[Intel-gfx] kernel BUG at drivers/gpu/drm/i915/i915_gem.c:1262!

[Chris Wilson]

On Sun, Jan 04, 2015 at 03:33:05PM +0100, Marcin Ślusarz wrote:
> [432102.923209] kernel BUG at drivers/gpu/drm/i915/i915_gem.c:1262!
> [432102.923237] invalid opcode: 0000 [#1] PREEMPT SMP 
> [432102.923262] Modules linked in: uas usb_storage dm_crypt ctr ccm arc4 iwldvm mac80211 iwlwifi cfg80211 rfcomm bnep uvcvideo rtsx_usb videobuf2_vmalloc videobuf2_memops videobuf2_core v4l2_common videodev snd_hda_codec_hdmi snd_hda_codec_idt snd_hda_codec_generic dell_laptop dell_wmi sparse_keymap snd_hda_intel snd_hda_controller snd_hda_codec dcdbas btusb snd_hwdep snd_pcm bluetooth snd_seq_midi snd_seq_midi_event x86_pkg_temp_thermal snd_rawmidi snd_seq coretemp ghash_clmulni_intel cryptd snd_timer snd_seq_device microcode snd soundcore serio_raw lpc_ich mac_hid parport_pc ppdev lp parport binfmt_misc hid_generic usbhid hid nouveau i915 psmouse mxm_wmi ttm i2c_algo_bit ahci libahci drm_kms_helper drm wmi video
> [432102.923617] CPU: 2 PID: 45 Comm: kswapd0 Tainted: G        W      3.17.0 #50
> [432102.923650] Hardware name: Dell Inc.          Inspiron 7720/04M3YM, BIOS A07 08/16/2012
> [432102.923685] task: ffff88013766a130 ti: ffff8801376c8000 task.ti: ffff8801376c8000
> [432102.923719] RIP: 0010:[<ffffffffa01235cc>]  [<ffffffffa01235cc>] i915_wait_seqno+0x7c/0xa0 [i915]
> [432102.923781] RSP: 0018:ffff8801376cbaf8  EFLAGS: 00010246
> [432102.923804] RAX: 0000000000000001 RBX: ffff880032870000 RCX: 0000000000000018
> [432102.923835] RDX: ffff88008d033908 RSI: 00000000005c4f93 RDI: ffff880032872680
> [432102.923870] RBP: ffff8801376cbb18 R08: 0000000000000000 R09: 0000000000000000
> [432102.923901] R10: 0000000000000000 R11: ffffea0000402c80 R12: ffff880032872680
> [432102.923932] R13: ffff88008d032c00 R14: 0000000000000001 R15: ffff880032870000
> [432102.923966] FS:  0000000000000000(0000) GS:ffff88013f280000(0000) knlGS:0000000000000000
> [432102.924002] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [432102.924030] CR2: 00007f3d465dc000 CR3: 0000000102ada000 CR4: 00000000001407e0
> [432102.924073] Stack:
> [432102.924087]  ffff88008d032c00 ffff88008d032cf0 ffff88008d032c00 ffff880032870000
> [432102.924133]  ffff8801376cbb30 ffffffffa0123627 ffff88008d032c00 ffff8801376cbb48
> [432102.924169]  ffffffffa0123664 ffff880104b4b000 ffff8801376cbb78 ffffffffa0126e48
> [432102.924205] Call Trace:
> [432102.924238]  [<ffffffffa0123627>] i915_gem_object_wait_rendering+0x37/0x60 [i915]
> [432102.924287]  [<ffffffffa0123664>] i915_gem_object_finish_gpu.part.53+0x14/0x30 [i915]
> [432102.924339]  [<ffffffffa0126e48>] i915_vma_unbind+0x118/0x240 [i915]
> [432102.924380]  [<ffffffffa012715c>] __i915_gem_shrink+0x1ec/0x2c0 [i915]
> [432102.924422]  [<ffffffffa01279f8>] i915_gem_shrinker_scan+0x68/0xa0 [i915]
> [432102.924465]  [<ffffffff811a3b78>] shrink_slab_node+0x138/0x2f0
> [432102.924494]  [<ffffffff811a556f>] ? shrink_slab+0x3f/0x160
> [432102.924521]  [<ffffffff811a55bb>] shrink_slab+0x8b/0x160
> [432102.924548]  [<ffffffff811a8848>] balance_pgdat+0x388/0x5a0
> [432102.924575]  [<ffffffff811a8be0>] kswapd+0x180/0x470
> [432102.924601]  [<ffffffff810d5830>] ? prepare_to_wait_event+0x110/0x110
> [432102.924635]  [<ffffffff811a8a60>] ? balance_pgdat+0x5a0/0x5a0
> [432102.924670]  [<ffffffff810b609c>] kthread+0xfc/0x120
> [432102.924693]  [<ffffffff810b5fa0>] ? kthread_create_on_node+0x230/0x230
> [432102.924730]  [<ffffffff81710e2c>] ret_from_fork+0x7c/0xb0
> [432102.924755]  [<ffffffff810b5fa0>] ? kthread_create_on_node+0x230/0x230
> [432102.924786] Code: 0f 1f 80 00 00 00 00 8b 93 a0 9d 00 00 41 0f b6 ce 44 89 ee 4c 89 e7 45 31 c9 45 31 c0 e8 ed ea ff ff 5b 41 5c 41 5d 41 5e 5d c3 <0f> 0b 45 84 f6 74 13 8b 83 a0 9d 00 00 c1 f8 1f 83 e0 06 83 e8 
> [432102.924965] RIP  [<ffffffffa01235cc>] i915_wait_seqno+0x7c/0xa0 [i915]
> [432102.925104]  RSP <ffff8801376cbaf8>
> [432102.935737] ---[ end trace 95ef281295f96d84 ]---
> 
> It's BUG_ON(!mutex_is_locked(&dev->struct_mutex)) in i915_wait_seqno.
> Warnings are from WARN_ON(!mutex_is_locked(&dev->struct_mutex)) in
> intel_fb_obj_invalidate.
> 
> dev->struct_mutex is locked from i915_gem_shrinker_scan, so something unlocks
> it when we reach intel_fb_obj_invalidate/i915_wait_seqno.

What does 'grep DEBUG_MUTEXES .config' say?
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre