16:48 <q4a> My 1st independent MR to Panfrost) https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20285

16:48 <alyssa> reading it now :-D

16:52 <alyssa> q4a: I agree that this fixes a real issue. I am trying to figure out whether there's a better fix.

16:54 <q4a> Cool. I found and fixed bug + my fix is not so bad)

16:54 <alyssa> trying to figure out what the actual gallium contract is here

16:55 <q4a> https://gitlab.freedesktop.org/mesa/mesa/blob/main/docs/gallium/context.rst ?

16:55 <alyssa> because I see different drivers doing different things

16:55 <alyssa> q4a: that docs says what we're doing now is correct

16:56 <alyssa> * :ref:`Sampler`: Texture sampler states are bound separately for fragment, vertex, geometry and compute shaders with the ``bind_sampler_states`` function. The ``start`` and ``num_samplers`` parameters indicate a range of samplers to change. NOTE: at this time, start is always zero and the CSO module will always replace all samplers at once (no sub-ranges). This may change in the future.

16:56 <alyssa> so clearly the docs are wrong :-)

17:00 <alyssa> yeah I genuinely can't tell what we're supposed to do

17:02 <q4a> write more tests or ask someone with other driver to test similar change?

17:02 <alyssa> well, the issue is what the actual contract is

17:02 <alyssa> if the Gallium docs are authoritative, then it is nine that is buggy, not panfrost

17:04 <q4a> alyssa: we need to ask David?

17:09 <alyssa> asked in #dri-devel, let's move discussion there because this isn't really specific to panfrost

17:11 <q4a> thanks