02:04 _1_Ganeshbboy: hy
13:00 NoReflex: hello! I'm getting some strange messages in the kernel message buffer: ACPI Warning: \_SB_.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20141107/nsarguments-95)
13:00 imirkin_: expected
13:00 imirkin_: (or at least, not unexpected)
13:00 imirkin_: and nothing to worry about
13:01 NoReflex: googling the message indicates it is harmless but in my case it locks up the interface for about one second
13:01 imirkin_: the message is not what locks things up :)
13:01 imirkin_: let me guess...
13:01 imirkin_: GK106 gpu which is powergated and doesn't come up with HUB_INIT timeout errors?
13:01 NoReflex: and from what I can tell there are two messages for each freeze
13:02 NoReflex: imirkin, my notebook is Asus UX52VS: Intel HD 4000 + nvidia 645M
13:02 imirkin_: the message is likely to be seen when the discrete gpu is about to be powered up or down
13:02 imirkin_: NoReflex: lspci -d 10de:
13:03 NoReflex: imirkin, 01:00.0 VGA compatible controller: NVIDIA Corporation GK107M [GeForce GT 645M] (rev a1)
13:03 imirkin_: i guessed wrong
13:03 imirkin_: :)
13:03 NoReflex: imirkin, if I reboot the problem usually goes away
13:03 imirkin_: why is the discrete gpu getting powered on/off?
13:03 imirkin_: are you plugging/unplugging external monitors?
13:04 imirkin_: or running things with DRI_PRIME=1?
13:04 NoReflex: from what I can tell it happens about once every few days (depending on something that is set during the boot) the issue may appear or not
13:04 NoReflex: I'm not running anything with DRI_PRIME and not plugging / unplugging any monitors to the notebook
13:05 imirkin_: are you using nouveau or blob drivers?
13:05 imirkin_: (pastebin your dmesg)
13:05 NoReflex: nouveau (mesa package in archlinux)
13:06 NoReflex: imirkin, http://pastebin.com/AVswMaiK
13:07 NoReflex: the are also two other messages: nouveau E[ PBUS][0000:01:00.0] MMIO write of 0x00000002 FAULT at 0x4188ac [ !ENGINE ]
13:07 imirkin_: looks like something's waking it up every 11 seconds
13:07 imirkin_: which is extremely odd, since... 11 isn't a common wait timeperiod
13:08 NoReflex: it's also odd that very likely if I reboot the problem won't manifest
13:09 imirkin_: NoReflex: do you have a system monitoring thing going on perhaps?
13:09 imirkin_: anyways, i have no clue why this is happening =/
13:09 imirkin_: sounds like sometimes when waking up, it suffers some sort of fail that freezes things up for a bit
13:09 imirkin_: probably holds some giant lock or something
13:09 NoReflex: the information that the messages appear when powering on / off can help me search further
13:10 imirkin_: if you don't care about battery life, you can boot with nouveau.runpm=0
13:10 imirkin_: which will disable the power-off-the-gpu logic
13:11 NoReflex: I have conky runnign every 3 seconds but nothing with nvidia there
13:11 imirkin_: things like running 'xrandr' are likely to wake the card up
13:11 NoReflex: can I boot with the card powered off since I won't use it?
13:11 imirkin_: or 'lspci'
13:11 imirkin_: perhaps you can do that in your bios, not sure
13:12 imirkin_: if you don't load nouveau, it'll stay powered on
13:12 imirkin_: however you can issue an acpi_call manually somehow
13:12 imirkin_: i forget the details
13:12 imirkin_: you could use something like bumblebee instead of nouveau
13:20 NoReflex: I will try to use bbswitch
13:20 NoReflex: thank you!
14:31 voxadam: I
14:32 voxadam: I've been fighting GPU lockups on my workstation since I installed a new distro. It
14:33 voxadam: It's running 4.0.0 final on a 550 Ti (NVCF).
14:33 voxadam: Most times I don't see anything in my logs.
14:36 voxadam: I did manage to capture something today.
14:36 voxadam: https://gist.github.com/voxadam/6bbf98e63e60f697c75d
14:36 voxadam: Scroll to Apr 14 01:08:21
14:37 imirkin_: that's a PDISP hang =/
14:37 imirkin_: which means that you're unlikely to get screen updates after it happens
14:37 imirkin_: even though everything else is working great
14:37 voxadam: Yep. I can ssh in just fine.
14:37 imirkin_: (including gpu rendering)
14:38 voxadam: Very rarely if I try to switch to a vc right as it's crashing I can even use the terminal fine.
14:38 imirkin_: coz i think it does in-place buffer updates
14:39 imirkin_: instead of rendering successive frames and displaying them
14:40 voxadam: There's something about KDE that triggers lockups really quickly but if I login to gnome I can use the system for hours sometimes.
14:46 voxadam: I suppose I should expect little bugs like this when running an alpha distro.
14:47 imirkin_: i doubt the distro has anything to do with it
14:47 imirkin_: i'm guessing the KDE compositor does something that upsets nouveau or the underlying hw
14:48 voxadam: Gnome triggers it too but not nearly as often. KDE does it in minutes, with Gnome it often takes hours.
14:48 imirkin_: clearly KDE is better ;)
14:49 voxadam: :)
14:50 tobijk: even if KDE is crap, it should not happen after all :D
14:50 voxadam: I suppose that as of today my OS is beta, not alpha, although I've been updating an alpha system so who knows what it really is.
14:50 Wonka: voxadam: tried rolling your own kernel?
14:50 voxadam: tobijk: Probably not.
14:50 voxadam: Wonka: I was thinking about it.
14:51 Wonka: voxadam: to rule out patches from your distro...
14:51 Wonka: just to be sure it isn't actually your distro.
14:51 voxadam: I needed to compile some extra modules that fedora doesn't include so why not go all the way...
14:52 tobijk: building a own kernel is pretty easy nowadays anyway
14:52 voxadam: I've done it for years.
14:52 Wonka: ack. if you deactivate some modules you won't ever need, it'll speed up the process a lot, though.
14:52 Wonka: distro kernel configs are juge
14:52 Wonka: huge
14:53 voxadam: I've always found configuring and compiling kernels somewhat relaxing.
14:53 tobijk: text adventure?!
14:55 voxadam: Someone should write a "make adventure-config" config interface for the Linux kernel.
14:55 voxadam: make nethack-config
14:55 Wonka: there's a maze of config options, all alike.
14:56 voxadam: It is pitch black. You are likely to be eaten by an angry kernel dev.
14:56 tobijk: make randconfig and start sorting after that :>
15:07 voxadam: It never cesases to amaze me that the gziped kernel source now tips the scales at 118 megs.
15:07 voxadam: When I started, around 1.2, it was a bit over 2 megs.
15:09 imirkin_: how times have changed
15:09 voxadam: Hmmm... Apparently some kind soul builds and maintains a vanial kernel for Fedora.
15:10 voxadam: Maybe I'll put off a custom build for now.