On Tue, Sep 20, 2016 at 9:10 AM, Christian König
> Am 20.09.2016 um 06:05 schrieb Deucher, Alexander:
>>> -----Original Message-----
>>> From: Michel Dänzer [mailto:mic...@daenzer.net]
>>> Sent: Monday, September 19, 2016 10:37 PM
>>> To: Huang, Ray; Deucher, Alexander
>>> Cc: email@example.com; Wang, Ken; Huan, Alvin
>>> Subject: Re: [PATCH 0/3] drm/amdgpu: implement raster configuration
>>> On 18/09/16 11:36 AM, Huang Rui wrote:
>>>> Hi all,
>>>> This patch set is to introduce the raster configuration for all gfx
>>> What's the motivation for doing this in the kernel? Mesa will write its
>>> own values to these registers anyway.
>> It's to fix the read register ioctl interface. The raster config
>> registers can be read back by the UMDs, but if the KMD hasn't programmed
>> them, they may have bogus values. It's mainly for the pro drivers which
>> read the raster config registers to get the baseline before programming
>> them. Mesa hardcodes the default raster config settings before harvesting.
>> We arguably should have done it in the kernel as well for mesa (my bad). At
>> least this way both sides are doing the right thing.
> Yeah, I wondered before as well why we don't do this in the kernel. What was
> the reason to move it into Mesa?
I just thought it made more sense to do it in mesa since it's part of
the 3D state and mesa needs to program it anyway (ideally we'd also
make adjustments to it to handle rendering to GTT and certain
crossfire configurations), but It makes more sense to put the base
harvest logic in the kernel driver.
> The patchset itself is Acked-by: Christian König <christian.koe...@amd.com>.
>> amd-gfx mailing list
> amd-gfx mailing list
amd-gfx mailing list