00:33airlied: Lyude, jani: https://lore.kernel.org/nouveau/20250429233929.24363-60-bskeggs@nvidia.com/T/#u either of you or point at anyone else that might be able to review that?
05:45DemiMarie: Under what conditions can the CSF FW of 10th-generation and later Arm Mali GPUs get into a bad state? Is it only when there is a firmware bug?
07:17daniels: yep
12:48zmike: enunes / anarsoul / anyone else from lima: can someone please give a look at the fails in https://gitlab.freedesktop.org/mesa/mesa/-/pipelines/1416643
13:48zmike: mupuf: https://gitlab.freedesktop.org/mesa/mesa/-/jobs/75556894
14:39eric_engestrom: zmike: for the lima fail, it's just that no dut was available within the deadline; it's either a farm infra issue, or a farm overload issue; either way the problem looks resolved as another job on that dut ran a few minutes ago: https://gitlab.freedesktop.org/mesa/mesa/-/jobs/75598440
14:39zmike: oh
14:39zmike: huh
14:40zmike: then maybe my current pipeline will pass
14:43eric_engestrom: I've retried the jobs, have a look :)
14:44zmike: I'm already running a pipeline, so probably that was unnecessary
14:44eric_engestrom: ack, cancelling mine
14:44zmike: thanks for investigating
14:44eric_engestrom: fwiw it looks like it fails `spec@ext_packed_depth_stencil@fbo-depth-gl_depth24_stencil8-blit`
14:45eric_engestrom: no idea if it's relevant to your change or if it's a flake
14:45zmike: the failing jobs I was seeing had a lot of different fails
14:46eric_engestrom: see https://gitlab.freedesktop.org/mesa/mesa/-/jobs/75599927; I cancelled it about 25% of the way through
14:46eric_engestrom: I see that `spec@ext_framebuffer_object@fbo-stencil-gl_stencil_index16-readpixels` failed too
14:47zmike: https://gitlab.freedesktop.org/mesa/mesa/-/jobs/75599495 and https://gitlab.freedesktop.org/mesa/mesa/-/jobs/75599494 seem to be failing just like I was seeing yesterday
14:56eric_engestrom: oh, then that's not the kind of failing in the job I looked at
14:56enunes: zmike: eric_engestrom: hi, I'll check it now
14:56eric_engestrom: so nothing I said applies :]
14:56zmike: enunes: awesome, thanks!
15:00enunes: zmike: looking at all previous jobs from the last days, I think this is only in your branch? and then it takes 1 hour to time out and starves the next pipeline of hardware to run jobs
15:00zmike: could be
15:01zmike: I only looked at the jobs which actually had runners
15:06enunes: so those crashes happen only with a badly created job submitted to the hardware, and currently each job like that will take 10s to time out which is why it takes so long
15:07zmike: oh huh
15:07enunes: so I guess we will need to look at the branch
15:07zmike: seems like
15:07enunes: any hint of what it may be?
15:08zmike: not really? I don't have any of that hardware
15:08zmike: but I've split out the lima changes into its own commit, so you should be able to see from there what is affected
15:37Lyude: airlied: yeah I can review that today
19:11cheako: See... it's not hard to get vulkan.org guides to be popular. https://github.com/Overv/VulkanTutorial/commit/d61a0f1e7c737baea17181e2904066ff04cacb1c