strongtz / i915-sriov-dkms

dkms module of Linux i915 driver with SR-IOV support
939 stars 114 forks source link

[BUG] Kernel 6.5.7-200.fc38.x86_64, invalidation response timeout- crash when using virtualized gpu #118

Open sihawken opened 11 months ago

sihawken commented 11 months ago

Does not work on kernel 6.5.7-200.fc38.x86_64

The desktop crashes when using 0000:00:02.1, set up using virt manager.

CPU: 12th Gen Intel® Core™ i7-1260P × 16

[ 3004.780435] [drm] Initialized i915 1.6.0 20201103 for 0000:00:02.1 on minor 0
[ 3015.412724] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 72
[ 3016.436819] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 73
[ 3017.460788] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 74
[ 3018.485044] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 75
[ 3019.509783] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 76
[ 3020.533098] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 77
[ 3021.557126] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 78
[ 3021.561824] i915 0000:00:02.1: [drm] *ERROR* GT0: rcs0 TLB invalidation did not complete in 4ms!
[ 3022.580909] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 79
[ 3022.585856] i915 0000:00:02.1: [drm] *ERROR* GT0: rcs0 TLB invalidation did not complete in 4ms!
[ 3023.604903] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 80
[ 3023.609872] i915 0000:00:02.1: [drm] *ERROR* GT0: rcs0 TLB invalidation did not complete in 4ms!
[ 3024.629169] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 81
[ 3024.633886] i915 0000:00:02.1: [drm] *ERROR* GT0: rcs0 TLB invalidation did not complete in 4ms!
[ 3025.653260] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 82
[ 3025.657902] i915 0000:00:02.1: [drm] *ERROR* GT0: rcs0 TLB invalidation did not complete in 4ms!
[ 3026.677200] i915 0000:00:02.1: [drm] *ERROR* tlb invalidation response timed out for seqno 83
[ 3026.681917] i915 0000:00:02.1: [drm] *ERROR* GT0: rcs0 TLB invalidation did not complete in 4ms!
resiliencer commented 10 months ago

I get the same error on 6.5.7-zen and on 6.5.9-zen (but seqno is 23) but it works for hour or maybe two and then hosts runs out of memory. Memory leak occurs somewhere.

githubpanguan commented 9 months ago

我在 6.5.7-zen 和 6.5.9-zen 上遇到相同的错误(但 seqno 是 23),但它可以工作一个小时或两个小时,然后主机内存不足。 某处发生内存泄漏。

Me too, the kernel memory overflowed, not the process

suki78 commented 3 months ago

any update ?

suki78 commented 2 months ago

ok so I figured it out, I found this commit that solved TLB invalidation kernel error https://github.com/intel/linux-intel-lts/commit/c75552e4fc1b556c4d33ec20ec1e5c99f666068e