07:00 Sid127: goob morning, funny people in my terminal
07:54 Sid127: love waiting for the kernel to compile
07:54 Sid127: # :(
08:03 fdobridge_: <t​om3026> do you use modprobed-db ? helps drasticly when it comes to bisecting like this
08:04 Sid127: I do
08:05 Sid127: it's ~250 modules, and I have a quad core skylake
08:06 Sid127: and the official arch pkgbuild is doing something funny after the kernel's done compiling e-e
08:09 Sid127: something something htmldocs, methinks
08:09 Sid127: I'll rip it out later
08:09 fdobridge_: <t​om3026> im gonna begin the bisect when nouveau makes my vkcube timeout instantly rc1 <-> rc4 is fine so time to approach upwards then bisect the broken rc :p
08:10 Sid127: I put the job timeout bug on the burner, because I even ran into it on rc1
08:11 Sid127: currently chasing an mmu fault
08:11 Sid127: that showed up somewhere between rc2 and rc3
08:39 fdobridge_: <S​id> I forgot to unset ccache max size
08:39 fdobridge_: <S​id> so my ccache dir was maxed out at the default 5 gigs
08:39 fdobridge_: <S​id> effectively ccache wasn't being used e-e
08:41 fdobridge_: <S​id> I'm so smart guys
08:59 fdobridge_: <S​id> is there a way I can only build the nouveau kernel module and live patch it 🐸
09:00 fdobridge_: <S​id> I mean, there should be
09:00 fdobridge_: <S​id> but I think that tangent will take me too much time to resolve
09:04 fdobridge_: <t​om3026> cool rc8 vkcube works, external monitor works. 6.7.0 doesnt
09:04 fdobridge_: <t​om3026> so its the last release doing something hrm
09:08 fdobridge_: <S​id> oof
09:08 fdobridge_: <S​id> I'm gonna take a short break for lunch and unwind a bit
09:57 fdobridge_: <t​om3026> @sid what size is your /sys/kernel/debug/dma_pools/pool_size_kernel on 6.7?
09:58 fdobridge_: <t​om3026> found out i can get rid of that trace at boot with setting coherent_pool=1M at boot for some reason they go to the type max without it being set, still bisecting nouveau tho 😛
09:58 fdobridge_: <S​id> gimme a moment, just rebooted for a new kernek build
10:00 fdobridge_: <S​id> 4194304
10:00 fdobridge_: <t​om3026> okay so uhm 4mb or similiar then i think
10:00 fdobridge_: <t​om3026> thanks
10:41 fdobridge_: <t​om3026> the heck, cant reproduce it now. is there some arch patches or config changes thats breaking it runs on self built 6.7 but as fast as i boot arch repo 6.7 it timeouts vkcube or freezes when external monitor is attached hrm
10:44 fdobridge_: <S​id> hang on
10:45 fdobridge_: <S​id> https://github.com/archlinux/linux/releases/download/v6.7-arch3/linux-v6.7-arch3.patch.zst
10:45 fdobridge_: <S​id> try applying this patch on top
11:51 fdobridge_: <S​id> ...I might be dumb
11:51 fdobridge_: <S​id> I spent so long bisecting, but forgot to test rc3 without arch patches first
12:05 fdobridge_: <r​hed0x> are there still any low hanging fruit to get into NVK development?
12:07 fdobridge_: <S​id> you could look at the issues tagged NVK
12:07 fdobridge_: <S​id> see if anything looks easy enough
12:10 fdobridge_: <b​ylaws> Advanced blend could kinda be fun
12:10 fdobridge_: <b​ylaws> I think it's pretty okay to do on Nvidia?
12:10 fdobridge_: <b​ylaws> Not like tilers where it's all shader side
12:15 fdobridge_: <e​sdrastarsis> maybe fix zink issues (Super Tux Kart and TF2)
12:28 fdobridge_: <S​id> I just wasted all day
12:28 fdobridge_: <S​id> no mmu fault on rc3
12:47 fdobridge_: <t​om3026> yeah ive wasted all day too, just cant find whats breaking on 6.7 😄
12:47 fdobridge_: <t​om3026> i bet its some race so it just occurs randomly or not
12:48 fdobridge_: <t​om3026> or UB
12:48 fdobridge_: <t​om3026> maybe 6.7.1 magically fixes it
12:49 fdobridge_: <S​id> did you try applying that arch patch on top?
12:49 fdobridge_: <t​om3026> yeah and guess what it works lol
12:49 fdobridge_: <S​id> ...wut
12:49 fdobridge_: <S​id> wtf
12:50 fdobridge_: <t​om3026> exactly, and i just have to reboot to the arch package and its instantly broken so just because i compiled it myself? different gcc flags? idk
12:50 fdobridge_: <S​id> possibly
12:50 fdobridge_: <S​id> what defconfig are you using?
12:50 fdobridge_: <t​om3026> the one linux-mainline on aur is shipping
12:51 fdobridge_: <S​id> so arch linux's 6.6.1 defconfig
12:51 fdobridge_: <S​id> hm
12:51 fdobridge_: <S​id> could you try with this one here
12:52 fdobridge_: <S​id> https://gitlab.archlinux.org/archlinux/packaging/packages/linux/-/blob/main/config?ref_type=heads
12:54 fdobridge_: <t​om3026> yeah maybe later feels like ive been compiling for the past 8 hours xD
12:54 fdobridge_: <S​id> okie
12:54 fdobridge_: <S​id> let me know if you run into your bug with that defconfig
13:04 fdobridge_: <S​id> `./diffconfig linux-mainline-defconfig linux-arch-defconfig`
13:04 fdobridge_: <S​id> https://cdn.discordapp.com/attachments/1034184951790305330/1198251791918305380/diffconfig?ex=65be39f1&is=65abc4f1&hm=f53b8e776e74407960883c4525a166b76d5876e8f24532e24457cd9106f21a1b&
13:10 fdobridge_: <S​id> currently building arch's kernel without their patches to see if I run into the mmu fault there
15:45 fdobridge_: <S​id> ok I am not dumb!
15:45 fdobridge_: <S​id> or maybe only partly
15:45 fdobridge_: <S​id> got an mmu fault on 6.7 after clearing the shader cache, not before though
17:00 fdobridge_: <S​id> hm, GPL MR doesn't apply on master cleanly anymore *again*
19:37 fdobridge_: <S​id> ok I give up getting this issue to reproduce consistently
19:37 fdobridge_: <S​id> shader cache cleared, no mmu fault again
20:24 fdobridge_: <k​arolherbst🐧🦀> mhhh... guess you either hit a shader cache collision or something isn't set up properly...
20:25 fdobridge_: <k​arolherbst🐧🦀> would be good to know if you could reproduce it between two commits, then we can probably check waht's going wrong there... but mhh.. not quite sure what it could be
20:44 fdobridge_: <S​id> thing is
20:45 fdobridge_: <S​id> I've hit it across multiple kernel builds, but rather inconsistently
20:45 fdobridge_: <S​id> the only surefire way to get it to happen is to try to launch any game using faith's nvk/pipelines branch
20:46 fdobridge_: <S​id> this is what I'm running into btw
20:46 fdobridge_: <S​id> ```
20:46 fdobridge_: <S​id> [Fri Jan 19 11:01:05 2024] nouveau 0000:01:00.0: gsp: mmu fault queued
20:46 fdobridge_: <S​id> [Fri Jan 19 11:01:05 2024] nouveau 0000:01:00.0: gsp: rc engn:00000001 chid:24 type:31 scope:1 part:233
20:46 fdobridge_: <S​id> [Fri Jan 19 11:01:05 2024] nouveau 0000:01:00.0: fifo:001001:0003:0018:[Metal.exe[82910]] errored - disabling channel
20:46 fdobridge_: <S​id> ```
20:47 fdobridge_: <k​arolherbst🐧🦀> does it happen with `MESA_SHADER_CACHE_DISABLE=1`? Though I think you need a debug build for that to work
20:48 fdobridge_: <k​arolherbst🐧🦀> ehh
20:48 fdobridge_: <k​arolherbst🐧🦀> apparently it doesn't matter
20:48 fdobridge_: <S​id> uhhhhhhhhhhh I'd check but I just booted up stardew valley e-e
20:48 fdobridge_: <k​arolherbst🐧🦀> I'm going to bed now anyway :ferrisCozy:
20:49 fdobridge_: <S​id> that makes two of us :D
20:49 fdobridge_: <k​arolherbst🐧🦀> 😄
20:49 fdobridge_: <S​id> I'm nice and cozy in my blanket, got my dualsense with me, laptop's to the side on sdv's main menu
20:50 fdobridge_: <S​id> purple and yellow fairy lights in my room
20:50 fdobridge_: <S​id> you know the vibe :)
20:50 fdobridge_: <k​arolherbst🐧🦀> nice
20:50 fdobridge_: <S​id> 0220, just winding down
20:50 fdobridge_: <k​arolherbst🐧🦀> I don't have fancy lights yet, though I should change that
20:50 fdobridge_: <S​id> I could grab a quick video and pop it into #the-unreal-world