[PATCH 0/2] Fix GPU screen handling for drivers with no normal screens

Aaron Plattner aplattner at nvidia.com
Thu Feb 12 13:21:51 PST 2015


On 01/05/2015 09:20 AM, Alex Deucher wrote:
> On Tue, Dec 30, 2014 at 12:13 PM, Aaron Plattner <aplattner at nvidia.com> wrote:
>> I noticed the problem fixed by the first patch while debugging a problem with
>> the Intel driver in a system using Prime.  See that patch for a description of
>> what is going wrong there.  Unfortunately, Prime configurations are relying on
>> that loophole because they need a GPU screen for the Intel device, and the X
>> server refuses to create one unless there's also a normal X screen using that
>> same driver.  The second patch fixes that by allowing drivers with no normal X
>> screens to have GPU screens created for them anyway.  In most cases, this will
>> end up using the modesetting driver for any platform devices that are not
>> claimed by another driver.
>>
>> Given the above, please consider these patches together.
>
> I'm not too familiar with the code in question, but the logic seems
> sound and the patches make sense.  Series is:
>
> Acked-by: Alex Deucher <alexander.deucher at amd.com>

Thanks Alex!

Now that the merge window is open, I'd like to try to get these into 
xserver 1.18 if possible.

>> Aaron Plattner (2):
>>    xfree86: Fix xf86_check_platform_slot's handling of PCI
>>    xfree86: Add GPU screens even if there are no active GDevs
>>
>>   hw/xfree86/common/xf86platformBus.c | 20 +++++++++++++++-----
>>   1 file changed, 15 insertions(+), 5 deletions(-)
>>
>> --
>> 2.2.1

-- 
Aaron


More information about the xorg-devel mailing list