ChanServ changed the topic of #etnaviv to: #etnaviv - the home of the reverse-engineered Vivante GPU driver - Logs https://oftc.irclog.whitequark.org/etnaviv
pcercuei has quit [Quit: dodo]
karolherbst has quit [Read error: Connection reset by peer]
karolherbst has joined #etnaviv
sravn has joined #etnaviv
pivi has quit [Quit: leaving]
pivi has joined #etnaviv
pcercuei has joined #etnaviv
pcercuei has quit [Read error: Connection reset by peer]
fomys has quit [Remote host closed the connection]
fomys has joined #etnaviv
<marex> austriancoder: I noticed 'MESA: error: Unhandled NIR tex src type: 3' when running deqp gles3 tests with mesa 25.0.0, do you know what that is about ?
<austriancoder> marex: happens with a txf nir instruction?
<marex> austriancoder: I dont know
pcercuei has joined #etnaviv
<marex> Test case 'dEQP-GLES3.functional.shaders.texture_functions.texture.sampler2darray_fixed_vertex'.. Pass (Pass)
<marex> Test case 'dEQP-GLES3.funcMESA: error: Unhandled NIR tex src type: 3
<austriancoder> ohh
<marex> I think I need to hand-build mesa here ... is there some trick to this ?
<marex> I mean ... hand-build mesa in debian rootfs for easy debugging and easy use with deqp that is
<marex> last time I tried to follow the readme, it wasn't as easy as I hoped it would be, so went back to OE builds, but that is slower
<austriancoder> apt-get build-dep mesa
<marex> that is exactly the readme I followed ... never got even kmscube to use the new mesa, so shrug
<marex> is there any way to make etnaviv somehow gracefully proceed instead of tearing down the entire application with it in case some error shows up?
<austriancoder> marex: meson devenv -C builddir kmscube .. it can not be simpler then this
<austriancoder> marex: remove the abort() from compile_error(..) macro - that's all you can do
<marex> why is it there in the first place ? should that be in release builds ? wouldn't it be better to keep it only in debug builds ?
<marex> re devenv, noted, thanks
<austriancoder> marex: there was not a super nice interface to report compiler errors back in mesa until 2-3 weeks. But you are using an unsupported feature that is not exposed to normal users. Btw. should you not see the information you want in the deqp result file?
<austriancoder> marex: as I slowly adding more devices for ci testing it might make sense to add a deqp3 run for e.g. gc7000 based one. would be nice to track progress.
<marex> austriancoder: this is mx8mp/gc7000
<marex> austriancoder: I am trying to figure out the state of gles3 support
<austriancoder> marex: I have two or three mx8mp's planed for gles3 and opencl testing