr_page_t *) ((STp->buffer)->b_data +
sizeof(osst_mode_parameter_header_t) + header->bdl);
STp->density = prm->density;
STp->capacity = ntohs(prm->segtrk) * ntohs(prm->trks);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Density %d, tape length: %dMB, drive buffer size: %dKB\n",
name, STp->density, STp->capacity / 32, drive_buffer_size);
#endif
return 0;
}
/* Step over EOF if it has been inadvertently crossed (ioctl not used because
it messes up the block number). */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int cross_eof(struct osst_tape *STp, struct osst_request ** aSRpnt, int forward)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
int result;
char * name = tape_name(STp);
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Stepping over filemark %s.\n",
name, forward ? "forward" : "backward");
#endif
if (forward) {
/* assumes that the filemark is already read by the drive, so this is low cost */
result = osst_space_over_filemarks_forward_slow(STp, aSRpnt, MTFSF, 1);
}
else
/* assumes this is only called if we just read the filemark! */
result = osst_seek_logical_blk(STp, aSRpnt, STp->logical_blk_num - 1);
if (result < 0)
printk(KERN_WARNING "%s:W: Stepping over filemark %s failed.\n",
name, forward ? "forward" : "backward");
return result;
}
/* Get the tape position. */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_get_frame_position(struct osst_tape *STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned char scmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
int result = 0;
char * name = tape_name(STp);
/* KG: We want to be able to use it for checking Write Buffer availability
* and thus don't want to risk to overwrite anything. Exchange buffers ... */
char mybuf[24];
char * olddata = STp->buffer->b_data;
int oldsize = STp->buffer->buffer_size;
if (STp->ready != ST_READY) return (-EIO);
memset (scmd, 0, MAX_COMMAND_SIZE);
scmd[0] = READ_POSITION;
STp->buffer->b_data = mybuf; STp->buffer->buffer_size = 24;
SRpnt = osst_do_scsi(*aSRpnt, STp, scmd, 20, DMA_FROM_DEVICE,
STp->timeout, MAX_RETRIES, 1);
if (!SRpnt) {
STp->buffer->b_data = olddata; STp->buffer->buffer_size = oldsize;
return (-EBUSY);
}
*aSRpnt = SRpnt;
if (STp->buffer->syscall_result)
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
result = ((SRpnt->sense[2] & 0x0f) == 3) ? -EIO : -EINVAL; /* 3: Write Error */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (result == -EINVAL)
printk(KERN_ERR "%s:E: Can't read tape position.\n", name);
else {
if (result == -EIO) { /* re-read position - this needs to preserve media errors */
unsigned char mysense[16];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
memcpy (mysense, SRpnt->sense, 16);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
memset (scmd, 0, MAX_COMMAND_SIZE);
scmd[0] = READ_POSITION;
STp->buffer->b_data = mybuf; STp->buffer->buffer_size = 24;
SRpnt = osst_do_scsi(SRpnt, STp, scmd, 20, DMA_FROM_DEVICE,
STp->timeout, MAX_RETRIES, 1);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reread position, reason=[%02x:%02x:%02x], result=[%s%02x:%02x:%02x]\n",
name, mysense[2], mysense[12], mysense[13], STp->buffer->syscall_result?"":"ok:",
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
SRpnt->sense[2],SRpnt->sense[12],SRpnt->sense[13]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
if (!STp->buffer->syscall_result)
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
memcpy (SRpnt->sense, mysense, 16);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
else
printk(KERN_WARNING "%s:W: Double error in get position\n", name);
}
STp->first_frame_position = ((STp->buffer)->b_data[4] << 24)
+ ((STp->buffer)->b_data[5] << 16)
+ ((STp->buffer)->b_data[6] << 8)
+ (STp->buffer)->b_data[7];
STp->last_frame_position = ((STp->buffer)->b_data[ 8] << 24)
+ ((STp->buffer)->b_data[ 9] << 16)
+ ((STp->buffer)->b_data[10] << 8)
+ (STp->buffer)->b_data[11];
STp->cur_frames = (STp->buffer)->b_data[15];
#if DEBUG
if (debugging) {
printk(OSST_DEB_MSG "%s:D: Drive Positions: host %d, tape %d%s, buffer %d\n", name,
STp->first_frame_position, STp->last_frame_position,
((STp->buffer)->b_data[0]&0x80)?" (BOP)":
((STp->buffer)->b_data[0]&0x40)?" (EOP)":"",
STp->cur_frames);
}
#endif
if (STp->cur_frames == 0 && STp->first_frame_position != STp->last_frame_position) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Correcting read position %d, %d, %d\n", name,
STp->first_frame_position, STp->last_frame_position, STp->cur_frames);
#endif
STp->first_frame_position = STp->last_frame_position;
}
}
STp->buffer->b_data = olddata; STp->buffer->buffer_size = oldsize;
return (result == 0 ? STp->first_frame_position : result);
}
/* Set the tape block */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_set_frame_position(struct osst_tape *STp, struct osst_request ** aSRpnt, int ppos, int skip)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned char scmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct st_partstat * STps;
int result = 0;
int pp = (ppos == 3000 && !skip)? 0 : ppos;
char * name = tape_name(STp);
if (STp->ready != ST_READY) return (-EIO);
STps = &(STp->ps[STp->partition]);
if (ppos < 0 || ppos > STp->capacity) {
printk(KERN_WARNING "%s:W: Reposition request %d out of range\n", name, ppos);
pp = ppos = ppos < 0 ? 0 : (STp->capacity - 1);
result = (-EINVAL);
}
do {
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Setting ppos to %d.\n", name, pp);
#endif
memset (scmd, 0, MAX_COMMAND_SIZE);
scmd[0] = SEEK_10;
scmd[1] = 1;
scmd[3] = (pp >> 24);
scmd[4] = (pp >> 16);
scmd[5] = (pp >> 8);
scmd[6] = pp;
if (skip)
scmd[9] = 0x80;
SRpnt = osst_do_scsi(*aSRpnt, STp, scmd, 0, DMA_NONE, STp->long_timeout,
MAX_RETRIES, 1);
if (!SRpnt)
return (-EBUSY);
*aSRpnt = SRpnt;
if ((STp->buffer)->syscall_result != 0) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: SEEK command from %d to %d failed.\n",
name, STp->first_frame_position, pp);
#endif
result = (-EIO);
}
if (pp != ppos)
osst_wait_ready(STp, aSRpnt, 5 * 60, OSST_WAIT_POSITION_COMPLETE);
} while ((pp != ppos) && (pp = ppos));
STp->first_frame_position = STp->last_frame_position = ppos;
STps->eof = ST_NOEOF;
STps->at_sm = 0;
STps->rw = ST_IDLE;
STp->frame_in_buffer = 0;
return result;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_write_trailer(struct osst_tape *STp, struct osst_request ** aSRpnt, int leave_at_EOT)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct st_partstat * STps = &(STp->ps[STp->partition]);
int result = 0;
if (STp->write_type != OS_WRITE_NEW_MARK) {
/* true unless the user wrote the filemark for us */
result = osst_flush_drive_buffer(STp, aSRpnt);
if (result < 0) goto out;
result = osst_write_filemark(STp, aSRpnt);
if (result < 0) goto out;
if (STps->drv_file >= 0)
STps->drv_file++ ;
STps->drv_block = 0;
}
result = osst_write_eod(STp, aSRpnt);
osst_write_header(STp, aSRpnt, leave_at_EOT);
STps->eof = ST_FM;
out:
return result;
}
/* osst versions of st functions - augmented and stripped to suit OnStream only */
/* Flush the write buffer (never need to write if variable blocksize). */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_flush_write_buffer(struct osst_tape *STp, struct osst_request ** aSRpnt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
int offset, transfer, blks = 0;
int result = 0;
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = *aSRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct st_partstat * STps;
char * name = tape_name(STp);
if ((STp->buffer)->writing) {
if (SRpnt == (STp->buffer)->last_SRpnt)
#if DEBUG
{ printk(OSST_DEB_MSG
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
"%s:D: aSRpnt points to osst_request that write_behind_check will release -- cleared\n", name);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
*aSRpnt = SRpnt = NULL;
#if DEBUG
} else if (SRpnt)
printk(OSST_DEB_MSG
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
"%s:D: aSRpnt does not point to osst_request that write_behind_check will release -- strange\n", name);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
osst_write_behind_check(STp);
if ((STp->buffer)->syscall_result) {
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Async write error (flush) %x.\n",
name, (STp->buffer)->midlevel_result);
#endif
if ((STp->buffer)->midlevel_result == INT_MAX)
return (-ENOSPC);
return (-EIO);
}
}
result = 0;
if (STp->dirty == 1) {
STp->write_count++;
STps = &(STp->ps[STp->partition]);
STps->rw = ST_WRITING;
offset = STp->buffer->buffer_bytes;
blks = (offset + STp->block_size - 1) / STp->block_size;
transfer = OS_FRAME_SIZE;
if (offset < OS_DATA_SIZE)
osst_zero_buffer_tail(STp->buffer);
if (STp->poll)
if (osst_wait_frame (STp, aSRpnt, STp->first_frame_position, -50, 120))
result = osst_recover_wait_frame(STp, aSRpnt, 1);
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = WRITE_6;
cmd[1] = 1;
cmd[4] = 1;
switch (STp->write_type) {
case OS_WRITE_DATA:
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Writing %d blocks to frame %d, lblks %d-%d\n",
name, blks, STp->frame_seq_number,
STp->logical_blk_num - blks, STp->logical_blk_num - 1);
#endif
osst_init_aux(STp, OS_FRAME_TYPE_DATA, STp->frame_seq_number++,
STp->logical_blk_num - blks, STp->block_size, blks);
break;
case OS_WRITE_EOD:
osst_init_aux(STp, OS_FRAME_TYPE_EOD, STp->frame_seq_number++,
STp->logical_blk_num, 0, 0);
break;
case OS_WRITE_NEW_MARK:
osst_init_aux(STp, OS_FRAME_TYPE_MARKER, STp->frame_seq_number++,
STp->logical_blk_num++, 0, blks=1);
break;
case OS_WRITE_HEADER:
osst_init_aux(STp, OS_FRAME_TYPE_HEADER, 0, 0, 0, blks=0);
break;
default: /* probably FILLER */
osst_init_aux(STp, OS_FRAME_TYPE_FILL, 0, 0, 0, 0);
}
#if DEBUG
if (debugging)
Fix common misspellings Fixes generated by 'codespell' and manually reviewed. Signed-off-by: Lucas De Marchi <lucas.demarchi@profusion.mobi>
2011-03-31 01:57:33 +00:00
printk(OSST_DEB_MSG "%s:D: Flushing %d bytes, Transferring %d bytes in %d lblocks.\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
name, offset, transfer, blks);
#endif
SRpnt = osst_do_scsi(*aSRpnt, STp, cmd, transfer, DMA_TO_DEVICE,
STp->timeout, MAX_RETRIES, 1);
*aSRpnt = SRpnt;
if (!SRpnt)
return (-EBUSY);
if ((STp->buffer)->syscall_result != 0) {
#if DEBUG
printk(OSST_DEB_MSG
"%s:D: write sense [0]=0x%02x [2]=%02x [12]=%02x [13]=%02x\n",
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
name, SRpnt->sense[0], SRpnt->sense[2],
SRpnt->sense[12], SRpnt->sense[13]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[0] & 0x70) == 0x70 &&
(SRpnt->sense[2] & 0x40) && /* FIXME - SC-30 drive doesn't assert EOM bit */
(SRpnt->sense[2] & 0x0f) == NO_SENSE) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->dirty = 0;
(STp->buffer)->buffer_bytes = 0;
result = (-ENOSPC);
}
else {
if (osst_write_error_recovery(STp, aSRpnt, 1)) {
printk(KERN_ERR "%s:E: Error on flush write.\n", name);
result = (-EIO);
}
}
STps->drv_block = (-1); /* FIXME - even if write recovery succeeds? */
}
else {
STp->first_frame_position++;
STp->dirty = 0;
(STp->buffer)->buffer_bytes = 0;
}
}
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Exit flush write buffer with code %d\n", name, result);
#endif
return result;
}
/* Flush the tape buffer. The tape will be positioned correctly unless
seek_next is true. */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_flush_buffer(struct osst_tape * STp, struct osst_request ** aSRpnt, int seek_next)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct st_partstat * STps;
int backspace = 0, result = 0;
#if DEBUG
char * name = tape_name(STp);
#endif
/*
* If there was a bus reset, block further access
* to this device.
*/
if( STp->pos_unknown)
return (-EIO);
if (STp->ready != ST_READY)
return 0;
STps = &(STp->ps[STp->partition]);
if (STps->rw == ST_WRITING || STp->dirty) { /* Writing */
STp->write_type = OS_WRITE_DATA;
return osst_flush_write_buffer(STp, aSRpnt);
}
if (STp->block_size == 0)
return 0;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reached flush (read) buffer\n", name);
#endif
if (!STp->can_bsr) {
backspace = ((STp->buffer)->buffer_bytes + (STp->buffer)->read_pointer) / STp->block_size -
((STp->buffer)->read_pointer + STp->block_size - 1 ) / STp->block_size ;
(STp->buffer)->buffer_bytes = 0;
(STp->buffer)->read_pointer = 0;
STp->frame_in_buffer = 0; /* FIXME is this relevant w. OSST? */
}
if (!seek_next) {
if (STps->eof == ST_FM_HIT) {
result = cross_eof(STp, aSRpnt, 0); /* Back over the EOF hit */
if (!result)
STps->eof = ST_NOEOF;
else {
if (STps->drv_file >= 0)
STps->drv_file++;
STps->drv_block = 0;
}
}
if (!result && backspace > 0) /* TODO -- design and run a test case for this */
result = osst_seek_logical_blk(STp, aSRpnt, STp->logical_blk_num - backspace);
}
else if (STps->eof == ST_FM_HIT) {
if (STps->drv_file >= 0)
STps->drv_file++;
STps->drv_block = 0;
STps->eof = ST_NOEOF;
}
return result;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_write_frame(struct osst_tape * STp, struct osst_request ** aSRpnt, int synchronous)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
int blks;
#if DEBUG
char * name = tape_name(STp);
#endif
if ((!STp-> raw) && (STp->first_frame_position == 0xbae)) { /* _must_ preserve buffer! */
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Reaching config partition.\n", name);
#endif
if (osst_flush_drive_buffer(STp, aSRpnt) < 0) {
return (-EIO);
}
/* error recovery may have bumped us past the header partition */
if (osst_get_frame_position(STp, aSRpnt) < 0xbb8) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Skipping over config partition.\n", name);
#endif
osst_position_tape_and_confirm(STp, aSRpnt, 0xbb8);
}
}
if (STp->poll)
if (osst_wait_frame (STp, aSRpnt, STp->first_frame_position, -48, 120))
if (osst_recover_wait_frame(STp, aSRpnt, 1))
return (-EIO);
// osst_build_stats(STp, &SRpnt);
STp->ps[STp->partition].rw = ST_WRITING;
STp->write_type = OS_WRITE_DATA;
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = WRITE_6;
cmd[1] = 1;
cmd[4] = 1; /* one frame at a time... */
blks = STp->buffer->buffer_bytes / STp->block_size;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Writing %d blocks to frame %d, lblks %d-%d\n", name, blks,
STp->frame_seq_number, STp->logical_blk_num - blks, STp->logical_blk_num - 1);
#endif
osst_init_aux(STp, OS_FRAME_TYPE_DATA, STp->frame_seq_number++,
STp->logical_blk_num - blks, STp->block_size, blks);
#if DEBUG
if (!synchronous)
STp->write_pending = 1;
#endif
SRpnt = osst_do_scsi(*aSRpnt, STp, cmd, OS_FRAME_SIZE, DMA_TO_DEVICE, STp->timeout,
MAX_RETRIES, synchronous);
if (!SRpnt)
return (-EBUSY);
*aSRpnt = SRpnt;
if (synchronous) {
if (STp->buffer->syscall_result != 0) {
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Error on write:\n", name);
#endif
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[0] & 0x70) == 0x70 &&
(SRpnt->sense[2] & 0x40)) {
if ((SRpnt->sense[2] & 0x0f) == VOLUME_OVERFLOW)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return (-ENOSPC);
}
else {
if (osst_write_error_recovery(STp, aSRpnt, 1))
return (-EIO);
}
}
else
STp->first_frame_position++;
}
STp->write_count++;
return 0;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
/* Lock or unlock the drive door. Don't use when struct osst_request allocated. */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
static int do_door_lock(struct osst_tape * STp, int do_lock)
{
osst: call scsi_set_medium_removal directly Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com> Reviewed-by: Hannes Reinecke <hare@suse.de>
2014-10-11 14:00:33 +00:00
int retval;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
printk(OSST_DEB_MSG "%s:D: %socking drive door.\n", tape_name(STp), do_lock ? "L" : "Unl");
#endif
osst: call scsi_set_medium_removal directly Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com> Reviewed-by: Hannes Reinecke <hare@suse.de>
2014-10-11 14:00:33 +00:00
retval = scsi_set_medium_removal(STp->device,
do_lock ? SCSI_REMOVAL_PREVENT : SCSI_REMOVAL_ALLOW);
if (!retval)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->door_locked = do_lock ? ST_LOCKED_EXPLICIT : ST_UNLOCKED;
osst: call scsi_set_medium_removal directly Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com> Reviewed-by: Hannes Reinecke <hare@suse.de>
2014-10-11 14:00:33 +00:00
else
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->door_locked = ST_LOCK_FAILS;
return retval;
}
/* Set the internal state after reset */
static void reset_state(struct osst_tape *STp)
{
int i;
struct st_partstat *STps;
STp->pos_unknown = 0;
for (i = 0; i < ST_NBR_PARTITIONS; i++) {
STps = &(STp->ps[i]);
STps->rw = ST_IDLE;
STps->eof = ST_NOEOF;
STps->at_sm = 0;
STps->last_block_valid = 0;
STps->drv_block = -1;
STps->drv_file = -1;
}
}
/* Entry points to osst */
/* Write command */
static ssize_t osst_write(struct file * filp, const char __user * buf, size_t count, loff_t *ppos)
{
ssize_t total, retval = 0;
ssize_t i, do_count, blks, transfer;
int write_threshold;
int doing_write = 0;
const char __user * b_point;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct st_modedef * STm;
struct st_partstat * STps;
struct osst_tape * STp = filp->private_data;
char * name = tape_name(STp);
[SCSI] osst: Use mutex instead of semaphore The OnStream SCSI Tape driver uses a semaphore as mutex. Use the mutex API instead of the (binary) semaphore. Signed-off-by: Matthias Kaehlcke <matthias.kaehlcke@gmail.com> Reviewed-by: Satyam Sharma <satyam@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Willem Riede <wrlk@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-08-10 21:50:44 +00:00
if (mutex_lock_interruptible(&STp->lock))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return (-ERESTARTSYS);
/*
* If we are in the middle of error recovery, don't let anyone
* else try and use this device. Also, if error recovery fails, it
* may try and take the device offline, in which case all further
* access to the device is prohibited.
*/
if( !scsi_block_when_processing_errors(STp->device) ) {
retval = (-ENXIO);
goto out;
}
if (STp->ready != ST_READY) {
if (STp->ready == ST_NO_TAPE)
retval = (-ENOMEDIUM);
else
retval = (-EIO);
goto out;
}
STm = &(STp->modes[STp->current_mode]);
if (!STm->defined) {
retval = (-ENXIO);
goto out;
}
if (count == 0)
goto out;
/*
* If there was a bus reset, block further access
* to this device.
*/
if (STp->pos_unknown) {
retval = (-EIO);
goto out;
}
#if DEBUG
if (!STp->in_use) {
printk(OSST_DEB_MSG "%s:D: Incorrect device.\n", name);
retval = (-EIO);
goto out;
}
#endif
if (STp->write_prot) {
retval = (-EACCES);
goto out;
}
/* Write must be integral number of blocks */
if (STp->block_size != 0 && (count % STp->block_size) != 0) {
lib/vsprintf.c: remove %Z support Now that %z is standartised in C99 there is no reason to support %Z. Unlike %L it doesn't even make format strings smaller. Use BUILD_BUG_ON in a couple ATM drivers. In case anyone didn't notice lib/vsprintf.o is about half of SLUB which is in my opinion is quite an achievement. Hopefully this patch inspires someone else to trim vsprintf.c more. Link: http://lkml.kernel.org/r/20170103230126.GA30170@avx2 Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Cc: Andy Shevchenko <andy.shevchenko@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-02-27 22:30:02 +00:00
printk(KERN_ERR "%s:E: Write (%zd bytes) not multiple of tape block size (%d%c).\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
name, count, STp->block_size<1024?
STp->block_size:STp->block_size/1024, STp->block_size<1024?'b':'k');
retval = (-EINVAL);
goto out;
}
if (STp->first_frame_position >= STp->capacity - OSST_EOM_RESERVE) {
printk(KERN_ERR "%s:E: Write truncated at EOM early warning (frame %d).\n",
name, STp->first_frame_position);
retval = (-ENOSPC);
goto out;
}
if (STp->do_auto_lock && STp->door_locked == ST_UNLOCKED && !do_door_lock(STp, 1))
STp->door_locked = ST_LOCKED_AUTO;
STps = &(STp->ps[STp->partition]);
if (STps->rw == ST_READING) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Switching from read to write at file %d, block %d\n", name,
STps->drv_file, STps->drv_block);
#endif
retval = osst_flush_buffer(STp, &SRpnt, 0);
if (retval)
goto out;
STps->rw = ST_IDLE;
}
if (STps->rw != ST_WRITING) {
/* Are we totally rewriting this tape? */
if (!STp->header_ok ||
(STp->first_frame_position == STp->first_data_ppos && STps->drv_block < 0) ||
(STps->drv_file == 0 && STps->drv_block == 0)) {
STp->wrt_pass_cntr++;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Allocating next write pass counter: %d\n",
name, STp->wrt_pass_cntr);
#endif
osst_reset_header(STp, &SRpnt);
STps->drv_file = STps->drv_block = 0;
}
/* Do we know where we'll be writing on the tape? */
else {
if ((STp->fast_open && osst_verify_position(STp, &SRpnt)) ||
STps->drv_file < 0 || STps->drv_block < 0) {
if (STp->first_frame_position == STp->eod_frame_ppos) { /* at EOD */
STps->drv_file = STp->filemark_cnt;
STps->drv_block = 0;
}
else {
/* We have no idea where the tape is positioned - give up */
#if DEBUG
printk(OSST_DEB_MSG
"%s:D: Cannot write at indeterminate position.\n", name);
#endif
retval = (-EIO);
goto out;
}
}
if ((STps->drv_file + STps->drv_block) > 0 && STps->drv_file < STp->filemark_cnt) {
STp->filemark_cnt = STps->drv_file;
STp->last_mark_ppos =
ntohl(STp->header_cache->dat_fm_tab.fm_tab_ent[STp->filemark_cnt-1]);
printk(KERN_WARNING
"%s:W: Overwriting file %d with old write pass counter %d\n",
name, STps->drv_file, STp->wrt_pass_cntr);
printk(KERN_WARNING
"%s:W: may lead to stale data being accepted on reading back!\n",
name);
#if DEBUG
printk(OSST_DEB_MSG
"%s:D: resetting filemark count to %d and last mark ppos,lbn to %d,%d\n",
name, STp->filemark_cnt, STp->last_mark_ppos, STp->last_mark_lbn);
#endif
}
}
STp->fast_open = 0;
}
if (!STp->header_ok) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Write cannot proceed without valid headers\n", name);
#endif
retval = (-EIO);
goto out;
}
if ((STp->buffer)->writing) {
if (SRpnt) printk(KERN_ERR "%s:A: Not supposed to have SRpnt at line %d\n", name, __LINE__);
osst_write_behind_check(STp);
if ((STp->buffer)->syscall_result) {
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Async write error (write) %x.\n", name,
(STp->buffer)->midlevel_result);
#endif
if ((STp->buffer)->midlevel_result == INT_MAX)
STps->eof = ST_EOM_OK;
else
STps->eof = ST_EOM_ERROR;
}
}
if (STps->eof == ST_EOM_OK) {
retval = (-ENOSPC);
goto out;
}
else if (STps->eof == ST_EOM_ERROR) {
retval = (-EIO);
goto out;
}
/* Check the buffer readability in cases where copy_user might catch
the problems after some tape movement. */
if ((copy_from_user(&i, buf, 1) != 0 ||
copy_from_user(&i, buf + count - 1, 1) != 0)) {
retval = (-EFAULT);
goto out;
}
if (!STm->do_buffer_writes) {
write_threshold = 1;
}
else
write_threshold = (STp->buffer)->buffer_blocks * STp->block_size;
if (!STm->do_async_writes)
write_threshold--;
total = count;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Writing %d bytes to file %d block %d lblk %d fseq %d fppos %d\n",
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
name, (int) count, STps->drv_file, STps->drv_block,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->logical_blk_num, STp->frame_seq_number, STp->first_frame_position);
#endif
b_point = buf;
while ((STp->buffer)->buffer_bytes + count > write_threshold)
{
doing_write = 1;
do_count = (STp->buffer)->buffer_blocks * STp->block_size -
(STp->buffer)->buffer_bytes;
if (do_count > count)
do_count = count;
i = append_to_buffer(b_point, STp->buffer, do_count);
if (i) {
retval = i;
goto out;
}
blks = do_count / STp->block_size;
STp->logical_blk_num += blks; /* logical_blk_num is incremented as data is moved from user */
i = osst_write_frame(STp, &SRpnt, 1);
if (i == (-ENOSPC)) {
transfer = STp->buffer->writing; /* FIXME -- check this logic */
if (transfer <= do_count) {
osst: update ppos instead of using file->f_pos osst_read()/osst_write() modify file->f_pos directly instead of the ppos given to them. The VFS later updates the file->f_pos and overwrites it with the value of ppos. Signed-off-by: Jan Blunck <jblunck@suse.de> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Willem Riede <osst@riede.org> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-26 21:44:44 +00:00
*ppos += do_count - transfer;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
count -= do_count - transfer;
if (STps->drv_block >= 0) {
STps->drv_block += (do_count - transfer) / STp->block_size;
}
STps->eof = ST_EOM_OK;
retval = (-ENOSPC); /* EOM within current request */
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: EOM with %d bytes unwritten.\n",
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
name, (int) transfer);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
}
else {
STps->eof = ST_EOM_ERROR;
STps->drv_block = (-1); /* Too cautious? */
retval = (-EIO); /* EOM for old data */
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: EOM with lost data.\n", name);
#endif
}
}
else
retval = i;
if (retval < 0) {
if (SRpnt != NULL) {
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
SRpnt = NULL;
}
STp->buffer->buffer_bytes = 0;
STp->dirty = 0;
if (count < total)
retval = total - count;
goto out;
}
osst: update ppos instead of using file->f_pos osst_read()/osst_write() modify file->f_pos directly instead of the ppos given to them. The VFS later updates the file->f_pos and overwrites it with the value of ppos. Signed-off-by: Jan Blunck <jblunck@suse.de> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Willem Riede <osst@riede.org> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-26 21:44:44 +00:00
*ppos += do_count;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
b_point += do_count;
count -= do_count;
if (STps->drv_block >= 0) {
STps->drv_block += blks;
}
STp->buffer->buffer_bytes = 0;
STp->dirty = 0;
} /* end while write threshold exceeded */
if (count != 0) {
STp->dirty = 1;
i = append_to_buffer(b_point, STp->buffer, count);
if (i) {
retval = i;
goto out;
}
blks = count / STp->block_size;
STp->logical_blk_num += blks;
if (STps->drv_block >= 0) {
STps->drv_block += blks;
}
osst: update ppos instead of using file->f_pos osst_read()/osst_write() modify file->f_pos directly instead of the ppos given to them. The VFS later updates the file->f_pos and overwrites it with the value of ppos. Signed-off-by: Jan Blunck <jblunck@suse.de> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Willem Riede <osst@riede.org> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-26 21:44:44 +00:00
*ppos += count;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
count = 0;
}
if (doing_write && (STp->buffer)->syscall_result != 0) {
retval = (STp->buffer)->syscall_result;
goto out;
}
if (STm->do_async_writes && ((STp->buffer)->buffer_bytes >= STp->write_threshold)) {
/* Schedule an asynchronous write */
(STp->buffer)->writing = ((STp->buffer)->buffer_bytes /
STp->block_size) * STp->block_size;
STp->dirty = !((STp->buffer)->writing ==
(STp->buffer)->buffer_bytes);
i = osst_write_frame(STp, &SRpnt, 0);
if (i < 0) {
retval = (-EIO);
goto out;
}
SRpnt = NULL; /* Prevent releasing this request! */
}
STps->at_sm &= (total == 0);
if (total > 0)
STps->eof = ST_NOEOF;
retval = total;
out:
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt != NULL) osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: Use mutex instead of semaphore The OnStream SCSI Tape driver uses a semaphore as mutex. Use the mutex API instead of the (binary) semaphore. Signed-off-by: Matthias Kaehlcke <matthias.kaehlcke@gmail.com> Reviewed-by: Satyam Sharma <satyam@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Willem Riede <wrlk@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-08-10 21:50:44 +00:00
mutex_unlock(&STp->lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return retval;
}
/* Read command */
static ssize_t osst_read(struct file * filp, char __user * buf, size_t count, loff_t *ppos)
{
ssize_t total, retval = 0;
ssize_t i, transfer;
int special;
struct st_modedef * STm;
struct st_partstat * STps;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct osst_tape * STp = filp->private_data;
char * name = tape_name(STp);
[SCSI] osst: Use mutex instead of semaphore The OnStream SCSI Tape driver uses a semaphore as mutex. Use the mutex API instead of the (binary) semaphore. Signed-off-by: Matthias Kaehlcke <matthias.kaehlcke@gmail.com> Reviewed-by: Satyam Sharma <satyam@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Willem Riede <wrlk@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-08-10 21:50:44 +00:00
if (mutex_lock_interruptible(&STp->lock))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return (-ERESTARTSYS);
/*
* If we are in the middle of error recovery, don't let anyone
* else try and use this device. Also, if error recovery fails, it
* may try and take the device offline, in which case all further
* access to the device is prohibited.
*/
if( !scsi_block_when_processing_errors(STp->device) ) {
retval = (-ENXIO);
goto out;
}
if (STp->ready != ST_READY) {
if (STp->ready == ST_NO_TAPE)
retval = (-ENOMEDIUM);
else
retval = (-EIO);
goto out;
}
STm = &(STp->modes[STp->current_mode]);
if (!STm->defined) {
retval = (-ENXIO);
goto out;
}
#if DEBUG
if (!STp->in_use) {
printk(OSST_DEB_MSG "%s:D: Incorrect device.\n", name);
retval = (-EIO);
goto out;
}
#endif
/* Must have initialized medium */
if (!STp->header_ok) {
retval = (-EIO);
goto out;
}
if (STp->do_auto_lock && STp->door_locked == ST_UNLOCKED && !do_door_lock(STp, 1))
STp->door_locked = ST_LOCKED_AUTO;
STps = &(STp->ps[STp->partition]);
if (STps->rw == ST_WRITING) {
retval = osst_flush_buffer(STp, &SRpnt, 0);
if (retval)
goto out;
STps->rw = ST_IDLE;
/* FIXME -- this may leave the tape without EOD and up2date headers */
}
if ((count % STp->block_size) != 0) {
printk(KERN_WARNING
lib/vsprintf.c: remove %Z support Now that %z is standartised in C99 there is no reason to support %Z. Unlike %L it doesn't even make format strings smaller. Use BUILD_BUG_ON in a couple ATM drivers. In case anyone didn't notice lib/vsprintf.o is about half of SLUB which is in my opinion is quite an achievement. Hopefully this patch inspires someone else to trim vsprintf.c more. Link: http://lkml.kernel.org/r/20170103230126.GA30170@avx2 Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Cc: Andy Shevchenko <andy.shevchenko@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-02-27 22:30:02 +00:00
"%s:W: Read (%zd bytes) not multiple of tape block size (%d%c).\n", name, count,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->block_size<1024?STp->block_size:STp->block_size/1024, STp->block_size<1024?'b':'k');
}
#if DEBUG
if (debugging && STps->eof != ST_NOEOF)
printk(OSST_DEB_MSG "%s:D: EOF/EOM flag up (%d). Bytes %d\n", name,
STps->eof, (STp->buffer)->buffer_bytes);
#endif
if ((STp->buffer)->buffer_bytes == 0 &&
STps->eof >= ST_EOD_1) {
if (STps->eof < ST_EOD) {
STps->eof += 1;
retval = 0;
goto out;
}
retval = (-EIO); /* EOM or Blank Check */
goto out;
}
/* Check the buffer writability before any tape movement. Don't alter
buffer data. */
if (copy_from_user(&i, buf, 1) != 0 ||
copy_to_user (buf, &i, 1) != 0 ||
copy_from_user(&i, buf + count - 1, 1) != 0 ||
copy_to_user (buf + count - 1, &i, 1) != 0) {
retval = (-EFAULT);
goto out;
}
/* Loop until enough data in buffer or a special condition found */
for (total = 0, special = 0; total < count - STp->block_size + 1 && !special; ) {
/* Get new data if the buffer is empty */
if ((STp->buffer)->buffer_bytes == 0) {
if (STps->eof == ST_FM_HIT)
break;
special = osst_get_logical_frame(STp, &SRpnt, STp->frame_seq_number, 0);
if (special < 0) { /* No need to continue read */
STp->frame_in_buffer = 0;
retval = special;
goto out;
}
}
/* Move the data from driver buffer to user buffer */
if ((STp->buffer)->buffer_bytes > 0) {
#if DEBUG
if (debugging && STps->eof != ST_NOEOF)
printk(OSST_DEB_MSG "%s:D: EOF up (%d). Left %d, needed %d.\n", name,
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
STps->eof, (STp->buffer)->buffer_bytes, (int) (count - total));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
/* force multiple of block size, note block_size may have been adjusted */
transfer = (((STp->buffer)->buffer_bytes < count - total ?
(STp->buffer)->buffer_bytes : count - total)/
STp->block_size) * STp->block_size;
if (transfer == 0) {
printk(KERN_WARNING
lib/vsprintf.c: remove %Z support Now that %z is standartised in C99 there is no reason to support %Z. Unlike %L it doesn't even make format strings smaller. Use BUILD_BUG_ON in a couple ATM drivers. In case anyone didn't notice lib/vsprintf.o is about half of SLUB which is in my opinion is quite an achievement. Hopefully this patch inspires someone else to trim vsprintf.c more. Link: http://lkml.kernel.org/r/20170103230126.GA30170@avx2 Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Cc: Andy Shevchenko <andy.shevchenko@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-02-27 22:30:02 +00:00
"%s:W: Nothing can be transferred, requested %zd, tape block size (%d%c).\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
name, count, STp->block_size < 1024?
STp->block_size:STp->block_size/1024,
STp->block_size<1024?'b':'k');
break;
}
i = from_buffer(STp->buffer, buf, transfer);
if (i) {
retval = i;
goto out;
}
STp->logical_blk_num += transfer / STp->block_size;
STps->drv_block += transfer / STp->block_size;
osst: update ppos instead of using file->f_pos osst_read()/osst_write() modify file->f_pos directly instead of the ppos given to them. The VFS later updates the file->f_pos and overwrites it with the value of ppos. Signed-off-by: Jan Blunck <jblunck@suse.de> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Willem Riede <osst@riede.org> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-26 21:44:44 +00:00
*ppos += transfer;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
buf += transfer;
total += transfer;
}
if ((STp->buffer)->buffer_bytes == 0) {
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Finished with frame %d\n",
name, STp->frame_seq_number);
#endif
STp->frame_in_buffer = 0;
STp->frame_seq_number++; /* frame to look for next time */
}
} /* for (total = 0, special = 0; total < count && !special; ) */
/* Change the eof state if no data from tape or buffer */
if (total == 0) {
if (STps->eof == ST_FM_HIT) {
STps->eof = (STp->first_frame_position >= STp->eod_frame_ppos)?ST_EOD_2:ST_FM;
STps->drv_block = 0;
if (STps->drv_file >= 0)
STps->drv_file++;
}
else if (STps->eof == ST_EOD_1) {
STps->eof = ST_EOD_2;
if (STps->drv_block > 0 && STps->drv_file >= 0)
STps->drv_file++;
STps->drv_block = 0;
}
else if (STps->eof == ST_EOD_2)
STps->eof = ST_EOD;
}
else if (STps->eof == ST_FM)
STps->eof = ST_NOEOF;
retval = total;
out:
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt != NULL) osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: Use mutex instead of semaphore The OnStream SCSI Tape driver uses a semaphore as mutex. Use the mutex API instead of the (binary) semaphore. Signed-off-by: Matthias Kaehlcke <matthias.kaehlcke@gmail.com> Reviewed-by: Satyam Sharma <satyam@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Willem Riede <wrlk@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-08-10 21:50:44 +00:00
mutex_unlock(&STp->lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return retval;
}
/* Set the driver options */
static void osst_log_options(struct osst_tape *STp, struct st_modedef *STm, char *name)
{
printk(KERN_INFO
"%s:I: Mode %d options: buffer writes: %d, async writes: %d, read ahead: %d\n",
name, STp->current_mode, STm->do_buffer_writes, STm->do_async_writes,
STm->do_read_ahead);
printk(KERN_INFO
"%s:I: can bsr: %d, two FMs: %d, fast mteom: %d, auto lock: %d,\n",
name, STp->can_bsr, STp->two_fm, STp->fast_mteom, STp->do_auto_lock);
printk(KERN_INFO
"%s:I: defs for wr: %d, no block limits: %d, partitions: %d, s2 log: %d\n",
name, STm->defaults_for_writes, STp->omit_blklims, STp->can_partitions,
STp->scsi2_logical);
printk(KERN_INFO
"%s:I: sysv: %d\n", name, STm->sysv);
#if DEBUG
printk(KERN_INFO
"%s:D: debugging: %d\n",
name, debugging);
#endif
}
static int osst_set_options(struct osst_tape *STp, long options)
{
int value;
long code;
struct st_modedef * STm;
char * name = tape_name(STp);
STm = &(STp->modes[STp->current_mode]);
if (!STm->defined) {
memcpy(STm, &(STp->modes[0]), sizeof(*STm));
modes_defined = 1;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Initialized mode %d definition from mode 0\n",
name, STp->current_mode);
#endif
}
code = options & MT_ST_OPTIONS;
if (code == MT_ST_BOOLEANS) {
STm->do_buffer_writes = (options & MT_ST_BUFFER_WRITES) != 0;
STm->do_async_writes = (options & MT_ST_ASYNC_WRITES) != 0;
STm->defaults_for_writes = (options & MT_ST_DEF_WRITES) != 0;
STm->do_read_ahead = (options & MT_ST_READ_AHEAD) != 0;
STp->two_fm = (options & MT_ST_TWO_FM) != 0;
STp->fast_mteom = (options & MT_ST_FAST_MTEOM) != 0;
STp->do_auto_lock = (options & MT_ST_AUTO_LOCK) != 0;
STp->can_bsr = (options & MT_ST_CAN_BSR) != 0;
STp->omit_blklims = (options & MT_ST_NO_BLKLIMS) != 0;
if ((STp->device)->scsi_level >= SCSI_2)
STp->can_partitions = (options & MT_ST_CAN_PARTITIONS) != 0;
STp->scsi2_logical = (options & MT_ST_SCSI2LOGICAL) != 0;
STm->sysv = (options & MT_ST_SYSV) != 0;
#if DEBUG
debugging = (options & MT_ST_DEBUGGING) != 0;
#endif
osst_log_options(STp, STm, name);
}
else if (code == MT_ST_SETBOOLEANS || code == MT_ST_CLEARBOOLEANS) {
value = (code == MT_ST_SETBOOLEANS);
if ((options & MT_ST_BUFFER_WRITES) != 0)
STm->do_buffer_writes = value;
if ((options & MT_ST_ASYNC_WRITES) != 0)
STm->do_async_writes = value;
if ((options & MT_ST_DEF_WRITES) != 0)
STm->defaults_for_writes = value;
if ((options & MT_ST_READ_AHEAD) != 0)
STm->do_read_ahead = value;
if ((options & MT_ST_TWO_FM) != 0)
STp->two_fm = value;
if ((options & MT_ST_FAST_MTEOM) != 0)
STp->fast_mteom = value;
if ((options & MT_ST_AUTO_LOCK) != 0)
STp->do_auto_lock = value;
if ((options & MT_ST_CAN_BSR) != 0)
STp->can_bsr = value;
if ((options & MT_ST_NO_BLKLIMS) != 0)
STp->omit_blklims = value;
if ((STp->device)->scsi_level >= SCSI_2 &&
(options & MT_ST_CAN_PARTITIONS) != 0)
STp->can_partitions = value;
if ((options & MT_ST_SCSI2LOGICAL) != 0)
STp->scsi2_logical = value;
if ((options & MT_ST_SYSV) != 0)
STm->sysv = value;
#if DEBUG
if ((options & MT_ST_DEBUGGING) != 0)
debugging = value;
#endif
osst_log_options(STp, STm, name);
}
else if (code == MT_ST_WRITE_THRESHOLD) {
value = (options & ~MT_ST_OPTIONS) * ST_KILOBYTE;
if (value < 1 || value > osst_buffer_size) {
printk(KERN_WARNING "%s:W: Write threshold %d too small or too large.\n",
name, value);
return (-EIO);
}
STp->write_threshold = value;
printk(KERN_INFO "%s:I: Write threshold set to %d bytes.\n",
name, value);
}
else if (code == MT_ST_DEF_BLKSIZE) {
value = (options & ~MT_ST_OPTIONS);
if (value == ~MT_ST_OPTIONS) {
STm->default_blksize = (-1);
printk(KERN_INFO "%s:I: Default block size disabled.\n", name);
}
else {
if (value < 512 || value > OS_DATA_SIZE || OS_DATA_SIZE % value) {
printk(KERN_WARNING "%s:W: Default block size cannot be set to %d.\n",
name, value);
return (-EINVAL);
}
STm->default_blksize = value;
printk(KERN_INFO "%s:I: Default block size set to %d bytes.\n",
name, STm->default_blksize);
}
}
else if (code == MT_ST_TIMEOUTS) {
value = (options & ~MT_ST_OPTIONS);
if ((value & MT_ST_SET_LONG_TIMEOUT) != 0) {
STp->long_timeout = (value & ~MT_ST_SET_LONG_TIMEOUT) * HZ;
printk(KERN_INFO "%s:I: Long timeout set to %d seconds.\n", name,
(value & ~MT_ST_SET_LONG_TIMEOUT));
}
else {
STp->timeout = value * HZ;
printk(KERN_INFO "%s:I: Normal timeout set to %d seconds.\n", name, value);
}
}
else if (code == MT_ST_DEF_OPTIONS) {
code = (options & ~MT_ST_CLEAR_DEFAULT);
value = (options & MT_ST_CLEAR_DEFAULT);
if (code == MT_ST_DEF_DENSITY) {
if (value == MT_ST_CLEAR_DEFAULT) {
STm->default_density = (-1);
printk(KERN_INFO "%s:I: Density default disabled.\n", name);
}
else {
STm->default_density = value & 0xff;
printk(KERN_INFO "%s:I: Density default set to %x\n",
name, STm->default_density);
}
}
else if (code == MT_ST_DEF_DRVBUFFER) {
if (value == MT_ST_CLEAR_DEFAULT) {
STp->default_drvbuffer = 0xff;
printk(KERN_INFO "%s:I: Drive buffer default disabled.\n", name);
}
else {
STp->default_drvbuffer = value & 7;
printk(KERN_INFO "%s:I: Drive buffer default set to %x\n",
name, STp->default_drvbuffer);
}
}
else if (code == MT_ST_DEF_COMPRESSION) {
if (value == MT_ST_CLEAR_DEFAULT) {
STm->default_compression = ST_DONT_TOUCH;
printk(KERN_INFO "%s:I: Compression default disabled.\n", name);
}
else {
STm->default_compression = (value & 1 ? ST_YES : ST_NO);
printk(KERN_INFO "%s:I: Compression default set to %x\n",
name, (value & 1));
}
}
}
else
return (-EIO);
return 0;
}
/* Internal ioctl function */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
static int osst_int_ioctl(struct osst_tape * STp, struct osst_request ** aSRpnt,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned int cmd_in, unsigned long arg)
{
int timeout;
long ltmp;
int i, ioctl_result;
int chg_eof = 1;
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = * aSRpnt;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct st_partstat * STps;
int fileno, blkno, at_sm, frame_seq_numbr, logical_blk_num;
int datalen = 0, direction = DMA_NONE;
char * name = tape_name(STp);
if (STp->ready != ST_READY && cmd_in != MTLOAD) {
if (STp->ready == ST_NO_TAPE)
return (-ENOMEDIUM);
else
return (-EIO);
}
timeout = STp->long_timeout;
STps = &(STp->ps[STp->partition]);
fileno = STps->drv_file;
blkno = STps->drv_block;
at_sm = STps->at_sm;
frame_seq_numbr = STp->frame_seq_number;
logical_blk_num = STp->logical_blk_num;
memset(cmd, 0, MAX_COMMAND_SIZE);
switch (cmd_in) {
case MTFSFM:
chg_eof = 0; /* Changed from the FSF after this */
case MTFSF:
if (STp->raw)
return (-EIO);
if (STp->linux_media)
ioctl_result = osst_space_over_filemarks_forward_fast(STp, &SRpnt, cmd_in, arg);
else
ioctl_result = osst_space_over_filemarks_forward_slow(STp, &SRpnt, cmd_in, arg);
if (fileno >= 0)
fileno += arg;
blkno = 0;
at_sm &= (arg == 0);
goto os_bypass;
case MTBSF:
chg_eof = 0; /* Changed from the FSF after this */
case MTBSFM:
if (STp->raw)
return (-EIO);
ioctl_result = osst_space_over_filemarks_backward(STp, &SRpnt, cmd_in, arg);
if (fileno >= 0)
fileno -= arg;
blkno = (-1); /* We can't know the block number */
at_sm &= (arg == 0);
goto os_bypass;
case MTFSR:
case MTBSR:
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Skipping %lu blocks %s from logical block %d\n",
name, arg, cmd_in==MTFSR?"forward":"backward", logical_blk_num);
#endif
if (cmd_in == MTFSR) {
logical_blk_num += arg;
if (blkno >= 0) blkno += arg;
}
else {
logical_blk_num -= arg;
if (blkno >= 0) blkno -= arg;
}
ioctl_result = osst_seek_logical_blk(STp, &SRpnt, logical_blk_num);
fileno = STps->drv_file;
blkno = STps->drv_block;
at_sm &= (arg == 0);
goto os_bypass;
case MTFSS:
cmd[0] = SPACE;
cmd[1] = 0x04; /* Space Setmarks */ /* FIXME -- OS can't do this? */
cmd[2] = (arg >> 16);
cmd[3] = (arg >> 8);
cmd[4] = arg;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Spacing tape forward %d setmarks.\n", name,
cmd[2] * 65536 + cmd[3] * 256 + cmd[4]);
#endif
if (arg != 0) {
blkno = fileno = (-1);
at_sm = 1;
}
break;
case MTBSS:
cmd[0] = SPACE;
cmd[1] = 0x04; /* Space Setmarks */ /* FIXME -- OS can't do this? */
ltmp = (-arg);
cmd[2] = (ltmp >> 16);
cmd[3] = (ltmp >> 8);
cmd[4] = ltmp;
#if DEBUG
if (debugging) {
if (cmd[2] & 0x80)
ltmp = 0xff000000;
ltmp = ltmp | (cmd[2] << 16) | (cmd[3] << 8) | cmd[4];
printk(OSST_DEB_MSG "%s:D: Spacing tape backward %ld setmarks.\n",
name, (-ltmp));
}
#endif
if (arg != 0) {
blkno = fileno = (-1);
at_sm = 1;
}
break;
case MTWEOF:
if ((STps->rw == ST_WRITING || STp->dirty) && !STp->pos_unknown) {
STp->write_type = OS_WRITE_DATA;
ioctl_result = osst_flush_write_buffer(STp, &SRpnt);
} else
ioctl_result = 0;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Writing %ld filemark(s).\n", name, arg);
#endif
for (i=0; i<arg; i++)
ioctl_result |= osst_write_filemark(STp, &SRpnt);
if (fileno >= 0) fileno += arg;
if (blkno >= 0) blkno = 0;
goto os_bypass;
case MTWSM:
if (STp->write_prot)
return (-EACCES);
if (!STp->raw)
return 0;
cmd[0] = WRITE_FILEMARKS; /* FIXME -- need OS version */
if (cmd_in == MTWSM)
cmd[1] = 2;
cmd[2] = (arg >> 16);
cmd[3] = (arg >> 8);
cmd[4] = arg;
timeout = STp->timeout;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Writing %d setmark(s).\n", name,
cmd[2] * 65536 + cmd[3] * 256 + cmd[4]);
#endif
if (fileno >= 0)
fileno += arg;
blkno = 0;
at_sm = (cmd_in == MTWSM);
break;
case MTOFFL:
case MTLOAD:
case MTUNLOAD:
case MTRETEN:
cmd[0] = START_STOP;
cmd[1] = 1; /* Don't wait for completion */
if (cmd_in == MTLOAD) {
if (STp->ready == ST_NO_TAPE)
cmd[4] = 4; /* open tray */
else
cmd[4] = 1; /* load */
}
if (cmd_in == MTRETEN)
cmd[4] = 3; /* retension then mount */
if (cmd_in == MTOFFL)
cmd[4] = 4; /* rewind then eject */
timeout = STp->timeout;
#if DEBUG
if (debugging) {
switch (cmd_in) {
case MTUNLOAD:
printk(OSST_DEB_MSG "%s:D: Unloading tape.\n", name);
break;
case MTLOAD:
printk(OSST_DEB_MSG "%s:D: Loading tape.\n", name);
break;
case MTRETEN:
printk(OSST_DEB_MSG "%s:D: Retensioning tape.\n", name);
break;
case MTOFFL:
printk(OSST_DEB_MSG "%s:D: Ejecting tape.\n", name);
break;
}
}
#endif
fileno = blkno = at_sm = frame_seq_numbr = logical_blk_num = 0 ;
break;
case MTNOP:
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: No-op on tape.\n", name);
#endif
return 0; /* Should do something ? */
break;
case MTEOM:
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Spacing to end of recorded medium.\n", name);
#endif
if ((osst_position_tape_and_confirm(STp, &SRpnt, STp->eod_frame_ppos) < 0) ||
(osst_get_logical_frame(STp, &SRpnt, -1, 0) < 0)) {
ioctl_result = -EIO;
goto os_bypass;
}
if (STp->buffer->aux->frame_type != OS_FRAME_TYPE_EOD) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: No EOD frame found where expected.\n", name);
#endif
ioctl_result = -EIO;
goto os_bypass;
}
ioctl_result = osst_set_frame_position(STp, &SRpnt, STp->eod_frame_ppos, 0);
fileno = STp->filemark_cnt;
blkno = at_sm = 0;
goto os_bypass;
case MTERASE:
if (STp->write_prot)
return (-EACCES);
ioctl_result = osst_reset_header(STp, &SRpnt);
i = osst_write_eod(STp, &SRpnt);
if (i < ioctl_result) ioctl_result = i;
i = osst_position_tape_and_confirm(STp, &SRpnt, STp->eod_frame_ppos);
if (i < ioctl_result) ioctl_result = i;
fileno = blkno = at_sm = 0 ;
goto os_bypass;
case MTREW:
cmd[0] = REZERO_UNIT; /* rewind */
cmd[1] = 1;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Rewinding tape, Immed=%d.\n", name, cmd[1]);
#endif
fileno = blkno = at_sm = frame_seq_numbr = logical_blk_num = 0 ;
break;
case MTSETBLK: /* Set block length */
if ((STps->drv_block == 0 ) &&
!STp->dirty &&
((STp->buffer)->buffer_bytes == 0) &&
((arg & MT_ST_BLKSIZE_MASK) >= 512 ) &&
((arg & MT_ST_BLKSIZE_MASK) <= OS_DATA_SIZE) &&
!(OS_DATA_SIZE % (arg & MT_ST_BLKSIZE_MASK)) ) {
/*
* Only allowed to change the block size if you opened the
* device at the beginning of a file before writing anything.
* Note, that when reading, changing block_size is futile,
* as the size used when writing overrides it.
*/
STp->block_size = (arg & MT_ST_BLKSIZE_MASK);
printk(KERN_INFO "%s:I: Block size set to %d bytes.\n",
name, STp->block_size);
return 0;
}
case MTSETDENSITY: /* Set tape density */
case MTSETDRVBUFFER: /* Set drive buffering */
case SET_DENS_AND_BLK: /* Set density and block size */
chg_eof = 0;
if (STp->dirty || (STp->buffer)->buffer_bytes != 0)
return (-EIO); /* Not allowed if data in buffer */
if ((cmd_in == MTSETBLK || cmd_in == SET_DENS_AND_BLK) &&
(arg & MT_ST_BLKSIZE_MASK) != 0 &&
(arg & MT_ST_BLKSIZE_MASK) != STp->block_size ) {
printk(KERN_WARNING "%s:W: Illegal to set block size to %d%s.\n",
name, (int)(arg & MT_ST_BLKSIZE_MASK),
(OS_DATA_SIZE % (arg & MT_ST_BLKSIZE_MASK))?"":" now");
return (-EINVAL);
}
return 0; /* FIXME silently ignore if block size didn't change */
default:
return (-ENOSYS);
}
SRpnt = osst_do_scsi(SRpnt, STp, cmd, datalen, direction, timeout, MAX_RETRIES, 1);
ioctl_result = (STp->buffer)->syscall_result;
if (!SRpnt) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Couldn't exec scsi cmd for IOCTL\n", name);
#endif
return ioctl_result;
}
if (!ioctl_result) { /* SCSI command successful */
STp->frame_seq_number = frame_seq_numbr;
STp->logical_blk_num = logical_blk_num;
}
os_bypass:
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: IOCTL (%d) Result=%d\n", name, cmd_in, ioctl_result);
#endif
if (!ioctl_result) { /* success */
if (cmd_in == MTFSFM) {
fileno--;
blkno--;
}
if (cmd_in == MTBSFM) {
fileno++;
blkno++;
}
STps->drv_block = blkno;
STps->drv_file = fileno;
STps->at_sm = at_sm;
if (cmd_in == MTEOM)
STps->eof = ST_EOD;
else if ((cmd_in == MTFSFM || cmd_in == MTBSF) && STps->eof == ST_FM_HIT) {
ioctl_result = osst_seek_logical_blk(STp, &SRpnt, STp->logical_blk_num-1);
STps->drv_block++;
STp->logical_blk_num++;
STp->frame_seq_number++;
STp->frame_in_buffer = 0;
STp->buffer->read_pointer = 0;
}
else if (cmd_in == MTFSF)
STps->eof = (STp->first_frame_position >= STp->eod_frame_ppos)?ST_EOD:ST_FM;
else if (chg_eof)
STps->eof = ST_NOEOF;
if (cmd_in == MTOFFL || cmd_in == MTUNLOAD)
STp->rew_at_close = 0;
else if (cmd_in == MTLOAD) {
for (i=0; i < ST_NBR_PARTITIONS; i++) {
STp->ps[i].rw = ST_IDLE;
STp->ps[i].last_block_valid = 0;/* FIXME - where else is this field maintained? */
}
STp->partition = 0;
}
if (cmd_in == MTREW) {
ioctl_result = osst_position_tape_and_confirm(STp, &SRpnt, STp->first_data_ppos);
if (ioctl_result > 0)
ioctl_result = 0;
}
} else if (cmd_in == MTBSF || cmd_in == MTBSFM ) {
if (osst_position_tape_and_confirm(STp, &SRpnt, STp->first_data_ppos) < 0)
STps->drv_file = STps->drv_block = -1;
else
STps->drv_file = STps->drv_block = 0;
STps->eof = ST_NOEOF;
} else if (cmd_in == MTFSF || cmd_in == MTFSFM) {
if (osst_position_tape_and_confirm(STp, &SRpnt, STp->eod_frame_ppos) < 0)
STps->drv_file = STps->drv_block = -1;
else {
STps->drv_file = STp->filemark_cnt;
STps->drv_block = 0;
}
STps->eof = ST_EOD;
} else if (cmd_in == MTBSR || cmd_in == MTFSR || cmd_in == MTWEOF || cmd_in == MTEOM) {
STps->drv_file = STps->drv_block = (-1);
STps->eof = ST_NOEOF;
STp->header_ok = 0;
} else if (cmd_in == MTERASE) {
STp->header_ok = 0;
} else if (SRpnt) { /* SCSI command was not completely successful. */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt->sense[2] & 0x40) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STps->eof = ST_EOM_OK;
STps->drv_block = 0;
}
if (chg_eof)
STps->eof = ST_NOEOF;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[2] & 0x0f) == BLANK_CHECK)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STps->eof = ST_EOD;
if (cmd_in == MTLOAD && osst_wait_for_medium(STp, &SRpnt, 60))
ioctl_result = osst_wait_ready(STp, &SRpnt, 5 * 60, OSST_WAIT_POSITION_COMPLETE);
}
*aSRpnt = SRpnt;
return ioctl_result;
}
/* Open the device */
osst: cdev lock_kernel() pushdown. Signed-off-by: Jonathan Corbet <corbet@lwn.net>
2008-05-15 18:23:19 +00:00
static int __os_scsi_tape_open(struct inode * inode, struct file * filp)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
unsigned short flags;
int i, b_size, new_session = 0, retval = 0;
unsigned char cmd[MAX_COMMAND_SIZE];
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct osst_tape * STp;
struct st_modedef * STm;
struct st_partstat * STps;
char * name;
int dev = TAPE_NR(inode);
int mode = TAPE_MODE(inode);
/*
* We really want to do nonseekable_open(inode, filp); here, but some
* versions of tar incorrectly call lseek on tapes and bail out if that
* fails. So we disallow pread() and pwrite(), but permit lseeks.
*/
filp->f_mode &= ~(FMODE_PREAD | FMODE_PWRITE);
write_lock(&os_scsi_tapes_lock);
if (dev >= osst_max_dev || os_scsi_tapes == NULL ||
(STp = os_scsi_tapes[dev]) == NULL || !STp->device) {
write_unlock(&os_scsi_tapes_lock);
return (-ENXIO);
}
name = tape_name(STp);
if (STp->in_use) {
write_unlock(&os_scsi_tapes_lock);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Device already in use.\n", name);
#endif
return (-EBUSY);
}
if (scsi_device_get(STp->device)) {
write_unlock(&os_scsi_tapes_lock);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Failed scsi_device_get.\n", name);
#endif
return (-ENXIO);
}
filp->private_data = STp;
STp->in_use = 1;
write_unlock(&os_scsi_tapes_lock);
STp->rew_at_close = TAPE_REWIND(inode);
if( !scsi_block_when_processing_errors(STp->device) ) {
return -ENXIO;
}
if (mode != STp->current_mode) {
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Mode change from %d to %d.\n",
name, STp->current_mode, mode);
#endif
new_session = 1;
STp->current_mode = mode;
}
STm = &(STp->modes[STp->current_mode]);
flags = filp->f_flags;
STp->write_prot = ((flags & O_ACCMODE) == O_RDONLY);
STp->raw = TAPE_IS_RAW(inode);
if (STp->raw)
STp->header_ok = 0;
/* Allocate data segments for this device's tape buffer */
if (!enlarge_buffer(STp->buffer, STp->restr_dma)) {
printk(KERN_ERR "%s:E: Unable to allocate memory segments for tape buffer.\n", name);
retval = (-EOVERFLOW);
goto err_out;
}
if (STp->buffer->buffer_size >= OS_FRAME_SIZE) {
for (i = 0, b_size = 0;
(i < STp->buffer->sg_segs) && ((b_size + STp->buffer->sg[i].length) <= OS_DATA_SIZE);
b_size += STp->buffer->sg[i++].length);
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
STp->buffer->aux = (os_aux_t *) (page_address(sg_page(&STp->buffer->sg[i])) + OS_DATA_SIZE - b_size);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
printk(OSST_DEB_MSG "%s:D: b_data points to %p in segment 0 at %p\n", name,
STp->buffer->b_data, page_address(STp->buffer->sg[0].page));
printk(OSST_DEB_MSG "%s:D: AUX points to %p in segment %d at %p\n", name,
STp->buffer->aux, i, page_address(STp->buffer->sg[i].page));
#endif
} else {
STp->buffer->aux = NULL; /* this had better never happen! */
printk(KERN_NOTICE "%s:A: Framesize %d too large for buffer.\n", name, OS_FRAME_SIZE);
retval = (-EIO);
goto err_out;
}
STp->buffer->writing = 0;
STp->buffer->syscall_result = 0;
STp->dirty = 0;
for (i=0; i < ST_NBR_PARTITIONS; i++) {
STps = &(STp->ps[i]);
STps->rw = ST_IDLE;
}
STp->ready = ST_READY;
#if DEBUG
STp->nbr_waits = STp->nbr_finished = 0;
#endif
memset (cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = TEST_UNIT_READY;
SRpnt = osst_do_scsi(NULL, STp, cmd, 0, DMA_NONE, STp->timeout, MAX_RETRIES, 1);
if (!SRpnt) {
retval = (STp->buffer)->syscall_result; /* FIXME - valid? */
goto err_out;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[0] & 0x70) == 0x70 &&
(SRpnt->sense[2] & 0x0f) == NOT_READY &&
SRpnt->sense[12] == 4 ) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
printk(OSST_DEB_MSG "%s:D: Unit not ready, cause %x\n", name, SRpnt->sense[13]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#endif
if (filp->f_flags & O_NONBLOCK) {
retval = -EAGAIN;
goto err_out;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt->sense[13] == 2) { /* initialize command required (LOAD) */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
memset (cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = START_STOP;
cmd[1] = 1;
cmd[4] = 1;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, 0, DMA_NONE,
STp->timeout, MAX_RETRIES, 1);
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
osst_wait_ready(STp, &SRpnt, (SRpnt->sense[13]==1?15:3) * 60, 0);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[0] & 0x70) == 0x70 &&
(SRpnt->sense[2] & 0x0f) == UNIT_ATTENTION) { /* New media? */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Unit wants attention\n", name);
#endif
STp->header_ok = 0;
for (i=0; i < 10; i++) {
memset (cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = TEST_UNIT_READY;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, 0, DMA_NONE,
STp->timeout, MAX_RETRIES, 1);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[0] & 0x70) != 0x70 ||
(SRpnt->sense[2] & 0x0f) != UNIT_ATTENTION)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
break;
}
STp->pos_unknown = 0;
STp->partition = STp->new_partition = 0;
if (STp->can_partitions)
STp->nbr_partitions = 1; /* This guess will be updated later if necessary */
for (i=0; i < ST_NBR_PARTITIONS; i++) {
STps = &(STp->ps[i]);
STps->rw = ST_IDLE; /* FIXME - seems to be redundant... */
STps->eof = ST_NOEOF;
STps->at_sm = 0;
STps->last_block_valid = 0;
STps->drv_block = 0;
STps->drv_file = 0 ;
}
new_session = 1;
STp->recover_count = 0;
STp->abort_count = 0;
}
/*
* if we have valid headers from before, and the drive/tape seem untouched,
* open without reconfiguring and re-reading the headers
*/
if (!STp->buffer->syscall_result && STp->header_ok &&
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
!SRpnt->result && SRpnt->sense[0] == 0) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = MODE_SENSE;
cmd[1] = 8;
cmd[2] = VENDOR_IDENT_PAGE;
cmd[4] = VENDOR_IDENT_PAGE_LENGTH + MODE_HEADER_LENGTH;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, cmd[4], DMA_FROM_DEVICE, STp->timeout, 0, 1);
if (STp->buffer->syscall_result ||
STp->buffer->b_data[MODE_HEADER_LENGTH + 2] != 'L' ||
STp->buffer->b_data[MODE_HEADER_LENGTH + 3] != 'I' ||
STp->buffer->b_data[MODE_HEADER_LENGTH + 4] != 'N' ||
STp->buffer->b_data[MODE_HEADER_LENGTH + 5] != '4' ) {
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Signature was changed to %c%c%c%c\n", name,
STp->buffer->b_data[MODE_HEADER_LENGTH + 2],
STp->buffer->b_data[MODE_HEADER_LENGTH + 3],
STp->buffer->b_data[MODE_HEADER_LENGTH + 4],
STp->buffer->b_data[MODE_HEADER_LENGTH + 5]);
#endif
STp->header_ok = 0;
}
i = STp->first_frame_position;
if (STp->header_ok && i == osst_get_frame_position(STp, &SRpnt)) {
if (STp->door_locked == ST_UNLOCKED) {
if (do_door_lock(STp, 1))
printk(KERN_INFO "%s:I: Can't lock drive door\n", name);
else
STp->door_locked = ST_LOCKED_AUTO;
}
if (!STp->frame_in_buffer) {
STp->block_size = (STm->default_blksize > 0) ?
STm->default_blksize : OS_DATA_SIZE;
STp->buffer->buffer_bytes = STp->buffer->read_pointer = 0;
}
STp->buffer->buffer_blocks = OS_DATA_SIZE / STp->block_size;
STp->fast_open = 1;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return 0;
}
#if DEBUG
if (i != STp->first_frame_position)
printk(OSST_DEB_MSG "%s:D: Tape position changed from %d to %d\n",
name, i, STp->first_frame_position);
#endif
STp->header_ok = 0;
}
STp->fast_open = 0;
if ((STp->buffer)->syscall_result != 0 && /* in all error conditions except no medium */
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
(SRpnt->sense[2] != 2 || SRpnt->sense[12] != 0x3A) ) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
memset(cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = MODE_SELECT;
cmd[1] = 0x10;
cmd[4] = 4 + MODE_HEADER_LENGTH;
(STp->buffer)->b_data[0] = cmd[4] - 1;
(STp->buffer)->b_data[1] = 0; /* Medium Type - ignoring */
(STp->buffer)->b_data[2] = 0; /* Reserved */
(STp->buffer)->b_data[3] = 0; /* Block Descriptor Length */
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 0] = 0x3f;
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 1] = 1;
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 2] = 2;
(STp->buffer)->b_data[MODE_HEADER_LENGTH + 3] = 3;
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Applying soft reset\n", name);
#endif
SRpnt = osst_do_scsi(SRpnt, STp, cmd, cmd[4], DMA_TO_DEVICE, STp->timeout, 0, 1);
STp->header_ok = 0;
for (i=0; i < 10; i++) {
memset (cmd, 0, MAX_COMMAND_SIZE);
cmd[0] = TEST_UNIT_READY;
SRpnt = osst_do_scsi(SRpnt, STp, cmd, 0, DMA_NONE,
STp->timeout, MAX_RETRIES, 1);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[0] & 0x70) != 0x70 ||
(SRpnt->sense[2] & 0x0f) == NOT_READY)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
break;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if ((SRpnt->sense[2] & 0x0f) == UNIT_ATTENTION) {
[SCSI] osst: fix warning drivers/scsi/osst.c: In function '__os_scsi_tape_open': drivers/scsi/osst.c:4705: warning: ISO C90 forbids mixed declarations and code Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: James Bottomley <jbottomley@parallels.com>
2011-05-23 22:29:13 +00:00
int j;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->pos_unknown = 0;
STp->partition = STp->new_partition = 0;
if (STp->can_partitions)
STp->nbr_partitions = 1; /* This guess will be updated later if necessary */
[SCSI] osst: wrong index used in inner loop Index i was already used in the outer loop. Fixes a potentially-infinite loop. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Roel Kluin <roel.kluin@gmail.com> Cc: Willem Riede <osst@riede.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: James Bottomley <jbottomley@parallels.com>
2011-05-23 22:29:12 +00:00
for (j = 0; j < ST_NBR_PARTITIONS; j++) {
STps = &(STp->ps[j]);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STps->rw = ST_IDLE;
STps->eof = ST_NOEOF;
STps->at_sm = 0;
STps->last_block_valid = 0;
STps->drv_block = 0;
STps->drv_file = 0 ;
}
new_session = 1;
}
}
}
if (osst_wait_ready(STp, &SRpnt, 15 * 60, 0)) /* FIXME - not allowed with NOBLOCK */
printk(KERN_INFO "%s:I: Device did not become Ready in open\n", name);
if ((STp->buffer)->syscall_result != 0) {
if ((STp->device)->scsi_level >= SCSI_2 &&
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
(SRpnt->sense[0] & 0x70) == 0x70 &&
(SRpnt->sense[2] & 0x0f) == NOT_READY &&
SRpnt->sense[12] == 0x3a) { /* Check ASC */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STp->ready = ST_NO_TAPE;
} else
STp->ready = ST_NOT_READY;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
SRpnt = NULL;
STp->density = 0; /* Clear the erroneous "residue" */
STp->write_prot = 0;
STp->block_size = 0;
STp->ps[0].drv_file = STp->ps[0].drv_block = (-1);
STp->partition = STp->new_partition = 0;
STp->door_locked = ST_UNLOCKED;
return 0;
}
osst_configure_onstream(STp, &SRpnt);
STp->block_size = STp->raw ? OS_FRAME_SIZE : (
(STm->default_blksize > 0) ? STm->default_blksize : OS_DATA_SIZE);
STp->buffer->buffer_blocks = STp->raw ? 1 : OS_DATA_SIZE / STp->block_size;
STp->buffer->buffer_bytes =
STp->buffer->read_pointer =
STp->frame_in_buffer = 0;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Block size: %d, frame size: %d, buffer size: %d (%d blocks).\n",
name, STp->block_size, OS_FRAME_SIZE, (STp->buffer)->buffer_size,
(STp->buffer)->buffer_blocks);
#endif
if (STp->drv_write_prot) {
STp->write_prot = 1;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Write protected\n", name);
#endif
if ((flags & O_ACCMODE) == O_WRONLY || (flags & O_ACCMODE) == O_RDWR) {
retval = (-EROFS);
goto err_out;
}
}
if (new_session) { /* Change the drive parameters for the new mode */
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: New Session\n", name);
#endif
STp->density_changed = STp->blksize_changed = 0;
STp->compression_changed = 0;
}
/*
* properly position the tape and check the ADR headers
*/
if (STp->door_locked == ST_UNLOCKED) {
if (do_door_lock(STp, 1))
printk(KERN_INFO "%s:I: Can't lock drive door\n", name);
else
STp->door_locked = ST_LOCKED_AUTO;
}
osst_analyze_headers(STp, &SRpnt);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
SRpnt = NULL;
return 0;
err_out:
if (SRpnt != NULL)
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
normalize_buffer(STp->buffer);
STp->header_ok = 0;
STp->in_use = 0;
scsi_device_put(STp->device);
return retval;
}
osst: cdev lock_kernel() pushdown. Signed-off-by: Jonathan Corbet <corbet@lwn.net>
2008-05-15 18:23:19 +00:00
/* BKL pushdown: spaghetti avoidance wrapper */
static int os_scsi_tape_open(struct inode * inode, struct file * filp)
{
int ret;
scsi: autoconvert trivial BKL users to private mutex All these files use the big kernel lock in a trivial way to serialize their private file operations, typically resulting from an earlier semi-automatic pushdown from VFS. None of these drivers appears to want to lock against other code, and they all use the BKL as the top-level lock in their file operations, meaning that there is no lock-order inversion problem. Consequently, we can remove the BKL completely, replacing it with a per-file mutex in every case. Using a scripted approach means we can avoid typos. file=$1 name=$2 if grep -q lock_kernel ${file} ; then if grep -q 'include.*linux.mutex.h' ${file} ; then sed -i '/include.*<linux\/smp_lock.h>/d' ${file} else sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file} fi sed -i ${file} \ -e "/^#include.*linux.mutex.h/,$ { 1,/^\(static\|int\|long\)/ { /^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex); } }" \ -e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \ -e '/[ ]*cycle_kernel_lock();/d' else sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \ -e '/cycle_kernel_lock()/d' fi Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: linux-scsi@vger.kernel.org Cc: "James E.J. Bottomley" <James.Bottomley@suse.de>
2010-06-02 12:28:52 +00:00
mutex_lock(&osst_int_mutex);
osst: cdev lock_kernel() pushdown. Signed-off-by: Jonathan Corbet <corbet@lwn.net>
2008-05-15 18:23:19 +00:00
ret = __os_scsi_tape_open(inode, filp);
scsi: autoconvert trivial BKL users to private mutex All these files use the big kernel lock in a trivial way to serialize their private file operations, typically resulting from an earlier semi-automatic pushdown from VFS. None of these drivers appears to want to lock against other code, and they all use the BKL as the top-level lock in their file operations, meaning that there is no lock-order inversion problem. Consequently, we can remove the BKL completely, replacing it with a per-file mutex in every case. Using a scripted approach means we can avoid typos. file=$1 name=$2 if grep -q lock_kernel ${file} ; then if grep -q 'include.*linux.mutex.h' ${file} ; then sed -i '/include.*<linux\/smp_lock.h>/d' ${file} else sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file} fi sed -i ${file} \ -e "/^#include.*linux.mutex.h/,$ { 1,/^\(static\|int\|long\)/ { /^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex); } }" \ -e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \ -e '/[ ]*cycle_kernel_lock();/d' else sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \ -e '/cycle_kernel_lock()/d' fi Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: linux-scsi@vger.kernel.org Cc: "James E.J. Bottomley" <James.Bottomley@suse.de>
2010-06-02 12:28:52 +00:00
mutex_unlock(&osst_int_mutex);
osst: cdev lock_kernel() pushdown. Signed-off-by: Jonathan Corbet <corbet@lwn.net>
2008-05-15 18:23:19 +00:00
return ret;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/* Flush the tape buffer before close */
[PATCH] vfs: add lock owner argument to flush operation Pass the POSIX lock owner ID to the flush operation. This is useful for filesystems which don't want to store any locking state in inode->i_flock but want to handle locking/unlocking POSIX locks internally. FUSE is one such filesystem but I think it possible that some network filesystems would need this also. Also add a flag to indicate that a POSIX locking request was generated by close(), so filesystems using the above feature won't send an extra locking request in this case. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 09:05:12 +00:00
static int os_scsi_tape_flush(struct file * filp, fl_owner_t id)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
int result = 0, result2;
struct osst_tape * STp = filp->private_data;
struct st_modedef * STm = &(STp->modes[STp->current_mode]);
struct st_partstat * STps = &(STp->ps[STp->partition]);
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
char * name = tape_name(STp);
if (file_count(filp) > 1)
return 0;
if ((STps->rw == ST_WRITING || STp->dirty) && !STp->pos_unknown) {
STp->write_type = OS_WRITE_DATA;
result = osst_flush_write_buffer(STp, &SRpnt);
if (result != 0 && result != (-ENOSPC))
goto out;
}
if ( STps->rw >= ST_WRITING && !STp->pos_unknown) {
#if DEBUG
if (debugging) {
printk(OSST_DEB_MSG "%s:D: File length %ld bytes.\n",
name, (long)(filp->f_pos));
printk(OSST_DEB_MSG "%s:D: Async write waits %d, finished %d.\n",
name, STp->nbr_waits, STp->nbr_finished);
}
#endif
result = osst_write_trailer(STp, &SRpnt, !(STp->rew_at_close));
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG "%s:D: Buffer flushed, %d EOF(s) written\n",
name, 1+STp->two_fm);
#endif
}
else if (!STp->rew_at_close) {
STps = &(STp->ps[STp->partition]);
if (!STm->sysv || STps->rw != ST_READING) {
if (STp->can_bsr)
result = osst_flush_buffer(STp, &SRpnt, 0); /* this is the default path */
else if (STps->eof == ST_FM_HIT) {
result = cross_eof(STp, &SRpnt, 0);
if (result) {
if (STps->drv_file >= 0)
STps->drv_file++;
STps->drv_block = 0;
STps->eof = ST_FM;
}
else
STps->eof = ST_NOEOF;
}
}
else if ((STps->eof == ST_NOEOF &&
!(result = cross_eof(STp, &SRpnt, 1))) ||
STps->eof == ST_FM_HIT) {
if (STps->drv_file >= 0)
STps->drv_file++;
STps->drv_block = 0;
STps->eof = ST_FM;
}
}
out:
if (STp->rew_at_close) {
result2 = osst_position_tape_and_confirm(STp, &SRpnt, STp->first_data_ppos);
STps->drv_file = STps->drv_block = STp->frame_seq_number = STp->logical_blk_num = 0;
if (result == 0 && result2 < 0)
result = result2;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt) osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (STp->abort_count || STp->recover_count) {
printk(KERN_INFO "%s:I:", name);
if (STp->abort_count)
printk(" %d unrecovered errors", STp->abort_count);
if (STp->recover_count)
printk(" %d recovered errors", STp->recover_count);
if (STp->write_count)
printk(" in %d frames written", STp->write_count);
if (STp->read_count)
printk(" in %d frames read", STp->read_count);
printk("\n");
STp->recover_count = 0;
STp->abort_count = 0;
}
STp->write_count = 0;
STp->read_count = 0;
return result;
}
/* Close the device and release it */
static int os_scsi_tape_close(struct inode * inode, struct file * filp)
{
int result = 0;
struct osst_tape * STp = filp->private_data;
if (STp->door_locked == ST_LOCKED_AUTO)
do_door_lock(STp, 0);
if (STp->raw)
STp->header_ok = 0;
normalize_buffer(STp->buffer);
write_lock(&os_scsi_tapes_lock);
STp->in_use = 0;
write_unlock(&os_scsi_tapes_lock);
scsi_device_put(STp->device);
return result;
}
/* The ioctl command */
scsi: Push down BKL into ioctl functions Push down the bkl into ioctl functions on the scsi layer. [jkacur: Forward declaration missing ';'. Conflicting declaraction in megaraid.h changed Fixed missing inodes declarations] Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: John Kacur <jkacur@redhat.com> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-26 22:24:01 +00:00
static long osst_ioctl(struct file * file,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
unsigned int cmd_in, unsigned long arg)
{
[SCSI] Signdness issue in drivers/scsi/osst.c another signdness warning from gcc 4.1 drivers/scsi/osst.c:5154: warning: comparison of unsigned expression < 0 is always false The problem is that blk is defined as unsigned, but all usages of it are normal int cases. osst_get_frame_position() and osst_get_sector() return ints and can return negative values. If blk stays an unsigned int, the error check is useless. Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-09-25 23:59:08 +00:00
int i, cmd_nr, cmd_type, blk, retval = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct st_modedef * STm;
struct st_partstat * STps;
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
struct osst_request * SRpnt = NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct osst_tape * STp = file->private_data;
char * name = tape_name(STp);
void __user * p = (void __user *)arg;
scsi: autoconvert trivial BKL users to private mutex All these files use the big kernel lock in a trivial way to serialize their private file operations, typically resulting from an earlier semi-automatic pushdown from VFS. None of these drivers appears to want to lock against other code, and they all use the BKL as the top-level lock in their file operations, meaning that there is no lock-order inversion problem. Consequently, we can remove the BKL completely, replacing it with a per-file mutex in every case. Using a scripted approach means we can avoid typos. file=$1 name=$2 if grep -q lock_kernel ${file} ; then if grep -q 'include.*linux.mutex.h' ${file} ; then sed -i '/include.*<linux\/smp_lock.h>/d' ${file} else sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file} fi sed -i ${file} \ -e "/^#include.*linux.mutex.h/,$ { 1,/^\(static\|int\|long\)/ { /^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex); } }" \ -e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \ -e '/[ ]*cycle_kernel_lock();/d' else sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \ -e '/cycle_kernel_lock()/d' fi Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: linux-scsi@vger.kernel.org Cc: "James E.J. Bottomley" <James.Bottomley@suse.de>
2010-06-02 12:28:52 +00:00
mutex_lock(&osst_int_mutex);
scsi: Push down BKL into ioctl functions Push down the bkl into ioctl functions on the scsi layer. [jkacur: Forward declaration missing ';'. Conflicting declaraction in megaraid.h changed Fixed missing inodes declarations] Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: John Kacur <jkacur@redhat.com> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-26 22:24:01 +00:00
if (mutex_lock_interruptible(&STp->lock)) {
scsi: autoconvert trivial BKL users to private mutex All these files use the big kernel lock in a trivial way to serialize their private file operations, typically resulting from an earlier semi-automatic pushdown from VFS. None of these drivers appears to want to lock against other code, and they all use the BKL as the top-level lock in their file operations, meaning that there is no lock-order inversion problem. Consequently, we can remove the BKL completely, replacing it with a per-file mutex in every case. Using a scripted approach means we can avoid typos. file=$1 name=$2 if grep -q lock_kernel ${file} ; then if grep -q 'include.*linux.mutex.h' ${file} ; then sed -i '/include.*<linux\/smp_lock.h>/d' ${file} else sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file} fi sed -i ${file} \ -e "/^#include.*linux.mutex.h/,$ { 1,/^\(static\|int\|long\)/ { /^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex); } }" \ -e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \ -e '/[ ]*cycle_kernel_lock();/d' else sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \ -e '/cycle_kernel_lock()/d' fi Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: linux-scsi@vger.kernel.org Cc: "James E.J. Bottomley" <James.Bottomley@suse.de>
2010-06-02 12:28:52 +00:00
mutex_unlock(&osst_int_mutex);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return -ERESTARTSYS;
scsi: Push down BKL into ioctl functions Push down the bkl into ioctl functions on the scsi layer. [jkacur: Forward declaration missing ';'. Conflicting declaraction in megaraid.h changed Fixed missing inodes declarations] Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: John Kacur <jkacur@redhat.com> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-26 22:24:01 +00:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#if DEBUG
if (debugging && !STp->in_use) {
printk(OSST_DEB_MSG "%s:D: Incorrect device.\n", name);
retval = (-EIO);
goto out;
}
#endif
STm = &(STp->modes[STp->current_mode]);
STps = &(STp->ps[STp->partition]);
/*
* If we are in the middle of error recovery, don't let anyone
* else try and use this device. Also, if error recovery fails, it
* may try and take the device offline, in which case all further
* access to the device is prohibited.
*/
scsi: split scsi_nonblockable_ioctl The calling conventions for this function are bad as it could return -ENODEV both for a device not currently online and a not recognized ioctl. Add a new scsi_ioctl_block_when_processing_errors function that wraps scsi_block_when_processing_errors with the a special case for the SG_SCSI_RESET ioctl command, and handle the SG_SCSI_RESET case itself in scsi_ioctl. All callers of scsi_ioctl now must call the above helper to check for the EH state, so that the ioctl handler itself doesn't have to. Reported-by: Robert Elliott <Elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com> Reviewed-by: Hannes Reinecke <hare@suse.de>
2014-10-11 14:25:31 +00:00
retval = scsi_ioctl_block_when_processing_errors(STp->device, cmd_in,
file->f_flags & O_NDELAY);
if (retval)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
goto out;
cmd_type = _IOC_TYPE(cmd_in);
cmd_nr = _IOC_NR(cmd_in);
#if DEBUG
printk(OSST_DEB_MSG "%s:D: Ioctl %d,%d in %s mode\n", name,
cmd_type, cmd_nr, STp->raw?"raw":"normal");
#endif
if (cmd_type == _IOC_TYPE(MTIOCTOP) && cmd_nr == _IOC_NR(MTIOCTOP)) {
struct mtop mtc;
int auto_weof = 0;
if (_IOC_SIZE(cmd_in) != sizeof(mtc)) {
retval = (-EINVAL);
goto out;
}
i = copy_from_user((char *) &mtc, p, sizeof(struct mtop));
if (i) {
retval = (-EFAULT);
goto out;
}
if (mtc.mt_op == MTSETDRVBUFFER && !capable(CAP_SYS_ADMIN)) {
printk(KERN_WARNING "%s:W: MTSETDRVBUFFER only allowed for root.\n", name);
retval = (-EPERM);
goto out;
}
if (!STm->defined && (mtc.mt_op != MTSETDRVBUFFER && (mtc.mt_count & MT_ST_OPTIONS) == 0)) {
retval = (-ENXIO);
goto out;
}
if (!STp->pos_unknown) {
if (STps->eof == ST_FM_HIT) {
if (mtc.mt_op == MTFSF || mtc.mt_op == MTFSFM|| mtc.mt_op == MTEOM) {
mtc.mt_count -= 1;
if (STps->drv_file >= 0)
STps->drv_file += 1;
}
else if (mtc.mt_op == MTBSF || mtc.mt_op == MTBSFM) {
mtc.mt_count += 1;
if (STps->drv_file >= 0)
STps->drv_file += 1;
}
}
if (mtc.mt_op == MTSEEK) {
/* Old position must be restored if partition will be changed */
i = !STp->can_partitions || (STp->new_partition != STp->partition);
}
else {
i = mtc.mt_op == MTREW || mtc.mt_op == MTOFFL ||
mtc.mt_op == MTRETEN || mtc.mt_op == MTEOM ||
mtc.mt_op == MTLOCK || mtc.mt_op == MTLOAD ||
mtc.mt_op == MTFSF || mtc.mt_op == MTFSFM ||
mtc.mt_op == MTBSF || mtc.mt_op == MTBSFM ||
mtc.mt_op == MTCOMPRESSION;
}
i = osst_flush_buffer(STp, &SRpnt, i);
if (i < 0) {
retval = i;
goto out;
}
}
else {
/*
* If there was a bus reset, block further access
* to this device. If the user wants to rewind the tape,
* then reset the flag and allow access again.
*/
if(mtc.mt_op != MTREW &&
mtc.mt_op != MTOFFL &&
mtc.mt_op != MTRETEN &&
mtc.mt_op != MTERASE &&
mtc.mt_op != MTSEEK &&
mtc.mt_op != MTEOM) {
retval = (-EIO);
goto out;
}
reset_state(STp);
/* remove this when the midlevel properly clears was_reset */
STp->device->was_reset = 0;
}
if (mtc.mt_op != MTCOMPRESSION && mtc.mt_op != MTLOCK &&
mtc.mt_op != MTNOP && mtc.mt_op != MTSETBLK &&
mtc.mt_op != MTSETDENSITY && mtc.mt_op != MTSETDRVBUFFER &&
mtc.mt_op != MTMKPART && mtc.mt_op != MTSETPART &&
mtc.mt_op != MTWEOF && mtc.mt_op != MTWSM ) {
/*
* The user tells us to move to another position on the tape.
* If we were appending to the tape content, that would leave
* the tape without proper end, in that case write EOD and
* update the header to reflect its position.
*/
#if DEBUG
printk(KERN_WARNING "%s:D: auto_weod %s at ffp=%d,efp=%d,fsn=%d,lbn=%d,fn=%d,bn=%d\n", name,
STps->rw >= ST_WRITING ? "write" : STps->rw == ST_READING ? "read" : "idle",
STp->first_frame_position, STp->eod_frame_ppos, STp->frame_seq_number,
STp->logical_blk_num, STps->drv_file, STps->drv_block );
#endif
if (STps->rw >= ST_WRITING && STp->first_frame_position >= STp->eod_frame_ppos) {
auto_weof = ((STp->write_type != OS_WRITE_NEW_MARK) &&
!(mtc.mt_op == MTREW || mtc.mt_op == MTOFFL));
i = osst_write_trailer(STp, &SRpnt,
!(mtc.mt_op == MTREW || mtc.mt_op == MTOFFL));
#if DEBUG
printk(KERN_WARNING "%s:D: post trailer xeof=%d,ffp=%d,efp=%d,fsn=%d,lbn=%d,fn=%d,bn=%d\n",
name, auto_weof, STp->first_frame_position, STp->eod_frame_ppos,
STp->frame_seq_number, STp->logical_blk_num, STps->drv_file, STps->drv_block );
#endif
if (i < 0) {
retval = i;
goto out;
}
}
STps->rw = ST_IDLE;
}
if (mtc.mt_op == MTOFFL && STp->door_locked != ST_UNLOCKED)
do_door_lock(STp, 0); /* Ignore result! */
if (mtc.mt_op == MTSETDRVBUFFER &&
(mtc.mt_count & MT_ST_OPTIONS) != 0) {
retval = osst_set_options(STp, mtc.mt_count);
goto out;
}
if (mtc.mt_op == MTSETPART) {
if (mtc.mt_count >= STp->nbr_partitions)
retval = -EINVAL;
else {
STp->new_partition = mtc.mt_count;
retval = 0;
}
goto out;
}
if (mtc.mt_op == MTMKPART) {
if (!STp->can_partitions) {
retval = (-EINVAL);
goto out;
}
if ((i = osst_int_ioctl(STp, &SRpnt, MTREW, 0)) < 0 /*||
(i = partition_tape(inode, mtc.mt_count)) < 0*/) {
retval = i;
goto out;
}
for (i=0; i < ST_NBR_PARTITIONS; i++) {
STp->ps[i].rw = ST_IDLE;
STp->ps[i].at_sm = 0;
STp->ps[i].last_block_valid = 0;
}
STp->partition = STp->new_partition = 0;
STp->nbr_partitions = 1; /* Bad guess ?-) */
STps->drv_block = STps->drv_file = 0;
retval = 0;
goto out;
}
if (mtc.mt_op == MTSEEK) {
if (STp->raw)
i = osst_set_frame_position(STp, &SRpnt, mtc.mt_count, 0);
else
i = osst_seek_sector(STp, &SRpnt, mtc.mt_count);
if (!STp->can_partitions)
STp->ps[0].rw = ST_IDLE;
retval = i;
goto out;
}
if (mtc.mt_op == MTLOCK || mtc.mt_op == MTUNLOCK) {
retval = do_door_lock(STp, (mtc.mt_op == MTLOCK));
goto out;
}
if (auto_weof)
cross_eof(STp, &SRpnt, 0);
if (mtc.mt_op == MTCOMPRESSION)
retval = -EINVAL; /* OnStream drives don't have compression hardware */
else
/* MTBSF MTBSFM MTBSR MTBSS MTEOM MTERASE MTFSF MTFSFB MTFSR MTFSS
* MTLOAD MTOFFL MTRESET MTRETEN MTREW MTUNLOAD MTWEOF MTWSM */
retval = osst_int_ioctl(STp, &SRpnt, mtc.mt_op, mtc.mt_count);
goto out;
}
if (!STm->defined) {
retval = (-ENXIO);
goto out;
}
if ((i = osst_flush_buffer(STp, &SRpnt, 0)) < 0) {
retval = i;
goto out;
}
if (cmd_type == _IOC_TYPE(MTIOCGET) && cmd_nr == _IOC_NR(MTIOCGET)) {
struct mtget mt_status;
if (_IOC_SIZE(cmd_in) != sizeof(struct mtget)) {
retval = (-EINVAL);
goto out;
}
mt_status.mt_type = MT_ISONSTREAM_SC;
mt_status.mt_erreg = STp->recover_erreg << MT_ST_SOFTERR_SHIFT;
mt_status.mt_dsreg =
((STp->block_size << MT_ST_BLKSIZE_SHIFT) & MT_ST_BLKSIZE_MASK) |
((STp->density << MT_ST_DENSITY_SHIFT) & MT_ST_DENSITY_MASK);
mt_status.mt_blkno = STps->drv_block;
mt_status.mt_fileno = STps->drv_file;
if (STp->block_size != 0) {
if (STps->rw == ST_WRITING)
mt_status.mt_blkno += (STp->buffer)->buffer_bytes / STp->block_size;
else if (STps->rw == ST_READING)
mt_status.mt_blkno -= ((STp->buffer)->buffer_bytes +
STp->block_size - 1) / STp->block_size;
}
mt_status.mt_gstat = 0;
if (STp->drv_write_prot)
mt_status.mt_gstat |= GMT_WR_PROT(0xffffffff);
if (mt_status.mt_blkno == 0) {
if (mt_status.mt_fileno == 0)
mt_status.mt_gstat |= GMT_BOT(0xffffffff);
else
mt_status.mt_gstat |= GMT_EOF(0xffffffff);
}
mt_status.mt_resid = STp->partition;
if (STps->eof == ST_EOM_OK || STps->eof == ST_EOM_ERROR)
mt_status.mt_gstat |= GMT_EOT(0xffffffff);
else if (STps->eof >= ST_EOM_OK)
mt_status.mt_gstat |= GMT_EOD(0xffffffff);
if (STp->density == 1)
mt_status.mt_gstat |= GMT_D_800(0xffffffff);
else if (STp->density == 2)
mt_status.mt_gstat |= GMT_D_1600(0xffffffff);
else if (STp->density == 3)
mt_status.mt_gstat |= GMT_D_6250(0xffffffff);
if (STp->ready == ST_READY)
mt_status.mt_gstat |= GMT_ONLINE(0xffffffff);
if (STp->ready == ST_NO_TAPE)
mt_status.mt_gstat |= GMT_DR_OPEN(0xffffffff);
if (STps->at_sm)
mt_status.mt_gstat |= GMT_SM(0xffffffff);
if (STm->do_async_writes || (STm->do_buffer_writes && STp->block_size != 0) ||
STp->drv_buffer != 0)
mt_status.mt_gstat |= GMT_IM_REP_EN(0xffffffff);
i = copy_to_user(p, &mt_status, sizeof(struct mtget));
if (i) {
retval = (-EFAULT);
goto out;
}
STp->recover_erreg = 0; /* Clear after read */
retval = 0;
goto out;
} /* End of MTIOCGET */
if (cmd_type == _IOC_TYPE(MTIOCPOS) && cmd_nr == _IOC_NR(MTIOCPOS)) {
struct mtpos mt_pos;
if (_IOC_SIZE(cmd_in) != sizeof(struct mtpos)) {
retval = (-EINVAL);
goto out;
}
if (STp->raw)
blk = osst_get_frame_position(STp, &SRpnt);
else
blk = osst_get_sector(STp, &SRpnt);
if (blk < 0) {
retval = blk;
goto out;
}
mt_pos.mt_blkno = blk;
i = copy_to_user(p, &mt_pos, sizeof(struct mtpos));
if (i)
retval = -EFAULT;
goto out;
}
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt) osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: Use mutex instead of semaphore The OnStream SCSI Tape driver uses a semaphore as mutex. Use the mutex API instead of the (binary) semaphore. Signed-off-by: Matthias Kaehlcke <matthias.kaehlcke@gmail.com> Reviewed-by: Satyam Sharma <satyam@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Willem Riede <wrlk@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-08-10 21:50:44 +00:00
mutex_unlock(&STp->lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
scsi: Push down BKL into ioctl functions Push down the bkl into ioctl functions on the scsi layer. [jkacur: Forward declaration missing ';'. Conflicting declaraction in megaraid.h changed Fixed missing inodes declarations] Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: John Kacur <jkacur@redhat.com> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-26 22:24:01 +00:00
retval = scsi_ioctl(STp->device, cmd_in, p);
scsi: autoconvert trivial BKL users to private mutex All these files use the big kernel lock in a trivial way to serialize their private file operations, typically resulting from an earlier semi-automatic pushdown from VFS. None of these drivers appears to want to lock against other code, and they all use the BKL as the top-level lock in their file operations, meaning that there is no lock-order inversion problem. Consequently, we can remove the BKL completely, replacing it with a per-file mutex in every case. Using a scripted approach means we can avoid typos. file=$1 name=$2 if grep -q lock_kernel ${file} ; then if grep -q 'include.*linux.mutex.h' ${file} ; then sed -i '/include.*<linux\/smp_lock.h>/d' ${file} else sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file} fi sed -i ${file} \ -e "/^#include.*linux.mutex.h/,$ { 1,/^\(static\|int\|long\)/ { /^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex); } }" \ -e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \ -e '/[ ]*cycle_kernel_lock();/d' else sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \ -e '/cycle_kernel_lock()/d' fi Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: linux-scsi@vger.kernel.org Cc: "James E.J. Bottomley" <James.Bottomley@suse.de>
2010-06-02 12:28:52 +00:00
mutex_unlock(&osst_int_mutex);
scsi: Push down BKL into ioctl functions Push down the bkl into ioctl functions on the scsi layer. [jkacur: Forward declaration missing ';'. Conflicting declaraction in megaraid.h changed Fixed missing inodes declarations] Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: John Kacur <jkacur@redhat.com> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-26 22:24:01 +00:00
return retval;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
out:
[SCSI] osst: changes required to move forward to block request On 02/07/2006 04:12:55 AM, Christoph Hellwig wrote: > On Mon, Feb 06, 2006 at 08:02:21PM -0500, Willem Riede wrote: > > > But I will certainly help retire scsi_request. And anything else that is > > needed to keep up with proper kernel style. Let me know what those are, if > > you would? I'll start looking at how st has changed, and will be back with > > any questions I may have. > > right now the above is the most urgent bit. What would be nice but not > required is a conversion to the sense handling helpers, similar to what > st got (aka using the *normalize_sense functions and then dealing with the > parsed sense buffer instead of the raw sense data) Ok, so here is my first take at satisfying this request. Be warned, that beyond compiling, and checking that the new module doesn't immediately blow up, there hasn't yet been a lot of testing. But this should allow you to comment on the changes, and move forward with dropping scsi_request from the kernel code. Signed-off-by: Willem Riede <osst@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-02-11 19:46:56 +00:00
if (SRpnt) osst_release_request(SRpnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] osst: Use mutex instead of semaphore The OnStream SCSI Tape driver uses a semaphore as mutex. Use the mutex API instead of the (binary) semaphore. Signed-off-by: Matthias Kaehlcke <matthias.kaehlcke@gmail.com> Reviewed-by: Satyam Sharma <satyam@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Willem Riede <wrlk@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-08-10 21:50:44 +00:00
mutex_unlock(&STp->lock);
scsi: autoconvert trivial BKL users to private mutex All these files use the big kernel lock in a trivial way to serialize their private file operations, typically resulting from an earlier semi-automatic pushdown from VFS. None of these drivers appears to want to lock against other code, and they all use the BKL as the top-level lock in their file operations, meaning that there is no lock-order inversion problem. Consequently, we can remove the BKL completely, replacing it with a per-file mutex in every case. Using a scripted approach means we can avoid typos. file=$1 name=$2 if grep -q lock_kernel ${file} ; then if grep -q 'include.*linux.mutex.h' ${file} ; then sed -i '/include.*<linux\/smp_lock.h>/d' ${file} else sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file} fi sed -i ${file} \ -e "/^#include.*linux.mutex.h/,$ { 1,/^\(static\|int\|long\)/ { /^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex); } }" \ -e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \ -e '/[ ]*cycle_kernel_lock();/d' else sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \ -e '/cycle_kernel_lock()/d' fi Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: linux-scsi@vger.kernel.org Cc: "James E.J. Bottomley" <James.Bottomley@suse.de>
2010-06-02 12:28:52 +00:00
mutex_unlock(&osst_int_mutex);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return retval;
}
#ifdef CONFIG_COMPAT
static long osst_compat_ioctl(struct file * file, unsigned int cmd_in, unsigned long arg)
{
struct osst_tape *STp = file->private_data;
struct scsi_device *sdev = STp->device;
int ret = -ENOIOCTLCMD;
if (sdev->host->hostt->compat_ioctl) {
ret = sdev->host->hostt->compat_ioctl(sdev, cmd_in, (void __user *)arg);
}
return ret;
}
#endif
/* Memory handling routines */
/* Try to allocate a new tape buffer skeleton. Caller must not hold os_scsi_tapes_lock */
static struct osst_buffer * new_tape_buffer( int from_initialization, int need_dma, int max_sg )
{
[PATCH] gfp_t: drivers/scsi Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-21 07:22:08 +00:00
int i;
gfp_t priority;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct osst_buffer *tb;
if (from_initialization)
priority = GFP_ATOMIC;
else
priority = GFP_KERNEL;
i = sizeof(struct osst_buffer) + (osst_max_sg_segs - 1) * sizeof(struct scatterlist);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
tb = kzalloc(i, priority);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (!tb) {
printk(KERN_NOTICE "osst :I: Can't allocate new tape buffer.\n");
return NULL;
}
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
tb->sg_segs = tb->orig_sg_segs = 0;
tb->use_sg = max_sg;
tb->in_use = 1;
tb->dma = need_dma;
tb->buffer_size = 0;
#if DEBUG
if (debugging)
printk(OSST_DEB_MSG
"osst :D: Allocated tape buffer skeleton (%d bytes, %d segments, dma: %d).\n",
i, max_sg, need_dma);
#endif
return tb;
}
/* Try to allocate a temporary (while a user has the device open) enlarged tape buffer */
static int enlarge_buffer(struct osst_buffer *STbuffer, int need_dma)
{
[PATCH] gfp_t: drivers/scsi Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-21 07:22:08 +00:00
int segs, nbr, max_segs, b_size, order, got;
gfp_t priority;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (STbuffer->buffer_size >= OS_FRAME_SIZE)
return 1;
if (STbuffer->sg_segs) {
printk(KERN_WARNING "osst :A: Buffer not previously normalized.\n");
normalize_buffer(STbuffer);
}
/* See how many segments we can use -- need at least two */
nbr = max_segs = STbuffer->use_sg;
if (nbr <= 2)
return 0;
priority = GFP_KERNEL /* | __GFP_NOWARN */;
if (need_dma)
priority |= GFP_DMA;
/* Try to allocate the first segment up to OS_DATA_SIZE and the others
big enough to reach the goal (code assumes no segments in place) */
for (b_size = OS_DATA_SIZE, order = OSST_FIRST_ORDER; b_size >= PAGE_SIZE; order--, b_size /= 2) {
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
struct page *page = alloc_pages(priority, order);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STbuffer->sg[0].offset = 0;
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
if (page != NULL) {
SG: Change sg_set_page() to take length and offset argument Most drivers need to set length and offset as well, so may as well fold those three lines into one. Add sg_assign_page() for those two locations that only needed to set the page, where the offset/length is set outside of the function context. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-24 09:20:47 +00:00
sg_set_page(&STbuffer->sg[0], page, b_size, 0);
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
STbuffer->b_data = page_address(page);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
break;
}
}
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
if (sg_page(&STbuffer->sg[0]) == NULL) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_NOTICE "osst :I: Can't allocate tape buffer main segment.\n");
return 0;
}
/* Got initial segment of 'bsize,order', continue with same size if possible, except for AUX */
for (segs=STbuffer->sg_segs=1, got=b_size;
segs < max_segs && got < OS_FRAME_SIZE; ) {
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
struct page *page = alloc_pages(priority, (OS_FRAME_SIZE - got <= PAGE_SIZE) ? 0 : order);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STbuffer->sg[segs].offset = 0;
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
if (page == NULL) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_WARNING "osst :W: Failed to enlarge buffer to %d bytes.\n",
OS_FRAME_SIZE);
#if DEBUG
STbuffer->buffer_size = got;
#endif
normalize_buffer(STbuffer);
return 0;
}
SG: Change sg_set_page() to take length and offset argument Most drivers need to set length and offset as well, so may as well fold those three lines into one. Add sg_assign_page() for those two locations that only needed to set the page, where the offset/length is set outside of the function context. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-24 09:20:47 +00:00
sg_set_page(&STbuffer->sg[segs], page, (OS_FRAME_SIZE - got <= PAGE_SIZE / 2) ? (OS_FRAME_SIZE - got) : b_size, 0);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
got += STbuffer->sg[segs].length;
STbuffer->buffer_size = got;
STbuffer->sg_segs = ++segs;
}
#if DEBUG
if (debugging) {
printk(OSST_DEB_MSG
"osst :D: Expanded tape buffer (%d bytes, %d->%d segments, dma: %d, at: %p).\n",
got, STbuffer->orig_sg_segs, STbuffer->sg_segs, need_dma, STbuffer->b_data);
printk(OSST_DEB_MSG
"osst :D: segment sizes: first %d at %p, last %d bytes at %p.\n",
STbuffer->sg[0].length, page_address(STbuffer->sg[0].page),
STbuffer->sg[segs-1].length, page_address(STbuffer->sg[segs-1].page));
}
#endif
return 1;
}
/* Release the segments */
static void normalize_buffer(struct osst_buffer *STbuffer)
{
int i, order, b_size;
for (i=0; i < STbuffer->sg_segs; i++) {
for (b_size = PAGE_SIZE, order = 0;
b_size < STbuffer->sg[i].length;
b_size *= 2, order++);
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
__free_pages(sg_page(&STbuffer->sg[i]), order);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
STbuffer->buffer_size -= STbuffer->sg[i].length;
}
#if DEBUG
if (debugging && STbuffer->orig_sg_segs < STbuffer->sg_segs)
printk(OSST_DEB_MSG "osst :D: Buffer at %p normalized to %d bytes (segs %d).\n",
STbuffer->b_data, STbuffer->buffer_size, STbuffer->sg_segs);
#endif
STbuffer->sg_segs = STbuffer->orig_sg_segs = 0;
}
/* Move data from the user buffer to the tape buffer. Returns zero (success) or
negative error code. */
static int append_to_buffer(const char __user *ubp, struct osst_buffer *st_bp, int do_count)
{
int i, cnt, res, offset;
for (i=0, offset=st_bp->buffer_bytes;
i < st_bp->sg_segs && offset >= st_bp->sg[i].length; i++)
scsi: osst: add missing indent on a for loop statement The for loop is statement is missing an indent, add it. Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2017-08-15 14:08:28 +00:00
offset -= st_bp->sg[i].length;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (i == st_bp->sg_segs) { /* Should never happen */
printk(KERN_WARNING "osst :A: Append_to_buffer offset overflow.\n");
return (-EIO);
}
for ( ; i < st_bp->sg_segs && do_count > 0; i++) {
cnt = st_bp->sg[i].length - offset < do_count ?
st_bp->sg[i].length - offset : do_count;
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
res = copy_from_user(page_address(sg_page(&st_bp->sg[i])) + offset, ubp, cnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (res)
return (-EFAULT);
do_count -= cnt;
st_bp->buffer_bytes += cnt;
ubp += cnt;
offset = 0;
}
if (do_count) { /* Should never happen */
printk(KERN_WARNING "osst :A: Append_to_buffer overflow (left %d).\n",
do_count);
return (-EIO);
}
return 0;
}
/* Move data from the tape buffer to the user buffer. Returns zero (success) or
negative error code. */
static int from_buffer(struct osst_buffer *st_bp, char __user *ubp, int do_count)
{
int i, cnt, res, offset;
for (i=0, offset=st_bp->read_pointer;
i < st_bp->sg_segs && offset >= st_bp->sg[i].length; i++)
offset -= st_bp->sg[i].length;
if (i == st_bp->sg_segs) { /* Should never happen */
printk(KERN_WARNING "osst :A: From_buffer offset overflow.\n");
return (-EIO);
}
for ( ; i < st_bp->sg_segs && do_count > 0; i++) {
cnt = st_bp->sg[i].length - offset < do_count ?
st_bp->sg[i].length - offset : do_count;
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
res = copy_to_user(ubp, page_address(sg_page(&st_bp->sg[i])) + offset, cnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (res)
return (-EFAULT);
do_count -= cnt;
st_bp->buffer_bytes -= cnt;
st_bp->read_pointer += cnt;
ubp += cnt;
offset = 0;
}
if (do_count) { /* Should never happen */
printk(KERN_WARNING "osst :A: From_buffer overflow (left %d).\n", do_count);
return (-EIO);
}
return 0;
}
/* Sets the tail of the buffer after fill point to zero.
Returns zero (success) or negative error code. */
static int osst_zero_buffer_tail(struct osst_buffer *st_bp)
{
int i, offset, do_count, cnt;
for (i = 0, offset = st_bp->buffer_bytes;
i < st_bp->sg_segs && offset >= st_bp->sg[i].length; i++)
offset -= st_bp->sg[i].length;
if (i == st_bp->sg_segs) { /* Should never happen */
printk(KERN_WARNING "osst :A: Zero_buffer offset overflow.\n");
return (-EIO);
}
for (do_count = OS_DATA_SIZE - st_bp->buffer_bytes;
i < st_bp->sg_segs && do_count > 0; i++) {
cnt = st_bp->sg[i].length - offset < do_count ?
st_bp->sg[i].length - offset : do_count ;
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
memset(page_address(sg_page(&st_bp->sg[i])) + offset, 0, cnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
do_count -= cnt;
offset = 0;
}
if (do_count) { /* Should never happen */
printk(KERN_WARNING "osst :A: Zero_buffer overflow (left %d).\n", do_count);
return (-EIO);
}
return 0;
}
/* Copy a osst 32K chunk of memory into the buffer.
Returns zero (success) or negative error code. */
static int osst_copy_to_buffer(struct osst_buffer *st_bp, unsigned char *ptr)
{
int i, cnt, do_count = OS_DATA_SIZE;
for (i = 0; i < st_bp->sg_segs && do_count > 0; i++) {
cnt = st_bp->sg[i].length < do_count ?
st_bp->sg[i].length : do_count ;
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
memcpy(page_address(sg_page(&st_bp->sg[i])), ptr, cnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
do_count -= cnt;
ptr += cnt;
}
if (do_count || i != st_bp->sg_segs-1) { /* Should never happen */
printk(KERN_WARNING "osst :A: Copy_to_buffer overflow (left %d at sg %d).\n",
do_count, i);
return (-EIO);
}
return 0;
}
/* Copy a osst 32K chunk of memory from the buffer.
Returns zero (success) or negative error code. */
static int osst_copy_from_buffer(struct osst_buffer *st_bp, unsigned char *ptr)
{
int i, cnt, do_count = OS_DATA_SIZE;
for (i = 0; i < st_bp->sg_segs && do_count > 0; i++) {
cnt = st_bp->sg[i].length < do_count ?
st_bp->sg[i].length : do_count ;
[SG] Update drivers to use sg helpers Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-10-22 19:19:53 +00:00
memcpy(ptr, page_address(sg_page(&st_bp->sg[i])), cnt);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
do_count -= cnt;
ptr += cnt;
}
if (do_count || i != st_bp->sg_segs-1) { /* Should never happen */
printk(KERN_WARNING "osst :A: Copy_from_buffer overflow (left %d at sg %d).\n",
do_count, i);
return (-EIO);
}
return 0;
}
/* Module housekeeping */
static void validate_options (void)
{
if (max_dev > 0)
osst_max_dev = max_dev;
if (write_threshold_kbs > 0)
osst_write_threshold = write_threshold_kbs * ST_KILOBYTE;
if (osst_write_threshold > osst_buffer_size)
osst_write_threshold = osst_buffer_size;
if (max_sg_segs >= OSST_FIRST_SG)
osst_max_sg_segs = max_sg_segs;
#if DEBUG
printk(OSST_DEB_MSG "osst :D: max tapes %d, write threshold %d, max s/g segs %d.\n",
osst_max_dev, osst_write_threshold, osst_max_sg_segs);
#endif
}
#ifndef MODULE
/* Set the boot options. Syntax: osst=xxx,yyy,...
where xxx is write threshold in 1024 byte blocks,
and yyy is number of s/g segments to use. */
static int __init osst_setup (char *str)
{
int i, ints[5];
char *stp;
stp = get_options(str, ARRAY_SIZE(ints), ints);
[SCSI] drivers/scsi: Use ARRAY_SIZE macro Use ARRAY_SIZE macro instead of sizeof(x)/sizeof(x[0]) and remove duplicates of the macro. Signed-off-by: Tobias Klauser <tklauser@nuerscht.ch> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-06-09 05:23:48 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (ints[0] > 0) {
for (i = 0; i < ints[0] && i < ARRAY_SIZE(parms); i++)
*parms[i].val = ints[i + 1];
} else {
while (stp != NULL) {
for (i = 0; i < ARRAY_SIZE(parms); i++) {
int len = strlen(parms[i].name);
if (!strncmp(stp, parms[i].name, len) &&
(*(stp + len) == ':' || *(stp + len) == '=')) {
*parms[i].val =
simple_strtoul(stp + len + 1, NULL, 0);
break;
}
}
[SCSI] drivers/scsi: Use ARRAY_SIZE macro Use ARRAY_SIZE macro instead of sizeof(x)/sizeof(x[0]) and remove duplicates of the macro. Signed-off-by: Tobias Klauser <tklauser@nuerscht.ch> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-06-09 05:23:48 +00:00
if (i >= ARRAY_SIZE(parms))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_INFO "osst :I: Illegal parameter in '%s'\n",
stp);
stp = strchr(stp, ',');
if (stp)
stp++;
}
}
return 1;
}
__setup("osst=", osst_setup);
#endif
[PATCH] mark struct file_operations const 6 Many struct file_operations in the kernel can be "const". Marking them const moves these to the .rodata section, which avoids false sharing with potential dirty data. In addition it'll catch accidental writes at compile time to these shared resources. Signed-off-by: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:55:34 +00:00
static const struct file_operations osst_fops = {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
.owner = THIS_MODULE,
.read = osst_read,
.write = osst_write,
scsi: Push down BKL into ioctl functions Push down the bkl into ioctl functions on the scsi layer. [jkacur: Forward declaration missing ';'. Conflicting declaraction in megaraid.h changed Fixed missing inodes declarations] Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: John Kacur <jkacur@redhat.com> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-26 22:24:01 +00:00
.unlocked_ioctl = osst_ioctl,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#ifdef CONFIG_COMPAT
.compat_ioctl = osst_compat_ioctl,
#endif
.open = os_scsi_tape_open,
.flush = os_scsi_tape_flush,
.release = os_scsi_tape_close,
osst: use noop_llseek() instead of default_llseek() __os_scsi_tape_open() suggests that llseek() doesn't work: "We really want to do nonseekable_open(inode, filp); here, but some versions of tar incorrectly call lseek on tapes and bail out if that fails. So we disallow pread() and pwrite(), but permit lseeks." Instead of using the fallback default_llseek() the driver should use noop_llseek() which leaves the file->f_pos untouched but succeeds. Signed-off-by: Jan Blunck <jblunck@suse.de> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Willem Riede <osst@riede.org> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-26 21:44:50 +00:00
.llseek = noop_llseek,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
};
static int osst_supports(struct scsi_device * SDp)
{
struct osst_support_data {
char *vendor;
char *model;
char *rev;
char *driver_hint; /* Name of the correct driver, NULL if unknown */
};
static struct osst_support_data support_list[] = {
/* {"XXX", "Yy-", "", NULL}, example */
SIGS_FROM_OSST,
{NULL, }};
struct osst_support_data *rp;
/* We are willing to drive OnStream SC-x0 as well as the
* * IDE, ParPort, FireWire, USB variants, if accessible by
* * emulation layer (ide-scsi, usb-storage, ...) */
for (rp=&(support_list[0]); rp->vendor != NULL; rp++)
if (!strncmp(rp->vendor, SDp->vendor, strlen(rp->vendor)) &&
!strncmp(rp->model, SDp->model, strlen(rp->model)) &&
!strncmp(rp->rev, SDp->rev, strlen(rp->rev)))
return 1;
return 0;
}
/*
* sysfs support for osst driver parameter information
*/
scsi: remove DRIVER_ATTR() usage It's better to use the DRIVER_ATTR_RW() and DRIVER_ATTR_RO() macros to explicitly show that this is a read/write or read/only sysfs file. So convert the remaining SCSI drivers that use the old style to use the newer macros. Bonus is that this removes some checkpatch.pl warnings :) This is part of a series to drop DRIVER_ATTR() from the tree entirely. Cc: "James E.J. Bottomley" <jejb@linux.vnet.ibm.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Kashyap Desai <kashyap.desai@broadcom.com> Cc: Sumit Saxena <sumit.saxena@broadcom.com> Cc: Shivasharan S <shivasharan.srikanteshwara@broadcom.com> Cc: Willem Riede <osst@riede.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2017-07-19 12:50:06 +00:00
static ssize_t version_show(struct device_driver *ddd, char *buf)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
return snprintf(buf, PAGE_SIZE, "%s\n", osst_version);
}
scsi: remove DRIVER_ATTR() usage It's better to use the DRIVER_ATTR_RW() and DRIVER_ATTR_RO() macros to explicitly show that this is a read/write or read/only sysfs file. So convert the remaining SCSI drivers that use the old style to use the newer macros. Bonus is that this removes some checkpatch.pl warnings :) This is part of a series to drop DRIVER_ATTR() from the tree entirely. Cc: "James E.J. Bottomley" <jejb@linux.vnet.ibm.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Kashyap Desai <kashyap.desai@broadcom.com> Cc: Sumit Saxena <sumit.saxena@broadcom.com> Cc: Shivasharan S <shivasharan.srikanteshwara@broadcom.com> Cc: Willem Riede <osst@riede.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2017-07-19 12:50:06 +00:00
static DRIVER_ATTR_RO(version);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
Replace remaining references to "driverfs" with "sysfs". Globally, s/driverfs/sysfs/g. Signed-off-by: Robert P. J. Day <rpjday@mindspring.com> Signed-off-by: Adrian Bunk <bunk@stusta.de>
2007-02-17 18:13:42 +00:00
static int osst_create_sysfs_files(struct device_driver *sysfs)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
Replace remaining references to "driverfs" with "sysfs". Globally, s/driverfs/sysfs/g. Signed-off-by: Robert P. J. Day <rpjday@mindspring.com> Signed-off-by: Adrian Bunk <bunk@stusta.de>
2007-02-17 18:13:42 +00:00
return driver_create_file(sysfs, &driver_attr_version);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
Replace remaining references to "driverfs" with "sysfs". Globally, s/driverfs/sysfs/g. Signed-off-by: Robert P. J. Day <rpjday@mindspring.com> Signed-off-by: Adrian Bunk <bunk@stusta.de>
2007-02-17 18:13:42 +00:00
static void osst_remove_sysfs_files(struct device_driver *sysfs)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
Replace remaining references to "driverfs" with "sysfs". Globally, s/driverfs/sysfs/g. Signed-off-by: Robert P. J. Day <rpjday@mindspring.com> Signed-off-by: Adrian Bunk <bunk@stusta.de>
2007-02-17 18:13:42 +00:00
driver_remove_file(sysfs, &driver_attr_version);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
/*
* sysfs support for accessing ADR header information
*/
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
static ssize_t osst_adr_rev_show(struct device *dev,
struct device_attribute *attr, char *buf)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
struct osst_tape * STp = (struct osst_tape *) dev_get_drvdata (dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ssize_t l = 0;
if (STp && STp->header_ok && STp->linux_media)
l = snprintf(buf, PAGE_SIZE, "%d.%d\n", STp->header_cache->major_rev, STp->header_cache->minor_rev);
return l;
}
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
DEVICE_ATTR(ADR_rev, S_IRUGO, osst_adr_rev_show, NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
static ssize_t osst_linux_media_version_show(struct device *dev,
struct device_attribute *attr,
char *buf)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
struct osst_tape * STp = (struct osst_tape *) dev_get_drvdata (dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ssize_t l = 0;
if (STp && STp->header_ok && STp->linux_media)
l = snprintf(buf, PAGE_SIZE, "LIN%d\n", STp->linux_media_version);
return l;
}
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
DEVICE_ATTR(media_version, S_IRUGO, osst_linux_media_version_show, NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
static ssize_t osst_capacity_show(struct device *dev,
struct device_attribute *attr, char *buf)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
struct osst_tape * STp = (struct osst_tape *) dev_get_drvdata (dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ssize_t l = 0;
if (STp && STp->header_ok && STp->linux_media)
l = snprintf(buf, PAGE_SIZE, "%d\n", STp->capacity);
return l;
}
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
DEVICE_ATTR(capacity, S_IRUGO, osst_capacity_show, NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
static ssize_t osst_first_data_ppos_show(struct device *dev,
struct device_attribute *attr,
char *buf)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
struct osst_tape * STp = (struct osst_tape *) dev_get_drvdata (dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ssize_t l = 0;
if (STp && STp->header_ok && STp->linux_media)
l = snprintf(buf, PAGE_SIZE, "%d\n", STp->first_data_ppos);
return l;
}
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
DEVICE_ATTR(BOT_frame, S_IRUGO, osst_first_data_ppos_show, NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
static ssize_t osst_eod_frame_ppos_show(struct device *dev,
struct device_attribute *attr,
char *buf)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
struct osst_tape * STp = (struct osst_tape *) dev_get_drvdata (dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ssize_t l = 0;
if (STp && STp->header_ok && STp->linux_media)
l = snprintf(buf, PAGE_SIZE, "%d\n", STp->eod_frame_ppos);
return l;
}
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
DEVICE_ATTR(EOD_frame, S_IRUGO, osst_eod_frame_ppos_show, NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
static ssize_t osst_filemark_cnt_show(struct device *dev,
struct device_attribute *attr, char *buf)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
struct osst_tape * STp = (struct osst_tape *) dev_get_drvdata (dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ssize_t l = 0;
if (STp && STp->header_ok && STp->linux_media)
l = snprintf(buf, PAGE_SIZE, "%d\n", STp->filemark_cnt);
return l;
}
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
DEVICE_ATTR(file_count, S_IRUGO, osst_filemark_cnt_show, NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[PATCH] class: convert drivers/scsi/* to use the new class api instead of class_simple Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2005-03-23 17:55:22 +00:00
static struct class *osst_sysfs_class;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
static int osst_sysfs_init(void)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
[PATCH] class: convert drivers/scsi/* to use the new class api instead of class_simple Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2005-03-23 17:55:22 +00:00
osst_sysfs_class = class_create(THIS_MODULE, "onstream_tape");
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
if (IS_ERR(osst_sysfs_class)) {
printk(KERN_ERR "osst :W: Unable to register sysfs class\n");
return PTR_ERR(osst_sysfs_class);
}
return 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
static void osst_sysfs_destroy(dev_t dev)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
device_destroy(osst_sysfs_class, dev);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
static int osst_sysfs_add(dev_t dev, struct device *device, struct osst_tape * STp, char * name)
{
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
struct device *osst_member;
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
int err;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
device create: scsi: convert device_create_drvdata to device_create Now that device_create() has been audited, rename things back to the original call to be sane. Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-07-22 03:03:34 +00:00
osst_member = device_create(osst_sysfs_class, device, dev, STp,
"%s", name);
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
if (IS_ERR(osst_member)) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_WARNING "osst :W: Unable to add sysfs class member %s\n", name);
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
return PTR_ERR(osst_member);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
err = device_create_file(osst_member, &dev_attr_ADR_rev);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
if (err)
goto err_out;
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
err = device_create_file(osst_member, &dev_attr_media_version);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
if (err)
goto err_out;
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
err = device_create_file(osst_member, &dev_attr_capacity);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
if (err)
goto err_out;
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
err = device_create_file(osst_member, &dev_attr_BOT_frame);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
if (err)
goto err_out;
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
err = device_create_file(osst_member, &dev_attr_EOD_frame);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
if (err)
goto err_out;
SCSI: convert struct class_device to struct device It's big, but there doesn't seem to be a way to split it up smaller... Signed-off-by: Tony Jones <tonyj@suse.de> Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Roland Dreier <rolandd@cisco.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-21 23:13:36 +00:00
err = device_create_file(osst_member, &dev_attr_file_count);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
if (err)
goto err_out;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
return 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
err_out:
osst_sysfs_destroy(dev);
return err;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
static void osst_sysfs_cleanup(void)
{
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
class_destroy(osst_sysfs_class);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
/*
* osst startup / cleanup code
*/
static int osst_probe(struct device *dev)
{
struct scsi_device * SDp = to_scsi_device(dev);
struct osst_tape * tpnt;
struct st_modedef * STm;
struct st_partstat * STps;
struct osst_buffer * buffer;
struct gendisk * drive;
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
int i, dev_num, err = -ENODEV;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (SDp->type != TYPE_TAPE || !osst_supports(SDp))
return -ENODEV;
drive = alloc_disk(1);
if (!drive) {
printk(KERN_ERR "osst :E: Out of memory. Device not attached.\n");
return -ENODEV;
}
/* if this is the first attach, build the infrastructure */
write_lock(&os_scsi_tapes_lock);
if (os_scsi_tapes == NULL) {
treewide: kmalloc() -> kmalloc_array() The kmalloc() function has a 2-factor argument form, kmalloc_array(). This patch replaces cases of: kmalloc(a * b, gfp) with: kmalloc_array(a * b, gfp) as well as handling cases of: kmalloc(a * b * c, gfp) with: kmalloc(array3_size(a, b, c), gfp) as it's slightly less ugly than: kmalloc_array(array_size(a, b), c, gfp) This does, however, attempt to ignore constant size factors like: kmalloc(4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The tools/ directory was manually excluded, since it has its own implementation of kmalloc(). The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ type TYPE; expression THING, E; @@ ( kmalloc( - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | kmalloc( - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression COUNT; typedef u8; typedef __u8; @@ ( kmalloc( - sizeof(u8) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(__u8) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(char) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(unsigned char) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(u8) * COUNT + COUNT , ...) | kmalloc( - sizeof(__u8) * COUNT + COUNT , ...) | kmalloc( - sizeof(char) * COUNT + COUNT , ...) | kmalloc( - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( - kmalloc + kmalloc_array ( - sizeof(TYPE) * (COUNT_ID) + COUNT_ID, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * COUNT_ID + COUNT_ID, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * (COUNT_CONST) + COUNT_CONST, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * COUNT_CONST + COUNT_CONST, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * (COUNT_ID) + COUNT_ID, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * COUNT_ID + COUNT_ID, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * (COUNT_CONST) + COUNT_CONST, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * COUNT_CONST + COUNT_CONST, sizeof(THING) , ...) ) // 2-factor product, only identifiers. @@ identifier SIZE, COUNT; @@ - kmalloc + kmalloc_array ( - SIZE * COUNT + COUNT, SIZE , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( kmalloc( - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kmalloc( - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kmalloc( - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kmalloc( - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( kmalloc( - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kmalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kmalloc( - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kmalloc( - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kmalloc( - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | kmalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ identifier STRIDE, SIZE, COUNT; @@ ( kmalloc( - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products, // when they're not all constants... @@ expression E1, E2, E3; constant C1, C2, C3; @@ ( kmalloc(C1 * C2 * C3, ...) | kmalloc( - (E1) * E2 * E3 + array3_size(E1, E2, E3) , ...) | kmalloc( - (E1) * (E2) * E3 + array3_size(E1, E2, E3) , ...) | kmalloc( - (E1) * (E2) * (E3) + array3_size(E1, E2, E3) , ...) | kmalloc( - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants, // keeping sizeof() as the second factor argument. @@ expression THING, E1, E2; type TYPE; constant C1, C2, C3; @@ ( kmalloc(sizeof(THING) * C2, ...) | kmalloc(sizeof(TYPE) * C2, ...) | kmalloc(C1 * C2 * C3, ...) | kmalloc(C1 * C2, ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * (E2) + E2, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * E2 + E2, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * (E2) + E2, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * E2 + E2, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - (E1) * E2 + E1, E2 , ...) | - kmalloc + kmalloc_array ( - (E1) * (E2) + E1, E2 , ...) | - kmalloc + kmalloc_array ( - E1 * E2 + E1, E2 , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-12 20:55:00 +00:00
os_scsi_tapes = kmalloc_array(osst_max_dev,
sizeof(struct osst_tape *),
GFP_ATOMIC);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (os_scsi_tapes == NULL) {
write_unlock(&os_scsi_tapes_lock);
printk(KERN_ERR "osst :E: Unable to allocate array for OnStream SCSI tapes.\n");
goto out_put_disk;
}
for (i=0; i < osst_max_dev; ++i) os_scsi_tapes[i] = NULL;
}
if (osst_nr_dev >= osst_max_dev) {
write_unlock(&os_scsi_tapes_lock);
printk(KERN_ERR "osst :E: Too many tape devices (max. %d).\n", osst_max_dev);
goto out_put_disk;
}
/* find a free minor number */
osst: fix read buffer overflow Check whether index is within bounds before testing the element. Signed-off-by: Roel Kluin <roel.kluin@gmail.com> Cc: Willem Riede <osst@riede.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-08-11 01:01:10 +00:00
for (i = 0; i < osst_max_dev && os_scsi_tapes[i]; i++)
;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if(i >= osst_max_dev) panic ("Scsi_devices corrupt (osst)");
dev_num = i;
/* allocate a struct osst_tape for this device */
[SCSI] kmalloc + memset conversion to kzalloc In NCR_D700, a4000t, aic7xxx_old, bvme6000, dpt_i2o, gdth, lpfc, megaraid, mvme16x osst, pluto, qla2xxx, zorro7xx Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-08-11 08:13:24 +00:00
tpnt = kzalloc(sizeof(struct osst_tape), GFP_ATOMIC);
if (!tpnt) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
write_unlock(&os_scsi_tapes_lock);
printk(KERN_ERR "osst :E: Can't allocate device descriptor, device not attached.\n");
goto out_put_disk;
}
/* allocate a buffer for this device */
i = SDp->host->sg_tablesize;
if (osst_max_sg_segs < i)
i = osst_max_sg_segs;
buffer = new_tape_buffer(1, SDp->host->unchecked_isa_dma, i);
if (buffer == NULL) {
write_unlock(&os_scsi_tapes_lock);
printk(KERN_ERR "osst :E: Unable to allocate a tape buffer, device not attached.\n");
kfree(tpnt);
goto out_put_disk;
}
os_scsi_tapes[dev_num] = tpnt;
tpnt->buffer = buffer;
tpnt->device = SDp;
drive->private_data = &tpnt->driver;
sprintf(drive->disk_name, "osst%d", dev_num);
tpnt->driver = &osst_template;
tpnt->drive = drive;
tpnt->in_use = 0;
tpnt->capacity = 0xfffff;
tpnt->dirty = 0;
tpnt->drv_buffer = 1; /* Try buffering if no mode sense */
tpnt->restr_dma = (SDp->host)->unchecked_isa_dma;
tpnt->density = 0;
tpnt->do_auto_lock = OSST_AUTO_LOCK;
tpnt->can_bsr = OSST_IN_FILE_POS;
tpnt->can_partitions = 0;
tpnt->two_fm = OSST_TWO_FM;
tpnt->fast_mteom = OSST_FAST_MTEOM;
tpnt->scsi2_logical = OSST_SCSI2LOGICAL; /* FIXME */
tpnt->write_threshold = osst_write_threshold;
tpnt->default_drvbuffer = 0xff; /* No forced buffering */
tpnt->partition = 0;
tpnt->new_partition = 0;
tpnt->nbr_partitions = 0;
tpnt->min_block = 512;
tpnt->max_block = OS_DATA_SIZE;
tpnt->timeout = OSST_TIMEOUT;
tpnt->long_timeout = OSST_LONG_TIMEOUT;
/* Recognize OnStream tapes */
/* We don't need to test for OnStream, as this has been done in detect () */
tpnt->os_fw_rev = osst_parse_firmware_rev (SDp->rev);
tpnt->omit_blklims = 1;
tpnt->poll = (strncmp(SDp->model, "DI-", 3) == 0) ||
(strncmp(SDp->model, "FW-", 3) == 0) || OSST_FW_NEED_POLL(tpnt->os_fw_rev,SDp);
tpnt->frame_in_buffer = 0;
tpnt->header_ok = 0;
tpnt->linux_media = 0;
tpnt->header_cache = NULL;
for (i=0; i < ST_NBR_MODES; i++) {
STm = &(tpnt->modes[i]);
STm->defined = 0;
STm->sysv = OSST_SYSV;
STm->defaults_for_writes = 0;
STm->do_async_writes = OSST_ASYNC_WRITES;
STm->do_buffer_writes = OSST_BUFFER_WRITES;
STm->do_read_ahead = OSST_READ_AHEAD;
STm->default_compression = ST_DONT_TOUCH;
STm->default_blksize = 512;
STm->default_density = (-1); /* No forced density */
}
for (i=0; i < ST_NBR_PARTITIONS; i++) {
STps = &(tpnt->ps[i]);
STps->rw = ST_IDLE;
STps->eof = ST_NOEOF;
STps->at_sm = 0;
STps->last_block_valid = 0;
STps->drv_block = (-1);
STps->drv_file = (-1);
}
tpnt->current_mode = 0;
tpnt->modes[0].defined = 1;
tpnt->modes[2].defined = 1;
tpnt->density_changed = tpnt->compression_changed = tpnt->blksize_changed = 0;
[SCSI] osst: Use mutex instead of semaphore The OnStream SCSI Tape driver uses a semaphore as mutex. Use the mutex API instead of the (binary) semaphore. Signed-off-by: Matthias Kaehlcke <matthias.kaehlcke@gmail.com> Reviewed-by: Satyam Sharma <satyam@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Willem Riede <wrlk@riede.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2007-08-10 21:50:44 +00:00
mutex_init(&tpnt->lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
osst_nr_dev++;
write_unlock(&os_scsi_tapes_lock);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
char name[8];
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/* Rewind entry */
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
err = osst_sysfs_add(MKDEV(OSST_MAJOR, dev_num), dev, tpnt, tape_name(tpnt));
if (err)
goto out_free_buffer;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/* No-rewind entry */
snprintf(name, 8, "%s%s", "n", tape_name(tpnt));
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
err = osst_sysfs_add(MKDEV(OSST_MAJOR, dev_num + 128), dev, tpnt, name);
if (err)
goto out_free_sysfs1;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[SCSI] fix up mismerge in osst Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-29 14:42:17 +00:00
sdev_printk(KERN_INFO, SDp,
[SCSI] move the mid-layer printk's over to shost/starget/sdev_printk This should eliminate (at least in the mid layer) to make numeric assumptions about any of the enumeration variables. As a side effect, it will also make all the messages consistent and line us up nicely for the error logging strategy (if it ever shows itself again). Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-02 16:45:08 +00:00
"osst :I: Attached OnStream %.5s tape as %s\n",
SDp->model, tape_name(tpnt));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return 0;
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
out_free_sysfs1:
osst_sysfs_destroy(MKDEV(OSST_MAJOR, dev_num));
out_free_buffer:
kfree(buffer);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
out_put_disk:
put_disk(drive);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
return err;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
};
static int osst_remove(struct device *dev)
{
struct scsi_device * SDp = to_scsi_device(dev);
struct osst_tape * tpnt;
[SCSI] Remove devfs support from the SCSI subsystem As devfs has been disabled from the kernel tree for a number of months now (5 to be exact), here's a patch against 2.6.16-rc1-git1 that removes support for it from the SCSI subsystem. The patch also removes the scsi_disk devfs_name field as it's no longer needed. Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-01-19 00:17:46 +00:00
int i;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if ((SDp->type != TYPE_TAPE) || (osst_nr_dev <= 0))
return 0;
write_lock(&os_scsi_tapes_lock);
for(i=0; i < osst_max_dev; i++) {
if((tpnt = os_scsi_tapes[i]) && (tpnt->device == SDp)) {
osst_sysfs_destroy(MKDEV(OSST_MAJOR, i));
osst_sysfs_destroy(MKDEV(OSST_MAJOR, i+128));
tpnt->device = NULL;
put_disk(tpnt->drive);
os_scsi_tapes[i] = NULL;
osst_nr_dev--;
write_unlock(&os_scsi_tapes_lock);
[PATCH] vfree and kfree cleanup in drivers/ This patch does a full cleanup of 'NULL checks before vfree', and a partial cleanup of calls to kfree for all of drivers/ - the kfree bit is partial in that I only did the files that also had vfree calls in them. The patch also gets rid of some redundant (void *) casts of pointers being passed to [vk]free, and a some tiny whitespace corrections also crept in. Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 07:26:54 +00:00
vfree(tpnt->header_cache);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (tpnt->buffer) {
normalize_buffer(tpnt->buffer);
kfree(tpnt->buffer);
}
kfree(tpnt);
return 0;
}
}
write_unlock(&os_scsi_tapes_lock);
return 0;
}
static int __init init_osst(void)
{
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
int err;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_INFO "osst :I: Tape driver with OnStream support version %s\nosst :I: %s\n", osst_version, cvsid);
validate_options();
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
err = osst_sysfs_init();
if (err)
return err;
err = register_chrdev(OSST_MAJOR, "osst", &osst_fops);
if (err < 0) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
printk(KERN_ERR "osst :E: Unable to register major %d for OnStream tapes\n", OSST_MAJOR);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
goto err_out;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
err = scsi_register_driver(&osst_template.gendrv);
if (err)
goto err_out_chrdev;
Replace remaining references to "driverfs" with "sysfs". Globally, s/driverfs/sysfs/g. Signed-off-by: Robert P. J. Day <rpjday@mindspring.com> Signed-off-by: Adrian Bunk <bunk@stusta.de>
2007-02-17 18:13:42 +00:00
err = osst_create_sysfs_files(&osst_template.gendrv);
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
if (err)
goto err_out_scsidrv;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return 0;
[SCSI] SCSI osst: add error handling to module init, sysfs - check all sysfs-related return codes, and propagate them back to callers - properly unwind errors in osst_probe(), init_osst(). This fixes a leak that occured if scsi driver registration failed, and fixes an oops if sysfs creation returned an error. (unrelated) - kzalloc() cleanup in new_tape_buf() Signed-off-by: Jeff Garzik <jeff@garzik.org> Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-10-04 09:23:04 +00:00
err_out_scsidrv:
scsi_unregister_driver(&osst_template.gendrv);
err_out_chrdev:
unregister_chrdev(OSST_MAJOR, "osst");
err_out:
osst_sysfs_cleanup();
return err;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
static void __exit exit_osst (void)
{
int i;
struct osst_tape * STp;
Replace remaining references to "driverfs" with "sysfs". Globally, s/driverfs/sysfs/g. Signed-off-by: Robert P. J. Day <rpjday@mindspring.com> Signed-off-by: Adrian Bunk <bunk@stusta.de>
2007-02-17 18:13:42 +00:00
osst_remove_sysfs_files(&osst_template.gendrv);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
scsi_unregister_driver(&osst_template.gendrv);
unregister_chrdev(OSST_MAJOR, "osst");
osst_sysfs_cleanup();
if (os_scsi_tapes) {
for (i=0; i < osst_max_dev; ++i) {
if (!(STp = os_scsi_tapes[i])) continue;
/* This is defensive, supposed to happen during detach */
[PATCH] vfree and kfree cleanup in drivers/ This patch does a full cleanup of 'NULL checks before vfree', and a partial cleanup of calls to kfree for all of drivers/ - the kfree bit is partial in that I only did the files that also had vfree calls in them. The patch also gets rid of some redundant (void *) casts of pointers being passed to [vk]free, and a some tiny whitespace corrections also crept in. Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 07:26:54 +00:00
vfree(STp->header_cache);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (STp->buffer) {
normalize_buffer(STp->buffer);
kfree(STp->buffer);
}
put_disk(STp->drive);
kfree(STp);
}
kfree(os_scsi_tapes);
}
printk(KERN_INFO "osst :I: Unloaded.\n");
}
module_init(init_osst);
module_exit(exit_osst);
Reference in a new issue Copy permalink