=== jono is now known as Guest91841 === RAOF_ is now known as RAOF [05:34] mlankhorst, tjaalton: When were you thinking of being all xserver 1.15-y? [05:36] RAOF: guess it depends on fglrx, they know of the ppa already since a month ago or such === tjaalton_ is now known as tjaalton === wgrant_ is now known as wgrant === psivaa_ is now known as psivaa [10:24] RAOF that [10:51] mlankhorst: is that patch of yours to start X with a specific GPU (on hybrid systems) in 14.04? [10:52] tseliot: well I think so [10:52] that's what the inactive screens were for [10:54] mlankhorst: do you mean xf86-inactive-gpuscreen.patch ? [10:54] yeah [10:55] mlankhorst: shouldn't we just use the discrete GPU with nvidia on desktop systems? [10:56] indeed, but that requires setting intel as gpu screen [10:57] which is why inactive is mapped to gpu screen, nvidia used a bug to do that [10:59] aren't both intel and nvidia gpu screens? [10:59] or does the first gpu become master screen? [11:02] mlankhorst: or maybe I misunderstood what you said [11:05] there's 1 master screen and 1 gpu screen [11:07] oh [11:12] mlankhorst: in your patch I only see that set the PLATFORM_PROBE_GPU_SCREEN flag but I don't see anything else. Maybe there's an additional patch? [11:14] no that's all you need [11:22] mlankhorst: but I remember that you could pass X the pci id [11:22] you wrote that patch [11:23] tseliot: correct, but still need to specify what pci-id belongs to primary, and which one to the gpu screen [11:24] mlankhorst: ok but what I don't understand is if we already have code to do that [11:25] tseliot: well nvidia used a bug in xorg-server to specify that intel is a gpu screen, I turned it into a feature :P [11:26] right [11:28] mlankhorst: but I also remember that you told me about some code to launch X with something like: /usr/bin/X :0 -auth... -primary=0:1:0 or something like that to select the card [11:29] oh that [11:29] I don't see it among our patches [11:29] I don't think I ever made a patch to set it on the command line [11:29] only in xorg.conf [11:29] oh the -gpu switch [11:30] I think I dropped that patch [11:30] yes, that was it [11:30] do you still have that patch somewhere? [11:31] no idea, why? [11:31] can't find it in the ubuntu tree at least [11:32] I don't want users to install nvidia and fail on desktop systems where crappy bioses which do not disable the integrated GPU when you plug in a discrete card [11:33] and you don't want to offload rendering in that case [11:35] as in bug #1126234 [11:35] bug 1126234 in ubuntu-drivers-common (Ubuntu) "Does not show nvidia driver if intel card is active" [High,Triaged] https://launchpad.net/bugs/1126234 [11:36] oh that [11:37] my system has that too [11:37] but what do you want to do in that case? [11:44] mlankhorst: I think we can simply detect that we're not dealing with a laptop, and, if that is the case, use the discrete card by default [11:45] mlankhorst: or add an option to do that, and my nvidia-prime package will set it in some configuration file [11:47] how do you know it's not a laptop, then? [11:53] mlankhorst: well, you can check if you have anything in /sys/class/power_supply/ or if you have /proc/acpi/button/lid [11:58] tseliot: I would like something more definitive :P [11:58] tseliot: anyway by default the primary gpu is used [11:59] which is set in the bios [11:59] graphics card initialization order or something [11:59] right now I do: if [ -n "`ls -A /sys/class/power_supply/`" ] && [ -d "/proc/acpi/button/lid" ]; then echo "Laptop detected"; fi [12:00] mlankhorst: right, and I want to be able to ignore that [12:00] i have an ups, wonder if it shows as a power supply.. [12:00] we cannot rely on the BIOS in some cases [12:00] not connected via usb currently [12:00] it should [12:00] in those cases you can use nvidia-prime [12:00] tjaalton: does it also show a lid? ;) [12:01] no :) [12:02] mlankhorst: a user would plug the monitor into the nvidia card and, surprise, he wouldn't see anything [12:02] * tseliot -> lunch [12:02] tseliot: yeah it's terrible, I know [12:02] so Mez's use case from yesterday [12:02] pretty much [12:14] hum, time to go trusty [12:14] on desktop too [12:14] I'm still on precise :p [12:15] well, you get to test the backports :) [12:15] true [12:15] I have precise on another machine [12:15] I will move after verifying lts-trusty [12:15] and latest dev release dualboot [12:16] I'm on lts-raring atm, will move to lts-saucy soon :P [12:21] good boy [12:24] hm seems to all be in precise-proposed now [12:24] oh cool [12:24] finally [12:32] # Xorg -version [12:32] X.Org X Server 1.14.5 [12:32] should probably log in again to use it ;-) [12:34] yay works [12:34] ship it [12:34] definitely [12:46] I didn't even mess up versions this time, I think [12:52] I deleted all lts-saucy packages from s-lts-backport and x-staging. only thing left in x-staging is xorg-server 1.15 for now [12:53] tseliot: any word from fglrx? [12:53] ->#private ;) [12:54] we have a call on thursday [12:54] ah k [13:09] mlankhorst: I think we can fix that use case [13:32] tseliot: should nvidia-173 dkms build on saucy? [13:40] what's the deal with libxcb, is it mergeable by now? [13:42] should be [13:51] tjaalton: does it fail? [13:52] tseliot: there's https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/1257392 where it's missing the kernel module, dunno why [13:52] Ubuntu bug 1257392 in xorg (Ubuntu) "Xorg crash" [Low,Incomplete] [13:56] tjaalton: maybe the kernel headers are not installed? [14:06] yeah thought about that === stgraber_ is now known as stgraber