03:13 sneil: hakzsam: GALLIUM_HUD="inst_executed" glxgears shows the graph and inst_executed = 0 (never varies). On the console I get:
03:13 sneil: gallium_hud: all queries are busy after 8 frames, can't add another query
03:13 sneil: (repeated probably once per frame)
03:15 sneil: that's git eeff7e11 minus d03bfb078 (I didn't think it mattered to your question so I haven't reset the checkout yet)
03:46 sneil: dboyan_: (yup, GM107 is correct). I tried 17.0.0 and still desktop freeze, before that I'll need to figure out a build issue (probably LLVM is too new)
06:28 Manoa: just played a wine game on nouveau for like 6 hours, not a single bug glich or crash :)
06:28 Manoa: windows: RIP 2017 :)
06:28 ozmage: ?
06:28 ozmage: i have this
06:28 ozmage: HexChat: 2.10.0 ** OS: Linux 3.2.0-4-486 i686 ** Distro: Debian 7.11 ** CPU: 1 x Intel(R) Pentium(R) M processor 1600MHz (GenuineIntel) @ 1.60GHz ** RAM: Physical: 502.5MiB, 54.9% free ** Disk: Total: 32.3GiB, 70.7% free ** VGA: NVIDIA Corporation NV28 [GeForce4 Ti 4200 Go AGP 8x] ** Sound: ICH4 - Intel 82801DB-ICH4 ** Ethernet: None found ** Uptime: 1h 45m 5s **
06:29 ozmage: and all i see is garbage on the screen
06:29 ozmage: distorted
06:29 ozmage: graphics
06:29 ozmage: any one have a fix for this ?
06:29 Manoa: 7.11 debien :) that's nice
06:29 ozmage: with nv ddx and nvidia leagcy works
06:31 Manoa: garbage when running 2D work on nouveau DDX ?
06:32 Manoa: you could try to disable 2D acceleration
06:32 Manoa: I think xorg.conf parameter is noaccel "true" inside the "device" section
06:33 Manoa: someone here might know how to help but you may need to provide some debug information like Xorg.0.log and dmesg log
06:38 ozmage: ok
06:38 ozmage: i would try later
06:38 ozmage: someone toll me that opensuse 13 works
06:38 ozmage: with nouveau and that gfx card
06:38 ozmage: ok
06:40 Manoa: hey 4200 Ti is a real classy
06:40 Manoa: I didn't expect for sutch a old card to survive for this long
09:15 DocMAX: is there a channel for nvidia module?
09:15 DocMAX: i want high resolution on console
13:03 imirkin: DocMAX: #nvidia
13:30 hakzsam: sneil: compute is broken on your card
13:30 hakzsam: skeggsb: do you remember this compute issue on gm107? I reported something similar a while back
13:31 hakzsam: sneil: I think you need to record a mmiotrace with the NV blob and send it to skeggsb
14:31 karolherbst: skeggsb: 108b1c41f002a3b3b1f262bd78be9a63f6bd1683 needs an upstream change in envyas :p
14:33 karolherbst: otherwise the statics are always removed again
14:37 quesada: ‎ Hi guys. If nouveau is not picking a monitor (xrandr says disconnected) is this a matter of xorg conf?
14:37 quesada: the monitor is connected phisically because is the onboard one for the laptop and the OS has been spitting stuff to it during booting :)
14:40 hakzsam: karolherbst: did you get the trace in the end?
14:41 karolherbst: hakzsam: ohh actually yes, let me check
14:41 hakzsam: ok
14:42 karolherbst: hakzsam: I see nothing
14:43 hakzsam: https://bugs.freedesktop.org/attachment.cgi?id=130271
14:43 hakzsam: you don't see this?
14:43 hakzsam: which is random
14:43 karolherbst: window is simply black
14:43 karolherbst: or maybe the compilation of shaders just needs a lot of time
14:44 karolherbst: anyway, the window is simply black
14:44 hakzsam: karolherbst: can you upload somewhere the framebuffer image at call 9482971?
14:44 karolherbst: ohhh now something apears
14:45 karolherbst: okay
14:47 quesada: NVIDIA GK106M [GeForce GTX 765M] if that matters
14:47 quesada: optimus laptop
14:48 karolherbst: hakzsam: okay, looks fine if I enable "opaque"
14:49 hakzsam: http://imgur.com/a/HXL09
14:49 hakzsam: radeonsi
14:49 karolherbst: hakzsam: https://i.imgur.com/NGlKAJ8.jpg
14:50 karolherbst: shouldI reflip it? :D
14:50 hakzsam: thanks
14:50 hakzsam: so it works fine with nouveau
14:50 karolherbst: yeah
14:50 hakzsam: cool
14:51 karolherbst: I suppose something is wrong witht he alpha part
14:51 karolherbst: ;)
14:52 hakzsam: some sort of decompression failures I think
14:53 hakzsam: which could explain why the output is random
14:53 karolherbst: maybe
14:53 karolherbst: or simply the data wasn't cleared?
14:53 karolherbst: because it's full alpha here
14:54 quesada: ‎ I know optimus is problematic overall. I don't need any fancy switching, just the screens working
14:54 karolherbst: quesada: reverse prime
14:54 quesada: karolherbst: let me look that up
14:55 karolherbst: and "fancy switching" is done by setting DRI_PRIME for an application ;)
14:55 hakzsam: karolherbst: can you upload with only "alpha"?
15:00 karolherbst: yeah
15:05 karolherbst: hakzsam: https://i.imgur.com/yGVpLeB.png
15:05 quesada: bumblebee is an abandoned project. The guys at Solus have a working replacement that they are actively working on. Solus only for now, but worth looking at it. They seem to be killing it. https://github.com/solus-project/linux-driver-management
15:06 hakzsam: karolherbst: thks
15:22 pmoreau: imirkin: Quick question as I am starting to cleanup the SPIR-V -> NVIR code: should I rename `nv50_ir_from_spirv.cpp` to `nv_ir_from_spirv.cpp`? As we seem to be transitionning to calling that, as it is not NV50 specific.
15:39 karolherbst: uhh that meson guy is here :O
17:01 quesada: I have an optimus. but xrandr --listproviders gives me only one provlder, called nouveau. Why?
17:10 imirkin: pmoreau: either way's fine
17:10 imirkin: quesada: i'm confused by your situation. as i'm lazy, could you summarize real quick what you're trying to achieve and what you're seeing? also pastebin dmesg and xorg logs.
17:16 quesada: imirkin: one sec
17:16 quesada: trying to have the laptop screen working. So far only external works
17:16 quesada: https://wiki.archlinux.org/index.php/PRIME only one output provider listed by xrandr
17:16 quesada: pastebin coming
17:17 imirkin: your internal screen is, most likely, attached to the intel iGPU, and is not switchable to nvidia
17:18 imirkin: my guess is that you've arranged things so that the nvidia gpu is used as the main one?
17:18 imirkin: if you have a custom xorg.conf, that would also be useful to see.
17:22 quesada: imirkin: https://pastebin.com/SA33u3ZH
17:26 quesada: imirkin: https://pastebin.com/9b3vkqyi
17:26 imirkin: that seems less happy...
17:26 quesada: yep
17:26 quesada: that is /var/log/Xorg.2.log
17:27 imirkin: you said X was starting though, no?
17:32 quesada: imirkin: yes
17:32 quesada: I'm on x
17:32 quesada: that may be old?
17:34 imirkin: well, i need the log form the running X instance.
17:34 imirkin: i believe you're on a distro infected by systemd, so i can't really help you with how to do that.
17:47 quesada: imirkin: arch
17:47 quesada: imirkin: http://sprunge.us/KITA
17:48 quesada: how do I connect that deactivated DP-1?
18:13 imirkin: quesada: chances are by plugging your laptop into the dock
18:13 imirkin: quesada: your internal display will be on either LVDS-1 or eDP-1
18:13 imirkin: you likely did something clever in your xorg.conf which is leading to your present predicament
18:13 imirkin: if you remove all the customization you did, chances are everything will work fine out of the box.
18:20 waylander32: http://www.nvidia.com/content/gtc-2010/pdfs/2238_gtc2010.pdf all read how that guy defined my rant
18:33 waylander32: you may do those synchronized jumps and bans around me, but any sane one when explained also knows, you name me troll, and say i tell lies and scam which i never have done, and you have never one a debate against me
18:37 Tom^: waylander32: and you think ranting and trying to harass this channel and its users is gonna win you any sort of acknowledge or level of acceptence at all? , if you do. you simply have lost already.
18:41 waylander32: i wish dx11 state tracker to be worked on again, wine has some late code additions
18:55 waylander32: Tom^: no, i do not want any sort of acceptance but i want my rights to live in my country, and function without regulations on my personals life, what were settled with utter scams and fabrications, if that never happens i will use force
18:55 imirkin: waylander32: that's not what this channel is about. you go on these long rants, and no one here cares. hence the bans.
19:02 waylander32: but i have at least proven that i can think and conclude based of available information, there are lots who care about my situation, but just maybe not here on that channel in fact
19:03 waylander32: but i want you to invest your time and skills on for doing something collaboratively productive and sane, instead of having a ban paranoia
19:04 imirkin: i wish i could as well
19:12 waylander32: i won't stand in your way, if you want me to leave i will, but some goals to target, is a big dx11 one, natively for gamers to linux, there are yet some more things to do, computer work never ends
19:17 waylander32: fixing glamor, and also basically an open gpu firmware comes to mind, all it is some quite a major reimplementing work, good luck on such missions for instance, you could make channels for upcoming new from ground up designed gpus too
19:17 waylander32: bye.
19:17 librin: good evening!
19:18 imirkin: and a good day to you, sir!
19:19 librin: appears I've hit a nouveau-related regression, but it's confusing and I'm not sure if it's nouveau's fault per se
19:22 librin: I've got a 120Hz-capable monitor, but after doing a long overdue reboot, it now flickers and has messed up contrast/brightness if I set it to any mode other than 60Hz
19:23 Tom^: librin: what cable are you using?
19:23 Tom^: 60+ requires displayport or dual link dvi
19:23 librin: dual link DVI
19:24 librin: I mean, it worked fine for years on Nouveau
19:24 librin: and I know it's a software issue, as I got my hands on windows (yuck) and it works just fine there with whatever mode, including still doing stereoscopic 3D just fine
19:25 librin: I tried downgrading my libdrm, mesa, xf86-video-nouveau and xrandr packages down from git HEAD to whatever revision way before I hit this, along with booting an older kernel that I had previously ran for months
19:26 librin: but that didn't help
19:26 librin: so I am confused
19:26 librin: looks like a regression, but downgrading everything that I can think of that could have caused this doesn't make it go away
19:26 librin:scratches head
19:32 librin: nothing out of ordinary in Xorg.0.log
19:33 librin: save for the constant spam of this one line, but it doesn't seen related at all:
19:33 librin: input-thread: InputThreadDoWork waiting for devices
19:34 librin: nothing out of ordinary in kernel logs, either
19:34 librin: Tom^: any ideas?
19:34 librin: I'm out of ideas where / what to look at
19:34 librin: :C
19:34 Tom^: hm no :/, maybe imirkin or the other gurus know
19:37 karolherbst: librin: verify that you load the older stuff
19:42 karolherbst: librin: you could also try the modesetting ddx
19:42 librin: karolherbst: well, I did reboot to make sure new "old" stuff gets loaded. After all, I only started seeing this after a reboot, so I assume at the very least I need to restart X
19:42 librin: how do I do that?
19:43 karolherbst: librin: uninstall xf86-video-nouveau
19:43 imirkin: librin: sounds like something unrelated is causing this
19:43 imirkin: librin: perhaps you now have some other piece of software that is causing lots of output reads
19:43 imirkin: er
19:43 imirkin: output status reads
19:43 imirkin: which might cause blinking
19:43 librin: nothing I'm aware of, though
19:44 imirkin: well, if you downgraded everythign you could think of
19:44 imirkin: then clearly the issue remains in something you didn't think of :)
19:45 librin: I don't remember installing anything new for quite a while, heck, hardly had the chance to used this computer at all
19:46 librin: hmmm, I'll try to "use" modesetting and see if it works
19:46 librin: now, to just figure out how to uninstall xf86-video-nouveau without gentoo, err, funtoo screaming at my face about dependencies
19:46 librin: :D
19:47 librin: karolherbst, imirkin: thanks for the help!
19:49 librin: oh! I forgot to mention that right after switching from more-than-60Hz to 60Hz mode, it still flickers in 60Hz mode, albeit not as much. Then the flicker gradually disappears over a course of a few minutes.
19:50 librin: the 60Hz doesn't get the contrast/brightness corruption at all, though.
19:50 librin: mode, I mean. 60Hz mode.
20:00 librin: errr, scratch that, I seemed to have had a brain fart and messed up on the simple procedure I've done many times before and the older revision packages I've built are in fact git HEAD
20:00 librin: ./punch.sh --target=self --face
20:01 librin: now I am going to *actually* revert
20:01 librin: B(
20:04 librin: welp, time test out
20:04 librin: see Ya!
20:32 pmoreau: imirkin: Any reason to keep NV50_PROGRAM_IR_GLSL & co around, instead of using PIPE_IR_GLSL? Also the ifdef'ed code line 1234 in codegen/nv50_ir.cpp? That code seems to have been merged as-is, and left commented out for the past 6 years…
20:35 imirkin: feel free to nuke it
20:35 imirkin: i believe calim and zhasha were working on the dx10/11 stuff in a branch, and it was for that
20:36 pmoreau: Ok
20:36 pmoreau: Will send a few patches then :-)
20:36 imirkin: i'd rather make a sm4/5 -> tgsi adapter than have a direct parser for that stuff
20:36 imirkin: but at that time, tgsi was nowhere close to representing those things
21:50 imirkin: pmoreau: your 3/4 didn't make it to my inbox... at least not yet
21:51 pmoreau: I did received it… so maybe in a few seconds/minutes?
21:52 pmoreau: It’s in the list archive, and on patchwork as well
21:57 imirkin: ok. probably just delayed
21:59 imirkin: ah there it goes
21:59 pmoreau: Here is the link to patchwork, if you want: https://patchwork.freedesktop.org/patch/154840/
22:00 pmoreau: Anyway, thanks for the quick reviews :-)
22:00 imirkin: np. do you have a fd.o account btw?
22:00 pmoreau: I do not
22:01 pmoreau: I was waiting for more than a one-liner patch before asking for one :-)
22:01 imirkin: hm yeah. you have like 11 commits at this point. fewer than i thought.
22:02 pmoreau: Yes, I have been mostly hacking in my corner
22:02 imirkin: wow. i have a lot more than i thought. i guess it adds up after a while.
22:03 pmoreau: With the clover serie, I should have enough commits to ask for the games
22:03 imirkin: hehe
22:03 pmoreau: Well, one patch here, one patch there, it can add up quickly :-)
22:03 imirkin: hopefully not the reason you're working on it - there'd be easier ways to get that
22:03 pmoreau: You have been contributing quite a bit to Nouveau on the Mesa-side
22:04 pmoreau: I think I have most of the Valve games already, probably a couple of the Feral ones
22:04 pmoreau: So no, I am not doing it for the games
22:04 imirkin: like 150 freedreno commits too. crazy. i don't remember doing those.
22:04 pmoreau: Ah ah ah
22:05 pmoreau: Someone impersonating you for those commits then? :-p
22:05 imirkin: must be.
22:05 pmoreau: Let’s revert all of them
22:05 imirkin: hehe. rob would be real happy.
22:06 pmoreau: I am sure he would :-D
22:13 karolherbst: pmoreau: how is your Nouveau image thing going by the way? I might need it with a few modifications in august
22:14 pmoreau: karolherbst: Badly… I haven’t fixed it yet in a while, even though it shouldn’t be that hard. I should just need to bump the glibc requirement for valgrind, and get a new PKGBUILD for Mesa.
22:14 pmoreau: karolherbst: I will definitely get back to it before August.
22:15 pmoreau: I should get back to the patch for converting to/from 64-bit integers, and send it along with a patch for folding 64-bit int constants.
22:16 pmoreau: IIRC the int 64 extension was merged, wasn’t it?
22:16 pmoreau: karolherbst: Which modifications would you need? I won’t make them right now, but at least write them down to remember them.
22:19 imirkin: pmoreau: yeah, int64 is in
22:19 imirkin: pmoreau: and supported by nouveau. i took your multiplication change.
22:20 pmoreau: Yup, I recall that
22:20 pmoreau: (that you took the multiplication patch)
22:20 pmoreau: Which should have ringed a bell that Nouveau supported it now, and that int64 support got merged
22:34 karolherbst: pmoreau: prime stuff
22:34 karolherbst: pmoreau: bumblebee + nouveau dual setup for reverse engineering
22:34 karolherbst: stuff like that
22:41 librin: ah, nope, even after reverting properly this time, with crazy amounts of verification that I in fact did, still the same issue
22:41 librin: hmmm...
22:42 librin: maybe I should try modesetting after all
22:54 librin: is there a way to switch to modesetting ddx without having to remove the nouveau one? e.g. like an xorg.conf switch
23:00 karolherbst: librin: removing it is the easiest way
23:13 librin: welp, here goes nothing
23:14 librin: thanks again and see Ya!