[02:25]  * hyperair spots a new kernel in xorg-edgers
[02:54] <Dandel> that reminds me... which ppa is the one that keeps up with the absolute latest kernels for ubuntu?
[03:00] <bryce> Dandel, http://kernel.ubuntu.com/~kernel-ppa/mainline/
[03:41] <Dandel> bryce: thanks. also do you know if the raring kernel will easily install on precise? ( not worried about module compat with closed source video card drivers )
[03:46] <bryce> Dandel, the kernel has no dependencies really, so generally should install fine on any distro.  Whether userspace works properly on top of that kernel is a different matter tho...
[07:17] <Dandel> ricotz: looks like piglit w/ opengl ( glx/egl), gles2, and gles1 should be doing well ( latest package build just fine for quantal )
[07:19] <Dandel> and add precise/raring ( 386)... my guess is that amd64 for those should also go well
[07:46] <tjaalton> upgrading my laptop to raring
[07:51] <Dandel> tjaalton: i see... well i'm leaving mine at the setups they are now... i'm actually going to cross reference mesa 9.0 on egl ( w/ x11), glx and gbm on my test cards when i get some time since the features are all now enabled ( if not this weekend, next weekend )
[07:52] <Dandel> i honestly expect the results to be near identical except for the minor differences that are expected.
[08:25] <mtron> hello people. Anyone i could talk to about hybrid-graphics support ? It's not a 'usual' Optimus Xorg bug but some netbook models have a Intel GMA 3150 GPU in the atom cpu and a directly wired ION2 Gpu to the LVDS port and HDMI port. It's possible to set the used GPU in software without bumblebee (i have some proof of concept bash scripts that work) so it would be great if it could be added to ubuntu :) Can someone tell me whom to talk to please?
[08:27] <tjaalton> no it can't
[08:28] <tjaalton> poulsbo doesn't work
[08:28] <tjaalton> period.
[08:28] <mtron> @tjaalton you're talking to me?
[08:28] <tjaalton> yes
[08:28] <mtron> yes it does.
[08:29] <mtron> i have it in front of me :)
[08:29] <tjaalton> ok then, but bumblebee will never get in the distro
[08:29] <mtron> on some specific models you don'T need it
[08:30] <tjaalton> so unless it supports the kernel dma_buf et al, you're busted
[08:30] <mtron> all you need to do is send a acpi call to activate a vga mode for the next boot cycle
[08:30] <tjaalton> still
[08:30] <tjaalton> hacks
[08:30] <mtron> there is a intel only mode, a nvidia mode and a dual gpu mode (optimus)
[08:30] <mtron> and a specific call for each mode
[08:31] <mtron> well i thought ubuntu might be more open to hacks than debian ;)
[08:32] <tjaalton> how so?
[08:33] <mtron> it was in the past. maybe your goals shifted since then...
[08:34] <tjaalton> goal is to get better quality instead of hacks, believe it or not
[08:35] <mtron> no offend intended. was just a question.
[08:35] <tjaalton> proper hybrid support is "around the corner" anyway, for hw/drivers that support it
[08:53] <mtron> @tjaalton i guess you misunderstood me. What i am talking about has nothing to do with dma_buf et all. It sits "below" this layer. The ONLY way to set a different VGA mode (on this specific netbook models i am talking about) is via setting this register in the acpi system. This is not a hack per-se but thats the way to do it (and that's the same way it is done on windows).  I talked to somebody from the dev team during UDS-P on IRC hybrid-graphics-support abo
[08:53] <mtron> ut this and he seemed interested. unfortunately i forgot who it was... 
[08:59] <tjaalton> if by vga mode you mean selecting which gfx chip to use, it's nothing new
[09:00] <mtron> no. again i am talking about something different :)
[09:01] <mtron> there are VGA modes on this laptop that can be set via a acpi call and are valid for the whole boot cycle
[09:01] <tjaalton> what's the point for those?
[09:01] <tjaalton> use case
[09:02] <mtron> in Mode 1 only the intel vga chip is active (nvidia gpu is off so low power consumtion)
[09:02] <mtron> mode 2 is only nvidia
[09:02] <tjaalton> right, what I was talking about
[09:02] <mtron> mode 3 is intel and nvidia
[09:02] <mtron> the traditional optimus mode
[09:02] <tjaalton> so exactly what my thinkpad is able to do from the bios, nothing new
[09:02] <mtron> yes but those models dont have a bios setting
[09:03] <tjaalton> some models can't, and need either userland tricks like this or proper support from the kernel
[09:03] <mtron> the only way to set the mode is via sending a acpi call
[09:03] <mtron> without this call only nvidia is activated (seems to be a fallback mode)
[09:04] <mtron> so you can't event get to the dual gpu mode without a acpi call
[09:05] <mtron> anyway thanks for your answer, maybe someone understands what i am talking about
[09:05] <tjaalton> eh, I do understand
[09:05] <tjaalton> why would an acpi call _from the userland_ be the only way to switch modes?
[09:06] <mtron> thats how it implemented by the hardware manufacture i guess. 
[09:06] <mtron> this calls are also used from windows to set the mode on this laptop
[09:07] <mtron> that's how we found them
[09:07] <tjaalton> mlankhorst: halp
[09:07] <tjaalton> it's not an uncommon use-case, so I'm pretty sure it's thought out already how it's going to work in the perfect world
[09:49] <mlankhorst> hey
[10:02] <MCR1> Dandel: Hi :) I promised to report about fglrx+3 monitors. I had no success...
[10:23] <Dandel> MCR1: ok. and was that in full wide screen?
[10:24] <MCR1> I first tried to use your xorg.conf - did not boot
[10:24] <Dandel> the one with 4096 by 4096?
[10:25] <Dandel> or the one that was set to match mesa.
[10:25] <MCR1> Then I removed all the contradicting stuff from my old xorg.conf and set virtual to 6144x6144
[10:25] <Dandel> ok.
[10:25] <MCR1> Now I am able to turn on/off every combination of 2 monitors
[10:25] <MCR1> but once I add the 3rd
[10:26] <MCR1> I get this error message (reproducing, one moment)
[10:26] <MCR1> The selected configuration for displays could not be applied
[10:27] <MCR1> could not set the configuration for CRTC 147
[10:27] <MCR1> Failed to apply configuration: %s
[10:27] <MCR1> GDBus.Error:org.gtk.GDBus.UnmappedGError.Quark._gnome_2drr_2derror_2dquark.Code2: could not set the configuration for CRTC 147
[10:27] <Dandel> paste xrandr output with all displays connected
[10:28] <MCR1> http://pastebin.com/xWZTF2DU
[10:29] <MCR1> Current xorg.conf: http://pastebin.com/BGywi9xL
[10:30] <Dandel> so this command fails?:  xrandr --output DFP1 --right-of  CRT2
[10:30] <Dandel> ** DFP2
[10:31] <MCR1> Dandel: I did it with the "Displays" GUI.
[10:31] <Dandel> give it a try through command line
[10:31] <Dandel> if it works, bug report to gnome
[10:31] <MCR1> ok
[10:36] <MCR1> Dandel: Just to be sure to make no mistakes - Could you please once again post both of the commands ?
[10:37] <Dandel> the command to enable the display is: xrandr --output DFP2 --auto
[10:38] <Dandel> then set the location: xrandr --output DFP2 --right-of CRT2
[10:40] <MCR1> Dandel: Does not seem to work - fails silently
[10:40] <MCR1> xrandr --output DFP2 --auto
[10:40] <MCR1> xrandr --output DFP2 --right-of CRT2
[10:40] <MCR1> xrandr --output DFP4 --auto
[10:41] <MCR1> DFP2 still does not get any signal
[10:41] <MCR1> black screen
[10:41] <Dandel> try this: xrandr --output DFP4 --off && xrandr --output DFP2 --auto
[10:42] <MCR1> this will probably work, but I do not want my main monitor to turn off ;)
[10:42] <MCR1> but I'll do it
[10:42] <Dandel> then turn off crt2
[10:43] <Dandel> from what i can see, it's artificially limiting the number of screens 
[10:43] <Dandel> and thus needs to be reported
[10:44] <MCR1> trying to reenable gives me this: xrandr --output DFP4 --auto
[10:44] <MCR1> xrandr: Configure crtc 2 failed
[10:44] <Dandel> yea.
[10:44] <MCR1> grmpf - the strange thing is that it worx perfectly with gallium
[10:45] <Dandel> gallium = mesa/open source
[10:45] <MCR1> yes, I know
[10:45] <Dandel> I will report the problem.
[10:45] <MCR1> thanx a lot
[10:45] <Dandel> in fact i already did.
[10:45] <MCR1> oh cool
[10:45] <Dandel> can you go ahead and regenerate the atisysteminfo.txt file and email it?
[10:45] <MCR1> hopefully AMD got some devs left...
[10:45] <Dandel> amd has devs left ( that's guaranteed )
[10:46] <MCR1> :)
[10:46] <Dandel> it's just that i don't know how big the team is ><;
[10:46] <MCR1> how can I generate that file and where can I find it ?
[10:47] <Dandel> sudo /usr/share/fglrx/atigetsysteminfo.sh
[10:47] <Dandel> and outputs to home folder
[10:50] <Dandel> just note that you can run the command over ssh without fail as long as you specify the display variable to :0
[10:51] <MCR1> ok, one moment - I will just reconfigure my screens to something useful - done...
[10:53] <MCR1> Dandel: Here it is: http://paste.ubuntu.com/1396648/
[10:55] <MCR1> Dandel: && thanx again 4 your help here...
[10:55] <Dandel> if i had a third display i'd test it more thoroughly ><;
[10:59] <Dandel> MCR1: i needed the config when you where loaded and running fglrx driver not radeon :/
[11:00] <MCR1> I am running fglrx
[11:00] <Dandel> no your not
[11:01] <Dandel> look at /var/log/Xorg.1.log
[11:01] <Dandel> search for: [     8.615] (II) LoadModule: "fglrx"
[11:01] <Dandel> within 4 lines: it fails saying module of fglrx cannot be found
[11:03] <Dandel> and then it reverts back to ati ( radeon driver )
[11:03] <MCR1> strange...
[11:03] <MCR1> I can open AMD Catalyst Control Center
[11:03] <Dandel> did you install through packages?
[11:03] <MCR1> nope, manually
[11:03] <MCR1> but I am sure fglrx is running
[11:04] <MCR1> I can configure via amdconfig also
[11:04] <MCR1> display: :0  screen: 0
[11:04] <MCR1> OpenGL vendor string: Advanced Micro Devices, Inc.
[11:04] <MCR1> OpenGL renderer string: ATI Radeon HD 5700 Series 
[11:04] <MCR1> OpenGL version string: 4.2.11991 Compatibility Profile Context
[11:05] <MCR1> output of fglrxinfo
[11:05] <MCR1> I did install latest AMD beta driver from their homepage
[11:05] <Dandel>  glxinfo | grep direct\ rendering
[11:05] <ricotz> MCR1, glxinfo!
[11:05] <ricotz> Dandel, ;)
[11:05] <MCR1> direct rendering: Yes
[11:06] <Dandel> MCR1: can you pastebin /var/log/Xorg.0.log
[11:06] <ricotz> MCR1, are you running kernel 3.7?
[11:06] <MCR1> no, 3.5 on Qu
[11:06] <ricotz> ok
[11:06] <MCR1> 3.7 would not work with fglrx or would it ?
[11:07] <ricotz> MCR1, you would need a patch if there werent a new release having it
[11:07] <ricotz> i don't use fglrx
[11:08] <Dandel> ricotz: i use fglrx so i am at least trying to debug it.
[11:08] <ricotz> glxinfo | grep vendor ?
[11:08] <MCR1> Dandel: http://paste.ubuntu.com/1396675/
[11:08] <MCR1> server glx vendor string: ATI
[11:08] <MCR1> client glx vendor string: ATI
[11:08] <MCR1> OpenGL vendor string: Advanced Micro Devices, Inc.
[11:08] <Dandel> now that looks as expected :D
[11:09] <Dandel> ricotz: it was amd auto-script failure.
[11:09] <MCR1> I knew I was running fglrx ;)
[11:10] <ricotz> MCR1, use a "proper" package to install it ;)
[11:10] <MCR1> ricotz: The packages are always outdated...
[11:11] <ricotz> i see, i am not really following those
[11:11] <ricotz> i guess 2:9.010-0ubuntu1~xedgers~quantal1  is outdated?
[11:12] <MCR1> no, sorry - this seems to be actual
[11:12] <MCR1> as I have the same version installed
[11:12] <ricotz> MCR1, have you tried that edgers package?
[11:13] <MCR1> to be honest - no
[11:13] <ricotz> ok
[11:13] <MCR1> but I am sure I've installed fglrx correctly as I am not doing that 4 the first time...
[11:13] <Dandel> MCR1: did you change builds since the last time?
[11:14] <ricotz> MCR1, ah there is a 12.11 beta8
[11:14] <MCR1> Dandel: no, I do not think so... latest beta was already installed when we talked last time
[11:15] <Dandel> so it's the 12.11 beta 8 that was up on their website?
[11:15] <MCR1> ricotz: Yes - that is the one I should use
[11:15] <MCR1> Yes
[11:15] <ricotz> MCR1, why "should"?#
[11:15] <MCR1> I am always installing with --buildandinstallpkg Ubuntu/quantal
[11:16] <MCR1> ricotz: IIRC 99.9%
[11:16] <ricotz> i don't know the changelog of it, just saw it
[11:17] <ricotz> MCR1, the packaging isnt the same as ubuntu's irc
[11:18] <MCR1> in Synaptic it shows the version as 2:9.010-0ubuntu1, which is the same version-num like from xorg edgers PPA (which I have enabled also)
[11:19] <MCR1> but in AMD version numbers it is 12.11 beta8
[11:19] <ricotz> MCR1, the edgers version is 12.11 beta1
[11:19] <MCR1> tell AMD that ;)
[11:20] <ricotz> i am just saying it isnt beta8, i guess i should update it
[11:20] <MCR1> I mean I do not know why they did not make them easily identifyable...
[11:21] <MCR1> in Synaptic they show as same version (except for ~xedgers~quantal1)
[11:22] <MCR1> ricotz: Ofc that would be cool, atlthough I do not think that they differ a lot...
[11:22] <MCR1> *although
[11:26] <ricotz> MCR1, they should be identifiable , 12.11 beta8 is 9.01.8, 12.11 beta is 9.01
[11:26] <ricotz> but yeah, the installers creating identically named source packages
[11:26] <MCR1> that is what I am saying -> they are not (at least not in synaptic)
[11:33] <ricotz> i will upload it as fglrx-installer 9.010.8
[11:38] <MCR1> sounds good :)
[11:53] <ricotz> should be available soon
[12:44] <ricotz> MCR1, package is there now
[12:53] <MCR1> ricotz: yeah, just noticed :)
[12:53] <MCR1> ricotz: The question is - Is it safe to upgrade ? or will it destroy my xorg.conf ?
[12:54] <ricotz> MCR1, i guess you would need to test it, keep a backup of it then
[12:54] <MCR1> ok, I'll test it - but later (will report ofc)
[12:55] <MCR1> because time is runnin' out... :)
[12:56] <ricotz> that is fine
[12:56] <MCR1> thx 4 updating - and great that xorg edgers is taking care of proprietary drivers now as well...
[13:00] <ricotz> nvidia blob is there for ages
[13:00] <ricotz> and fglrx got updated occasionally
[13:00] <ricotz> so this isnt new ;)
[13:01] <MCR1> I remember times when adding xorg edgers will kill your proprietary driver 100% ;)
[13:01] <MCR1> *would kill
[13:01] <ricotz> they are blobs so we cant hold back new xserver versions forever
[13:02] <ricotz> especially fglrx takes forever to support newer video-abis
[13:02] <MCR1> sure, I know that it is not the fault of xorg, but the prop. drivers not supporting new versions of xorg and the kernel...
[13:02] <ricotz> nvidia is way faster to catch up
[13:27] <mlankhorst> yeah fortunately
[13:30] <jcristau> "way faster than fglrx" isn't really saying much
[13:31] <ricotz> not the old stories again
[13:40] <mlankhorst> phoronix, putting numbers in graphs :-)
[14:31] <mlankhorst> ok quick review of the debian rules and the output looks sane..