03:02marcus: hi all. what exactly is modesettings ddx about? can it be activated by using nomodeset boot parameter?
03:08pq: marcus, it is generic Xorg display driver, which depends on kernel modesetting, so nomodeset would disable it.
03:12marcus: is there a way to force the usage of this driver instead of nouveau?
03:16pq: marcus, in your xorg.conf
03:16RSpliet: marcus: may I ask: why would you want such a thing? having trouble with nouveau?
03:17pq: it won't replace Nouveau...
03:17pq: it will just use Nouveau's KMS and Nouveau's Mesa GL acceleration, instead of the hand-written acceleration routines.
03:20pq: marcus, what we're saying is, if you have trouble with Nouveau now, it is unlikely that the modesetting ddx would solve any of them.
03:42marcus: it has been suggested to me to do so: https://bugs.freedesktop.org/show_bug.cgi?id=90932#c1
04:15pq: ah, right then
06:18imirkin: pq: for maxwell, no exa yet
06:19pq: is there ever going to be, even?
06:20imirkin: i have patches which get it 99% of the way there
06:20imirkin: just being really lazy about the last 1%
06:20imirkin: i don't think ben wants there to be though
06:29imirkin: marcus: you can either uninstall xf86-video-nouveau or use Driver "modesetting" in the device section of you xorg.conf
06:50imirkin: skeggsb_: btw, you're not *seriously* waiting for me to review your libdrm patches, right? i've forgotten about them long long ago. besides, hakzsam is probably a better person to review them since he's actually used the functionality
06:51hakzsam: I'll try to have a look this week but I'm not sure If I'm the best person to do it ;)
06:52imirkin: best doesn't have to be good... just better than everyone else :p
07:18hakzsam: hehe :)
11:34jsabeaudry: I just switched to nouveau and noticed my lcd backlight does not turn off, the screen goes blank the the backlight is still on
11:34imirkin_: jsabeaudry: laptop or desktop?
11:35jsabeaudry: imirkin_, laptop
11:35imirkin_: eDP panel?
11:36jsabeaudry: possibly, how can I tell?
11:36imirkin_: run 'xrandr'
11:37jsabeaudry: yup eDP!
11:37imirkin_: old kernel?
11:37imirkin_: bleh. no, that's pretty new.
11:37imirkin_: a while ago we didn't have dpms support on DP
11:38imirkin_: can you force dpms? i.e. 'sleep 1 && xset dpms force off'
11:38jsabeaudry: Screen goes blank but lcd backlight remains on
11:39jsabeaudry: The brightness control however work fine and I can turn off the backlight that way
11:40jsabeaudry: if I lower it all the way
11:40imirkin_: well, putting the screen to sleep is supposed to be a separate operation
11:41imirkin_: file a bug, including a boot with nouveau.debug=VBIOS=trace,I2C=debug,DISP=debug,DRM=debug drm.debug=0xe on the kernel cmdline
11:41imirkin_: not just a boot, but including that 'xset dpms force off' deal
13:20mupuf: imirkin_: well put imirkin_!
13:20imirkin_: about the horizon?
13:20mupuf: "best does not have to mean good"
13:20mupuf: yeah, I'm late to the game
13:21mupuf: very busy month! Working 30h/week on learning finnish
13:21imirkin_: how's that going?
13:22mupuf: Great, but I would already be able to speak in any other language but Finnish. Finnish is ... weird
13:22mupuf: but I like it
13:22mupuf: see, you can get your citizenship without even knowing how to form the plural in all cases
13:23mupuf: and if you like learning 10 times the same word, you would love finnish :D
13:23imirkin_: not just an 'add s' situation?
13:23RSpliet: plural is for greedy people
13:23mupuf: nope :D
13:23mupuf: the future is also interesting ... there is none!
13:23imirkin_: russian (and i suspect all slavic languages) has 2 plurals
13:23imirkin_: one for 2-4 items, and one for 5+
13:24mupuf: ah ah, fun
13:24mupuf: they have different words to say getting in, being in, getting out
13:24mupuf: and actually, that would be 6 words, because they differentiate between being into and on
13:25mupuf: and no way of knowing if you go inside the shop or onto the shop. You just need to know
13:25imirkin_: and 100 diff words for snow i'm sure
13:25imirkin_: like english has for rain
13:25mupuf: but if it is part of a sentence (the usual) all the adjectives and nouns need to be using the same form
13:26jsabeaudry: imirkin_, https://bugs.freedesktop.org/show_bug.cgi?id=90999 should I add anything else?
13:26imirkin_: jsabeaudry: looking now
13:26imirkin_: awesome that includes the bits i was hoping it would
13:27mupuf: there are rules to form those cases (3 among 16 others), but there are something like 12 cases + irregularities for the nouns
13:27imirkin_: jsabeaudry: if it's not too much trouble, also include your vbios (/sys/kernel/debug/dri/0/vbios.rom)
13:27jsabeaudry: :) had to make that kernel buffer very fat to get everything
13:27imirkin_: oh yeah, i think there's a cmdline option to set it
13:27mupuf: jsabeaudry: yeah, that happens
13:27mupuf: imirkin_: there is, indeed
13:28imirkin_: jsabeaudry: can i assume that you suspended it for ~3 seconds before moving the mouse or something?
13:30jsabeaudry: imirkin_, yes around 1-3 seconds
13:30jsabeaudry: vbios attached
13:30imirkin_: ok cool. thanks
13:30imirkin_: unfortunately i don't actually know anything about this, but this should hopefully be enough for skeggsb_ to look at
13:30jsabeaudry: could it be related to the fact that there are two video cards in this machine?
13:30imirkin_: yeah... what's the deal there?
13:30tobijk: finnish sounds fun :-)
13:31imirkin_: is it vgaswitcheroo-controlled?
13:31jsabeaudry: from what I know it is not supported by vgaswitcheroo but I have not looked at it thoroughly
13:32jsabeaudry: but it is that kind of setup
13:32imirkin_: jsabeaudry: so where's the second gpu?
13:32imirkin_: or do you have to switch in OSX or something?
13:33jsabeaudry: I never switch, I run off A/C most of the time and I still get 4hours battery life on the discreet gpu
13:33imirkin_: right, i just mean in general
13:34jsabeaudry: I don't know how I can switch
13:34imirkin_: ah ok :)
13:34imirkin_: anyways... yeah, that sort of thing could be in play
13:35imirkin_: mjg59 tends to know about such random things... this is re a MacbookPro11,3
13:35imirkin_: mjg59: any idea if turning off the panel (for dpms) requires something extra-special on there?
13:38mjg59: imirkin_: Not that I know of
13:39mjg59: If it's a dual-GPU system then the eDP link is going via gmux
13:39mjg59: So there's always potential for that to confuse things
13:40imirkin_: does apple-gmux support it that you know?
13:44imirkin_: jsabeaudry: by 'nv' driver, you mean nvidia blob driver, or xf86-video-nv? i'm assuming the former
13:47mjg59: imirkin_: gmux has backlight control, but there's no callback code
13:47mjg59: In the past I've seen it pass through DPMS state without any problem
13:48mupuf: tobijk: if you like challenges, go for it! The country is lovely and its inhabitants are great!
13:49tobijk: mupuf: hehe maybe its easier for germans? hmm
13:49mupuf: possibly, german helps me a bit.
13:49imirkin_: tobijk: btw, thoughts on this? https://github.com/imirkin/mesa/commit/de037bdd5d7ed05740e3c155b80d5ce9c7ceaf52
13:49tobijk: let me see...
13:51imirkin_: mjg59: hm ok. well some sort of fail in nouveau apparently. or perhaps that one doesn't pass things through.
13:53jsabeaudry: imirkin_, yes, the former
13:54tobijk: imirkin_: looks a bit funny with that define, but well allright :)
13:54imirkin_: jsabeaudry: this will take a lot more effort, but it may also be useful to have an mmiotrace of the dpms thing
13:54imirkin_: jsabeaudry: this is a good guide: https://wiki.ubuntu.com/X/MMIOTracing
13:54imirkin_: tobijk: yeah, but i cover all the combos :p
13:55imirkin_: and it's still less code in the end
13:55tobijk: imirkin_: no piglit regressions i guess? the you can add my r-b
13:55briocalter: kernel: nouveau E[ PFIFO][0000:02:00.0] read fault at 0x00175c0000 [UNSUPPORTED_KIND] from CE2/GR_CE on channel 0x007f6da000 [unknown]
13:56briocalter: froze after launching game on 660 TI
13:56imirkin_: briocalter: you had the gzdoom issue right?
13:56briocalter: that's another game tho :)
13:56imirkin_: briocalter: try running it with MESA_EXTENSION_OVERRIDE=-GL_ARB_buffer_storage
13:56imirkin_: (gzdoom that is)
13:56briocalter: sure, sec
13:57briocalter: still with garbled graphics =/
13:58imirkin_: skeggsb_: --^ interesting btw, note that it's an unknown channel. i bet that the background copy doesn't keep a ref when it should... or it should just not happen when the context is gone or... something
13:58imirkin_: briocalter: can you capture a new trace with that?
13:58imirkin_: unfortunately apitrace doesn't handle buffer storage
13:58imirkin_: which is why it was coming out all black
14:00tobijk: hey a new xserver release, pleasant surprise
14:00briocalter: imirkin_, https://drive.google.com/file/d/0B5zFVV8gOfs7V05KQXBIWXFIWGc/view?usp=sharing
14:01briocalter: it gets a little better with that override, but not quite
14:01briocalter: today tested ioq3, everything is fine
14:01imirkin_: briocalter: ok cool. that renders correctly on i965
14:02imirkin_: ... and shits its pants with llvmpipe
14:04imirkin_: briocalter: hmmm... well it seems to replay ok on my GK208
14:04imirkin_: briocalter: with my latest fix at least (which is already in master)
14:04briocalter: lemme update and retry
14:20briocalter: imirkin_, master works flawlessly now
14:20imirkin_: briocalter: even with ARB_buffer_storage?
14:20imirkin_: (i.e. without disabling it)
14:21briocalter: yes, both
14:21imirkin_: oh ok, cool. i thought you had tried it with my patch and said it didn't work
14:22briocalter: what was the fix?
14:22imirkin_: just clamping the size to 64k
14:23briocalter: weird, yesterday I applied your patch and still didn't work
14:23imirkin_: perhaps a build failure of sorts? who knows.
14:23briocalter: reverting and updating solved
14:23imirkin_: this is what i pushed: http://cgit.freedesktop.org/mesa/mesa/commit/?id=8b24388647f626a5cad10fd48e61335ed26a8560
14:25briocalter: cool, nj
14:26briocalter: thanks for the fix
14:26imirkin_: briocalter: what game are you getting the other issue in?
14:27briocalter: Faster than Light, on steam
14:27imirkin_: free to play?
14:27briocalter: nope =/
14:27briocalter: no source also
14:28imirkin_: hm, yeah. even though $5 isn't that much, my policy is that i don't pay money for the privilege of fixing bugs :)
14:29briocalter: is apitrace enough?
14:29imirkin_: in theory, yes.
14:29imirkin_: in practice, you're unlikely to be able to capture that error with a trace
15:05imirkin_: Karlton: i kinda suspect at least part of that FGFS failure has to do with flat/smooth shading model switches
15:06imirkin_: unfortunately that's getting into the land of GL that i largely don't understand. and nouveau has a bunch of failures on it too :(
15:23GREAT: hello please where can i get a shell scanner
15:36GREAT: please brothers try help me out with a linux shell scanner server
15:36GREAT: please brothers try help me out with a linux shell scanner server
15:48Karlton: imirkin_: ah, well I noticed some boxy artifacts appear on the aircraft even when I turn ALS off
15:48imirkin_: Karlton: yeah, i see boxy artifacts and the green in your trace
15:49Karlton: the green stuff goes away when I turn off ALS
15:50imirkin_: and it's not actually related to flatshading
15:50imirkin_: ugh, flightgear uses call lists, which apitrace doesn't handle in particularly great ways =/
15:56tobijk: like terasology, we have a serial behavior :/
16:59ryouma: hi everybody, is it a known problem that "product: RV610 video device [Radeon HD 2400 PRO]" no longer works on xorg due to a "modesetting" issue? upgrading debian wheezy to jessie has left me with xorg that does not launch.
17:00imirkin_: did you mean to ask in #radeon?
17:01ryouma: imirkin_: i'm sorry, i have no idea. i am quite ignorant, but i thought nouveau was the free driver for this card? or did i get the video card manufacturer wrong?
17:01imirkin_: nouveau is the open-source driver for nvidia gpu's
17:01tobijk: ryouma: for that card it should be radeon
17:01tobijk: nouveau is for nvidia cards
17:01tobijk: radeon for ati/amd cards
17:01ryouma: tobijk: imirkin_ thanks :)
20:08mogorva: Hi! I compiled mesa git on Fedora 22 and installed it in a predefined prefix and set up LD_LIBRARY_PATH accordingly. Now I have lots of 'Mesa: User error ...' messages in the terminal when running native Linux games or games under Wine. This is the output when running the Linux game 'GL-117': http://pastebin.com/S5y0SCms
20:08mogorva: many if not all games under Wine produce this: http://pastebin.com/smPRXMHK
20:08mogorva: my gfx card: VGA compatible controller: NVIDIA Corporation G92 [GeForce GTS 250] (rev a2) (prog-if 00 [VGA controller])
20:09mogorva: i don't see those messages with the distro shipped Mesa. Maybe I set up something incorrectly when compiling mesa?
20:10imirkin: probably using a debug build which prints those
20:11imirkin: and the game is just buggy
20:11mogorva: imirkin: i built mesa with --enable-debug option
20:11lucas__: working as intended then :)
20:11imirkin: yeah, otherwise it wouldn't print those errors :)
20:11mogorva: lucas__: ah okay, so nothing to worry about I guess
20:11imirkin: Mesa: User error: GL_INVALID_OPERATION in glEnd
20:11imirkin: i wonder how it achieves that...
20:12imirkin: no vertices submitted maybe?
20:13imirkin: GL_INVALID_OPERATION is generated if glEnd is executed without being preceded by a glBegin
20:13imirkin: well, either it or mesa is messing up bigtime
20:14dardevelin: imirkin, heyia, sup hero ?
20:15dardevelin: imirkin, you probably don't remember me. so here is a little recap. (I am the guy that once had a Nvidia 9300GS ) and reported a bug.
20:15dardevelin: that was fixed and works . the error was related to a SCHED lock and the kernel didn't boot
20:16dardevelin: guess what. now I have a new pc which has an nvidia gtx 850m
20:16imirkin: ... and
20:17dardevelin: and I have SCHED lock issues
20:17dardevelin: and sometimes doesn't boot
20:17dardevelin: isn't that awesome ?
20:17imirkin: which gpu is it?
20:17dardevelin: gtx 850m
20:17imirkin: (the marketing model number tells me little to nothing)
20:17imirkin: GKxxx or GMxxx
20:17dardevelin: i think GM107
20:17dardevelin: if not mistaken
20:18imirkin: lspci -nn -d 10de:
20:18dardevelin: alright let me check
20:22dardevelin: imirkin, sorry i had to reboot it panic-ed doing that
20:22dardevelin: and the couldn't boot
20:22imirkin: when doing the lspci? that's not encouraging... is this an optimus thing?
20:22dardevelin: i noticed that I have higher chances of having a successful boot if I change on the grub entry the load_video to after insmod gzio
20:23imirkin: try running with nouveau.runpm=0
20:23dardevelin: when booting ? on grub ?
20:23imirkin: kernel cmdline
20:24dardevelin: i edit the grub line before booting and then do nouveau.runpm=0 on the kernel line or you actually mean drop to kernel shell ?
20:24dardevelin: i don't know how to continue to boot from the kernel shell
20:33imirkin: try to read some guides online?
20:35dardevelin: imirkin, sorry
20:35dardevelin: i crashed again since i booted
20:36dardevelin: i was dumb enough to run lspci again
20:36dardevelin: you said nouveau.runpm=0
20:36dardevelin: kernel cmdline
20:36dardevelin: btw I am running 4.1-rc8
20:37dardevelin: i built kernels in the hopes the newest had a fix since someone suggested that a patch had been applied
20:40dardevelin-andro: Imirkin its me. Just making sure i don't lose what you say on crashes
20:41imirkin: this chan is logged
20:43dardevelin: imirkin, well but if you type and i crash i won't read it
20:43dardevelin: and will have to go to the log or ask you to repeat
20:43dardevelin: that is why i joined with dardevelin-andro
20:44dardevelin: alright having that said the kernel does always boot if I change insmod gzio to first line and the load_video
20:45dardevelin: that is my dmesg where it says GM107 (NV117)
20:45dardevelin: i should also mention that the first time i came in here the card was reporting something like 1petabyte of ram
20:46dardevelin: the newer kernels don't do this and show correct VRAM
20:46dardevelin: ps: I am using debian sid, without non-free and without contrib repositories. however I did provided the firmware and wifi driver during debian install
20:46dardevelin: so those are installed
20:47imirkin: yeah, i think a fix went in to fix that up for the secondary gpu's
20:47imirkin: nouveau E[ VBIOS][0000:01:00.0] 0x062c: unknown opcode 0xaf
20:47imirkin: ooh fun
20:47imirkin: W[ VBIOS][0000:01:00.0] DCB version 0x83 unknown
20:47imirkin: that looks like just mega-fail all around
20:48dardevelin: i remember having to extract something from bios in last card
20:48imirkin: the vbios image is probably corrupt
20:48dardevelin: how does one check for that ?
20:50dardevelin: imirkin, ^
20:53dardevelin: imirkin, by the way I am using uefi but my bios does have the option to activate cmus
20:53dardevelin: I don't think it would be needed, but since you mention vbios (could it be this ) ?
20:54imirkin: mmm... surprisingly tricky in your situation
20:54imirkin: easiest would be to just make an acpidump i guess?
20:55dardevelin: imirkin, you mean acipi --dump or something ? I only see acpi listed
20:56dardevelin: i have acpi acpi_listen (non root) and the added acpid as sudo
20:56dardevelin: btw i can run lspci -v without crashing
20:58dardevelin: imirkin, i should also note that if i remember correctly when I had the petabytes being reported this wasn't an issue (the lspci command)
20:58dardevelin: I can't be 100% sure, but this is the impression i have from the things I was requested to do
20:58dardevelin: in here
21:00imirkin: yeah, that's probably just reading the cached info
21:00dardevelin: I see
21:00imirkin: i dunno why the acpi thing is failing though
21:01dardevelin: no is not failing I don't see the command you say available
21:01imirkin: you should just boot with nouveau.modeset=0 -- that nvidia adapter won't do you much good anyways
21:01dardevelin: i can install it though
21:01dardevelin: which nvidia adapter won't do much good ?
21:01dardevelin:is installing acpidump
21:02dardevelin: or rather acpi-tools which has acpidump
21:02imirkin: the one in your laptop
21:02dardevelin: why is that ? sorry I am not sure I follow
21:02imirkin: just use the intel gpu, it'll work better
21:03dardevelin: is the card crap ?
21:03imirkin: i'd disable the nvidia one in the bios if that's an option
21:03imirkin: card's fine, nouveau just won't work very well with it
21:03imirkin: among other things, no reclocking
21:03imirkin: so it'll be slower than the intel gpu
21:03dardevelin: I though about it and nop. this asus is great in many ways ( allows to add my own keys on uefi and everything )
21:04dardevelin: but don't have the option on the bios for the card
21:04dardevelin: only info about pre allocated memory for the video card (if not mistaken)
21:04dardevelin: imirkin, i pasted the dump by the way (in case you did not notice )
21:05imirkin: no, i noticed it. i'd have to figure out how to process it.
21:06dardevelin: do you need this as a single assembled binary ?
21:06dardevelin: i remember last time that i gave you a vbios.rom or something in those lines
21:10dardevelin: would lspci -nn -d 10de: with modeset.nouveau=0 ?
21:10dardevelin: imirkin, ^
21:10dardevelin: i mean, if you think the result of that command would still be useful
21:10dardevelin: I can also try an older kernel see if that older kernels allows for it
21:11imirkin: no, i was just asking about it to see if it was GK10x or GM107
21:11dardevelin: oh ok
21:12dardevelin: so in that sense wouldn't envytools work to get the vbios.rom ?
21:12dardevelin: last time you told me to compile a github repo and run then in a sequence to extract
21:12dardevelin: imirkin, _
21:13mogorva: the nouveau driver produces those messages in dmesg when running certain games in Wine: http://pastebin.com/efDStCPE <- are those something to worry about?
21:17imirkin: dardevelin: not sure that it knows how to read it out of acpi, unfortunately
21:18dardevelin: isn't there a way to tell it to dump from the card though ?
21:18dardevelin: or bios itself ?
21:18dardevelin: i don't know exactly how we did it last time with the other card
21:18dardevelin: but I think is worth a shot if you think it makes sense
21:19dardevelin: I will also try use other older kernels that did the error but booted
21:19imirkin: mogorva: yeah... that opcode is a "join", which means that we generated bad code =/
21:19dardevelin: because this 'newer' ones don't boot
21:19mogorva: neither the game crashes nor the system locks up btw.
21:19dardevelin: the weird part to me is that changing insmod gzio to first line and load_video to second line on the grub entry
21:19dardevelin: actually boots
21:20imirkin: mogorva: wow, it really gets mad at you
21:20imirkin: mogorva: what's "kb.exe"?
21:20mogorva: the King's Bounty game
21:20imirkin: don't suppose it's free-to-play is it?
21:21mogorva: actually all the games in the KB series have this error
21:21imirkin: also is this with wined3d or with gallium nine?
21:21mogorva: there's a demo for the 1st KB game, i'll check if the problem is reproducible with that
21:22mogorva: with wined3d, i can't make wine+nine work :(
21:22dardevelin: imirkin, any idea which package might have vdpau module ?
21:23dardevelin: i have vdpau-va-driver
21:23dardevelin: is the only dependency still missing for envytools to build
21:24dardevelin: would it be vdpauinfo ?!?
21:29imirkin: mogorva: actually someone sent a patch which might be related to the stack underflow errors -- https://bugs.freedesktop.org/attachment.cgi?id=116339
21:34dardevelin: i built without building vpow
21:43mogorva: imirkin: i'll try the patch later. so far i couldn't reproduce the problem with the demo of KB
21:44imirkin: mogorva: ok. you could also try to do an apitrace, although it's a bit of a pain with wine iirc
21:45mogorva: imirkin: well i'm not familiar with apitrace, i switched from the binary drivers to nouveau 2 days ago :)
21:46dardevelin: nvagetbios gets it corrupted
21:47imirkin: mogorva: it's a generic tool... https://github.com/apitrace/apitrace
21:57mogorva: when i see such errors from nouveau in dmesg, is it advised to restart X (or the whole system) or they can be ignored until something more serious happens?