On 1 December 2016 at 12:00, Nicolai Hähnle <nhaeh...@gmail.com> wrote:
> Congratulations on a huge amount of work! Obviously I can't say much about
> the driver itself. Some things that I noticed for the renderonly library.
> On 30.11.2016 14:44, Christian Gmeiner wrote:
>> This a very lightweight library to add basic support for
>> renderonly GPUs. It does all the magic regarding in/exporting
>> buffers etc. This library will likely break android support and
>> hopefully will get replaced with a better solution based on gbm2.
> Some more comments would be _really_ helpful. What is the purpose of a
> "scanout" object? What does for_prime vs. for_resource mean? What does
> intermediate_rendering mean and what is it good for?
> The lifecycle of the renderonly object itself looks wrong to me:
> renderonly_screen_create calls the actual driver's screen_create, but the
> actual driver's screen_create frees the renderonly struct? Please make it
> consistent: Either have a proper wrapper or (better, since this is so
> lightweight) have the driver's screen_create call the
> renderonly_screen_create.
I think Nicolai has it spot on here - both on the comment and wrapping side.
Please include some documentation (even the GBM2 snippet from the
cover letter) here. Be that in the summary and/or code.

mesa-dev mailing list

Reply via email to