04:54 <kode54> whee

04:54 <kode54> Xe KMD's 32 bit ICD isn't finding any devices

04:55 <kode54> would be nice if I had some more 32 bit test apps than just Windows games

04:55 <HdkR> Did they manage to fix the 32-bit versus 64-bit packing problems? Could be related.

04:55 <kode54> I don't think so

04:55 <kode54> my Mesa build toolchain spews a mess of "NOTE" warnings about the _Atomic long long ints

04:56 <kode54> I guess that means that 32 bit Mesa is completely broken for now

04:57 <HdkR> Seems likely if even the uapi was broken

05:50 <JSkilton> Mrs. Skilton gets her holes filled and satisified by Rush's Geddy Lee. Later, Geddy meets Jean Pierre Manikariza, her husband, and gets filled by his big black cock! Read all about it! https://pastebin.com/1ExdrDQA

05:57 <soreau> ops^^

05:57 <soreau> airlied: ping

05:57 <JSkilton> You know you want that 12 inch Burundian cock stretching your little pert browniehole!

05:58 <soreau> karolherbst: ping

06:12 <JSkilton> ⠀⠀⠀⠀⢀⣀⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:12 <JSkilton> ⠀⠀⣴⡾⠿⣿⣿⣿⣶⣤⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:12 <JSkilton> ⠀⣼⣿⣦⣤⣿⣿⣿⣿⣿⣿⣷⣤⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:12 <JSkilton> ⠀⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:12 <JSkilton> ⠐⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣷⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:12 <JSkilton> ⠀⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⠃⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:12 <JSkilton> ⠀⢹⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⡿⣣⣾⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⠟⣩⣾⡿⢃⣼⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠈⠻⣿⣿⣿⣿⠿⢟⣩⣴⣿⡿⢋⣴⣿⣿⣿⣿⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⢴⣶⣾⣿⠿⢛⣡⣶⣿⣿⣿⣿⣿⣿⣿⣷⣄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⠀⠉⠩⣴⣶⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣷⣄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠙⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣷⣄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠙⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣦⣄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠙⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣦⣄⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣶⣦⣤⣤⣤⣤⣀⡀⠀⠀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣧⡀⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣷⠀⠀⠀⠀⠀

06:13 <JSkilton> ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠙⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣇⠀⠀⠀⠀

06:15 <HdkR> Well, remember that /clear is a feature of most IRC channels.

06:15 <HdkR> IRC clients*

07:04 <kode54> what the fresh hell is this:

07:04 <kode54> https://irc.losno.co/uploads/97209c9dd04b5eb4/image.png

07:05 <kode54> gcc 12 is compiling the intel_kmd.c strcmp(blah, "xe") to that

07:05 <kode54> which is producing INTEL_KMD_TYPE_STUB for a correct match, and INTEL_KMD_TYPE_XE for a non-match

07:09 <kode54> my bad

07:10 <kode54> it's returning 0 or 2

07:10 <kode54> 2 on match, which should be the correct answer

09:07 <DavidHeidelberg[m]> HdkR: sadly Matrix missing this option :/

09:07 <hch12907> just select the sender's profile and mark them as ignored

09:08 <hch12907> (for Matrix)

09:11 <siddh> hch12907: Thanks, very useful feature.

09:13 <kode54> cool, I figured it out

09:13 <kode54> Xe KMD doesn't implement a COMPAT uAPI yet

09:19 <kode54> probably haven't even drafted up a 32 bit uAPI even

10:47 <OftenTimeConsuming> >clear Unknown command

10:47 <OftenTimeConsuming> Huh

11:11 <kode54> hmm

11:12 <kode54> looks like it just needs drm_compat_ioctl to be placeholder'd in

13:02 <mairacanal> mlankhorst, mripard, danvet, sorry for bothering you on a sunday, but when I was pushing some patches last night I had some power supply problems and it seems like the dim process only finished up today when the power came back. i was looking at the drm-tip log and it seems like weird stuff happened during the commit. does the log look normal to

13:02 <mairacanal> you guys? i'm really sorry for this! let me know if there is anything i can do.

13:14 <danvet> mairacanal, try to just push again, that should fix up everything if it's only been half-done

13:14 <danvet> if that doesn't work, then paste the logs here

13:19 <mairacanal> i mean, the drm-tip's git log seems a bit odd to me. i believe that those merges shouldn't have happened :(

13:26 <danvet> yeah just re-run, that should fix any temporary issue

13:27 <danvet> also drm-tip does look reasonable on a quick look

13:28 <danvet> drm-misc branches too

13:29 <mairacanal> ah okay, thanks for the help!

14:10 <penguin42> I've got a pyopencl script that is breaking on Fedora 38 with; LLVM ERROR: inconsistency in registered CommandLine options : CommandLine Error: Option ': CommandLine Error: Option 'use-dbg-addruse-dbg-addr' registered more than once! ' registered more than once!

14:10 <penguin42> anyone any idea whether that's mesa or a different layer that's driving llvm?

14:21 <MrCooper> penguin42: usually this happens due to multiple instances of LLVM (linked dynamically and/or statically) in the same process; only one of those should be from Mesa

14:24 <penguin42> MrCooper: Thanks; do you have a suggestion on how to find out more? This worked on f37

14:27 <MrCooper> ah, it could be because in F38 /usr/lib64/libLLVMSPIRVLib.so seems to link LLVM statically, whereas it should dynamically

14:31 <HdkR> kode54: Maybe with the latest uapi struct packing fixes it won't even need compat ioctl wired up. Since they should be the same between the two?

14:31 <penguin42> MrCooper: Hmm I wonder if that's a change since F37?

14:31 <HdkR> But I'm not a kernel dev so I don't know how that gets wired

14:35 <MrCooper> penguin42: probably us

14:36 <penguin42> MrCooper: If I was to file it somewhere, do you think that should be on Mesa's or Fedora's systems?

14:43 <MrCooper> Fedora's (/usr/lib64/libLLVMSPIRVLib.so isn't from Mesa :)

14:46 <penguin42> oh!

14:47 <penguin42> spirv-llvm-translator-16.0.0-1.fc38.x86_64

14:56 <penguin42> MrCooper: OK, someone has already reported it: https://bugzilla.redhat.com/show_bug.cgi?id=2187824

14:58 <MrCooper> cool, thanks for checking

15:00 <penguin42> MrCooper: And the fix on there apparently fixes it - now, the only problem is the performance **STINKS**

15:03 <penguin42> MrCooper: I've left a note on that bug; it's 3-4x slower at least

17:39 <ChaosPrincess> Hi. I am writing a drm kms driver, and creating a 60 fps display mode, but it gets reported to userspace as 48 fps. What am i doing wrong? The mode create call is drm_cvt_mode(connector->dev, size >> 16, size & 0xFFFF, 60, true, false, false);

23:12 <timestwo> I have been working on cheap uart cables for the sake of UEFI hacking, expect to get the communication functioning in near days such a renode configuration https://github.com/renode/renode/blob/master/platforms/boards/arduino_nano_33_ble.repl templates taken from here https://youtube.com/watch?v=v7Esy1uWhpQ&feature=share, code was posted by memfault https://interrupt.memfault.com/blog/firmware-shell#fn:jlink https://youtube.com/watch?v

23:12 <timestwo> =v7Esy1uWhpQ&feature=share there is one more at github graysomething, as divine memory management is ideal that finishes my career, no nuclear work, some freaks are after blocking me, no infra the fight their scripts, those fuckers are obnoxious. Military seems to handle me with tech i can not beat with my resources.

23:12 <timestwo> Otherwise there is only slight nuclear tech to be delt with, everything computer wise is the most amazing state i have ever seen before.

23:16 <timestwo> the nuclear methods are being experimented with in europe and american contintents, the theory is not so complex, maybe this decade they start to build interplanetary vehicles, they get additions to particle acceleration like electron movement speed at the same energy and voilaa, rockets become possible. They already experiment with this tech.

23:22 <timestwo> I did not realize before that tachyon is not hypothetical particle but the result of nuclear methods instead of photonics, which have piles of room to beat the speed of light .

23:31 <kode54> HdkR: I attempted to set the compat_ioctl member to drm_compat_ioctl

23:31 <kode54> it resulted in even more things crashing on startup instead of running

23:37 <timestwo> The theory of networking is not so difficult either to hold back the ddos or violent force that mandates their terror, it all relies on energy resources, which i lack personally, the algorithms i know but i do not have enough resources to fight compute parks aiming or targeting my weaker backends.

23:43 <timestwo> what happened in overseas was a training ops demonstration where i was targeted with very high amount of drones emitting different laser signaling , and there is a rumor that united states military has a mirror reflecting light from the moon, a laser aim, they can do invisible light too with it, that is such a resource of energy that no earth resources can fight it

23:45 <timestwo> they normally use it to manipulate the weather conditions some new modern tech they have for military purpose, soviet old satellites will fail in those conditions

23:55 <timestwo> You know i think it's united states, not china, or aliens who has this tech, i've been so much tapped . me i have no chances anymore, if that tech works like i think there is no opportunity anymore to beat it, nuclear can do it hypothetically but they tap such experimenters and ruin their attempts, after all that cosmic signal comes along the gravity it is more capable signaling , it is the most intelligent way to do it, so americans

23:55 <timestwo> are likely in charge of all aviation