[01:23] <Sarvatt> looks like i'm not alone getting insane memory usage with intel 2.11
[01:29] <ripps> I know the moment I say this I'm going to Jinx it, but since installed kernel 2.6.33 from the kernel-ppa I havent' had any Xorg cpu hogging.
[01:36] <Sarvatt> tjaalton: have you seen http://cgit.freedesktop.org/~whot/xf86-input-wacom/log/?h=waltop ?
[01:43] <Sarvatt> ripps: did you also update ati to 6.13 at the same time? :D
[01:43] <ripps> Sarvatt: no, I rebooted and tried that before installing 2.6.33
[01:43] <ripps> So, no the ati update didn't fix it
[01:43] <Sarvatt> still had it with 6.13 and the lucid kernel then?
[01:43] <ripps> Sarvatt: yes
[01:46] <ripps> whoa, irssi just crashed all of sudden there..... hope that wasn't an omen
[01:46] <Sarvatt> i was going to say do a xlsclients and start killing off things until you find which it is, most likely something in there
[01:46] <Sarvatt> what mainline kernel did you install?
[01:46] <Sarvatt> 2.6.33.2?
[01:48] <ripps> Sarvatt: inux-image-2.6.33-020633-generic
[01:48] <Sarvatt> do you see anything flooding ~/.xsession-errors when its slow? or dmesg?
[01:49] <ripps> Sarvatt: fitst thing I checked, I didn't see anything
[01:50] <Sarvatt> chromium has been acting kind of similar for me the past week of daily builds but its not making the X process CPU spike after closing it.. I was going OOM after a few hours and thrashing wildly even with 3GB ram, thats why I was asking about that
[01:51] <ripps> Sarvatt: eh, I switched back to google-chrome, it's html5 video is a bit uglier, but it seems a little faster
[01:51] <Sarvatt> but i did have flash using all of my cpu still doing what you said after i closed it one time
[01:51] <Sarvatt> yeah I switched to the in-archive chromium and its alot better
[01:51] <Sarvatt> think its from the beta channel
[01:53] <ripps> I recently helped fta resolve a bug with sse2 with chromium, because chromium was crashing on any html5 video, but even with that fixed, I still think the equivalent version of chrome when compared to chromium is still a bit faster.
[01:53] <ripps> It's probably just how it's compiled. Maybe google uses profiled builds
[01:53] <Sarvatt> interesting if its really fixed by 2.6.33 and not the lucid kernel though, definitely not gpu related if so
[01:54] <Sarvatt> you said mouse options affected it at some point?
[01:54] <RAOF> Sarvatt: I've got a hacked-up nouveau/linux-2.6 building locally.  Mostly.
[01:55] <ripps> Sarvatt: well, I didn't have this problem when I first installed 2.6.32-19. It first appeared after the large number of updates the day before. So I'm guess that an update in xserver earlier caused it.
[01:55] <ripps> Sarvatt: well I was playing around with some stuff, and after disabling mousetweaks the cpu usage of Xorg dropped to around 5% instead of 20%, but a few minutes later it went back up.
[01:59] <Sarvatt> well hopefully it was a client app that was screwed up and fixed, i think gnome-keyring gwibber desktopcouch and gnome-control-center were screwed up pretty bad recently but not sure if any of those had spiking X cpu usage as a side affect.. if you cant reproduce it after a day or so on that mainline kernel i'd be interested if you still have the problem on -19 or -20
[02:00] <Sarvatt> dont know what generation ati you're using but there were a bunch of fixes in mesa 7.7.1 too
[02:01] <ripps> I'm using r300 ati radeon 9600
[02:02] <ripps> I'm wondering if the 2.6.33 has something in it's scheduler that prevents whatever's happening in Xorg to occur
[02:04] <ripps> On an unrelated note, I couldn't get the 2.6.34-rc3 kernel to work because it would turn my monitor off when X started. I couldn't figure out a way around it.
[02:05] <Sarvatt> did you use the lucid or karmic one? i can't use 2.6.34-x-karmic at all, panics right when it starts :( there's always radeon.modeset=0
[02:06] <ripps> the karmic ppa only had header packages
[02:06] <Sarvatt> check your /var/log/dmesg.0 after booting it, and rebooting into a good kernel?
[02:08] <ripps> Sarvatt: the archived dmesg log doesn't say anything that stands out to me
[02:09] <ripps> and the Xorg.1.log says that it couldn't find any screen(s)
[02:10] <ripps> it seems my monitor wasn't initialized
[02:14] <ripps> y'know, every once in a while, I get an audio stutter and Xorg goes up to 15% for a moment, but than it seems to recover and go back to ~3%. I wonder if the problem is still occuring but the kernel is able to recover more elegantly than lucids
[03:07] <Sarvatt> 2752 objects
[03:07] <Sarvatt> -2109001728 object bytes
[03:07] <Sarvatt> kaaay
[03:11] <Sarvatt> thats with stock lucid packages too, hmm..
[03:12] <Sarvatt> ahh my mesa texture tiling setting was carried over after the downgrade
[03:20] <Sarvatt> ripps: try installing linux-backports-modules-alsa and use the lucid kernel?
[03:21] <ripps> Sarvatt: really? okay, I don't know how it could be related...
[03:24] <ripps> Hmmm... I'm getting a lot of gdk errors for gmpc in my .xsession-errors log... think that could also be a cause?
[03:25] <Sarvatt> what errors?
[03:25] <ripps> (gmpc:8052): GdkPixbuf-CRITICAL **: gdk_pixbuf_get_pixels: assertion `GDK_IS_PIXBUF (pixbuf)' failed
[03:26] <ripps> I have gmpc start at boot, and leave running a seperate workspace
[03:28] <JanC> hm, question, could anybody recommend a (cheap) graphics card that supports OpenGL 2 with open open source drivers in lucid?  ☺
[03:29] <ripps> Does any oss driver have complete opengl 2?
[03:32] <JanC> I wish I knew...
[03:33] <johanbr> according to http://www.x.org/wiki/RadeonFeature , the radeon driver has that on R600/R700
[03:34] <johanbr> on the other hand, most of the 3d features say "Mostly done"
[03:37] <JanC> I suppose some of the newest Intel drivers might support it on their newest hardware too
[03:37] <johanbr> I hear the intel drivers have been in pretty bad shape lately, though
[03:38] <JanC> but intel doesn't sell "cards"
[03:39] <ScottK> johanbr: My experience has been Intel was 'unfortunate' in Jaunty/Karmic, but pretty good so far in Lucid.
[03:39] <JanC> sointel has been good on karmic for me
[03:39] <johanbr> ahh... happy to hear that's improved
[03:40] <JanC> only jaunty was bad
[03:40] <JanC> and I have 3 different intel systems...
[03:42] <ripps> Hmm... according to xlsclients, ubuntuone-login and ubuntuone-preferences are in use, is there any way to stop them?
[03:44] <ripps> hmm killing them seems make things a little smoother
[03:46] <ripps> Well, I'm going back to the lucid kernel to see if I can isolate use xlsclients to see if there's a particular app that's causing cpu hogging.
[03:51] <Sarvatt> 965+ intel, R600+ ati with mesa 7.8+
[03:55] <ripps> My new favorite tool: watch. easy to check changes to various logs at a glance from my screen terminal
[04:02] <JanC> Sarvatt: was that an answer about my OpenGL 2 question?  if so, I was looking at recommendations about R600+ ATI cards that actually work well I suppose...  ;)
[04:39] <bjsnider> JanC, i'm going to regret asking this question i guess, but why the "open source drivers" condition?
[07:00] <JanC> bjsnider: because I prefer that, even if only to give a "message" to manufacturers  ☺
[07:03] <JanC> and I don't really need it, I just wanted to try out an application that apparently needs OpenGL 2
[07:50] <tjaalton> Sarvatt: yes, have you seen http://sourceforge.net/mailarchive/forum.php?thread_name=alpine.DEB.2.00.1004091001160.30959%40deckard.hut.fi&forum_name=linuxwacom-devel :)
[07:50] <tjaalton> Sarvatt: merged the waltoptablet kernel driver with wacom
[07:50] <RAOF> JanC: I recently got a radeon 43something or other for $60 that Just Works™ and supports OpenGL 2.0 with the free drivers.
[17:45] <ripps> Does anybody know how to read straces of Xorg? What exactly is normal?
[17:49] <ScottK> bryceh: Any objection t me uploading a fix for Bug #560814?
[20:24] <Sarvatt> wow, didn
[20:24] <Sarvatt> t realize it was so easy to edit insyde efi bioses
[20:24] <Sarvatt> ahh wrong channel sorry :)
[20:29] <Sarvatt> ugh so it looks like theres quite a few lenovo laptops with intel needing powersave=0 too.. maybe we should have just made it opt-in globally
[20:34] <Ng> what actually happens when you set that?
[20:44] <Sarvatt> depends on the chipset, it disables things like self refresh CxSR and framebuffer compression
[20:44] <Sarvatt> Ng: do you have one of the machines experiencing the problem?
[20:45] <Sarvatt> can you boot with drm.debug=0x04 and watch dmesg to see what happens when it flickers? that'll help alot figuring out what feature is causing it
[20:50] <Ng> Sarvatt: my X301 (Gm45) intermittently has flashing issues. I can't really say whether powersave=0 changes anything yet, it's so annoyingly rare :(
[20:52] <Sarvatt> yeah all the affected ones i've seen still are gm45 too, all lenovo too oddly
[20:53] <Sarvatt> can you describe the flicker at all?
[20:53] <Ng> I have a vast selection bias towards lenovos, since a lot of canonicalers buy them, but fwiw one of them is an IdeaPad, which I think is quite different internally to a Thinkpad, but I'm not sure
[20:53] <Sarvatt> ideapad shouldn't be flickering at all anymore..
[20:54] <Ng> it could be that it stopped, I'll check with ivanka tomorrow
[20:54] <Sarvatt> those should have been fixed in the -18 or -19 kernel
[20:55] <Ng> I'm booted without powersave atm and I've not seen it for a while, but it's not that predictable
[20:55] <Sarvatt> Ng: if you can boot with drm.debug=0x04 on your x301 and send along your dmesg it would help
[20:56] <Sarvatt> like does the whole screen shift for a fraction of a second? or is it isolated to part of the screen? does it black out or just shift the screen contents over a little?
[20:56] <Ng> Sarvatt: just attach the output after booting, or when it actually flashes?
[20:56] <Sarvatt> and does it happen when the system is idle or right after you do something?
[20:56] <Ng> it's like the bottom 1/3-1/4 goes slightly out of sync and corrupted for like 1 or 2 frames
[20:56] <Sarvatt> either, preferably the latter but you said it was hard to trigger
[20:57] <Ng> hmm, I'm not sure about the idleness factor
[21:00] <Sarvatt> any idea if its isolated to something like during video playback?
[21:01] <Ng> I'm pretty sure it's not related to video playback, but perhaps I had flash open in browser tabs at the time. certainly not something like totem though
[21:04] <Sarvatt> Ng: do you have another monitor plugged in?
[21:04] <Ng> Sarvatt: it generally seems to only happen when I have no monitor attached. At work I have an LCD hooked up to the DisplayPort connector and it's *never* happened then
[21:05] <Sarvatt> self refresh is disabled when you have more than one monitor plugged in
[21:05] <Sarvatt> Ng: yeah a dmesg just after boot with drm.debug=0x04 would be handy
[21:06] <Ng> ok, lemmie just close everything down and get that now
[21:06] <Sarvatt> appreciate it Ng
[21:07] <Ng> very much likewise :)
[21:14] <Sarvatt> heyo tormod
[21:15] <Ng> Sarvatt: added
[21:15] <Sarvatt> Ng: bug number?
[21:16] <tormod> konnichi wa Sarvatt :)
[21:19] <stgraber> hey tormod !
[21:19] <Ng> Sarvatt: bug #538648
[21:20] <Sarvatt> thanks, odd that it didnt show up in my ubuntu-x-swat mail
[21:20]  * ScottK says "What the heck" and uploads.
[21:21] <ScottK> bryceh: Nevermind.  Uploaded it.
[21:22] <tormod> bonsoir stgraber :)
[21:24] <Ng> Sarvatt: what influence does the idleness of the machine have? some part of the chip being powered down or so?
[21:29] <Ng> ooh
[21:29] <Ng> it just flickered
[21:29] <Ng> and again
[21:29] <Ng> the drm debug thing is only logging cursor changes, and it doesn't seem to be related to that
[21:32] <stgraber> tormod: it's still the afternoon here ;)
[21:33] <Sarvatt> thats odd, Ng can you try drm.debug=0x05 then?
[21:34] <Ng> Sarvatt: sure, lemmie just close everything down again ;)
[21:34] <Sarvatt> looks like self refresh and fifo watermark updates are 0x01 not 0x04 so 0x05 will get both of those
[21:34] <Sarvatt> Ng: actually just drm.debug=0x01 would be enough
[21:34] <Ng> ok
[21:34] <tormod> stgraber, guess you're in Quebec then :)
[21:34] <Sarvatt> sorry about that, I thought that info was given in 0x04
[21:35] <stgraber> tormod: yep, will be in Switzerland for two weeks between the 24th of April and the 9th of May. Not sure if there's anything getting planned for the release there though.
[21:37] <Ng> Sarvatt: wow that's a lot of output :D
[21:38] <Sarvatt> yeah :( save it now if you can before it overflows then save the rest after it happens (if it does)
[21:38] <tormod> stgraber, not much cooking for an RP here (yet) except something in Winterthur
[21:39] <Ng> Sarvatt: looks like it overflowed pretty quickly after booting, as soon as my desktop appeared I launched a terminal and did dumped dmesg to a file
[21:39] <Ng> and it's lacking kernel boot messages
[21:39] <Ng> but I'm guessing it's all gone to syslog
[21:39] <Sarvatt> thats ok you'll have the boot stuff in /var/log/dmesg
[21:41] <Ng> syslog from boot->now is 116MB :o
[21:41] <Ng> I'm not sure I can leave this on, but hopefully it'll flicker soon
[21:42] <Sarvatt> well can ya just run intel_reg_dumper and attach that, and attach the /var/log/dmesg from this boot to the bug too?
[21:43] <Sarvatt> i'll upstream it after ya put those on there and I get home
[21:44] <Ng> Sarvatt: sure thing
[21:49] <Ng> Sarvatt: done
[21:49] <Sarvatt> i dont know why g4x_update_wm and i965_update_wm self refresh info goes to DRM_DEBUG when i9xx_update_wm and older goes to DRM_DEBUG_KMS, thats what threw me off
[22:18] <kklimonda> hey, what to do if crt monitor sends invalid edid info?
[22:21] <kklimonda> http://pastebin.org/146979 Xorg.0.log is anyone is interested
[23:12] <Ng> Sarvatt: just had a glitch and I hadn't rebooted :)
[23:16] <Ng> snagged dmesg and gpu regs and added them to the bug
[23:16] <Ng> and now I'm definitely rebooting to get this debug mode cleared ;)
[23:51] <Sarvatt> Ng: just to be clear, it only happens with compiz enabled for you too?
[23:51]  * Sarvatt is filling out the upstream bug report and using your info
[23:52] <Ng> Sarvatt: I've only seen it with compiz because I haven't tried without compiz :/
[23:52] <Sarvatt> ah ok, too bad theres nothing in that new dmesg you posted since it was only a seconds worth :D
[23:53] <Ng> woah
[23:53] <Ng> hrm
[23:53] <Ng> syslog will have it even though I've rebooted
[23:53] <Ng> current syslog is 1.9GB :O
[23:53] <Sarvatt> sheesh! :D
[23:53] <Ng> I might wait until I'm in the office tomorrow to upload that ;)
[23:55] <Sarvatt> grep it for [drm:g4x or [drm:intel ?
[23:58] <Ng> grepping out drm:drm_ioctl and gzipping gets it down to 869K