Re: Alder lake supported? (graphics)

On 2024-01-18 15:42, Chris wrote:
> On 2024-01-18 15:28, Chris wrote:
>> On 2024-01-18 08:47, Jan Beich wrote:
>>> Chris <bsd-lists@bsdforge.com> writes:
>>> 
>>>> On 2024-01-16 19:02, Jan Beich wrote:
>>>> 
>>>>> Chris <bsd-lists@bsdforge.com> writes:
>>>>> 
>>>>>> I upgraded to an alder lake based machine and installed 14.
>>>>>> But I can't seem to get the intel graphics loaded (drm-515-kmod).
>>>>>> It simply freezes at load.
>>>>>> Are Alder lake graphics supported?
>>>>> Try drm-61-kmod instead (with gpu-firmware-intel-kmod-alderlake >=
>>>>> 20230625).
>>>>> Reported success in
>>>>> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=270888#c8
>>> [...]
>>>> I'm on 14. Commenting that conditional indicates I don't have a necessary
>>>> file (linux/iosys-map.h). So looks like I'll we'll have to wait. Or I'll
>>>> need to track 15. :(
>>> 
>>> On current@ list -CURRENT is expected. Due to backward compatibility it's
>>> possible to run -CURRENT kernel with -RELEASE userland (world + packages).
>>> For example, poudriere (as used by the package cluster) relies on this
>>> to build binary packages for older FreeBSD versions on the same machine.
>>> 
>>> Alternatively, https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=274770#c5
>>> proposed a branch which removes .require_force_probe for both ADL-S and 
>>> ADL-P.
>>> It builds fine on 14.0-RELEASE "as is" e.g., see ports/ diff below.
>>> 
>>> diff --git a/graphics/drm-515-kmod/Makefile.version 
>>> b/graphics/drm-515-kmod/Makefile.version
>>> index 4a7c27611bc8..469071220731 100644
>>> --- a/graphics/drm-515-kmod/Makefile.version
>>> +++ b/graphics/drm-515-kmod/Makefile.version
>>> @@ -1,5 +1,5 @@
>>>  # drm-kmod common version definition
>>>  #
>>>  # This will be included from consumers such as nvidia-drm
>>> -DRM_KMOD_DISTVERSION=	5.15.118
>>> -DRM_KMOD_GH_TAGNAME=	drm_v5.15.118_4
>>> +DRM_KMOD_DISTVERSION=	5.15.focal
>>> +DRM_KMOD_GH_TAGNAME=	97a4ad4364
>>> diff --git a/graphics/drm-515-kmod/distinfo 
>>> b/graphics/drm-515-kmod/distinfo
>>> index 3599fc42317b..cadc6be14456 100644
>>> --- a/graphics/drm-515-kmod/distinfo
>>> +++ b/graphics/drm-515-kmod/distinfo
>>> @@ -1,3 +1,3 @@
>>> -TIMESTAMP = 1703336317
>>> -SHA256 (freebsd-drm-kmod-5.15.118-drm_v5.15.118_4_GH0.tar.gz) =
>>> 58e2fc195979e2361346ca57cc158e44413e5de26b83b951a631d09849caf90c
>>> -SIZE (freebsd-drm-kmod-5.15.118-drm_v5.15.118_4_GH0.tar.gz) = 26092371
>>> +TIMESTAMP = 1698298780
>>> +SHA256 (freebsd-drm-kmod-5.15.focal-97a4ad4364_GH0.tar.gz) =
>>> fc6a94a74aea714bb25ccf788b8361de4db348ef1893fc391d00bd346e828732
>>> +SIZE (freebsd-drm-kmod-5.15.focal-97Thank you very mua4ad4364_GH0.tar.gz) 
>>> = 26126042
>> Thank you very much for all your time and efforts in your replies, Jan.
>> I examined the the forum thread and applied your patch to a current pull of 
>> ports.
>> After deinstalling my current install of 515.118, I performed a make 
>> install. Which
>> proceeded as expected. A kld_list resulted in a hard lock, in the same way 
>> 515.118 did.
>> Requiring the use of the power button to power it down. Then a boot to 
>> single-user for
>> an fsck(8). Nothing reported in the logs. A boot without loading via 
>> rc.conf was went
>> without incident. A kldload i915kms locked it up hard. Again, nothing 
>> reported in the
>> logs. I'm keeping recent copies of /var/logs/messages in case there is 
>> anything anyone
>> might be interested in looking at.
>> At this point I've resolved to boot a recent 15 and format my current 
>> slices and unpack
>> a recent snapshot of 15 in hopes of getting decent graphics support on 
>> this. I'm happy
>> to test or further report anything that might be helpful for others.
>> 
> I forgot to mention. When calling kldload i915kms
> the screen returned only (transcribed):
> 
> iic0: < I2C generic I/O > iicbus0
> iic1: < I2C generic I/O > iicbus1
> Lindebugfs registered
> 
UPDATE
It's not i915kms that causes the problem. Turns out to be the 
graphics/gpu-firmware-intel-kmod.
I opened a pr (https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=276451).

Thanks again for all your help, Jan.

--Chris