APPLIED [O] Re: NACK: [PULL][O]: drm/xe: Backport support for LNL/BMG from 6.12

Timo Aaltonen tjaalton at ubuntu.com
Wed Sep 4 12:56:48 UTC 2024


Thibault Ferrante kirjoitti 4.9.2024 klo 14.42:
> 
> On 03-09-2024 18:46, Timo Aaltonen wrote:
>>
>>
>> This is the first batch of enabling new/upcoming Lunar Lake / 
>> Battlemage in the oracular kernel. There are two derivative kernels 
>> that need this, so having it in the generic kernel makes sense.
>> > Nearly all clean cherry-picks, and all from linux-next.
>>
>>
>> -- 
>>
>> BugLink: https://bugs.launchpad.net/bugs/2076209
>>
>> [Impact]
>>
>> Intel Lunar Lake & Battlemage GPU's don't work in 24.10.
>>
>> [Fix]
>>
>> Backport commits from linux-next to 6.11 in two batches, the second 
>> one dropping xe.force_probe requirement.
> 
> Should we accept this patchset before having the second one ? Or should 
> we wait the second part ?

I wouldn't want to leave it all at the last minute... and have it 
running locally on LNL and it works fine as far as I can see.

>>
>> [Test case]
>>
>> Boot the new kernel, check from dmesg that the drm driver is 
>> initialized correctly.
>>
>> [Where problems could occur]
>>
>> The backport is mostly confined in drm/xe, which isn't used by default 
>> on any hardware before LNL/BMG, so chance of regressions should be 
>> limited. Also, this backport is validated by the vendor.
>>
>>
>> ----------------------------------------------------------------
>> The following changes since commit 
>> 431c1646e1f86b949fa3685efc50b660a364c2b6:
>>
>>    Linux 6.11-rc6 (2024-09-01 19:46:02 +1200)
>>
>> are available in the Git repository at:
>>
>>    https://git.launchpad.net/~tjaalton/linux lnl
>>
>> for you to fetch changes up to 24a51165336d3fb3effa66ef84ce4cd414694cf0:
> As I see, it's 372d041d7b06821dc5ff063ab04bf797b8349747 the tip of lnl. 
> I believe 24a51165336d3fb3effa66ef84ce4cd414694cf0 is from a previous 
> version.

Indeed, it got mangled a bit more for submission after starting to write 
the email.

>>
>>    drm/xe: Fix total initialization in xe_ggtt_print_holes() 
>> (2024-09-03 17:40:19 +0300)
>>
>> ----------------------------------------------------------------
>> Akshata Jahagirdar (5):
>>        drm/xe/migrate: Handle clear ccs logic for xe2 dgfx
>>        drm/xe/migrate: Add helper function to program identity map
>>        drm/xe/xe2: Introduce identity map for compressed pat for vram
> Contains:
> (backported picked from commit 2b808d6b2919cb2fe92901e5087da7b4ed4b9e07 
> linux-next)
> [ Conflicts:
>          drivers/gpu/drm/xe/tests/xe_migrate.c ]
> 
> 
> It has a typo in the backported line and the explanation isn't one.
> Maybe reworded as:
> (backported from commit 2b808d6b2919cb2fe92901e5087da7b4ed4b9e07 linux- 
> next)
> [ Conflicts in drivers/gpu/drm/xe/tests/xe_migrate.c
>    due to skipping multiple refactor commits mainly targeting drm/xe/ 
> kunit ]

Yes there's a typo, the "backported..." was mine, but the Conflicts are 
from upstream, so dunno if we should be changing those, but I did amend 
it anyway since you already had a suggestion :)

with these changes I've applied this.

-- 
t




More information about the kernel-team mailing list