mirror of
https://github.com/SCST-project/scst.git
synced 2026-05-20 20:21:30 +00:00
svn+ssh://yanb123@svn.code.sf.net/p/scst/svn/trunk
........
r5456 | bvassche | 2014-04-23 08:15:57 +0300 (Wed, 23 Apr 2014) | 1 line
scst/README: Update implicit ALUA section
........
r5457 | bvassche | 2014-04-23 11:57:17 +0300 (Wed, 23 Apr 2014) | 1 line
ib_srpt: RHEL 5.9 build fix
........
r5458 | bvassche | 2014-04-23 11:59:18 +0300 (Wed, 23 Apr 2014) | 1 line
ib_srpt: Clean up the CM event handling code
........
r5459 | bvassche | 2014-04-23 12:02:29 +0300 (Wed, 23 Apr 2014) | 1 line
ib_srpt: Clean up the CM event handling code (part 2)
........
r5460 | bvassche | 2014-04-23 12:04:15 +0300 (Wed, 23 Apr 2014) | 1 line
ib_srpt, RDMA/CM: Avoid hanging sessions due to a cable pull
........
r5461 | bvassche | 2014-04-23 12:20:52 +0300 (Wed, 23 Apr 2014) | 1 line
ib_srpt: RHEL 5.9 build fix (part 2)
........
r5462 | bvassche | 2014-04-23 16:38:54 +0300 (Wed, 23 Apr 2014) | 1 line
ib_srpt: Clean up the CM event handling messages (part 3)
........
r5463 | vlnb | 2014-04-24 03:07:22 +0300 (Thu, 24 Apr 2014) | 3 lines
Cleanups
........
r5464 | vlnb | 2014-04-24 05:30:52 +0300 (Thu, 24 Apr 2014) | 35 lines
scst_lib: Avoid integer overflows
This patch fixes the following kernel oops:
[3696]: scst: scst_parse_cmd:826:Warning: expected transfer length 522240 for opcode 0x08 (handler vcdrom, target iscsi) doesn't match decoded value -2048
[3696]: scst_parse_cmd:828:Suspicious CDB:
(h)___0__1__2__3__4__5__6__7__8__9__A__B__C__D__E__F
0: 08 1f ff ff ff 00 ......
BUG: unable to handle kernel paging request at ffff88283597f0c8
IP: [<ffffffff8135c3af>] sg_init_table+0x5f/0x90
Call Trace:
[<ffffffffa021eb48>] sgv_pool_alloc+0x3b8/0xbf0 [scst]
[<ffffffffa0204c16>] scst_alloc_space+0xb6/0x290 [scst]
[<ffffffffa01e2d28>] scst_prepare_space+0x3b8/0x6e0 [scst]
[<ffffffffa01ef4a5>] scst_process_active_cmd+0x455/0x7e0 [scst]
[<ffffffffa01efb22>] scst_cmd_init_done+0x2f2/0x5c0 [scst]
[<ffffffffa0156ac2>] scst_cmd_init_stage1_done.constprop.37+0x12/0x20 [iscsi_scst]
[<ffffffffa015d9fa>] scsi_cmnd_start+0x25a/0x550 [iscsi_scst]
[<ffffffffa015e4b8>] cmnd_rx_start+0x148/0x1a0 [iscsi_scst]
[<ffffffffa0161598>] process_read_io+0x3b8/0x800 [iscsi_scst]
[<ffffffffa0161aa7>] scst_do_job_rd+0xc7/0x220 [iscsi_scst]
[<ffffffffa016208d>] istrd+0x16d/0x2e0 [iscsi_scst]
[<ffffffff81075bad>] kthread+0xed/0x110
[<ffffffff816dd5bc>] ret_from_fork+0x7c/0xb0
and causes the following message to be reported instead:
[11269]: scst: scst_generic_parse:7402:***WARNING***: bufflen 16777215, data_len 16777215 or out_bufflen 0 too large for device disk12 (block size 2048)
scst_generic_parse:CDB:
(h)___0__1__2__3__4__5__6__7__8__9__A__B__C__D__E__F
0: 08 1f ff ff ff 00 ......
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
........
r5465 | vlnb | 2014-04-24 05:33:07 +0300 (Thu, 24 Apr 2014) | 5 lines
scst_targ: Clarify a comment
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
........
r5466 | vlnb | 2014-04-24 05:39:05 +0300 (Thu, 24 Apr 2014) | 12 lines
scst: Avoid that smatch complains about dead code
The panic() function that implements BUG() has been declared with attribute
noreturn in some RHEL 5 kernel headers and also in the RHEL 6 kernel headers.
Smatch warns about dead code if any code follows a function that has been
declared with attribute noreturn. Hence add a few preprocessor statements to
suppress the smatch warning when building against an upstream kernel or when
building against RHEL 6 or later.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
........
r5467 | vlnb | 2014-04-24 06:15:12 +0300 (Thu, 24 Apr 2014) | 3 lines
Fix queuing of UA for aborted by PREEMPT AND ABORT, if TAS is 0, + some clarifications
........
r5468 | vlnb | 2014-04-24 07:03:51 +0300 (Thu, 24 Apr 2014) | 6 lines
Fix for TARGET RESET race
It can happen, when a device added after blocking, so unblocking then
will make dev->block_count of the new device negative.
........
r5469 | bvassche | 2014-04-24 11:41:15 +0300 (Thu, 24 Apr 2014) | 1 line
scst/README: Update multipathd information in implicit ALUA section
........
r5470 | bvassche | 2014-04-24 13:37:58 +0300 (Thu, 24 Apr 2014) | 1 line
ib_srpt: OFED 3.12 build fix
........
r5471 | bvassche | 2014-04-24 13:38:37 +0300 (Thu, 24 Apr 2014) | 1 line
ib_srpt: Add support in the Makefile for MLNX OFED and for OFED 3.x
........
r5472 | bvassche | 2014-04-24 14:11:26 +0300 (Thu, 24 Apr 2014) | 1 line
ib_srpt: Make srpt_disconnect_ch() close sessions properly that have not yet reached the connected state
........
r5473 | bvassche | 2014-04-24 15:27:00 +0300 (Thu, 24 Apr 2014) | 1 line
ib_srpt, Makefile: Introduce the OFED_KERNEL_DIR variable
........
r5474 | bvassche | 2014-04-24 21:43:41 +0300 (Thu, 24 Apr 2014) | 1 line
ib_srpt: Unload properly with RDMA/CM disabled
........
r5475 | bvassche | 2014-04-24 21:45:23 +0300 (Thu, 24 Apr 2014) | 1 line
scstadmin, regression tests: Follow-up for r5409
........
r5476 | bvassche | 2014-04-24 21:49:52 +0300 (Thu, 24 Apr 2014) | 16 lines
scst: Avoid that reassigning a session triggers a kernel crash
This patch fixes the following kernel bug:
BUG: unable to handle kernel NULL pointer dereference at 0000000000000064
IP: [<ffffffffa069e59c>] scst_alloc_add_tgt_dev+0x9c/0x540 [scst]
Call Trace:
[<ffffffffa06a1b57>] scst_check_reassign_sessions+0x367/0x3b0 [scst]
[<ffffffffa06a1cb7>] scst_acg_add_acn+0x117/0x1a0 [scst]
[<ffffffffa06a5e42>] scst_acg_ini_mgmt_store_work_fn+0x152/0x370 [scst]
[<ffffffffa06a93a6>] sysfs_work_thread_fn+0xa6/0x2f0 [scst]
[<ffffffff8108dd72>] kthread+0xd2/0xf0
[<ffffffff816912bc>] ret_from_fork+0x7c/0xb0
Reported-by: Zhen Xu <zhenxu.zj@gmail.com>
........
r5477 | vlnb | 2014-04-25 02:07:58 +0300 (Fri, 25 Apr 2014) | 3 lines
Minor logging cleanup
........
r5478 | vlnb | 2014-04-25 02:44:25 +0300 (Fri, 25 Apr 2014) | 3 lines
Cleanup
........
r5479 | vlnb | 2014-04-25 05:03:04 +0300 (Fri, 25 Apr 2014) | 3 lines
Saved mode pages added
........
r5480 | vlnb | 2014-04-25 05:47:31 +0300 (Fri, 25 Apr 2014) | 3 lines
Follow up for the previous commit
........
r5481 | vlnb | 2014-04-26 04:56:36 +0300 (Sat, 26 Apr 2014) | 3 lines
Processing of QErr and TMF_ONLY added
........
r5482 | vlnb | 2014-04-26 05:12:17 +0300 (Sat, 26 Apr 2014) | 3 lines
Cleanups
........
r5483 | bvassche | 2014-04-26 09:32:32 +0300 (Sat, 26 Apr 2014) | 7 lines
scst: Fix recently introduced checkpatch complaints about whitespace
Fix two instances of the following checkpatch errors:
ERROR: code indent should use tabs where possible
ERROR: spaces required around that ':' (ctx:VxW)
........
r5484 | bvassche | 2014-04-27 09:26:29 +0300 (Sun, 27 Apr 2014) | 6 lines
scst_const.h: Fix a checkpatch complaint about whitespace
Avoid that checkpatch reports the following warning message:
WARNING: please, no space before tabs
........
r5485 | bvassche | 2014-04-28 12:29:25 +0300 (Mon, 28 Apr 2014) | 5 lines
scst: Export scst_path_put()
This patch makes the code that was added via r5479 build against kernel
version 2.6.38 and before.
........
r5487 | vlnb | 2014-04-28 23:31:14 +0300 (Mon, 28 Apr 2014) | 3 lines
Cleanups
........
r5488 | vlnb | 2014-04-29 00:38:22 +0300 (Tue, 29 Apr 2014) | 3 lines
Returned sense cleanups
........
r5489 | vlnb | 2014-04-29 01:30:03 +0300 (Tue, 29 Apr 2014) | 3 lines
Let REPORT SUPPORTED OPERATION CODES be handled by dev handler as well
........
r5490 | vlnb | 2014-04-29 03:13:24 +0300 (Tue, 29 Apr 2014) | 6 lines
It's wrong to clean reservation on failed RESERVE commands
With multiple outstanding commands it can open a race window leading to
loose of a valid reservation
........
r5491 | vlnb | 2014-04-29 04:32:13 +0300 (Tue, 29 Apr 2014) | 3 lines
SCSI logging improvements
........
r5492 | vlnb | 2014-04-29 04:45:43 +0300 (Tue, 29 Apr 2014) | 3 lines
Missed hunk in the previous commit
........
r5493 | vlnb | 2014-04-29 05:09:29 +0300 (Tue, 29 Apr 2014) | 3 lines
Minor logging improvement
........
r5494 | bvassche | 2014-04-29 15:35:41 +0300 (Tue, 29 Apr 2014) | 1 line
ib_srpt, README: Fix RDMA/CM login instructions
........
r5495 | bvassche | 2014-04-29 15:39:50 +0300 (Tue, 29 Apr 2014) | 4 lines
ib_srpt: Make LUN masking work again
This patch fixes a regression that was introduced in r5425.
........
r5496 | vlnb | 2014-04-30 04:04:27 +0300 (Wed, 30 Apr 2014) | 3 lines
Extended INQUIRY page added
........
r5497 | vlnb | 2014-04-30 04:07:41 +0300 (Wed, 30 Apr 2014) | 3 lines
Cleanup
........
r5498 | bvassche | 2014-04-30 08:30:45 +0300 (Wed, 30 Apr 2014) | 1 line
ib_srpt: Fix two recently introduced checkpatch complaints about whitespace
........
r5499 | bvassche | 2014-04-30 09:01:17 +0300 (Wed, 30 Apr 2014) | 1 line
nightly build: Update kernel versions
........
r5500 | vlnb | 2014-05-02 05:50:12 +0300 (Fri, 02 May 2014) | 3 lines
Fix CDROM empty case
........
r5501 | vlnb | 2014-05-02 05:50:34 +0300 (Fri, 02 May 2014) | 3 lines
Cleanups
........
r5502 | vlnb | 2014-05-02 05:56:19 +0300 (Fri, 02 May 2014) | 9 lines
scst_sysfs: Save the value of the 'preferred' attribute only if it has been set
The default value of the 'preferred' attribute is 0 (disabled). Hence
it is only necessary that scstadmin saves the value of that attribute
if it is not zero.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
........
r5503 | vlnb | 2014-05-07 02:22:32 +0300 (Wed, 07 May 2014) | 3 lines
Cleanups and logging improvements
........
r5504 | vlnb | 2014-05-07 05:13:11 +0300 (Wed, 07 May 2014) | 5 lines
Fix COMMAND DATA LENGTH in All_commands parameter data of REPORT SUPPORTED OPERATION CODES
Reported by Sebastian Herbszt <herbszt@gmx.de>
........
r5505 | bvassche | 2014-05-07 11:38:47 +0300 (Wed, 07 May 2014) | 5 lines
scst: Fix the procfs build
Move the definition of scst_get_opcode_name() up such that it occurs
outside #ifndef CONFIG_SCST_PROC / #endif. See also r5491.
........
r5506 | bvassche | 2014-05-07 11:57:04 +0300 (Wed, 07 May 2014) | 4 lines
scst_vdisk: Build fix for kernel versions < 2.6.37
See also r5420 / r5479.
........
r5507 | bvassche | 2014-05-07 16:42:56 +0300 (Wed, 07 May 2014) | 1 line
nightly build: Update kernel versions
........
r5508 | vlnb | 2014-05-08 05:28:49 +0300 (Thu, 08 May 2014) | 7 lines
Avoid that the code for dumping the PR state triggers a race condition
Callers of scst_pr_dump_prs() must hold dev_pr_mutex.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
........
git-svn-id: http://svn.code.sf.net/p/scst/svn/branches/iser@5509 d57e44dd-8a1f-0410-8b47-8ef2f437770f