[Bug 103251] New: [CI] igt at gem_userptr_blits@sync-unmap-cycles - dmesg-warn - WARNING: possible circular locking dependency detected

Fri Oct 13 06:31:13 UTC 2017

https://bugs.freedesktop.org/show_bug.cgi?id=103251

            Bug ID: 103251
           Summary: [CI] igt at gem_userptr_blits@sync-unmap-cycles -
                    dmesg-warn - WARNING: possible circular locking
                    dependency detected
           Product: DRI
           Version: DRI git
          Hardware: Other
                OS: All
            Status: NEW
          Severity: normal
          Priority: medium
         Component: DRM/Intel
          Assignee: intel-gfx-bugs at lists.freedesktop.org
          Reporter: marta.lofstedt at intel.com
        QA Contact: intel-gfx-bugs at lists.freedesktop.org
                CC: intel-gfx-bugs at lists.freedesktop.org

a new lockdep? on CI_DRM_3225 HSW-shards:

Note, a fix for BUG 102886 was integrated to CI_DRM_3215, a fix for BUG 102939
was integrated to CI_DRM_3202. 

So, this looks like a new thing, therefore I file a new bug. Feel free to
re-open and duplicate to the old ones. 

I have checked dmesgs from surrounding runs where we don't hit the issues, but
there is no other splat before, so this looks more sporadic than the previous
ones.

https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3225/shard-hsw6/igt@gem_userptr_blits@sync-unmap-cycles.html

this is also after CI_DRM_3215, but I believe I had not archived BUG 102939,
before it hit. So it was missed by cibuglog.
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3216/shard-snb1/igt@gem_userptr_blits@map-fixed-invalidate-gup.html

[   20.805315] ======================================================
[   20.805316] WARNING: possible circular locking dependency detected
[   20.805319] 4.14.0-rc4-CI-CI_DRM_3225+ #1 Tainted: G     U         
[   20.805320] ------------------------------------------------------
[   20.805322] kworker/6:1H/1438 is trying to acquire lock:
[   20.805324]  (&mm->mmap_sem){++++}, at: [<ffffffffa01c8e01>]
__i915_gem_userptr_get_pages_worker+0x141/0x240 [i915]
[   20.805355] 
               but now in release context of a crosslock acquired at the
following:
[   20.805357]  ((complete)&this_flusher.done){+.+.}, at: [<ffffffff8190b06d>]
wait_for_completion+0x1d/0x20
[   20.805363] 
               which lock already depends on the new lock.

[   20.805365] 
               the existing dependency chain (in reverse order) is:
[   20.805367] 
               -> #1 ((complete)&this_flusher.done){+.+.}:
[   20.805372]        __lock_acquire+0x1420/0x15e0
[   20.805374]        lock_acquire+0xb0/0x200
[   20.805376]        wait_for_common+0x58/0x210
[   20.805378]        wait_for_completion+0x1d/0x20
[   20.805381]        flush_workqueue+0x1af/0x540
[   20.805400]        i915_gem_userptr_mn_invalidate_range_start+0x13c/0x150
[i915]
[   20.805404]        __mmu_notifier_invalidate_range_start+0x76/0xc0
[   20.805406]        unmap_vmas+0x7d/0xa0
[   20.805408]        unmap_region+0xae/0x110
[   20.805410]        do_munmap+0x276/0x3f0
[   20.805411]        vm_munmap+0x67/0x90
[   20.805413]        SyS_munmap+0xe/0x20
[   20.805415]        entry_SYSCALL_64_fastpath+0x1c/0xb1
[   20.805416] 
               -> #0 (&mm->mmap_sem){++++}:
[   20.805419]        down_read+0x3e/0x70
[   20.805435]        __i915_gem_userptr_get_pages_worker+0x141/0x240 [i915]
[   20.805438]        process_one_work+0x233/0x660
[   20.805440]        worker_thread+0x4e/0x3b0
[   20.805441]        kthread+0x152/0x190
[   20.805442] 
               other info that might help us debug this:

[   20.805445]  Possible unsafe locking scenario by crosslock:

[   20.805447]        CPU0                    CPU1
[   20.805448]        ----                    ----
[   20.805449]   lock(&mm->mmap_sem);
[   20.805451]   lock((complete)&this_flusher.done);
[   20.805453]                                lock(&mm->mmap_sem);
[   20.805455]                               
unlock((complete)&this_flusher.done);
[   20.805457] 
                *** DEADLOCK ***

[   20.805460] 2 locks held by kworker/6:1H/1438:
[   20.805461]  #0:  (&(&pool->lock)->rlock){-.-.}, at: [<ffffffff8109c94c>]
process_one_work+0x2dc/0x660
[   20.805465]  #1:  (&x->wait#10){....}, at: [<ffffffff810cd69d>]
complete+0x1d/0x60
[   20.805469] 
               stack backtrace:
[   20.805472] CPU: 6 PID: 1438 Comm: kworker/6:1H Tainted: G     U         
4.14.0-rc4-CI-CI_DRM_3225+ #1
[   20.805474] Hardware name: MSI MS-7924/Z97M-G43(MS-7924), BIOS V1.12
02/15/2016
[   20.805480] Call Trace:
[   20.805483]  dump_stack+0x68/0x9f
[   20.805486]  print_circular_bug+0x235/0x3c0
[   20.805488]  ? HARDIRQ_verbose+0x10/0x10
[   20.805490]  check_prev_add+0x430/0x840
[   20.805492]  ? ret_from_fork+0x27/0x40
[   20.805494]  lock_commit_crosslock+0x3ee/0x660
[   20.805496]  ? lock_commit_crosslock+0x3ee/0x660
[   20.805498]  complete+0x29/0x60
[   20.805500]  pwq_dec_nr_in_flight+0x9c/0xa0
[   20.805502]  ? _raw_spin_lock_irq+0x40/0x50
[   20.805504]  process_one_work+0x335/0x660
[   20.805506]  worker_thread+0x4e/0x3b0
[   20.805508]  kthread+0x152/0x190
[   20.805509]  ? process_one_work+0x660/0x660
[   20.805511]  ? kthread_create_on_node+0x40/0x40
[   20.805513]  ret_from_fork+0x27/0x40
[   22.417432] random: crng init done

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20171013/681e0b50/attachment.html>