Kernel v6.0.18: Resume from hibernate fails, system hangs; bisected

Kernel v6.0.18: Resume from hibernate fails, system hangs; bisected

[Rainer Fiebig]

Hi! Since kernel 6.0.18 resume from hibernate fails, the system hangs
and a hard reset is necessary.  The CPU is a Ryzen 5600G, the system is
Linux From Scratch-11.1.

I found this in the system log:

[...]
Jan 12 19:30:03 LUX kernel: [   50.248036] amdgpu 0000:30:00.0: [drm]
*ERROR* [CRTC:67:crtc-0] flip_done timed out
Jan 12 19:30:03 LUX kernel: [   50.248040] amdgpu 0000:30:00.0: [drm]
*ERROR* [CRTC:70:crtc-1] flip_done timed out
Jan 12 19:30:14 LUX kernel: [   60.488034] amdgpu 0000:30:00.0: [drm]
*ERROR* flip_done timed out
Jan 12 19:30:14 LUX kernel: [   60.488040] amdgpu 0000:30:00.0: [drm]
*ERROR* [CRTC:67:crtc-0] commit wait timed out
^@^@^@^@^@^@^@^@^@^[...]@^@^@^@^@^@^@^@^@^@^@^@^@^@Jan 12 19:31:20 LUX
syslogd 1.5.1: restart.
[...]


Bisecting the problem turned up this:

~/Downloads/linux-stable-BLFS-11.1> git bisect bad
306df163069e78160e7a534b892c5cd6fefdd537 is the first bad commit
commit 306df163069e78160e7a534b892c5cd6fefdd537
Author: Alex Deucher <alexander.deucher@amd.com>
Date:   Wed Dec 7 11:08:53 2022 -0500

    drm/amdgpu: make display pinning more flexible (v2)

    commit 81d0bcf9900932633d270d5bc4a54ff599c6ebdb upstream.

    Only apply the static threshold for Stoney and Carrizo.
    This hardware has certain requirements that don't allow
    mixing of GTT and VRAM.  Newer asics do not have these
    requirements so we should be able to be more flexible
    with where buffers end up.
    [...]


Let me know if you need more info. Thanks.

Rainer Fiebig
-- 
The truth always turns out to be simpler than you thought.
Richard Feynman

Re: Kernel v6.0.18: Resume from hibernate fails, system hangs; bisected

[Linux kernel regression tracking (Thorsten Leemhuis)]

[ccing Alex and the regressions list]

Hi, this is your Linux kernel regression tracker.

On 12.01.23 22:03, Rainer Fiebig wrote:
> Hi! Since kernel 6.0.18

Note, 6.0.y is now EOL, you should move to 6.1.y.

> resume from hibernate fails, the system hangs
> and a hard reset is necessary.  The CPU is a Ryzen 5600G, the system is
> Linux From Scratch-11.1.

FWIW, there is a report with a problems that looks somewhat similar here:

https://bugzilla.kernel.org/show_bug.cgi?id=216917

But the reporter doesn't care anymore, as 6.1 works.

Ciao, Thorsten (wearing his 'the Linux kernel's regression tracker' hat)
--
Everything you wanna know about Linux kernel regression tracking:
https://linux-regtracking.leemhuis.info/about/#tldr
If I did something stupid, please tell me, as explained on that page.

> I found this in the system log:
> 
> [...]
> Jan 12 19:30:03 LUX kernel: [   50.248036] amdgpu 0000:30:00.0: [drm]
> *ERROR* [CRTC:67:crtc-0] flip_done timed out
> Jan 12 19:30:03 LUX kernel: [   50.248040] amdgpu 0000:30:00.0: [drm]
> *ERROR* [CRTC:70:crtc-1] flip_done timed out
> Jan 12 19:30:14 LUX kernel: [   60.488034] amdgpu 0000:30:00.0: [drm]
> *ERROR* flip_done timed out
> Jan 12 19:30:14 LUX kernel: [   60.488040] amdgpu 0000:30:00.0: [drm]
> *ERROR* [CRTC:67:crtc-0] commit wait timed out
> ^@^@^@^@^@^@^@^@^@^[...]@^@^@^@^@^@^@^@^@^@^@^@^@^@Jan 12 19:31:20 LUX
> syslogd 1.5.1: restart.
> [...]
> 
> 
> Bisecting the problem turned up this:
> 
> ~/Downloads/linux-stable-BLFS-11.1> git bisect bad
> 306df163069e78160e7a534b892c5cd6fefdd537 is the first bad commit
> commit 306df163069e78160e7a534b892c5cd6fefdd537
> Author: Alex Deucher <alexander.deucher@amd.com>
> Date:   Wed Dec 7 11:08:53 2022 -0500
> 
>     drm/amdgpu: make display pinning more flexible (v2)
> 
>     commit 81d0bcf9900932633d270d5bc4a54ff599c6ebdb upstream.
> 
>     Only apply the static threshold for Stoney and Carrizo.
>     This hardware has certain requirements that don't allow
>     mixing of GTT and VRAM.  Newer asics do not have these
>     requirements so we should be able to be more flexible
>     with where buffers end up.
>     [...]
> 
> 
> Let me know if you need more info. Thanks.
> 
> Rainer Fiebig

Re: Kernel v6.0.18: Resume from hibernate fails, system hangs; bisected

[Rainer Fiebig]

Am 13.01.23 um 10:20 schrieb Linux kernel regression tracking (Thorsten
Leemhuis):
> [ccing Alex and the regressions list]
> 
> Hi, this is your Linux kernel regression tracker.
> 
> On 12.01.23 22:03, Rainer Fiebig wrote:
>> Hi! Since kernel 6.0.18
> 
> Note, 6.0.y is now EOL, you should move to 6.1.y.
> 
>> resume from hibernate fails, the system hangs
>> and a hard reset is necessary.  The CPU is a Ryzen 5600G, the system is
>> Linux From Scratch-11.1.
> 
> FWIW, there is a report with a problems that looks somewhat similar here:
> 
> https://bugzilla.kernel.org/show_bug.cgi?id=216917
> 
> But the reporter doesn't care anymore, as 6.1 works.
> 
> Ciao, Thorsten (wearing his 'the Linux kernel's regression tracker' hat)
Thanks for the info.  6.1.5 works here, too.

I clung to 6.0 as I use VirtualBox-6.1.38 and according to the changelog
it doesn't support linux-6.1.  VirtualBox-7.0.x supports linux-6.1 but
doesn't start on my system - and I couldn't yet figure out why.

But although the changelog for Virtual-Box 6.1.38 says otherwise, it
*does* support kernel 6.1 which was a nice surprise on Friday 13th. ;)

Rainer

Re: Kernel v6.0.18: Resume from hibernate fails, system hangs; bisected

[Greg KH]

On Thu, Jan 12, 2023 at 10:03:33PM +0100, Rainer Fiebig wrote:
> Hi! Since kernel 6.0.18 resume from hibernate fails, the system hangs
> and a hard reset is necessary.  The CPU is a Ryzen 5600G, the system is
> Linux From Scratch-11.1.
> 
> I found this in the system log:
> 
> [...]
> Jan 12 19:30:03 LUX kernel: [   50.248036] amdgpu 0000:30:00.0: [drm]
> *ERROR* [CRTC:67:crtc-0] flip_done timed out
> Jan 12 19:30:03 LUX kernel: [   50.248040] amdgpu 0000:30:00.0: [drm]
> *ERROR* [CRTC:70:crtc-1] flip_done timed out
> Jan 12 19:30:14 LUX kernel: [   60.488034] amdgpu 0000:30:00.0: [drm]
> *ERROR* flip_done timed out
> Jan 12 19:30:14 LUX kernel: [   60.488040] amdgpu 0000:30:00.0: [drm]
> *ERROR* [CRTC:67:crtc-0] commit wait timed out
> ^@^@^@^@^@^@^@^@^@^[...]@^@^@^@^@^@^@^@^@^@^@^@^@^@Jan 12 19:31:20 LUX
> syslogd 1.5.1: restart.
> [...]
> 
> 
> Bisecting the problem turned up this:
> 
> ~/Downloads/linux-stable-BLFS-11.1> git bisect bad
> 306df163069e78160e7a534b892c5cd6fefdd537 is the first bad commit
> commit 306df163069e78160e7a534b892c5cd6fefdd537
> Author: Alex Deucher <alexander.deucher@amd.com>
> Date:   Wed Dec 7 11:08:53 2022 -0500
> 
>     drm/amdgpu: make display pinning more flexible (v2)
> 
>     commit 81d0bcf9900932633d270d5bc4a54ff599c6ebdb upstream.

Can you resend this and cc: all of the people involved in this commit as
well as the drm mailing list?

Also, does 6.1.y work properly for you?  6.0.y is now end-of-life and
you shouldn't be using it anymore.

thanks,

greg k-h

Re: Kernel v6.0.18: Resume from hibernate fails, system hangs; bisected

[Rainer Fiebig]

Am 13.01.23 um 11:07 schrieb Greg KH:
> On Thu, Jan 12, 2023 at 10:03:33PM +0100, Rainer Fiebig wrote:
>> Hi! Since kernel 6.0.18 resume from hibernate fails, the system hangs
>> and a hard reset is necessary.  The CPU is a Ryzen 5600G, the system is
>> Linux From Scratch-11.1.
>>
>> I found this in the system log:
>>
>> [...]
>> Jan 12 19:30:03 LUX kernel: [   50.248036] amdgpu 0000:30:00.0: [drm]
>> *ERROR* [CRTC:67:crtc-0] flip_done timed out
>> Jan 12 19:30:03 LUX kernel: [   50.248040] amdgpu 0000:30:00.0: [drm]
>> *ERROR* [CRTC:70:crtc-1] flip_done timed out
>> Jan 12 19:30:14 LUX kernel: [   60.488034] amdgpu 0000:30:00.0: [drm]
>> *ERROR* flip_done timed out
>> Jan 12 19:30:14 LUX kernel: [   60.488040] amdgpu 0000:30:00.0: [drm]
>> *ERROR* [CRTC:67:crtc-0] commit wait timed out
>> ^@^@^@^@^@^@^@^@^@^[...]@^@^@^@^@^@^@^@^@^@^@^@^@^@Jan 12 19:31:20 LUX
>> syslogd 1.5.1: restart.
>> [...]
>>
>>
>> Bisecting the problem turned up this:
>>
>> ~/Downloads/linux-stable-BLFS-11.1> git bisect bad
>> 306df163069e78160e7a534b892c5cd6fefdd537 is the first bad commit
>> commit 306df163069e78160e7a534b892c5cd6fefdd537
>> Author: Alex Deucher <alexander.deucher@amd.com>
>> Date:   Wed Dec 7 11:08:53 2022 -0500
>>
>>     drm/amdgpu: make display pinning more flexible (v2)
>>
>>     commit 81d0bcf9900932633d270d5bc4a54ff599c6ebdb upstream.
> 
> Can you resend this and cc: all of the people involved in this commit as
> well as the drm mailing list?
CCed those involved in the commit but couldn't find the address of the
drm list.

> 
> Also, does 6.1.y work properly for you?  6.0.y is now end-of-life and
> you shouldn't be using it anymore.
Yes, a cursory test seems to indicate that 6.1.5 is OK, I'll be using
6.1 from now on.

Thanks.

Rainer Fiebig

Re: Kernel v6.0.18: Resume from hibernate fails, system hangs; bisected

[Rainer Fiebig]

Am 13.01.23 um 11:07 schrieb Greg KH:
> On Thu, Jan 12, 2023 at 10:03:33PM +0100, Rainer Fiebig wrote:
>> Hi! Since kernel 6.0.18 resume from hibernate fails, the system hangs
>> and a hard reset is necessary.  The CPU is a Ryzen 5600G, the system is
>> Linux From Scratch-11.1.
>>
>> I found this in the system log:
>>
>> [...]
>> Jan 12 19:30:03 LUX kernel: [   50.248036] amdgpu 0000:30:00.0: [drm]
>> *ERROR* [CRTC:67:crtc-0] flip_done timed out
>> Jan 12 19:30:03 LUX kernel: [   50.248040] amdgpu 0000:30:00.0: [drm]
>> *ERROR* [CRTC:70:crtc-1] flip_done timed out
>> Jan 12 19:30:14 LUX kernel: [   60.488034] amdgpu 0000:30:00.0: [drm]
>> *ERROR* flip_done timed out
>> Jan 12 19:30:14 LUX kernel: [   60.488040] amdgpu 0000:30:00.0: [drm]
>> *ERROR* [CRTC:67:crtc-0] commit wait timed out
>> ^@^@^@^@^@^@^@^@^@^[...]@^@^@^@^@^@^@^@^@^@^@^@^@^@Jan 12 19:31:20 LUX
>> syslogd 1.5.1: restart.
>> [...]
>>
>>
>> Bisecting the problem turned up this:
>>
>> ~/Downloads/linux-stable-BLFS-11.1> git bisect bad
>> 306df163069e78160e7a534b892c5cd6fefdd537 is the first bad commit
>> commit 306df163069e78160e7a534b892c5cd6fefdd537
>> Author: Alex Deucher <alexander.deucher@amd.com>
>> Date:   Wed Dec 7 11:08:53 2022 -0500
>>
>>     drm/amdgpu: make display pinning more flexible (v2)
>>
>>     commit 81d0bcf9900932633d270d5bc4a54ff599c6ebdb upstream.
> 
> Can you resend this and cc: all of the people involved in this commit as
> well as the drm mailing list?
> 
> Also, does 6.1.y work properly for you?  6.0.y is now end-of-life and
> you shouldn't be using it anymore.
5.15.87/88 (and perhaps other longterm kernels) are also affected.
Haven't bisected this time but as the symptoms and error messages are
the same it's fairly safe to assume that the cause is also the same.

6.1.5 is fine, though.

Rainer Fiebig