02:20mlankhorst: hakzsam: yes, why?
02:21hakzsam: mlankhorst, hi, could you nvascan 0x419e00 and show me the output, please?
02:22mlankhorst: sure, sec
02:47hakzsam: mupuf, --^
02:48hakzsam: mlankhorst, and btw, if you have the blob and cuda/cupti on your gk20a, it would be very good for me to see how pm are implemented :)
02:49mupuf: And we can monitor BLCG, as expected!
02:50mupuf: this could be exposed by mesa already, right?
02:50mupuf: Block-level clock gating
02:51hakzsam: this is for gk20a, and I still have to check on gk104-
02:52hakzsam: I'm not sure if the signal selection will be identical on gk104-
02:53mupuf: I would think that this would be more stable than the signal number in pcounter
02:54hakzsam: it would be good to have some traces on gk20a to confirm this :)
02:54hakzsam: btw, MP counters are quite different on gm107, but it's not my top priority
02:54mupuf: well, I guess I should get my hands on one at some point
02:54mupuf: yeah, I remember you telling me
07:07mlankhorst: hakzsam: ah, I forgot the hostname
07:38zero_note: Hi everybody
07:42zero_note: I've just installed Debian 8 on my Lenovo notebook (I needed something more stable than Ubuntu)
07:43buhman: I think you mean 'newer'
07:44zero_note: I was reading dmseg output to verifying correct network card firmware installation and I've noticed that nouveau drivers cannot detect my nvidia chipset
07:45zero_note: buhman, Debian "newer" than Ubuntu? from which perspective?
07:45buhman: that sort of thing
07:47zero_note: buhman, ah, when I think to "newer" I think at kernel, workspace, applications version
07:50zero_note: to say more, my nvidia is a 840M model (NV110 family (Maxwell) , precisely NV118 (GM204))
07:50buhman: 3.16 is probably too old
07:50buhman: but you might provide what you're looking at
07:53zero_note: dmesg | grep nouveau gives me -> nouveau ![ DEVICE][0000:03:00.0] unknown Maxwell chipset
07:55zero_note: sorry, maybe there is a more appropriate way or place to paste this information, am I right?
07:57RSpliet: make sure to paste the full log, but for Maxwell to do anything at all (which, still isn't much) you need a 4.0 kernel
07:59zero_note: if it does not break any of the rules of the channel, I will use pastebin
08:04zero_note: in that case, here is my dmesg output http://pastebin.com/7JyzdhyZ
08:09RSpliet: yeah, try again with a 4.0 kernel
08:09RSpliet: and having rules on which paste websites you can and cannot use would be silly :-)
08:14imirkin: zero_note: you have a GM108 -- not supported by nouveau atm
08:14imirkin: zero_note: https://bugs.freedesktop.org/show_bug.cgi?id=89558
08:21zero_note: ok, thank you all for your time and information, I'll try with 4.0 kernel
08:23imirkin: won't help
08:24imirkin: you could try to apply the change i suggest in comment 1 of that bug and see if you fare better than the other guy...
08:35zero_note: imirkin, ok, I'll let you know
08:48effractur: do i need to sumbit a bug report about that openfirmware issue?
08:49imirkin: effractur: wouldn't hurt
09:43Dokugogagoji: Hi, is there anyone who could help me with an issue having when using nouveau?
09:44imirkin_: depends on the issue...
09:44imirkin_: describe your issue and pastebin dmesg, and, if relevant, xorg log and glxinfo output
09:52Dokugogagoji: Hi imirkin_, thanks for the explanation. Well, the issue are random moving blank stripes appearing for a few seconds generally when swtiching to X to console, changing video resolutions or launching certain applications like video players or games. After around 5 seconds, the video is stable. Tweaking xorg parameters (disabling acceleration, switching GLVXBlank option) has no effect .It does not happen with vesa/vga module.
09:52Dokugogagoji: I've looked the xorg and dmesg but nothing relavant to me. I'm pastebin them.
09:52imirkin_: i use those as a shortcut to a whole bunch of questions about your configuration
09:52imirkin_: i don't anticipate seeing actual problems in them
09:59Dokugogagoji: imirkin_, I understand. Thanks :) dmesg: http://pastebin.com/c4a4uEZA xorg: http://pastebin.com/a5z9mmAi glxinfo: http://pastebin.com/bdTmgh0q
09:59Dokugogagoji: It's an old GeForce FX5500 with a SyncMaster 753s hooked up.
10:00imirkin_: do you really have a 1024x768 screen, or does it just not detect edid?
10:01Dokugogagoji: yes, 1024x768. It's old stuff.
10:01imirkin_: ah hm. looks like 1024x768@85 is the preferred mode, but it'll do up to 1280x1024@60
10:02imirkin_: and it even has a 1152x864 mode =]
10:02imirkin_: that was my favourite for a while
10:04imirkin_: anyways, i've never heard of such an issue... sounds like we do something improperly with the relevant PLLs...
10:04imirkin_: and it ends up switching modes between X and console
10:05imirkin_: you could confirm that by running fbset -i in X and on the console and comparing
10:07Dokugogagoji: Yup, but setting those refresh rates are not reall eyes friendly as you can see the screen vibrating. :D Ok doing that.
10:07imirkin_: yeah, i remember those well. i couldn't handle 60 or even 75Hz refresh rates
10:08imirkin_: 85 was good for me though
10:11Dokugogagoji: from X: http://pastebin.com/eb8XrXdA from console: http://pastebin.com/iVpeMWte same.
10:12Dokugogagoji: I should try if those black stripes are recorded when running a screen capture
10:13imirkin_: that doesn't print what i thought it printed
10:14imirkin_: like... why are the timings empty? gr.
10:15imirkin_: also it's a pile of lies... tells me i have 1024x768
10:15imirkin_: nevermind :)
10:15Dokugogagoji: I'm pretty ignorant here but are timing read by the video card bios? Could you be something due to video clock?
10:16imirkin_: clearly, it's *something*
10:16imirkin_: all the modesetting logic for old cards is... fragile to say the least
10:25buhman: I suppose gk106 is too old to have any hope of ever being fixed :(
10:26imirkin_: gk106 works fine. *your* gk106 doesn't.
10:26imirkin_: i know you're frustrated, but please avoid spreading misinformation
10:26buhman: heh, 'spreading'? ;p
10:27imirkin_: not everyone in this channel knows all the details.
10:28Karlton: my gk106 works :D
10:28imirkin_: as for that situation being fixed, my understanding is that ben's getting a problem laptop. i assume he'll have it fixed within 30s of receiving it.
10:34Dokugogagoji: (ok got a simple run with simplescreenreader, but the issues are not shown in the recording.)
10:34imirkin_: oh yeah, it's clearly some sort of analog artifact
10:38Dokugogagoji: @Karlton, what is your issue with gk106? (I have seen some HP laptops having video artifcats with random coloured lines for a fallacy BIOS)
10:39Dokugogagoji: @imirkin_ I'll check if I have the same artifact with another monitor or using the DVI port.
10:44hakzsam: mlankhorst, :/
10:46imirkin_: calim: oops? :) http://cgit.freedesktop.org/mesa/mesa/commit/?id=db269ae495425849804fb1d05cfe42b0d3d304b3
10:47Karlton: Dokugogagoji: I said mine works. It works perfectly except for a few quirks in some games, and like many other cards, not being able to reclock all the way
10:54hakzsam: imirkin_, could you trace the blob for me on your gk208? :)
10:55imirkin_: not right now...
10:55imirkin_: if you email me instructions, i'll try to remember to do it though
10:55hakzsam: okay, thanks I'm going to send you an email
11:01Dokugogagoji: Tried with an ancient 800x600 monitor and well, no moving black borders but the screen was jumping up and down in the same moments the black borders appeared to the other screen.
11:04Dokugogagoji: ...and took the monitor from a doublescaled Amiga. I am living in a vintage museum ahah.
11:05imirkin_: why not complain that nouveau doesn't support Riva 128 while you're at it...
11:08imirkin_: i *should* be able to just unload nouveau and load up the blob...
11:09imirkin_: since i have no screens connected to that gk208, it's just sitting on the side
11:09imirkin_: [came bundled with the desktop... apparently the built-in intel isn't good enough? but it has a DP port while the gk208 doesn't...]
11:11Dokugogagoji: No vintage Nvidia cards sadly, but I have an handful batch of Ati Rage, Matrox and S3 cards so I can leave my issues to other groups :D Btw, I'll try to play with the issue better to squeeze some more useful info from it. Thanks for the help :)
11:13buhman: imirkin_: if I wanted to get started, wouldn't riva 128 be low hanging fruit? (well-known how it works internally, etc..)
11:13imirkin_: buhman: get started with what?
11:14buhman: nouveau development of course ;p
11:14imirkin_: buhman: its 3d accel is unusable for GL
11:14buhman: isn't it glide or something ridiculous?
11:15imirkin_: oh wait
11:15imirkin_: i lied
11:15Dokugogagoji: I got it working with Direct3D in windows 95
11:15imirkin_: nv1 had the crazy thing
11:15Dokugogagoji: so probably around DirectX 6 was fine
11:15buhman: yeah; looks like nvidia wrote gl drivers for it
11:15buhman: so .. ;p
11:15imirkin_: Nvidia's "NV1" chip had been designed for a fundamentally different type of rendering technology, called quadratic texture mapping, a technique not supported by Direct3D. The RIVA 128 was instead designed to accelerate Direct3D to the utmost extent possible.
11:16imirkin_: NV1 aka Diamond Edge 3D :)
11:17imirkin_: and it's all nice and documented: https://github.com/envytools/envytools/blob/master/rnndb/graph/nv3_3d.xml
11:17imirkin_: but... might be a bit of a waste of time :)
11:17hakzsam: imirkin_, yeah, should be good :)
11:18imirkin_: iirc the pushbuf submission changed between nv3 and nv4
11:18imirkin_: that's why nouveau only supports nv04+
11:19Dokugogagoji: Well, the sastifying moments to run a 3D demo on linux on a TNT2 driver could pay off. :p
11:19imirkin_: although with the recent refactorings, might be possible to shoehorn it in
11:19imirkin_: Dokugogagoji: TNT2 works fine with nouveau + nouveau_vieux
11:19imirkin_: a handful of features that could be supported aren't
11:19imirkin_: and a handful of things don't work (i'm looking at you, clip planes)
11:21imirkin_: and 3d textures..
11:22imirkin_: although iirc i weaseled out of that quite nicely with http://cgit.freedesktop.org/mesa/mesa/commit/?id=5524af81365d51757ef68a8e62dbd426d0b07184
11:25buhman: woah, glsl 410
11:26imirkin_: buhman: only on my integration branch
11:26buhman: oh, so 'allow' doesn't actually make the magic happen
11:26imirkin_: hehe, no.
11:26imirkin_: that makes the magic happen though
12:50Dokugogagoji: @imirkin_, sorry, are you here?
12:51Dokugogagoji: :) I have solved it..somehow...
12:51Dokugogagoji: or better workarounded it
12:54Dokugogagoji: it seemd due to the video card that when changing display modes, it tries to change output destination too. in the drmmode_display.c , I have forced drmmode->mode_res->count_connectors=1 and drmmode->mode_res_count_crtcs=1 to have only the VGA intiialized.
12:54Dokugogagoji: and now the issues does not arise.
12:55Dokugogagoji: though I have no idea why it is trying to change output yet. I'll check better tomorrow..and iirc, it detected three outputs but the video card has only DVI and VGA outputs.
12:58imirkin_: there are a number of confusing factors
12:58imirkin_: the third connector is probably the TV bonnector
12:59imirkin_: might not be hooked up on your board, but vbiss advertises it iirc
12:59Dokugogagoji: yes, it was DVi, VGA and TV
12:59imirkin_: there's also the concept of crtc
12:59imirkin_: which is separate from output connector
12:59imirkin_: there are 2 CRTC's
13:00imirkin_: which can be hooked up however you want to any of the connectors
13:01Dokugogagoji: Ah ok, thanks for the info
13:01imirkin_: this is why you can only do up to dual-screen with older nvidia hw
13:03Dokugogagoji: I'll try to keep the three connectors and forcing 1 CRT (and viceversa) to check where the issue is. ...Not having clues, I have tried with those two counters ' cause I have thought "VGa and VESA has only ouput, nouveau three so it's only the piece of info I have to differentiate them". and so I hacked it. :D
13:03imirkin_: whatever works!
13:03Dokugogagoji: Yup, it's jsut the computer science magic in the end
13:04Dokugogagoji: ok have to go. Thanks for the help.
13:36imirkin_: Karlton: did you ever file a bug for your terasology fail?
13:37imirkin_: now that i have access to a GK208, figured i might take a look, instead of tilling at the tess windmill
13:39mlankhorst: not a danish wind mill? :P https://mblankhorst.nl/etc/IMG_5580.JPG
13:44imirkin_: i do like danishes...
13:45imirkin_: bbl... i guess it should be easy enough for me to repro myself...
15:51skeggsb_: imirkin, buhman: yes, i'm having one of the problem laptops shipped to me, i'm still waiting for it to arrive though...
15:57skeggsb_: there's been some delays in it being sent, but, it should be today, i was told yesterday at the latest, but it didn't show up :p
16:09imirkin: skeggsb_: you *do* realize it'll have no problems when you try to use it, right? :)
16:24skeggsb_: imirkin: airlied already made that joke :P
16:25imirkin: who's joking? :p
18:58skeggsb_: imirkin: you're correct
18:58skeggsb_: imirkin: they shipped me a gk107 version of the laptop, not the broken gk106...
21:09imirkin: skeggsb_: doh!
23:54mlankhorst: ok, back to work