On Fri, Jun 30, 2017 at 12:40 PM, Lionel Landwerlin
<lionel.g.landwer...@intel.com> wrote:
> On 30/06/17 17:14, Alex Deucher wrote:
>>
>> On Fri, Jun 30, 2017 at 12:02 PM, Eric Anholt <e...@anholt.net> wrote:
>>>
>>> Alex Deucher <alexdeuc...@gmail.com> writes:
>>>
>>>> On Thu, Jun 29, 2017 at 11:42 AM, Eric Anholt <e...@anholt.net> wrote:
>>>>>
>>>>> I want to remove vc4's dependency on headers from libdrm as well, but
>>>>> storing multiple copies of drm_fourcc.h in our tree would be silly.
>>>>>
>>>>> v2: Update Android.mk as well, move distcheck drm*.h references to
>>>>> top-level noinst_HEADERS.
>>>>>
>>>>> Reviewed-by: Lionel Landwerlin <lionel.g.landwer...@intel.com>
>>>>> Reviewed-by: Daniel Stone <dani...@collabora.com>
>>>>> ---
>>>>> Makefile.am | 4 ++++
>>>>> {src/intel/drm => include/drm-uapi}/README | 0
>>>>> {src/intel/drm => include/drm-uapi}/drm.h | 0
>>>>> {src/intel/drm => include/drm-uapi}/drm_fourcc.h | 0
>>>>> {src/intel/drm => include/drm-uapi}/drm_mode.h | 0
>>>>> {src/intel/drm => include/drm-uapi}/i915_drm.h | 0
>>>>> src/intel/Android.vulkan.mk | 2 +-
>>>>> src/intel/Makefile.am | 1 -
>>>>> src/intel/Makefile.drm.am | 22
>>>>> ----------------------
>>>>> src/intel/Makefile.sources | 6 ------
>>>>> src/intel/Makefile.vulkan.am | 2 +-
>>>>> src/mesa/drivers/dri/i965/Android.mk | 4 ++--
>>>>> src/mesa/drivers/dri/i965/Makefile.am | 2 +-
>>>>> 13 files changed, 9 insertions(+), 34 deletions(-)
>>>>> rename {src/intel/drm => include/drm-uapi}/README (100%)
>>>>> rename {src/intel/drm => include/drm-uapi}/drm.h (100%)
>>>>> rename {src/intel/drm => include/drm-uapi}/drm_fourcc.h (100%)
>>>>> rename {src/intel/drm => include/drm-uapi}/drm_mode.h (100%)
>>>>> rename {src/intel/drm => include/drm-uapi}/i915_drm.h (100%)
>>>>> delete mode 100644 src/intel/Makefile.drm.am
>>>>
>>>>
>>>> I don't mean to pick on this patch specifically, but maybe it would
>>>> still make sense to depend on libdrm for the drm headers? If not do
>>>> we want similar restrictions on updating these as we have for libdrm?
>>>
>>> Yes, we certainly have the same restrictions on updating headers (pull
>>> only things that have landed in airlied's drm-next, or possibly
>>> drm-misc-next if acked by airlied) as libdrm does. That's
>>> "participating in kernel DRM development" rules, not libdrm rules.
>>
>> I'm not arguing about the fact that stuff has to land in drm-next,
>> etc. first, but there are pretty strict additional requirements as to
>> how they are updated:
>> https://cgit.freedesktop.org/mesa/drm/tree/include/drm/README
>> Do we want to enforce similar requirements in mesa?
>
>
> I think that's the idea, and what's in the README file of this commit.
> Maybe we need to update to to actually put the url of airlied's tree?
>
>>
>>> I don't think it makes sense to depend on libdrm if all you're using
>>> From libdrm is the header that you can just put in the tree.
>>
>> I though we agreed that libdrm was supposed to be the canonical source
>> for these headers in userspace. It just seems like we are going to
>> end up with a proliferation of these headers as various projects
>> decide to include them directly.
>
>
> I guess most driver developers didn't have an opinion at the time or didn't
> pay attention to the patch introducing
> https://cgit.freedesktop.org/mesa/drm/tree/include/drm/README
>
> Here is a pointer to the discussion with a different set of people, with
> Eric's answer which makes the best argument for this (in my opinion) :
>
> https://lists.freedesktop.org/archives/mesa-dev/2017-May/154491.html
I don't necessarily agree with the rules for updating the headers in
libdrm. I agree with Eric's statement that they should always be
compatible. I like to be able to update the UAPI headers along with
the libdrm patch that uses them. That said, I get yelled at every
time I try and do that, so I feel like we should apply that equally or
get rid of it.
Alex
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
