My host OS is centos7.3. When start VM with KVMGT bu did not work. uname -a : Linux localhost.localdomain 4.3.0-rc6-vgt+ #2 SMP Thu May 25 03:20:03 EDT 2017 x86_64 x86_64 x86_64 GNU/Linux dmesg : [ 1.346718] [drm] Initialized drm 1.1.0 20060810 [ 2.959335] [drm] Virtual GPU for Intel GVT-g detected. [ 2.959961] [drm] Memory usable by graphics device = 4096M [ 2.960552] [drm] Replacing VGA console driver [ 98.970262] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013). [ 98.970466] [drm] Driver supports precise vblank timestamp query. [ 99.013552] [drm] VGT ballooning configuration: [ 99.013744] [drm] Mappable graphic memory: base 0x0 size 98304KiB [ 99.013978] [drm] Unmappable graphic memory: base 0x10000000 size 393216KiB [ 99.014248] [drm] balloon space: range [ 0x28000000 - 0xfffff000 ] 3538940 KiB. [ 99.014582] [drm] balloon space: range [ 0x6000000 - 0x10000000 ] 163840 KiB. [ 99.014701] [drm] VGT balloon successfully [ 99.072665] [drm] Initialized i915 1.6.0 20151010 for 0000:00:02.0 on minor 0 [ 99.196066] [<ffffffffa0113247>] drm_atomic_commit+0x37/0x60 [drm] [ 100.560671] [drm] RC6 on [ 388.517792] [drm] stuck on render ring [ 390.851502] [drm] GPU HANG: ecode 9:0:0x00000000, in gnome-shell [5718], reason: VGT device reset, action: reset [ 390.851832] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. [ 390.852143] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel [ 390.852500] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue. [ 390.852839] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it. [ 390.853045] [drm] GPU crash dump saved to /sys/class/drm/card0/error [ 395.527120] [drm] RC6 on
Created attachment 131508 [details] dmesg output
Created attachment 131509 [details] /sys/class/drm/card0/error
Hello chenlizhou, We are working on the gvt-staging source code which is already based on 4.12. Did you backport the GVT source code from 4.12 to 4.3.0-rc6+ by yourself? Is it possible to directly try our gvt-staging kernel instead of backporting? We have a wiki for how to setup: https://github.com/01org/gvt-linux/wiki/GVTg_Setup_Guide And what test did you run to reproduce this issue?
Long time no response, request to close. if you still have this issue, please reopen it and help to share update information. thanks.
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.