02:29 gfxstrand[d]: Assigned SM32 to Marge. :transtada128x128:
04:03 gfxstrand[d]: Ancient panfrost is timing out. Looks like we've got to wait for Europe to wake up before we'll get SM32 merged.
04:03 gfxstrand[d]: Shouldn't be too many hours.
06:59 gfxstrand[d]: Okay, so shared memory was working on Friday...
07:29 blisto[d]: Sharing is caring
07:57 airlied[d]: running cts for 2.5 hours on blackwell Pass: 4259, Fail: 3010, Crash: 2635, Skip: 11146, Timeout: 1, Missing: 208, Flake: 741, Duration: 2:40:11, Remaining: 340:51:44
07:57 airlied[d]: not seeing any illegal instructions, just a bunch of Z warnings, mmu faults and general channel hangs
08:03 kitsukonagi: yes, it's not linuxes fault that much, i went over all the dma controllers drivers, and they are fine in dmaengine, it's just mobile intel chipsets do not have dma controllers, only pci/pcie, spdk is for desktops too, and jtag i am not sure about, so i better maintain my numbers and cryptography ways.
08:57 gfxstrand[d]: *grumble grumble*
08:58 gfxstrand[d]: I really wish I could figure out WHY it's faulting when I set shared memory bits.
08:58 gfxstrand[d]: Like, what bit of hardware is trying to access null and failing?!?
09:10 gfxstrand[d]: Okay, I need to do a full sweep of the blob for this. It looks like tables changed or HW rules changed or something
15:44 toenutofiro: I have accessed 3 consecutive indexes , and working with fourth, but i was not entire on llvm cause i did not want to share anything special, I do not want to disencourage people not to work or encourage to give up, cause actually it works in certain ways, just the clue is you have to use algortihmically reference values, and interpolate around them getting to delta that way, i had went
15:44 toenutofiro: over the sigma repetitions already, but i need to quit talking here with dead people. I am amazed how much terror abortion leftovers can do in the world.
15:45 snowycoder[d]: gfxstrand[d]: What tables does the blob driver have?
15:56 gfxstrand[d]: snowycoder[d]: Mapping from shared memory sizes to HW reservation sizes.
17:09 hentai: Can someone enlighten me, what exactly are the firmwares in /lib/firmware/nouveau?
17:10 hentai: Do they have any function apart from video decoding?
18:11 gfxstrand[d]: Yeah, the GPU won't work without them.
18:12 gfxstrand[d]: They handle everything from command submission to command parsing for some chips to video to power management and can control.
18:13 gfxstrand[d]: *fan
18:14 mhenning[d]: are you sure? isn't most firmware stuff under /lib/firmware/nvidia ?
18:14 mhenning[d]: I'm not actually sure what goes in /lib/firmware/nouveau
18:14 gfxstrand[d]: Oh, maybe I misread that.
18:15 gfxstrand[d]: I don't know what goes in nouveau/, either. Maybe that's where the open R/E'd firmwares live?
18:16 gfxstrand[d]: But the ones in nvidia/ are pretty essential.
18:17 gfxstrand[d]: (Sorry. I'm working on like 4h of sleep. I should take a nap.)
20:55 airlied[d]: oh vkcube doesn't die at least, just spins a black cube
21:13 leftmostcat[d]: gfxstrand[d]: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34372 has been bitrotting; I can close it, break it up, or update it. Do you have a preference?
21:13 gfxstrand[d]: I'll look in a few hours
21:13 leftmostcat[d]: No rush at all.
21:14 gfxstrand[d]: Nah, it's cool. This is a good week to review stuff like that.
21:15 gfxstrand[d]: And writing rustdoc, apparently.
21:15 snowycoder[d]: I'll also drop a very small MR when you have time: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34940 .
21:15 snowycoder[d]: It's in preparation of Kepler image storage
22:42 gfxstrand[d]: snowycoder[d]: I like the idea. I'm just not sure how I feel about the execution. Like, I really like the idea of a better way of converting the struct into a `u32` safely. I just don't know that I really like the decl-style macro. 😕
23:07 jja2000[d]: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34914 huh, interesting stuff
23:13 gfxstrand[d]: leftmostcat[d]: Can you rebase? There's some merge conflicts
23:22 leftmostcat[d]: gfxstrand[d]: Oh, yep, sorry, misunderstood. Thought you were gonna take a look to see if it was worth resolving. Will do that now.
23:24 gfxstrand[d]: Most of it looks good. I'm reading it right now.