Created attachment 64163 [details] dmesg System Environment: -------------------------- Arch: x86_64 Platform: Ivybridge Libdrm: (master)libdrm-2.4.37-2-g881e69f185d2450ca35964caf254b04e7e8838c1 Mesa: (master)33202b4876a88b6f54ca7022eadd2875a2d3508a Xserver:(master)xorg-server-1.12.0-233-gdae317e7265007b38012244722e3b3a06e904ed5 Libva: (staging)f12f80371fb534e6bbf248586b3c17c298a31f4e Libva_intel_driver:(staging)82fa52510a37ab645daaa3bb7091ff5096a20d0b Kernel: (drm-intel-next-queued) 127b87d76a345166840e2fbbb2c7501dd388e16b Bug detailed description: ------------------------- When nightly testing run piglit cases, call trace appears in dmesg. It happens on ivybridge and sandybrigde with drm-intel-next-queued kernel. It doesn't happen on drm-intel-fixes kernel. Call Trace: [ 5791.888795] [<ffffffffa009b8b2>] i915_gem_object_wait_rendering+0x34/0x51 [i915] [ 5791.888839] [<ffffffff811fa2e6>] ? list_del+0xe/0x2e [ 5791.888874] [<ffffffffa009c417>] i915_gem_object_finish_gpu+0x39/0x45 [i915] [ 5791.888920] [<ffffffffa009c718>] i915_gem_object_unbind+0x45/0x19f [i915] [ 5791.888965] [<ffffffffa009f73a>] i915_gem_evict_something+0x277/0x2c0 [i915] [ 5791.889011] [<ffffffffa009a682>] i915_gem_object_bind_to_gtt+0x1e0/0x38c [i915] [ 5791.889058] [<ffffffffa009cad3>] i915_gem_fault+0xd4/0x1ee [i915] [ 5791.889095] [<ffffffff810cb444>] __do_fault+0xa7/0x3bb [ 5791.889126] [<ffffffff810cd973>] handle_pte_fault+0x28f/0x6b9 [ 5791.889159] [<ffffffff810cee29>] handle_mm_fault+0x196/0x1ab [ 5791.889192] [<ffffffff813cb17d>] do_page_fault+0x3ad/0x3d2 [ 5791.889225] [<ffffffff810fac14>] ? do_vfs_ioctl+0x444/0x485 [ 5791.889258] [<ffffffff810c6918>] ? vm_mmap_pgoff+0x78/0x8b [ 5791.889290] [<ffffffff813c862f>] page_fault+0x1f/0x30 [ 5791.889339] Code: 38 5b 41 5c 41 5d 41 5e 41 5f 5d c3 55 85 f6 48 89 e5 41 55 41 54 41 89 f4 53 48 89 fb 51 48 8b 47 18 4c 8b a8 48 03 00 00 75 02 <0f> 0b 41 0f b6 b5 20 28 00 00 4c 89 ef e8 a2 f4 ff ff 85 c0 75 [ 5791.889445] RIP [<ffffffffa009b83c>] i915_wait_seqno+0x1f/0x61 [i915] [ 5791.889481] RSP <ffff8800a3c31ab8>
Looks to be the same flushing list removal issue I encountered. Patches posted to list. Will attach squashed version shortly.
Created attachment 64164 [details] [review] Remove the defunct flushing list
Created attachment 64168 [details] [review] Remove the defunct flushing list (v2)
Should be mostly fixed with commit b0e5586160189327221725fe99ee1844efc14907 Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Fri Jul 13 14:14:08 2012 +0100 drm/i915: Insert a flush between batches if the breadcrumb was dropped Please report back if you still have such issues with running piglit on latest -queued.
It still happens on -queued kernel commit:06b863619d431f39269b
Second guess: this is the context blow up.
This should be fixed on latest dinq with commit de72b43f3691fe6949694c8ffb282467a9b67f87 Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Sun Jul 15 12:34:22 2012 +0100 drm/i915: Flush the context object from the CPU caches upon switching
Verified. It has been fixed on -queued kernel commit b17a616d43882fe1.
A patch referencing this bug report has been merged in Linux v3.6-rc1: commit 09cf7c9a1299c386525d31305fd50591acf49a9a Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Fri Jul 13 14:14:08 2012 +0100 drm/i915: Insert a flush between batches if the breadcrumb was dropped
Closing old verified.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.