Re: acpi issues on FreeBSD-current_r350103 on Thinkpad A485

Hey,

thanks for the feedback and your work, didn't think this would be 
interesting for anyone but the laptop owners.


On ds., jul. 20 2019, Greg V. wrote:

> On July 20, 2019 1:54:47 AM GMT+03:00, Evilham 
> <contact_at_evilham.com> wrote:
>
>> it even suspends and resumes back to X
>
> Wow, that's great news! Desktop Ryzen+Vega doesn't (not that I 
> need suspend very much on desktop haha)

Indeed, I was pleasantly surprised too

>>- xbacklight doesn't work, neither does intel-backlight because
>>  it's AMD
>
> Since it's a Thinkpad, do the brightness keys work anyway? Does 
> acpi_ibm work?

Forgot to mention brightness keys, they also don't work.
I just tried acpi_ibm and it also didn't help.

>>Serious issue:
>>I was just debugging this right now, more infos with a proper 
>>bug
>>report will come, but I think the system encounters a deadlock
>>sometimes with the drm-kmod / amdgpu which results in a kernel
>>panic.
>
> If you're on the packaged drm-kmod v4.16, it's amazing that 
> Raven GPU works at all. You should try drm-v5.0 from git.
>
>>kld_list="amdgpu"

Huh, interesting, I'm trying to compile drm-v5.0 right now.
This comment and Pete's made me re-visit the graphics settings, 
specifics below.


> It even works when loaded this early? Interesting. Do you also 
> not have the EFI framebuffer conflict? i.e. without disabling 
> vt.syscons, everything just works reliably?

Yes, with my previous laptop that was necessary, but this one has 
no need for those settings, it works just fine without and font 
size is according to the screen (i.e. it's not huge).



On ds., jul. 20 2019, Pete Wright wrote:

> On 7/19/19 3:54 PM, Evilham wrote:
>>
>> Serious issue:
>> I was just debugging this right now, more infos with a proper 
>> bug
>> report will come, but I think the system encounters a deadlock
>> sometimes with the drm-kmod / amdgpu which results in a kernel 
>> panic.
>> It is a serious issue, but it allows me to use the computer for 
>> work,
>> it doesn't happen every couple hours, but it does happen a 
>> couple
>> times a day.
>>
>> FWIW, this is part of the crashlog:
>>
>> WARNING !drm_modeset_is_locked(&plane->mutex) failed at
>> /wrkdirs/usr/ports/graphics/drm-fbsd12.0-kmod/work/kms-drm-6365030/drivers/gpu/drm/drm_atomic_helper.c:821
>> [Multiple times...]
>> kernel trap 22 with interrupts disabled
>>  kernel trap 22 with interrupts
>>  disabled
>> kernel trap 22 with interrupts disabled
>> kernel trap 22 with interrupts disabled
>>  panic: spin lock held too long
>>
>
> interesting. can you post this kernel panic, and any backtraces 
> you are
> able to get here:
>
> https://github.com/FreeBSDDesktop/kms-drm/issues
>
> also, are you using the xf86-video-amdgpu driver, or the stock
> modesetting driver to X?


That was my plan for when I manage to fully isolate that it is on 
drm-kmod, thank you for confirming the path!

I noticed I did have xf86-video-amdgpu installed, but just removed 
it and all traces of drm-kmod as well as the kld_list="amdgpu" bit 
and X actually works without it and without the xf86-video 
packages.

HDMI output won't work though, I guess that makes sense.

We'll see if working like this the system doesn't crash at all, if 
it does it may be related to #231760 and not to drm-kmod.

If it does not crash without drm-kmod, I'll try my build of 
drm-kmod v5.0 before opening an issue, that'd be more information 
:-).

I did have the question as to why the packaged version is 4.16 but 
didn't quite find an answer to that on the wikis / documentation.

Thank you again,
--
Evilham