Bug#954938: Occasional spice console freezes on boot after kernel upgrade

[Frank Otto <f.otto@ucl.ac.uk>]

Observing the same behaviour with version 4.19+105+deb10u5.
But here it is not occasional but happens on every boot.
The hung kworker causes a constant system load of 1.
After switching the video device from qxl to virtio, the problem disappears.



On Wed, 25 Mar 2020 16:01:02 +0100 IB Development Team <dev@ib.pl> wrote:
> Package: linux-image-amd64
> Version: 4.19+105+deb10u3
>
> After upgrading kernel from 4.19+105+deb10u1 to 4.19+105+deb10u3 in
> debian 10 vm (qemu, kvm, virt-manager, qxl graphics) there are
> occassional spice text console image corruptions during system boot
> (boot with all text messages displayed, without quiet nor splash kernel
> parameters).
>
> Corrupted console contains many random characters witch different
> colors and does not update (freezes just before switching to high
> resolution text mode). 
>
> Boot continues normally and one can unfreeze first console switching to
> second console and back (Ctrl+Alt+F2 and then Ctrl+Alt+F1).
>
> After a few minutes syslog starts to receive messages like below
> (repeating every 2 minutes):
>
> 2020-03-25 15:50:04 myhost kernel: [  484.596112] INFO: task
> kworker/1:1:42 blocked for more than 120 seconds.
> 2020-03-25 15:50:04 myhost kernel: [  484.596116]       Not tainted
> 4.19.0-8-amd64 #1 Debian 4.19.98-1
> 2020-03-25 15:50:04 myhost kernel: [  484.596118] "echo 0 >
> /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> 2020-03-25 15:50:04 myhost kernel: [  484.596120]
> kworker/1:1     D    0    42      2 0x80000000
> 2020-03-25 15:50:04 myhost kernel: [  484.596149] Workqueue: events
> drm_fb_helper_dirty_work [drm_kms_helper]
> 2020-03-25 15:50:04 myhost kernel: [  484.596152] Call Trace:
> 2020-03-25 15:50:04 myhost kernel: [  484.596163]  ?
> __schedule+0x2a2/0x870
> 2020-03-25 15:50:04 myhost kernel: [  484.596168]  schedule+0x28/0x80
> 2020-03-25 15:50:04 myhost kernel:
> [  484.596177]  qxl_ring_push+0x15e/0x1e0 [qxl]
> 2020-03-25 15:50:04 myhost kernel: [  484.596184]  ?
> finish_wait+0x80/0x80
> 2020-03-25 15:50:04 myhost kernel:
> [  484.596190]  qxl_push_command_ring_release+0x64/0x90 [qxl]
> 2020-03-25 15:50:04 myhost kernel:
> [  484.596197]  qxl_draw_opaque_fb+0x25a/0x3f0 [qxl]
> 2020-03-25 15:50:04 myhost kernel: [  484.596203]  ?
> __switch_to_asm+0x35/0x70
> 2020-03-25 15:50:04 myhost kernel:
> [  484.596208]  qxlfb_framebuffer_dirty+0xeb/0x180 [qxl]
> 2020-03-25 15:50:04 myhost kernel: [  484.596214]  ?
> syscall_return_via_sysret+0x33/0x83
> 2020-03-25 15:50:04 myhost kernel:
> [  484.596227]  drm_fb_helper_dirty_work+0x14e/0x160 [drm_kms_helper]
> 2020-03-25 15:50:04 myhost kernel:
> [  484.596236]  process_one_work+0x1a7/0x3a0
> 2020-03-25 15:50:04 myhost kernel:
> [  484.596241]  worker_thread+0x30/0x390
> 2020-03-25 15:50:04 myhost kernel: [  484.596245]  ?
> create_worker+0x1a0/0x1a0
> 2020-03-25 15:50:04 myhost kernel: [  484.596249]  kthread+0x112/0x130
> 2020-03-25 15:50:04 myhost kernel: [  484.596252]  ?
> kthread_bind+0x30/0x30
> 2020-03-25 15:50:04 myhost kernel:
> [  484.596257]  ret_from_fork+0x35/0x40