Bug 99227 - [KBL] GPU HANG: ecode 9:0:0x85dffffb, in Cities.x64 [48013], reason: Hang on render ring, action: reset (GPU hang when playing Cities: Skyline for a few minutes (<10 minutes))
Summary: [KBL] GPU HANG: ecode 9:0:0x85dffffb, in Cities.x64 [48013], reason: Hang on ...
Status: RESOLVED INVALID
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: unspecified
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Intel 3D Bugs Mailing List
QA Contact: Intel 3D Bugs Mailing List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2016-12-30 04:31 UTC by Saikrishna Arcot
Modified: 2017-02-05 17:33 UTC (History)
1 user (show)

See Also:
i915 platform: KBL
i915 features: GPU hang

Attachments
GPU hang crash dump with i915.enable_fbc=1 (787.73 KB, text/plain)
2016-12-30 04:31 UTC, Saikrishna Arcot 		Details
GPU hang crash dump without i915.enable_fbc=1 (780.11 KB, text/plain)
2016-12-30 04:31 UTC, Saikrishna Arcot 		Details
View All

Description Saikrishna Arcot 2016-12-30 04:31:21 UTC 
Created attachment 128690 [details]
GPU hang crash dump with i915.enable_fbc=1

Ubuntu 16.10, with Linux kernel 4.9 (compiled by myself). When playing Cities: Skyline for a few minutes, a GPU hang occurs.

Two crash dumps have been attached. The first one (gpuHang-fbc-enabled) is with the i915.enable_fbc=1 set in the kernel command line. The second one (gpuHang-fbc-disabled) is without that flag set.
Comment 1 Saikrishna Arcot 2016-12-30 04:31:56 UTC 
Created attachment 128691 [details]
GPU hang crash dump without i915.enable_fbc=1
Comment 2 yann 2017-01-03 14:49:01 UTC 
There are constant improvements pushed in kernel and Mesa that will benefit to your system, so please re-test with latest kernel & Mesa (12 or 13) and mark as REOPENED if you can reproduce (and attach fresh gpu error dump & kernel log, please capture and upload an apitrace(https://github.com/apitrace/apitrace) so that we can easily reproduce as well.) and RESOLVED/* if you cannot reproduce.
 
In parallel, assigning to Mesa product.

Kernel: 4.9.0arcot
Platform: KBL-ULT (pci id: 0x5916, pci revision: 0x02, pci subsystem: 1558:2410)
Mesa: [Please confirm your mesa version]

From these error dumps, hung is both happening in render ring batch with 0x7b000004 (PIPE_CONTROL) as IPEHR -same sequence in batch, no impact from fbc, so please use default options/parameters.

Batch extract (around active head at 0xef25a638):

0xef25a5c4:      0x78090005: 3DSTATE_VERTEX_ELEMENTS
0xef25a5c8:      0x02400000:    buffer 0: invalid, type 0x0040, src offset 0x0000 bytes
0xef25a5cc:      0x11130000:    (X, Y, Z, 1.0), dst offset 0x00 bytes
0xef25a5d0:      0x0285000c:    buffer 0: invalid, type 0x0085, src offset 0x000c bytes
0xef25a5d4:      0x11230000:    (X, Y, 0.0, 1.0), dst offset 0x00 bytes
0xef25a5d8:      0x02850014:    buffer 0: invalid, type 0x0085, src offset 0x0014 bytes
0xef25a5dc:      0x11230000:    (X, Y, 0.0, 1.0), dst offset 0x00 bytes
0xef25a5e0:      0x78490001: 3D UNKNOWN: 3d_965 opcode = 0x7849
0xef25a5e4:      0x00000000: MI_NOOP
0xef25a5e8:      0x00000000: MI_NOOP
0xef25a5ec:      0x78490001: 3D UNKNOWN: 3d_965 opcode = 0x7849
0xef25a5f0:      0x00000001: MI_NOOP
0xef25a5f4:      0x00000000: MI_NOOP
0xef25a5f8:      0x78490001: 3D UNKNOWN: 3d_965 opcode = 0x7849
0xef25a5fc:      0x00000002: MI_NOOP
0xef25a600:      0x00000000: MI_NOOP
Bad length 7 in (null), expected 6-6
0xef25a604:      0x7b000005: 3DPRIMITIVE: fail sequential
0xef25a608:      0x00000104:    vertex count
0xef25a60c:      0x0000ea60:    start vertex
0xef25a610:      0x00000000:    instance count
0xef25a614:      0x00000001:    start instance
0xef25a618:      0x00000000:    index bias
0xef25a61c:      0x00000000: MI_NOOP
Bad count in PIPE_CONTROL
0xef25a620:      0x7a000004: PIPE_CONTROL: no write, no depth stall, no RC write flush, no inst flush
0xef25a624:      0x00101001:    destination address
0xef25a628:      0x00000000:    immediate dword low
0xef25a62c:      0x00000000:    immediate dword high
Bad count in PIPE_CONTROL
0xef25a638:      0x7a000004: PIPE_CONTROL: no write, no depth stall, no RC write flush, no inst flush
0xef25a63c:      0x00000408:    destination address
0xef25a640:      0x00000000:    immediate dword low
0xef25a644:      0x00000000:    immediate dword high
0xef25a650:      0x78210000: 3D UNKNOWN: 3d_965 opcode = 0x7821
Comment 3 Saikrishna Arcot 2017-01-03 15:14:39 UTC 
The GPU error dump was on Mesa 12.0.3. I'll see if I can reproduce the crash with Mesa 13.
Comment 4 Saikrishna Arcot 2017-02-05 17:33:07 UTC 
This might have been fixed in a more recent version of Mesa.

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.