Bug 110927

The test went from failing (https://bugs.freedesktop.org/show_bug.cgi?id=110519) to failing with a WARNING:

https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6272/shard-skl9/igt@gem_exec_schedule@semaphore-resolve.html

Starting subtest: semaphore-resolve
(gem_exec_schedule:2891) igt_aux-CRITICAL: Test assertion failure function sig_abort, file ../lib/igt_aux.c:502:
(gem_exec_schedule:2891) igt_aux-CRITICAL: Failed assertion: !"GPU hung"
Subtest semaphore-resolve failed.

<6> [759.522330] i915 0000:00:02.0: GPU HANG: ecode 9:1:0xfffffffe, in gem_exec_schedu [2891], hang on rcs0
<6> [759.522623] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
<6> [759.522645] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
<6> [759.522664] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
<6> [759.522681] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
<6> [759.522700] [drm] GPU crash dump saved to /sys/class/drm/card0/error
<7> [759.532843] [drm:i915_reset_device [i915]] resetting chip
<5> [759.538823] i915 0000:00:02.0: Resetting chip for hang on rcs0
<4> [759.540201] 
<4> [759.540227] ======================================================
<4> [759.540259] WARNING: possible circular locking dependency detected
<4> [759.540297] 5.2.0-rc4-CI-CI_DRM_6272+ #1 Tainted: G     U           
<4> [759.540327] ------------------------------------------------------
<4> [759.540361] kworker/0:1/2700 is trying to acquire lock:
<4> [759.540392] 000000005112029f (wakeref#2/1){+.+.}, at: __intel_wakeref_get_first+0x24/0xa0 [i915]
<4> [759.540717] 
but task is already holding lock:
<4> [759.540748] 000000000de18043 (i915.reset){+.+.}, at: i915_reset+0x57/0x3d0 [i915]
<4> [759.541087] 
which lock already depends on the new lock.

<4> [759.541126] 
the existing dependency chain (in reverse order) is:
<4> [759.541160] 
-> #4 (i915.reset){+.+.}:
<4> [759.541551]        i915_request_wait+0x16f/0x940 [i915]
<4> [759.541922]        i915_gem_wait_for_idle+0xf5/0x570 [i915]
<4> [759.542275]        i915_gem_shrink+0x4b0/0x630 [i915]
<4> [759.542609]        i915_gem_shrink_all+0x2c/0x50 [i915]
<4> [759.542917]        i915_drop_caches_set+0x1de/0x270 [i915]
<4> [759.542958]        simple_attr_write+0xb0/0xd0
<4> [759.542996]        full_proxy_write+0x51/0x80
<4> [759.543031]        vfs_write+0xbd/0x1b0
<4> [759.543062]        ksys_write+0x8f/0xe0
<4> [759.543093]        do_syscall_64+0x55/0x1c0
<4> [759.543129]        entry_SYSCALL_64_after_hwframe+0x49/0xbe
<4> [759.543159] 
-> #3 (&dev->struct_mutex/1){+.+.}:
<4> [759.543522]        i915_gem_shrinker_taints_mutex+0x52/0xe0 [i915]
<4> [759.543889]        i915_address_space_init+0x59/0x120 [i915]
<4> [759.544255]        i915_ggtt_init_hw+0x50/0x150 [i915]
<4> [759.544526]        i915_driver_load+0xebb/0x18b0 [i915]
<4> [759.544801]        i915_pci_probe+0x3f/0x1a0 [i915]
<4> [759.544836]        pci_device_probe+0x9e/0x120
<4> [759.544870]        really_probe+0xea/0x3c0
<4> [759.544901]        driver_probe_device+0x10b/0x120
<4> [759.544936]        device_driver_attach+0x4a/0x50
<4> [759.544970]        __driver_attach+0x97/0x130
<4> [759.545001]        bus_for_each_dev+0x74/0xc0
<4> [759.545032]        bus_add_driver+0x13f/0x210
<4> [759.545066]        driver_register+0x56/0xe0
<4> [759.545097]        do_one_initcall+0x58/0x300
<4> [759.545132]        do_init_module+0x56/0x1f6
<4> [759.545164]        load_module+0x24d1/0x2990
<4> [759.545196]        __se_sys_finit_module+0xd3/0xf0
<4> [759.545229]        do_syscall_64+0x55/0x1c0
<4> [759.545264]        entry_SYSCALL_64_after_hwframe+0x49/0xbe
<4> [759.545293] 
-> #2 (fs_reclaim){+.+.}:
<4> [759.545344]        fs_reclaim_acquire.part.24+0x24/0x30
<4> [759.545381]        kmem_cache_alloc_trace+0x2a/0x290
<4> [759.545730]        i915_gem_object_get_pages_stolen+0xa2/0x130 [i915]
<4> [759.546084]        ____i915_gem_object_get_pages+0x1d/0xa0 [i915]
<4> [759.546436]        __i915_gem_object_get_pages+0x59/0xb0 [i915]
<4> [759.546788]        _i915_gem_object_create_stolen+0xd4/0x100 [i915]
<4> [759.547143]        i915_gem_object_create_stolen_for_preallocated+0xf0/0x550 [i915]
<4> [759.547548]        intel_alloc_initial_plane_obj.isra.119+0xc6/0x1d0 [i915]
<4> [759.547954]        intel_modeset_init+0x8b8/0x1a20 [i915]
<4> [759.548223]        i915_driver_load+0xdb1/0x18b0 [i915]
<4> [759.548497]        i915_pci_probe+0x3f/0x1a0 [i915]
<4> [759.548533]        pci_device_probe+0x9e/0x120
<4> [759.548567]        really_probe+0xea/0x3c0
<4> [759.548599]        driver_probe_device+0x10b/0x120
<4> [759.548639]        device_driver_attach+0x4a/0x50
<4> [759.548673]        __driver_attach+0x97/0x130
<4> [759.548706]        bus_for_each_dev+0x74/0xc0
<4> [759.548737]        bus_add_driver+0x13f/0x210
<4> [759.548771]        driver_register+0x56/0xe0
<4> [759.548803]        do_one_initcall+0x58/0x300
<4> [759.548836]        do_init_module+0x56/0x1f6
<4> [759.548869]        load_module+0x24d1/0x2990
<4> [759.548902]        __se_sys_finit_module+0xd3/0xf0
<4> [759.548936]        do_syscall_64+0x55/0x1c0
<4> [759.548970]        entry_SYSCALL_64_after_hwframe+0x49/0xbe
<4> [759.549001] 
-> #1 (&obj->mm.lock){+.+.}:
<4> [759.549046]        __mutex_lock+0x8c/0x960
<4> [759.549391]        i915_gem_object_pin_map+0x2d/0x2a0 [i915]
<4> [759.549701]        __engine_unpark+0x42/0x80 [i915]
<4> [759.549992]        __intel_wakeref_get_first+0x40/0xa0 [i915]
<4> [759.550369]        i915_request_create+0x101/0x240 [i915]
<4> [759.550709]        i915_gem_do_execbuffer+0xb07/0x20f0 [i915]
<4> [759.551050]        i915_gem_execbuffer2_ioctl+0x11b/0x430 [i915]
<4> [759.551090]        drm_ioctl_kernel+0x83/0xf0
<4> [759.551123]        drm_ioctl+0x2f3/0x3b0
<4> [759.551153]        do_vfs_ioctl+0xa0/0x6e0
<4> [759.551184]        ksys_ioctl+0x35/0x60
<4> [759.551214]        __x64_sys_ioctl+0x11/0x20
<4> [759.551246]        do_syscall_64+0x55/0x1c0
<4> [759.551280]        entry_SYSCALL_64_after_hwframe+0x49/0xbe
<4> [759.551309] 
-> #0 (wakeref#2/1){+.+.}:
<4> [759.551363]        lock_acquire+0xa6/0x1c0
<4> [759.551394]        __mutex_lock+0x8c/0x960
<4> [759.551679]        __intel_wakeref_get_first+0x24/0xa0 [i915]
<4> [759.551999]        reset_prepare_engine+0x9/0x30 [i915]
<4> [759.552317]        reset_prepare+0x29/0x40 [i915]
<4> [759.552636]        i915_reset+0x9f/0x3d0 [i915]
<4> [759.552953]        i915_reset_device+0xf2/0x170 [i915]
<4> [759.553272]        i915_handle_error+0x231/0x370 [i915]
<4> [759.553580]        i915_hangcheck_elapsed+0x41c/0x530 [i915]
<4> [759.553620]        process_one_work+0x245/0x610
<4> [759.553651]        worker_thread+0x37/0x380
<4> [759.553684]        kthread+0x119/0x130
<4> [759.553716]        ret_from_fork+0x3a/0x50
<4> [759.553743] 
other info that might help us debug this:

<4> [759.553784] Chain exists of:
  wakeref#2/1 --> &dev->struct_mutex/1 --> i915.reset

<4> [759.553852]  Possible unsafe locking scenario:

<4> [759.553884]        CPU0                    CPU1
<4> [759.553909]        ----                    ----
<4> [759.553934]   lock(i915.reset);
<4> [759.553959]                                lock(&dev->struct_mutex/1);
<4> [759.553996]                                lock(i915.reset);
<4> [759.554028]   lock(wakeref#2/1);
<4> [759.554059] 
 *** DEADLOCK ***

<4> [759.554095] 4 locks held by kworker/0:1/2700:
<4> [759.554120]  #0: 00000000c7971d7e ((wq_completion)events_long){+.+.}, at: process_one_work+0x1bf/0x610
<4> [759.554182]  #1: 000000003ad9146c ((work_completion)(&(&i915->gpu_error.hangcheck_work)->work)){+.+.}, at: process_one_work+0x1bf/0x610
<4> [759.554252]  #2: 000000005f0d6d67 (&dev_priv->gpu_error.wedge_mutex){+.+.}, at: i915_reset_device+0xe4/0x170 [i915]
<4> [759.554599]  #3: 000000000de18043 (i915.reset){+.+.}, at: i915_reset+0x57/0x3d0 [i915]
<4> [759.554941] 
stack backtrace:
<4> [759.554979] CPU: 0 PID: 2700 Comm: kworker/0:1 Tainted: G     U            5.2.0-rc4-CI-CI_DRM_6272+ #1
<4> [759.555025] Hardware name: Google Caroline/Caroline, BIOS MrChromebox 08/27/2018
<4> [759.555345] Workqueue: events_long i915_hangcheck_elapsed [i915]
<4> [759.555381] Call Trace:
<4> [759.555413]  dump_stack+0x67/0x9b
<4> [759.555453]  print_circular_bug+0x1c8/0x2b0
<4> [759.555493]  __lock_acquire+0x1ce9/0x24c0
<4> [759.555537]  ? wake_up_klogd+0x4a/0x60
<4> [759.555570]  ? __bfs+0xe8/0x220
<4> [759.555608]  ? lock_acquire+0xa6/0x1c0
<4> [759.555643]  lock_acquire+0xa6/0x1c0
<4> [759.555931]  ? __intel_wakeref_get_first+0x24/0xa0 [i915]
<4> [759.555973]  __mutex_lock+0x8c/0x960
<4> [759.556257]  ? __intel_wakeref_get_first+0x24/0xa0 [i915]
<4> [759.556564]  ? intel_engine_stop_cs+0x1f/0xb0 [i915]
<4> [759.556853]  ? __intel_wakeref_get_first+0x24/0xa0 [i915]
<4> [759.556899]  ? lock_acquire+0xa6/0x1c0
<4> [759.557207]  ? intel_engine_cs_mock_selftests+0x20/0x20 [i915]
<4> [759.557497]  ? __intel_wakeref_get_first+0x24/0xa0 [i915]
<4> [759.557786]  __intel_wakeref_get_first+0x24/0xa0 [i915]
<4> [759.557828]  ? _raw_spin_unlock_irqrestore+0x4c/0x60
<4> [759.558149]  reset_prepare_engine+0x9/0x30 [i915]
<4> [759.558470]  reset_prepare+0x29/0x40 [i915]
<4> [759.558789]  i915_reset+0x9f/0x3d0 [i915]
<4> [759.559113]  i915_reset_device+0xf2/0x170 [i915]
<4> [759.559439]  ? i915_gem_set_wedged+0x60/0x60 [i915]
<4> [759.559483]  ? queue_work_node+0x70/0x70
<4> [759.559763]  i915_handle_error+0x231/0x370 [i915]
<4> [759.559812]  ? string+0x40/0x50
<4> [759.560133]  i915_hangcheck_elapsed+0x41c/0x530 [i915]
<4> [759.560178]  ? __drm_printfn_info+0x20/0x20
<4> [759.560222]  ? __lock_acquire+0x530/0x24c0
<4> [759.560268]  ? debug_object_deactivate+0x137/0x160
<4> [759.560315]  ? lock_acquire+0xa6/0x1c0
<4> [759.560349]  ? process_one_work+0x1bf/0x610
<4> [759.560387]  process_one_work+0x245/0x610
<4> [759.560430]  worker_thread+0x37/0x380
<4> [759.560465]  ? process_one_work+0x610/0x610
<4> [759.560499]  kthread+0x119/0x130
<4> [759.560532]  ? kthread_park+0x80/0x80
<4> [759.560568]  ret_from_fork+0x3a/0x50
<7> [759.561198] [drm:i915_reset_request [i915]] client gem_exec_schedu[2891]: gained 1 ban score, now 1