00:03NoReflex: back; I now have the second provider as well: Provider 1: id: 0x5f cap: 0x5, Source Output, Source Offload crtcs: 0 outputs: 0 associated providers: 0 name:nouveau
00:07NoReflex: the next section of the wiki tells me to use "xrandr --setprovideroffloadsink provider sink" in order to specify who sends the output where
00:08imirkin: just use "1" for provider, and "0" for sink. that works in like 99% of situations
00:09NoReflex: and that this settings is lost once the X server reboots and I could put it in a startup script but that will "reduce battery life and increase heat"
00:10NoReflex: shouldn't the battery life and heat be only influenced if the card is actually used for rendering?
00:10airlied: well the RAM gets powereed on
00:11imirkin: i think only if the crtc is actually turned on
00:12airlied: yes it should poweroff if you don't scanout
00:18NoReflex1: I'm running some tests now with glxspheres
00:20NoReflex1: do I still need virtualgl installed?
00:21airlied: nope, but does hdmi audio work?
00:21imirkin: DRI_PRIME=1 glxinfo
00:23NoReflex: sorry I got disconnected because my notebook locked up
00:24NoReflex: it locks up when I try something like DRI_PRIME=1 glxspheres64 16384
00:24NoReflex: I don't know if the audio works yet
00:24NoReflex: I only have 0:IGD:+:Pwr:0000:00:02.0
00:24NoReflex: 1:DIS: :DynOff:0000:01:00.0
00:25NoReflex: in vgaswitcheroo/switch .....
00:25imirkin: perhaps you have one of the keplers that doesn't power up properly
00:25NoReflex: I've read that I should also have DIS-Audio there ...
00:26NoReflex: maybe it only powers up if I connect a HDMI cable...
00:32NoReflex: I've connected an external display via HDMI and while the video shows up OK no audio ...
00:32imirkin: check what eld says
00:33imirkin: also what kernel are you on?
00:34NoReflex: Linux ionut-pc 3.18.6-1-ARCH #1 SMP PREEMPT Sat Feb 7 08:44:05 CET 2015 x86_64 GNU/Linux
00:34imirkin: hm ok. that should have the kepler hdmi fixes... i think
00:35imirkin: check that there are eld contents
00:35NoReflex: what is eld? does is have something to do with edid data?
00:35imirkin: ssssort of
00:36imirkin: it's in /sys somewhere
00:37imirkin: oh, or not... /proc/asound/card0/eld#0.0
00:38NoReflex: find /sys -name *eld* does not find anything useful from what I can tell
00:38NoReflex: oh, ok
00:38imirkin: there's a bunch of diff ones, and you probably have multiple cards
00:38imirkin: so you have to find the right one
00:38imirkin: anyways, instead of monitor_present 0, look for one with monitor_present 1 :)
00:39NoReflex: find only returns: /proc/asound/card0/eld#3.0
00:40NoReflex: it contains: http://privatepaste.com/314f73c99a
00:40imirkin: so... looks like hdmi audio should work fine
00:42imirkin: are you sure your mixer settings aren't messed up?
00:42NoReflex: HDMI audio does not appear in pavucontrol (I'm using pulseaudio)
00:42imirkin: and/or that you're outputting to the right place
00:43imirkin: oh, well, that's problem #1
00:43imirkin: i've never gotten pulseaudio to work... too complex (and totally unnecessary)
00:44NoReflex: aplay -l gives:
00:44NoReflex: card 0: PCH [HDA Intel PCH], device 3: HDMI 0 [HDMI 0]
00:44NoReflex: Subdevices: 1/1
00:44NoReflex: Subdevice #0: subdevice #0
00:46NoReflex: using "aplay -D plughw:0,3 /usr/share/sounds/alsa/Front_Right.wav" does not produce any sound via hdmi ....
00:48NoReflex: perhaps I'll just use a simple audio cable to connect the output from my notbook to the display ...
00:49NoReflex: is there a way to enable tearfree for the intel driver without using the xorg.conf method? my screen flickers badly without it
00:56NoReflex: perhaps using modprobe.conf or another method?
02:42NoReflex: after some reading it seems that option tearfree is not really desirable but instead I should fix my window manager
02:43NoReflex: luckily xfce 4.12 is just around the corner :)
06:03xexaxo: mlankhorst: sweet stuff. no warnings with the WARN_CFLAGS, plus the test works like a charm.
06:04xexaxo: plus as you mentioned valgrind is quite happy with it :)
09:13mlankhorst: I doubt many people care to review it, so I'll commit
09:26joi: mlankhorst: Ben might care and IIRC he's on vacation
09:31mlankhorst: hm true, I'll give him a monday. :P
09:51joi: mlankhorst: btw, you could tell in 3/4 which patch is "previous", because it surely is not 2/4, (and do s/cound/could/)
10:18mlankhorst: joi: huh it is previous, are you on libdrm master?
10:24joi: master: "tests/radeon: set the list* functions as inline", master/nouveau: "android: remove duplicate libdrm in LOCAL_SHARED_LIBRARIES", master/nouveau/author=you: "nouveau: Only export public functions."
10:24joi: none of which is a possible candidate
10:39DouglasK: Question, does Nouveau support the Nvidia GTX860M chipset?
10:39DouglasK: I looked over on the codename page, and it's listed, but they seem recalcitrant about saying one card or another works.
10:42imirkin: DouglasK: depends which chip it is... and your definition of 'support'
10:42DouglasK: if it works in 2d basic, that'll do for now.
10:43imirkin: ok, well, with patches you should be able to make any shipping nvidia chipset do that
10:44DouglasK: *nod* good to hear. I'll work at it tonight when I get home. At work on said laptop now.
10:44imirkin: however note that most of what people tend to think of '2d basic', is actually 'using 3d accel hardware'
10:45imirkin: if you're on that laptop now, can you do 'lspci -nn -d 10de:'
10:45DouglasK: I'm still on Windows that came with it ...
10:45imirkin: oh :)
10:45DouglasK: The PCI ide is though: PCI\VEN_10DE&DEV_1392&SUBSYS_157D1043&REV_A2
10:46imirkin: ok, let's see...
10:46imirkin: ok. GM107
10:46imirkin: so with ctxsw firmware from the blob, you should get full 3d accel for that (and recent kernel/mesa versions obviously)
10:47imirkin: without that firmware, you get no accel at all, so just a 2d framebuffer
10:53DouglasK: Thank you ... making notes. :)
10:54imirkin: you can see firmware-extraction instructions at http://nouveau.freedesktop.org/wiki/NVC0_Firmware/
10:57DouglasK: Perfect. Looks like it's doable.
10:58imirkin: [sadly my script doesn't work for the maxwell firmware]
10:58imirkin: so you have to do the mmiotrace dance
10:59DouglasK: I'm good with that. If nothing else, I am stubborn.
11:00DouglasK: Once I extract it, is there somewhere I can submit it for inclusion in the standard packages or some such?
11:00imirkin: not without consulting a lawyer
11:00imirkin: it's the same fw for all gm107's... we just can't redistribute it
11:01imirkin: my little extract_firmware.py script gets around that by having the end user run it against the nvidia-supplied driver files
11:02imirkin: but they changed how they packed firmware in newer blob versions, and older ones didn't have maxwell
11:02imirkin: i tried figuring it out, but... it used to be in these nice easy-to-find gzip'd blobs inside of the nvidia.ko blob
11:03imirkin: now they're just in there without any sort of "packing" around it
11:03DouglasK: Yeah, that sounds at best challenging.
11:03imirkin: i suppose i could write sufficiently accurate definitions for finding them
11:03imirkin: but... it's tedious and could be of questionable legality
12:09mlankhorst: up to nvidia to solve :p
12:09imirkin: well, we could also write our own fw
12:09imirkin: and by 'we', i mean people who are not me
12:09mlankhorst: how do you want to sign it?
12:10imirkin: for GM107, no sig
12:11imirkin: for GM20x, might be trickier
13:05aloiece: hei anyone here?
13:06aloiece: I've got a question about making the backlight controls on my macbook working
13:08imirkin: which macbook?
13:09aloiece: macbook pro 6.1
13:09aloiece: mid 2010
13:09imirkin: oh. ought to work... probably
13:09imirkin: there was an issue with some old powerbooks that remains unresolved
13:09aloiece: prefer running with nouveau rather than nvidia, but I can't seem to get the screen brightnest to work
13:09aloiece: fresh install
13:10imirkin: pastebin dmesg
13:12aloiece: lot of stuff
13:12aloiece: what are we looking for?
13:13imirkin: a bunch of things, all of which are in dmesg :)
13:14aloiece: unfortunately not a programmer =( =
13:15imirkin: ok, so this is an optimus setup of some sort?
13:16imirkin: or rather, pre-optimus?
13:16imirkin: i.e. the plain switcheroo thing
13:16aloiece: but after (from my judgement) a lot of research
13:17imirkin: so unless you've switched to using the nvidia card, backlight control should be on the i915 driver
13:17imirkin: or the acpi video or something like that
13:17aloiece: there isn't a way to run with both nvidia and intel at the same time like osx does
13:17imirkin: also the backlight situation in both i915 and also the arbitration between acpi_video and vendor has changed a bunch
13:17imirkin: 3.13 is fairly old
13:18imirkin: you could try booting with
13:18imirkin: to see if that helps
13:19aloiece: ok that goes in the quiet splash line in @@ right?
13:19aloiece: quotationmakrs i meant
13:20imirkin: not sure what "the quiet splash" line is
13:20imirkin: but it needs to be on the kernel cmdline
13:20imirkin: however you have that set up
13:20aloiece: ok be right back
13:29aloiece: imikrin: hei, back. adpi_backlight=vendor not doing the job
13:32aloiece: imirkin: hei, back. adpi_backlight=vendor not doing the job
13:33aloiece: (sry mispelled)
13:33imirkin: hopefully you did acpi_backlight=vendor
13:33imirkin: and not adpi_backlight=vendor
13:34aloiece: yes, horrible typing... but when it's important I triple check ;)
13:35aloiece: learnt the importance of syntax the hard way =p
13:37imirkin: ok... maybe you need the inverse then
13:37imirkin: i forget what it's called
13:37imirkin: not-vendor :)
13:38aloiece: so acpi_backlight=not-vendor?
13:38imirkin: no, not really
13:38imirkin: it's something like system... someone would have to read the docs
13:39aloiece: hmmm someone, which docs? nouveau?
13:39imirkin: you might also try a recent kernel, problem may just be fixed
13:39aloiece: a more recent low-latency kernel?
13:43imirkin: 3.19 is the latest
13:44aloiece: and can i bother you with one more question? if I install a new kernel and it doesn't boot, will the old one still boot or is everything messed up then?
13:44imirkin: all depends on what you do during the 'install' stage
13:46imirkin: it is usual to be able to boot multiple kernels by selecting one in grub
13:46imirkin: or whatever boot manager you use
13:46imirkin: however that hardly prevents you from screwing something up and overwriting an old kernel's "stuff"
13:47aloiece: using grub, yes
13:47aloiece: how do you recomment installing it? synaptic?
13:47aloiece: or download .deb files?
13:48aloiece: or wget?
13:48imirkin: sorry, i've never used any of those methods
13:48imirkin: they're too... unclear... for me. i do everything much more explicitly
13:48imirkin: it's simpler for me, but probably not appropriate for beginners
13:50aloiece: right, never built anything really. except tiny stuff with how-tos
13:52imirkin: i think that there are ppa's that have recent kernels
13:52imirkin: generally speaking the people who make the packages know what they're doing, as long as you're getting them from a 'reputable' source
13:53imirkin: but i say this without ever having used ubuntu or debian or any of those :)
13:53aloiece: hehe right
13:53aloiece: out of curiosity, what ar eyou using?
13:54imirkin: but i build my own kernels (with my bare hands!)
13:54imirkin: not using distro helpers or whatever
13:54aloiece: you work in IT?
13:54imirkin: i guess. never really understood what 'IT' was
13:55imirkin: but what i do can be largely summed up to 'pressing buttons on computers'
13:55aloiece: just slamming it out there... IT... i guess that a vague "not-description"
13:55aloiece: did you study that?
13:55aloiece: pressing buttons?
13:55imirkin: yes, i majored in pressing buttons
13:56aloiece: I'm thinking about going in for it as well
13:56imirkin: the key is always in knowing which buttons to press
13:56imirkin: and equally importantly, which buttons not to press
14:41xexaxo: next thing you know is you'll be surrounded by big red button with signs "Do NOT press" :-P
14:43imirkin: yeah, like the "shut off all power in the room" button in a datacenter? those are fun.
14:43imirkin: esp when the janitor accidentally backs into it
14:43clements_: we have one of those..its disabled now
14:43clements_: for good reason.. :)
14:43imirkin: and then there's my favourite... the 'oh shit' button
14:44imirkin: which gives you an extra 30s to get out of the room before the air is replaced with some non-oxygen gas
14:44clements_: it gives you a massive headache.. just saying
14:45clements_: and it looks like a KISS concert in your headend
14:45xexaxo: sounds like someone was having fun...
14:46clements_: air handler belt smoked and set off the fire suppression
21:15Hoolootwo: I have a geforce 8400 GS which is acting up on one computer but not another. on one, it shows a pink (magenta?) screen, and on the other it works fine
21:17Hoolootwo: the one which is screwing up gives a lot of errors in the kernel log
21:18imirkin: perhaps pastebin those errors?
21:18imirkin: is one of the computers special in any way?
21:19imirkin: [like, powerpc or other big-endian arch]
21:20Hoolootwo: one is 64-bit, but I don't think that is causing the issue
21:21imirkin: can you pastebin the full dmesg?
21:21Hoolootwo: the weird one does have some aversion to graphics cards though, being a server
21:21imirkin: or did it wrap or something?
21:22Hoolootwo: that was just the nouveau-specific parts
21:29imirkin: sorry, no great ideas spring to mind
21:30Hoolootwo: hmm okay
21:30imirkin: you could maybe boot with nouveau.config=NvForcePost=1
21:36Hoolootwo: well I get a desktop,
21:36Hoolootwo: but it's completely pink except for a few areas of black
21:36Hoolootwo: I can login apparently
21:37Hoolootwo: would trying to put the boot console on a different video card help possibly?
21:46Hoolootwo: hmm nope
21:55Hoolootwo: so I put the other card as the primary card, and it came back with http://www.rafb.me/results/iybMzx60.html
21:55imirkin: [ 9.586176] nouveau E[ DEVICE][0000:0a:00.0] unknown chipset, 0xffffffff
21:56imirkin: wait, is it a PCI card?
21:56imirkin: not PCIe?
21:56imirkin: hmmmm. what if you do 'nouveau.config=NvMSI=0'
21:59Hoolootwo: same error