Skip to contentSkip to conversation

cfq bug still not fixed on x86

DISCUSSION walt 2 participants 3 messages Jan 31, 2012 – Feb 1, 2012

lore
Share & tools
lore

walt

author
#0 lore raw

Hi guys. Linus recently commited this bugfix from Jens:

commit 54b466e44b1c7809144bbd8cd6be3f85877ca46f

Author: Jens Axboe
Date:   Tue Jan 17 21:26:11 2012 +0100

cfq-iosched: fix use-after-free of cfqq

That fixed the problem on my amd64x2 machine, but my old UP x86 amd machine is still producing the same backtrace when udevd segfaults during bootup.

I'm wondering if the patch neglects some 32-bit or non-SMP code?

Thanks for all your hard work on open-source :)

#1 lore raw

can you try the debug patch from Tejun:

http://marc.info/?l=linux-kernel&m=132683711206544&w=2

2012/2/1 walt <[email protected]>:

Quoted 21 lines

Hi guys.  Linus recently commited this bugfix from Jens:

commit 54b466e44b1c7809144bbd8cd6be3f85877ca46f
Author: Jens Axboe
Date:   Tue Jan 17 21:26:11 2012 +0100

   cfq-iosched: fix use-after-free of cfqq

That fixed the problem on my amd64x2 machine, but my old UP x86 amd machine is still producing the same backtrace when udevd segfaults during bootup.

I'm wondering if the patch neglects some 32-bit or non-SMP code?

Thanks for all your hard work on open-source :)

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

author
#2 lore raw
Quoted 3 lines · Shaohua Li

can you try the debug patch from Tejun:

http://marc.info/?l=linux-kernel&m=132683711206544&w=2

udevd still segfaults during bootup, but the backtrace looks quite different to me. I've pasted it at the bottom:

Quoted 12 lines

2012/2/1 walt <[email protected]>:

Quoted 11 lines
Quoted 11 lines

Hi guys. Linus recently commited this bugfix from Jens:

commit 54b466e44b1c7809144bbd8cd6be3f85877ca46f

Author: Jens Axboe
Date:   Tue Jan 17 21:26:11 2012 +0100

cfq-iosched: fix use-after-free of cfqq

That fixed the problem on my amd64x2 machine, but my old UP x86 amd machine is still producing the same backtrace when udevd segfaults during bootup.

Feb  1 07:17:52 k2 kernel: Pid: 951, comm: scsi_id Tainted: G        W    3.3.0-rc2-dirty #128
Feb  1 07:17:52 k2 kernel: Call Trace:
Feb  1 07:17:52 k2 kernel:  [<c10326c5>] __schedule_bug+0x50/0x55
Feb  1 07:17:52 k2 kernel:  [<c12aecb3>] __schedule+0x57/0x39a
Feb  1 07:17:52 k2 kernel:  [<c11bd5cf>] ? scsi_init_sgtable+0x56/0x6f
Feb  1 07:17:52 k2 kernel:  [<c11bd781>] ? scsi_init_io+0x17/0x80
Feb  1 07:17:52 k2 kernel:  [<c11bd8f4>] ? scsi_setup_blk_pc_cmnd+0x74/0xe4
Feb  1 07:17:52 k2 kernel:  [<c11ca734>] ? sd_prep_fn+0x73c/0x753
Feb  1 07:17:52 k2 kernel:  [<c11d2100>] ? ata_scsiop_mode_sense+0x25a/0x25a
Feb  1 07:17:52 k2 kernel:  [<c12af374>] schedule+0x47/0x49
Feb  1 07:17:52 k2 kernel:  [<c12ae112>] schedule_timeout+0x17/0x13d
Feb  1 07:17:52 k2 kernel:  [<c1148b73>] ? blk_peek_request+0x81/0x151
Feb  1 07:17:52 k2 kernel:  [<c11adeb3>] ? put_device+0xf/0x11
Feb  1 07:17:52 k2 kernel:  [<c12af23d>] wait_for_common+0xb4/0x121
Feb  1 07:17:52 k2 kernel:  [<c1032d93>] ? try_to_wake_up+0xe7/0xe7
Feb  1 07:17:52 k2 kernel:  [<c12af32b>] wait_for_completion+0x12/0x14
Feb  1 07:17:52 k2 kernel:  [<c114b7f4>] blk_execute_rq+0x73/0x8b
Feb  1 07:17:52 k2 kernel:  [<c114b694>] ? blk_rq_map_user+0x1b1/0x1b1
Feb  1 07:17:52 k2 kernel:  [<c114b96f>] ? blk_recount_segments+0x16/0x24
Feb  1 07:17:52 k2 kernel:  [<c10970cf>] ? bio_phys_segments+0x11/0x17
Feb  1 07:17:52 k2 kernel:  [<c1148ceb>] ? blk_rq_bio_prep+0x23/0x70
Feb  1 07:17:52 k2 kernel:  [<c114b3b4>] ? blk_rq_append_bio+0x14/0x3b
Feb  1 07:17:52 k2 kernel:  [<c114b615>] ? blk_rq_map_user+0x132/0x1b1
Feb  1 07:17:52 k2 kernel:  [<c114eb12>] sg_io+0x24c/0x339
Feb  1 07:17:52 k2 kernel:  [<c114f098>] scsi_cmd_ioctl+0x1a0/0x351
Feb  1 07:17:52 k2 kernel:  [<c1098d5b>] ? __blkdev_get+0x246/0x2da
Feb  1 07:17:52 k2 kernel:  [<c1099093>] ? blkdev_get+0x2a4/0x2a4
Feb  1 07:17:52 k2 kernel:  [<c1098fae>] ? blkdev_get+0x1bf/0x2a4
Feb  1 07:17:52 k2 kernel:  [<c10854b4>] ? _d_rehash+0x39/0x3c
Feb  1 07:17:52 k2 kernel:  [<c114f274>] scsi_cmd_blk_ioctl+0x2b/0x37
Feb  1 07:17:52 k2 kernel:  [<c11c881c>] sd_ioctl+0x78/0x98
Feb  1 07:17:52 k2 kernel:  [<c11c87a4>] ? sd_check_events+0xc0/0xc0
Feb  1 07:17:52 k2 kernel:  [<c114c803>] __blkdev_driver_ioctl+0x1f/0x2b
Feb  1 07:17:52 k2 kernel:  [<c114d191>] blkdev_ioctl+0x6af/0x6ce
Feb  1 07:17:52 k2 kernel:  [<c106809b>] ? handle_pte_fault+0x200/0x4a8
Feb  1 07:17:52 k2 kernel:  [<c1097d77>] block_ioctl+0x2c/0x34
Feb  1 07:17:52 k2 kernel:  [<c1097d77>] ? block_ioctl+0x2c/0x34
Feb  1 07:17:52 k2 kernel:  [<c1097d4b>] ? blkdev_get_block+0x35/0x35
Feb  1 07:17:52 k2 kernel:  [<c1082e2b>] do_vfs_ioctl+0x459/0x48f
Feb  1 07:17:52 k2 kernel:  [<c1016215>] ? do_page_fault+0x304/0x333
Feb  1 07:17:52 k2 kernel:  [<c1076180>] ? do_sys_open+0xac/0x170
Feb  1 07:17:52 k2 kernel:  [<c1082e8f>] sys_ioctl+0x2e/0x4b
Feb  1 07:17:52 k2 kernel:  [<c12b004c>] sysenter_do_call+0x12/0x22
Feb  1 07:17:52 k2 kernel: scsi_id[951]: segfault at b758b060 ip b758b060 sp bf8220f8 error 4
Feb  1 07:17:52 k2 kernel: note: scsi_id[951] exited with preempt_count 1