1. 25 Oct, 2013 20 commits
  2. 23 Oct, 2013 2 commits
    • Aaron Lu's avatar
      [SCSI] sd: call blk_pm_runtime_init before add_disk · 10c580e4
      Aaron Lu authored
      Sujit has found a race condition that would make q->nr_pending
      unbalanced, it occurs as Sujit explained:
      
      "
      sd_probe_async() ->
      	add_disk() ->
      		disk_add_event() ->
      			schedule(disk_events_workfn)
      	sd_revalidate_disk()
      	blk_pm_runtime_init()
      return;
      
      Let's say the disk_events_workfn() calls sd_check_events() which tries
      to send test_unit_ready() and because of sd_revalidate_disk() trying to
      send another commands the test_unit_ready() might be re-queued as the
      tagged command queuing is disabled.
      
      So the race condition is -
      
      Thread 1 			  |		Thread 2
      sd_revalidate_disk()		  |	sd_check_events()
      ...nr_pending = 0 as q->dev = NULL|	scsi_queue_insert()
      blk_runtime_pm_init()		  | 	blk_pm_requeue_request() ->
      				  |	nr_pending = -1 since
      				  |	q->dev != NULL
      "
      
      The problem is, the test_unit_ready request doesn't get counted the
      first time it is queued, so the later decrement of q->nr_pending in
      blk_pm_requeue_request makes it unbalanced.
      
      Fix this by calling blk_pm_runtime_init before add_disk so that all
      requests initiated there will all be counted.
      Signed-off-by: default avatarAaron Lu <aaron.lu@intel.com>
      Reported-and-tested-by: default avatarSujit Reddy Thumma <sthumma@codeaurora.org>
      Cc: stable@vger.kernel.org
      Signed-off-by: default avatarJames Bottomley <JBottomley@Parallels.com>
      10c580e4
    • Chad Dupuis's avatar
      [SCSI] qla2xxx: Fix request queue null dereference. · 36008cf1
      Chad Dupuis authored
      If an invalid IOCB is returned on the response queue then the index into the
      request queue map could be invalid and could return to us a bogus value. This
      could cause us to try to deference an invalid pointer and cause an exception.
      
      If we encounter this condition, simply return as no context can be established
      for this response.
      Signed-off-by: default avatarChad Dupuis <chad.dupuis@qlogic.com>
      Signed-off-by: default avatarSaurav Kashyap <saurav.kashyap@qlogic.com>
      Signed-off-by: default avatarJames Bottomley <JBottomley@Parallels.com>
      36008cf1
  3. 16 Oct, 2013 1 commit
    • Khalid Aziz's avatar
      [SCSI] BusLogic: Fix an oops when intializing multimaster adapter · 6541932e
      Khalid Aziz authored
      This fixes an oops caused by buslogic driver when initializing a BusLogic
      MultiMaster adapter. Initialization code used scope of a variable
      incorrectly which created a NULL pointer. Oops message is below:
      
      BUG: unable to handle kernel NULL pointer dereference at 0000000c
      IP: [<c150c137>] blogic_init_mm_probeinfo.isra.17+0x20a/0x583
      *pde = 00000000
      Oops: 002 [#1] PREEMPT SMP
      Modules linked in:
      CPU: 1 PID: 1 Comm: swapper/0 Not tainted 3.11.1.puz1 #1
      Hardware name:    /Canterwood, BIOS 6.00 PG 05/16/2003
      task: f7050000 ti: f7054000 task.ti: f7054000
      EIP: 0060:[<c150c137>] EFLAGS: 00010246 CPU:1
      EIP is at blogic_init_mm_probeinfo.isra.17+0x20a/0x583
      EAX: 00000013 EBX: 00000000 ECX: 00000000 EDX: f8001000
      ESI: f71cb800 EDI: f7388000 EBP: 00007800 ESP: f7055c84
       DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
      CR0: 8005003b CR2: 0000000c CR3: 0154f000 CR4: 000007d0
      Stack:
       0000001c 00000000 c11a59f6 f7055c98 00008130 ffffffff ffffffff 00000000
       00000003 00000000 00000000 00000000 00000013 f8001000 00000001 000003d0
       00000000 00000000 00000000 c14e3f84 f78803c8 00000000 f738c000 000000e9
      Call Trace:
       [<c11a59f6>] ? pci_get_subsys+0x33/0x38
       [<c150c4fb>] ? blogic_init_probeinfo_list+0x4b/0x19e
       [<c108d593>] ? __alloc_pages_nodemask+0xe3/0x623
       [<c108d593>] ? __alloc_pages_nodemask+0xe3/0x623
       [<c10fb99e>] ? sysfs_link_sibling+0x61/0x8d
       [<c10b0519>] ? kmem_cache_alloc+0x8b/0xb5
       [<c150cce5>] ? blogic_init+0xa1/0x10e8
       [<c10fc0a8>] ? sysfs_add_one+0x10/0x9d
       [<c10fc18a>] ? sysfs_addrm_finish+0x12/0x85
       [<c10fca37>] ? sysfs_do_create_link_sd+0x9d/0x1b4
       [<c117c272>] ? blk_register_queue+0x69/0xb3
       [<c10fcb68>] ? sysfs_create_link+0x1a/0x2c
       [<c1181a07>] ? add_disk+0x1a1/0x3c7
       [<c138737b>] ? klist_next+0x60/0xc3
       [<c122cc3a>] ? scsi_dh_detach+0x68/0x68
       [<c1213e36>] ? bus_for_each_dev+0x51/0x61
       [<c1000356>] ? do_one_initcall+0x22/0x12c
       [<c10f3688>] ? __proc_create+0x8c/0xba
       [<c150cc44>] ? blogic_setup+0x5f6/0x5f6
       [<c14e94aa>] ? repair_env_string+0xf/0x4d
       [<c14e949b>] ? do_early_param+0x71/0x71
       [<c103efaa>] ? parse_args+0x21f/0x33d
       [<c14e9a54>] ? kernel_init_freeable+0xdf/0x17d
       [<c14e949b>] ? do_early_param+0x71/0x71
       [<c1388b64>] ? kernel_init+0x8/0xc0
       [<c1392222>] ? ret_from_kernel_thread+0x6/0x28
       [<c1392227>] ? ret_from_kernel_thread+0x1b/0x28
       [<c1388b5c>] ? rest_init+0x6c/0x6c
      Code: 89 44 24 10 0f b6 44 24 3d 89 44 24 0c c7 44 24 08 00 00 00 00 c7 44 24 04 38 62 46 c1 c7 04 24 02 00 00 00 e8 78 13 d2 ff 31 db <89> 6b 0c b0 20 89 ea ee
       c7 44 24 08 04 00 00 00 8d 44 24 4c 89
      EIP: [<c150c137>] blogic_init_mm_probeinfo.isra.17+0x20a/0x583 SS:ESP 0068:f7055c84
      CR2: 000000000000000c
      ---[ end trace 17f45f5196d40487 ]---
      Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000009
      Signed-off-by: default avatarKhalid Aziz <khalid.aziz@oracle.com>
      Cc: <stable@vger.kernel.org> # 3.11.x
      Reported-by: default avatarPierre Uszynski <pierre@rahul.net>
      Tested-by: default avatarPierre Uszynski <pierre@rahul.net>
      Signed-off-by: default avatarJames Bottomley <JBottomley@Parallels.com>
      6541932e
  4. 13 Oct, 2013 17 commits