Ever since we swapped fi-ctg-l9400 for another one (https://bugs.freedesktop.org/show_bug.cgi?id=101089), we started randomly hitting an assert when running igt@gem_tiled_fence_blits@basic: (gem_tiled_fence_blits:3343) CRITICAL: Test assertion failure function check_bo, file gem_tiled_fence_blits.c:96: (gem_tiled_fence_blits:3343) CRITICAL: Failed assertion: linear[i] == start_val (gem_tiled_fence_blits:3343) CRITICAL: Expected 0x0008ec80, found 0x0008ec90 at offset 0x0003b200 Subtest basic failed. **** DEBUG **** (gem_tiled_fence_blits:3343) INFO: Using 3 1MiB buffers (gem_tiled_fence_blits:3343) CRITICAL: Test assertion failure function check_bo, file gem_tiled_fence_blits.c:96: (gem_tiled_fence_blits:3343) CRITICAL: Failed assertion: linear[i] == start_val (gem_tiled_fence_blits:3343) CRITICAL: Expected 0x0008ec80, found 0x0008ec90 at offset 0x0003b200 **** END **** Statistics: Failure rate 2/5 run(s) (40%) Full logs: https://intel-gfx-ci.01.org/CI/CI_DRM_2672/fi-ctg-l9400/igt@gem_tiled_fence_blits@basic.html
Single bit failure. I would start by doing a memtest86 for 48 hours or so and check the RAM is error free.
Results of the memcheck: After a little bit less than 2 days, still no errors. Meanwhile, out of the 4 runs that were made on this machine, we got consistent errors. So, the bug may be in the hardware, but not in the memory.
Drat, RAM was replaceable. Hmm, do you have a complete list of verification failures like this? I want to check the pattern between tiled/gpu/gtt etc.
(In reply to Chris Wilson from comment #3) > Drat, RAM was replaceable. Hmm, do you have a complete list of verification > failures like this? I want to check the pattern between tiled/gpu/gtt etc. Not so sure what you are asking. Here are is all the info I have: https://intel-gfx-ci.01.org/CI/fi-ctg-l9400.html (the last 4 runs are on the new machine).
Closing the bug since it used to be very reproducable, but completely stopped being so for more than a month.
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.