12:01tagr: skeggsb: any thoughts on this: https://patchwork.freedesktop.org/patch/323927/
12:02tagr: skeggsb: if that looks good to you, do you think you could give an Acked-by so that I can take it through drm-misc-next where it fixes a regression?
17:29mooch2: hey, since nvidia's now open-sourcing parts of their gpu documentation on github, who do you think i should ask at nvidia for a copy of their riva 128 documentation?
17:30ajax: i expect they'd have a difficult time locating it tbh
17:34mooch2: oh? why?
17:35ajax: why would they have kept it
17:35mooch2: maybe it's on one of their computers or something?
17:35mooch2: i dunno, man
17:35pmoreau: mooch2: You could ask on the mailing list they set up, or maybe ask tagr.
17:35mooch2: i just hope they did tho
17:36mooch2: tagr: hey, i'm somebody who's looking to add emulation of early nvidia cards to a PC emulator known as 86Box! I've tried to do so in the past, and only got as far as making Windows 3.11 send bad commands
17:36mooch2: i need better documentation in order to make everything work!
17:37mooch2: so, can you please check to see if you have any docs relating to the nv1, riva 128, and riva tnt please?
17:40ajax: you may find the docs in https://github.com/envytools/envytools/ to be useful
17:45ajax: huh, utahglx had an nv3 driver
17:46mooch2: ajax, i already looked through that
17:46mooch2: hell, i contributed to it even!
17:47ajax: heh, afraid i'm out of pointers then
17:48ajax: but like, i tried to ask intel for the i740 docs probably... eight years ago? a bit more?
17:48ajax: and they straight up didn't exist anymore
17:48ajax: so i wouldn't be surprised if the same was true of nv0X
17:52mooch2: oof ;-;
17:52mooch2: i wonder why companies so often throw out docs for old tech
17:53mooch2: it's hurting the preservation of this hardware!
18:03RSpliet: they never even finished their Win2000 driver, did they?
18:04mooch2: for the riva 128?
18:04mooch2: i dunno
18:04RSpliet: "A driver for RIVA 128 is also included in Windows 2000 and XP, but lacks 3D support. A beta driver with OpenGL support was once leaked by Nvidia but was canceled later, and there is no Windows 2000 driver for RIVA 128 on Nvidia's driver site today. Neither the beta driver nor the ones come with Windows 2000/XP could support Direct3D. "
20:18airlied: mooch2: they often archive old docs and designs to tape, which ends up in bins :-P
20:18mooch2: oof :c
20:21kodomo: Hi all! Nouveau-noob question: Do I get it correctly that the MX150 is not supported by nouveau? (I do not see it listed in the XOrg logfile output and when I compile it into the kernel, I get segfaults in dmesg + an EE-msg in Xorg.log)
20:21kodomo: More specifically: Failed to initialise context object: 2D_NVC0 (0)
20:24kodomo: [running 4.19.66-gentoo w/ XOrg 1.20.4 on a Lenovo T480s]
20:30gnarface: kodomo: consult feature matrix for answers: https://nouveau.freedesktop.org/wiki/FeatureMatrix/
20:31pmoreau: kodomo: Which chipset is the MX150? (If you run `lspci -d 10de:`)
20:33kodomo: gnurou: did - and did not find the MX150 mentioned... but the code table is a little cryptic - so I'm not a 100% sure I read it right...
20:33kodomo: pmoreau: 01:00.0 3D controller: NVIDIA Corporation GP108M [GeForce MX150] (rev a1)
20:35pmoreau: Should be supported, though buggy
20:35kodomo: pmoreau: ok - I didn't find it in https://nouveau.freedesktop.org/wiki/CodeNames/
20:36kodomo: But if it is, then it does indeed look buggy to me ^^
20:36pmoreau: Not all cards are there, because we don’t know of all of them
20:36pmoreau: It’s more chipsets that are or not supported, than individual cards
20:37pmoreau: (I’ll have a look in a bit)
20:37gnarface: GP108 is listed on there, but no GP108M
20:37gnarface: i wouldn't know if it matters
20:38kodomo: np - thank you very much - any input you'd like to have from my side?
20:39gnarface: i wonder if it's actually a NV108 or a GF108?
20:39gnarface: is that possible?
20:39gnarface: the "MX150" moniker sounds very old to me, too old to be actually from the Pascal family
20:40pmoreau: Mobile version is probably just a downcloked version
20:40gnarface: kodomo: typically i think the first thing they ask you to do is try it with a newer kernel, and then get some sort of (mmiotrace?) crash dump
20:41gnarface: i don't know if you can just use gdb or you need envytools for this
20:41kodomo: I don't know about chipset families and code names - but I am aware that there were two different variants of the MX150 (I've got the revised version... drastically reduced max power consumption, albeit slower)
20:42gnarface: well we're talking about like a 1-2 decade gap in time between the possible culprits going on sale, so the time and place you bought it should really answer the question i think
20:43gnarface: hell, just which type of motherboard slot it plugs into will probably answer the question
20:44kodomo: gnarface: t480s is (or at least was beginning this year) an active laptop line from Lenovo... It's most definitely from this decade (unless Lenovo has a highly twisted sense of humor ;) )
20:45gnarface: hmm. maybe it is Pascal after all then. weird they'd label it "MX150" though.
20:45kodomo: maybe this helps
20:46pmoreau: MX150 is a GT 1030 apparently
20:48pmoreau: There is this opened bug report, about an MX150 too: https://bugs.freedesktop.org/show_bug.cgi?id=108043
20:49kodomo: basic 2D seems to be marked WIP for NV130 - and that sounds a bit like it connected to (one of) the error message(s) I got (the above one)... so that would sound plausible to me.
20:50pmoreau: (Those tables aren’t necessarily kept up-to-date)
20:52kodomo: Also: When I do 'startx -- :2', I do get a black screen... which reverts to the console after a while and then informs me that it was not possible to connect to the X server... so I do feel like my situation may connect to that bug report, as well.
20:52kodomo: So I won't get my hopes up on running the MX150 with nouveau, I guess...
20:53pmoreau: Could you please share the output of dmesg for you? Also, which kernel are you running?
20:54kodomo: [running 4.19.66-gentoo w/ XOrg 1.20.4 on a Lenovo T480s]
20:54gnarface: kodomo: that's not a dual-gpu laptop, is it?
20:55pmoreau: Hum, I’ll need to double-check whether any fixes were merged since 4.19.
20:55kodomo: I can paste /var/log/messages and /var/log/Xorg output. But I'd need first to recompile the kernel to include nouveau. Having nouveau in the kernel broke suspend functionality I rely upon - so I took it out, again...
20:56kodomo: gnarface: it is...
20:56gnarface: kodomo: that could be a source of additional problems. it might not have even been using the nvidia gpu during your previous tests.
20:57gnarface: kodomo: i'm pretty sure it's a bad idea to statically compile the module in either way
20:58kodomo: gnarface: ack about the static inclusion.
20:59gnarface: there is a thing you can try to enable the nvidia gpu called "DRI_PRIME"
20:59kodomo: w.r.t. the 'used' GPU: The working X definitely ran under Intel HD620 - as did the console (though I ignore whether the console isn't agnostic... it should be in my mind)... I created an Xorg config file to manually launch a second X session to experiment
21:01gnarface: DRI_PRIME support is hit and miss, but i think in this situation it is a step you're missing, so it is worth a try in case it works (unless someone here knows that it specifically won't work with that card - i don't actually know)
21:04kodomo: looking for DRI_PRIME...
21:04kodomo: ...I do find CONFIG_VGA_SWITCHEROO, which I have not enabled... I could still try *scratch*
21:05gnarface: i think DRI_PRIME is just an environment variable for userspace, but you might still need CONFIG_VGA_SWITCHEROO too
21:07kodomo: Either way, I'm going to enable additional MMU debugging for my next attempt... maybe I'll see something enlightening...
21:10kodomo: I'll give it another shot tomorrow when I find a minute for it
21:10kodomo: Thank you both for answering
21:10gnarface: good luck with it
21:11gnarface: these guys are really busy but if you hang around long enough they eventually respond
21:12pmoreau: I saw some new commits fixing issues with firmwares, for Pascal, but they aren’t part of any release yet.
21:12kodomo: so it's worthwile trying from time to time? Do they go into 4.19? Or only subsequent kernels?
21:13gnarface: my guess is 5.x now
21:13pmoreau: They should be in 5.2-rc5 though, so you could try compiling it.
21:13pmoreau: No idea whether they were backported or not.
21:13kodomo: ack - will do that as a subsequent step