07:35mlankhorst: yeah I've grabbed that patch and some hack I did to uniqify all vm->phys translations, nvc0 is remarkably stable now :P
07:36mlankhorst: only issue I still have with suspend from time to time is that my network card falls off the pci bus
07:40glennk: thats what you get for hanging your server off the ceiling
11:43pmoreau: So many Nouveau bugs on the kernel bugzilla…
11:43imirkin: yeah, i ignore them
11:44imirkin: one bugs repo is enough for me
11:44imirkin: if they can't be arsed to look at our reporting bugs page, they won't be able to do anything useful with the help i'd give them either
11:45pmoreau: Maybe we should add some kernel info message "Please reports bugs on bugs.freedeskop.org" like some other drivers do
11:46imirkin: if it's possible, sure!
11:46imirkin: like kernel debug?
11:46imirkin: i think that's a little much
11:46pmoreau: Or some link to the wiki
11:47pmoreau: Just in case they read the dmesg output :D
11:47imirkin: if they're not willing to do the research
11:47imirkin: i don't really care
11:48pmoreau: Fair enough
11:48imirkin: perhaps ben has a diff viewpoint
11:48RSpliet: I'm sure Ben is obliged to pick up bugs from rhbz
11:48imirkin: seems likely
11:48pmoreau: We could also import the ones from distribution bugzillas, and LibreOffice, and …
11:49RSpliet: as far as I know, LibreOffice and FreeDesktop share the same database
11:49buhman: what's wrong with the explanation in http://nouveau.freedesktop.org/wiki/Bugs/ ?
11:49RSpliet: buhman: nothing, pebkac
11:53pmoreau: imirkin: Is fp64 part of OpenGL 4.0 specification?
11:54poisson_rouge: Hi, is anyone around?
11:55pmoreau:hides behind RSpliet
11:55poisson_rouge: found :F
11:55RSpliet: but srsly, don't ask to ask, just ask
11:55RSpliet: (and don't you dare apologising now!)
11:55poisson_rouge: ok :)
11:56poisson_rouge: I get weird colored lines on my computer sometimes, and I have to restart it for it to go away
11:56pmoreau: Which card do you have, and which kernel version?
11:56poisson_rouge: I think it might because of gpu drivers, but I am not sure how to check it out :/
11:56poisson_rouge: gtx 660
11:56imirkin: horizontal or vertical?
11:56poisson_rouge: and... good question, just a sec :o
11:57poisson_rouge: I'd say vertical but I'm not sure
11:57imirkin: assuming that those are the only two options
11:57imirkin: it should be pretty easy to pick one out :p
11:57poisson_rouge: I can't remember for sure ^^
11:57imirkin: anyways, do you have an HDMI screen and you're seeing a pink line along the right hand side edge?
11:58poisson_rouge: and my kernel tells me 3.13.0-37-generic, is that possible ?
11:58imirkin: is the screen hooked up with HDMI?
11:58tobijk: imirkin: lines across :P
11:58poisson_rouge: I think so, ill check
11:58poisson_rouge: yes it is!
11:59poisson_rouge: and for the pink line, I can't remember ._.
11:59imirkin: poisson_rouge: update your kernel. there was an issue in the way we were setting up HDMI on kepler chips (of which gtx 660 is one), which resulted in colored lines on one of the edges of the screen
11:59poisson_rouge: the lines are all over though
11:59imirkin: oh hm
11:59imirkin: could be a diff issue then
11:59poisson_rouge: and ok, I will do that
11:59RSpliet: either way, update your kernel :p
11:59imirkin: but we were definitely doing it wrong before
11:59poisson_rouge: I will ^^
12:00poisson_rouge: diff issue?
12:00imirkin: definitely presenting differently than what i heard
12:00imirkin: but... who knows
12:00poisson_rouge: huh, ok
12:00imirkin: there were also other issues connected with the GTX 660 that got fixed
12:00poisson_rouge: well I will try that anyway
12:00poisson_rouge: oh, cool, ok!
12:00imirkin: like it had a funny memory <-> partition layout... sometimes
12:00imirkin: and i think a couple of other things
12:01imirkin: how those problems map into actual visual issues? who knows.
12:01poisson_rouge: I mean it's not like it's *super* problematic, I just log off and on and it's fine, but...
12:01poisson_rouge: ok :o
12:01poisson_rouge: ty :)
12:02imirkin: try a 3.18 kernel, or 3.19
12:02imirkin: you should even get semi-reliable reclocking on those
12:02poisson_rouge: ok ^^
12:02imirkin: (if you boot with nouveau.pstate=1)
12:02RSpliet: imirkin: don't make promises now :D GDDR5 reclocking is still in undesirable shape
12:03poisson_rouge: and, huh, tbh, i've never done that before, but its not too hard if I follow a guide right?
12:03poisson_rouge: syupid question, sry ._.
12:03imirkin: RSpliet: it's in a lot more desirable shape than it was in 3.13
12:03imirkin: RSpliet: in that it works reliably except for the highest memory speeds
12:06poisson_rouge: wow, looks simple after all ^^ ty for everything :)
12:06poisson_rouge: have a nice day!
14:44Rudolf: hello there, someone with this problem in kernel 3.19: http://pastebin.com/iTV2tXTD
14:45imirkin: Rudolf: yeah, apparently we broke loading fw from firmware
14:45imirkin: skeggsb: --^
14:45imirkin: apparently we broke loading vbios from acpi
14:46Rudolf: imirkin: a bug waiting for solution? nothing to do right now?
14:46imirkin: yeah basically
14:47imirkin: Rudolf: do you have a successful boot log from an older kernel?
14:47Rudolf: imirkin: ok, do you know the page to this bug?
14:47Rudolf: imirkin: no, but i can reboot into 3.18.X and take one
14:47imirkin: Rudolf: i forget... it was either in an email thread or there's a bug open for it
14:48Rudolf: imirkin: i was unable to find in bugs.freedesktop.org
14:49Rudolf: imirkin: thank you, i'll follow it
17:32whompy: imirkin: did you get anyone to test that nv50 patch?
17:33imirkin: whompy: not yet... but i got mupuf to plug a nv50 into reator
17:33imirkin: so i'll test it out later on... unless you're offering
17:33whompy: I was planning to test nouveau on my laptop anyway today, so I can give it a whirl
17:33whompy: I'm about to rebuild anyway
17:33imirkin: ok. there was a small bug in the version i posted
17:33whompy: anything in particular you'd like to see?
17:34imirkin: new patch version: https://github.com/imirkin/mesa/commit/0cc61c295ca82f6c3a8df62c7a50f65636468338.patch
17:34imirkin: update piglit and run all the bin/arb_pipeline_statistics_query-* binaries with -fbo -auto
17:43imirkin: mupuf: btw, you have a G200 right?
17:43imirkin: i might try to do doubles on it :)
19:44whompy: imirkin: nv50 arb_pipeline_statistics_query-* -fbo -auto results with patch: http://ix.io/grA
19:45imirkin: whompy: heh, you need to run each of them one by one
19:45imirkin: whompy: for i in bin/arb_pipeline_statistics_query-*; do echo $i; $i -fbo -auto; done
19:46whompy: stupid question: how do I pipe all output for the test? I'm just getting the last line the easy way
19:46whompy: Score. That'll do sir.
19:47imirkin: ( for i in bin/arb_pipeline_statistics_query-*; do echo $i; $i -fbo -auto; done) > foo.out
19:48whompy: Hah, parentheses. Of course.
19:51imirkin: yay, thanks!
19:51imirkin: which gpu is this btw?
19:53imirkin: whompy: --^
19:54whompy: GT216 (240M)
19:55whompy: I'll need to dink with power management... I think you or mupuf mentioned it's pretty decent on this chipset.
19:55whompy: or family, rather.
19:56imirkin: yeah, should work on GT21x
19:56imirkin: just boot with nouveau.pstate=1
19:57whompy: Ah, that's the one. My google fu is weak tonight.
19:58imirkin: whompy: if you want a tested-by credit on the commit, pm me your name + email. or i guess i can just use your irc handle...
20:16imirkin: ok pushed... nv50 has pipeline statistics now
20:30airlied: wow I merge stuff then I find things straight away, I should merge stuff more often :-P
20:30airlied: imirkin: http://paste.fedoraproject.org/187927/24406614/ though I don't think its fully correct, just found again while playing with attrib64bit
20:31imirkin: + case GLSL_TYPE_DOUBLE:+ return this->matrix_columns * 2;
20:31imirkin: i don't think that's right
20:31imirkin: i did that at first, but then decided against it
20:31airlied: no its not
20:32airlied: it probably needs the number of components taken into a/c
20:32imirkin: i couldn't _quite_ crystallize what it was counting
20:33imirkin: but it didn't seem to be anything size-related
20:33imirkin: it was more like "how many diff values are there"
20:35imirkin: airlied: there's not a ton of clarity about what all these various things are
20:35imirkin: airlied: but i think that var->data.location *is* size-related
20:35imirkin: so it needs to knwo about the dual-slot stuff
20:36airlied: so far it only seems to matter on vertx inputs
23:23bps_: Hi, is there any way to dump the efi rom of the video card with nouveau? When I dump vbios.rom it doesn't seem to contain the efi version
23:24imirkin: bps_: vbios.rom just contains the vbios
23:24imirkin: bps_: if you want to dump the option rom, you can do that by futzing with pci in sysfs -- see the dumping instructions in the nouveau wiki, that's one of the options
23:24bps_: imirkin: http://nouveau.freedesktop.org/wiki/DumpingVideoBios/ suggests that it does the same thing as catting rom in pci sysfs?
23:25bps_: imirkin: which gives me an i/o error
23:26imirkin: i'm talking about the "using /sys" method
23:26imirkin: if you're getting an error, that means there's no option rom
23:26imirkin: which would be quite odd
23:27imirkin: if you're booting off the card
23:27imirkin: i guess i don't know that much about efi, perhaps it works differently
23:28bps_: imirkin: This is actually a two card laptop with a mux
23:28bps_: imirkin: I'm trying to get the rom for vga passthrough
23:28imirkin: ah, efi rom is probably completely separate then
23:29bps_: imirkin: Well most discussion seems to think the /sys method should work
23:29imirkin: it can work any way it feels like on a laptop -- manufacturer has control of all aspects of the thing
23:31bps_: imirkin: Do you know if this rom is something the driver should have access to?
23:32imirkin: all the driver needs is what gets dumped out in vbios.rom
23:32imirkin: in /sys/kernel/debug/dri/*/vbios.rom
23:33bps_: Does the method for getting that differ in pure EFI systems?
23:33imirkin: there are like 5 diff methods for retrieving them
23:33imirkin: and none of them depend on efi in any way
23:33imirkin: in a laptop like yours, it's likely that we get the vbios from acpi
23:34imirkin: i.e. there's an acpi method that dumps out the vbios from... who knows where
23:34bps_: gotcha okay
23:34imirkin: [in fact i couldn't tell you where acpi tables come from in the first place... set up by the bios one way or another i guess]