tjaaltonbryce: I'm tempted to use git for -intel.. since we need to pull in a number of commits from master for it to build08:26
tjaaltonI've got mesa & libdrm ready, but intel needs some work still08:27
brycetjaalton: I'd be disappointed to see that08:32
tjaaltonok then08:35
tjaaltonmerging by hand is not that straightforward, since grab-merge.sh doesn't work with experimental08:35
tjaaltonwell, it's MoM that doesn't, but anyway08:36
tjaaltonlibdrm & mesa uploaded09:14
tjaaltonand intel too09:23
tjaaltonAlexia_Death: what patch did you need to xorg-server?09:57
sorenI'm running Jaunty on my laptop, and my X performance alternates between decent and abysmal. In the latter case, the mouse curser "jitters" somewhat when I move it around the screen, and X eats almost an entire CPU core. Disabling compiz didn't help at all.10:40
sorenI would say it's decent around 15% of the time, and absolutely horrible the other 85% of the time.10:41
sorenOh, the graphics card is a intel 945, by the way.10:42
tjaaltonsoren: I've uploaded a new libdrm/mesa/intel, so report back if those are any better or worse10:44
sorentjaalton: When was this?10:51
sorentjaalton: Ah, 48 minutes ago, according to launchpad :)10:53
sorenI'll wait a bit.10:53
crevettesoren, 1 hour ago10:53
tjaaltonright, it'll take a while since libdrm needs to be built & published first10:55
sorenRight. I'll give the ones in xorg-edgers a chance until then..10:55
* soren restarts X10:55
tjaaltonhmm, I wonder if you are seeing bug 30730610:57
ubottuLaunchpad bug 307306 in gnome-power-manager "upgrade to 2: makes session utterly slow" [High,In progress] https://launchpad.net/bugs/30730610:57
tjaaltonlunch ->10:58
sorentjaalton: I've just taken gnom-epower-manager out of the equation, and it's still really, really slow.11:06
tseliotsoren: what's the problem?11:11
tseliotor what are the symptoms?11:12
tseliotdoes killing the gnome-settings-daemon help?11:12
sorentseliot: Mouse pointer jitters as I move it around the screen, and performance is absolutely horrible.11:13
sorenAbout 85% of the time.11:14
tseliotsoren: kill the gnome-settings-daemon11:14
sorenRight now, I'm experiencing the lovely 15% of the time, where everything is fine, so I'm trying to get some work done while it lasts :)11:14
tseliotupstream hasn't adopted my latest patches yet11:14
tseliotsoren: or, if you want, you can apply the patch attached in comment 10 directly: http://bugzilla.gnome.org/show_bug.cgi?id=56816011:16
ubottuGnome bug 568160 in libgnome-desktop "Gnome Settings daemon causes high CPU usage with an expensive call" [Major,Resolved: fixed]11:16
sorentseliot: Thanks for the hint. I'll look at it when it starts happening again.11:17
tjaaltonsigh, missing comma in libdrm's debian/control11:45
tjaaltonthe result was that libdrm-dev didn't depend on the new linux-libc-dev11:47
tjaaltonand why do the buildd's use an older version of xserver-xorg-dev..11:50
tjaaltonwould've thought they always fetched the latest one11:50
tjaaltonhmm, actually it is the latest one11:53
tjaaltonsince I didn't upload the merge yet, duh11:53
tjaaltonnew synaptics merge uploaded12:32
sorentjaalton: The buildd's should always use the most recent version of everything.12:55
tjaaltonsoren: yes, it was my mistake12:55
sorenRight, I just wanted to confirm anyway. :)12:55
tjaaltonI'll remember now :)12:57
tjaaltontoo bad that the ports archs have an outdated kernel, so basically nothing will build on them until they are updated to .2812:57
tjaaltonnothing X related.. drivers, server etc12:58
tjaaltontseliot: you aware of the nvidia bug where after a while (~days) of using compiz the contents of new windows are not refreshed?14:45
tseliottjaalton: no, what bug is it?14:46
tjaaltonI thought it was a bug fixed long ago, but still happens in 180.22 (was using 169.12 on hardy before..)14:46
tjaaltonwell, I just described it, but I guess no-one else is using a 30" monitor :)14:46
tjaaltonso I guess the video memory is exhausted or something14:47
tseliotwhat resolution are you using?14:48
tjaaltonand sometimes when I unlock the screensaver, the whole screen is behaving like that. every redraw is painfully slow14:48
tjaaltonbut that can be fixed by visiting a vt briefly14:49
tseliota memory leak, perhaps?14:49
tseliotalso, can you try to reinstall nvidia-glx-VERSION and restart X14:50
tjaaltonrestarting X helps, but I don't think a reinstall is needed14:51
tjaaltonguess I need to file this14:51
tseliotan update of X made the system very slow here, even without breaking direct rendering14:51
tseliotit's because of a bug which I have to fix14:52
tjaaltonwhich bug?14:52
ubottuLaunchpad bug 309116 in nvidia-graphics-drivers-96 "when removing nvidia-glx-177 /usr/lib/libGL.so is gone" [Medium,In progress]14:53
tseliotit's easy to fix14:54
tjaaltonhmm, the nvidia problem might be due to "powermizer"15:13
tseliottoo bad it's closed source, otherwise I wouldn't mind fixing it myself15:41
tjaaltonbug 270377 for instance15:41
tjaaltonand several threads on nvnews15:41
ubottuError: Could not parse data returned by Launchpad: Unknown host. (https://launchpad.net/bugs/270377/+text)15:42
ubottuError: Could not parse data returned by Launchpad: Unknown host. (https://launchpad.net/bugs/270377/+text)15:42
tjaaltonwrong package though15:43
tseliottjaalton: did you subscribe Aaron Plattner to the bug15:45
tjaaltonlooks like he's well aware of it :)15:45
tjaaltonat least should be15:45
tjaaltonthe driver could install the modprobe file to work around it15:46
tjaaltonduh, the amd64 buildd seems to be busy or broken somehow15:48
tjaaltonbusy it seems..15:50
tseliotwhat modprobe file?15:50
tjaaltona file to add options for the module15:50
ubottuLaunchpad bug 270377 in nvidia-graphics-drivers-180 "content of window do not redraw automatically (intrepid)" [Undecided,Confirmed]15:51
tseliotdoes it affect only some specific models?15:53
tjaaltonI don't know15:53
tseliotas I would like to implement some quirks module for Jockey sooner or later15:55
tseliottjaalton: I've just tried my patch for the "nv" driver to add the support for my card and it works16:47
tseliotI'll send it to Aaron Plattner16:47
tseliotshall I also file a bug report about it on launchpad?16:47
tseliotjust to keep track of things?16:48
mnem0tjaalton: I was hit by that intel waitForVBlank freeze bug and I had tested the "dont enable vblanks on disabled pipes" drm patch which worked great on my machine... but today when the .28-5 kernel installed my intel G45 machine got unbootable --> https://bugs.launchpad.net/ubuntu/+source/linux-meta/+bug/32052516:54
ubottuLaunchpad bug 320525 in linux-meta "jaunty unbootable on intel G45 since .28-5 kernel update" [Undecided,New]16:54
tjaaltonmnem0: checkint18:01
tjaaltonmnem0: uh.. .NEF18:02
tjaaltonplease attach only jpegs or so18:02
mnem0tjaalton: sry I converted it before but then I attached the wrong file.. I attached jpg as well now19:00
mnem0plus xorg logs and dmesg19:00
mnem0will check without usplash in a sec19:00
bryce<bryce> jdstrand: hmm19:12
bryce<jdstrand> bryce: failsafe mode does give me this:19:12
bryce<jdstrand> (EE) open /dev/fb0: No such file or directory19:12
tjaaltonbryce: what's that?19:18
brycejdstrand ran into that after upgrading to latest X bits (but not -intel)19:19
brycehe's run into a situation where it doesn't start up, and failsafe mode is also failing with corrupted screen19:20
jcristaulogs or it didn't happen ;)19:20
bryceI suggested he update to the newest -intel19:20
brycejcristau: heh19:21
jdstrandbryce: amd64 -instal is still in 'needs building'19:21
brycejdstrand: price of admission is posting your Xorg.0.log[.old] files somewhere we can see them19:21
tjaaltonthe thing is that the other amd64 buildd is stuck with qt4-x1119:21
tjaaltonso amd64 is left behind19:21
brycetjaalton: do the changes require a rebuild of the driver (or maybe just -intel)?19:22
jdstrandI can post the logs, but it seems maybe I just don't have the necessary bits cause of poor upgrade timing19:22
tjaaltonbryce: if it failed to build, then just retrying is enough19:23
tjaaltonbut "needs building" means that the queue is just too long19:23
tjaaltonyellow(amd64 buildd) has been building qt4-x11 for three days now19:24
brycetjaalton: no I mean, would using a new xserver/mesa with old -intel cause problems (e.g. ABI mismatches)?19:24
tjaaltonI know for a fact that with the old intel and new libdrm X won't start19:24
brycethree days??19:24
jdstrandXorg.0.log: http://paste.ubuntu.com/108701/19:24
tjaaltonbryce: yes19:25
brycetjaalton: aha19:25
tjaaltonand without the new mesa compiz fails19:25
jdstrandXorg.0.log.old: http://paste.ubuntu.com/108702/19:25
tjaaltonbut this is jaunty so I didn't feel like adding a bunch of Breaks :/19:25
brycetjaalton: do you think it'd make any difference if he were to manually rebuild the old 2.5 -intel against the new X bits?  e.g. is it just ABI breakage, or API stuff too?19:26
tjaaltonbryce: no idea.. I reported that upstream and they asked if 2.6 works. when it did, they closed the bug :)19:26
tjaaltonbetter just grab the new source and build that19:26
tjaaltonyeah, the logfile matches mine19:27
bryceok cool19:27
tjaaltonmnem0: maybe the same issue for you then19:27
jdstrandbryce: apparently I did *not* downgrade libdrm2, just libdrm-intel119:27
brycejdstrand: aha, that could do it19:27
jdstrandbryce: I put both at 2.4.1-0ubuntu9 and it started19:27
mnemoaha, did you downgrade libdrm-intel1 by mistake or what happened?19:28
tjaaltonthat works19:28
jdstrandmnemo: both upgraded without the new -intel. during triage, I only downgraded the one19:28
tjaaltondamn g-s-d keeps crashing on me19:28
tjaaltonbryce: btw, the new mesa now uses vblank for intel, because the drm drivers were fixed. It means that we'll get a ton of bugreports saying 'glxgears showing only 50fps!!!1!' :P19:30
tjaaltonit's been the default since 7.1, but it was fixed only recently (jbarnes)19:31
tjaaltonso we had it disabled19:31
* bryce nods19:31
brycewas wondering about that19:31
brycekeithp mentioned it to me a while back and wondered when it'd hit ubuntu :-)19:32
tjaaltonI git-bisected it last summer19:33
tjaaltonmy laptop froze when the display was turned off19:33
tjaaltonby the screensaver19:33
brycetjaalton: do fps == refresh rate exactly now?19:40
jcristauin general, fps <= refresh rate. in glxgears, it's ==19:40
tjaaltonat least in glxgears19:40
bryceso regarding the worry of user complaints I think it should be straightforward to explain19:41
pwnguinspeaking of glxgears19:41
bryceif we see a lot of bug reports to this affect I can probably rig up an automated reply19:41
tjaaltonbryce: yes; 'glxgears is not a benchmark'19:41
bryceer, s/affect/effect/19:41
pwnguindoesn't it require -iacknowledgethisisnotabnechmark option?19:41
jcristaupwnguin: not anymore, sadly19:42
pwnguinjcristau: any reason not to put it back?19:42
bryceusers don't appear to know what a benchmark is anyway19:42
mnemocould someone explain to me what it means to have vblank enabled versus disabled? and also why would that result in lower fps in glxgears? and finally who is that lower FPS count not a problem?19:42
tjaaltonmnemo: you don't see the "extra" fps anyway19:42
brycemnemo: sure19:42
bryceyour monitor can only show a certain number of frames per second19:43
pwnguinvblank is the interrupt that tells the hardware a new frame can be drawn19:43
brycethis is called your "refresh rate"19:43
mnemoright vblank used to be when the electron beam returned to the other side of the screen or somethin right?19:43
pwnguinenabling vblank handling means you only need to render when a new frame is available to display19:43
brycefor LCD's, 60 hz is common19:43
mnemorights to LCD is 60 FPS sort of?19:43
pwnguinwhen its disabled, the fps count is lying to you19:44
bryceso the xserver/-intel is now being **really clever** and timing its screen draws to match what your hardware is actually capable of19:44
bryceso if it were putting out 120fps, your monitor would only show every other one anyway19:44
mnemoaahh so with vblank enabled im actually saving some CPU by not wasting time rendering frames that wont be shown anyway?19:45
pwnguinmatching fps to refresh rate should reduce power consumption19:45
bryceso why bother?  save some processor cycles and only send the ones that are actually going to get drawn19:45
tjaaltonmm, dogfood19:45
bryceit's very cool19:45
pwnguinmnemo: well, it gets a bit complicated, but basically yes19:45
mnemothat seems like a very nice feature indeed19:45
pwnguinif look up double-buffering in wikipedia19:45
brycebut yeah this means glxgears is going to show much different fps measurements19:45
mnemommkay makes sense19:45
tjaaltonI actually get over a 1000 now without vblank :)19:45
pwnguinglxgears is a shitty measurement though19:45
bryceI think we can say if your glxgears fps match your monitor refresh rate, you're running optimally19:46
tjaaltonon 96519:46
pwnguinglxgears numbers track "buffer clearing" speed19:46
brycetjaalton: heh, you realize people are going to turn off vblank now, as a "performance improvement"  ;-)19:46
pwnguinie how fast your hardware can clear out the old frame for a new one19:46
tjaaltonbryce: I bet.. "here's my xorg.conf, you can use it"19:46
bryceit'll be worth chuckles19:47
tjaaltonthere were posts like that on nvnews.net.. showing everything like what modules to load and stuff19:47
pwnguinglxgears has no textures, barely any geometry, no lighting, or anything relevant to the apps users actually care about19:47
pwnguinyou could probably replace glxgears with "grep (ati|nvidia|fglrx) /var/log/Xorg.0.log" and not lose any information19:48
brycewe could be clever and hack glxgears to replace the printouts of xxx fps with % refresh rate19:49
bryceso it'd print 100% 101% 99% 100% ...19:50
tjaalton"don't touch the mouse!"19:50
tjaalton(to keep it ~100%)19:50
tjaaltonoh, aaronp posted on the "x-server breakage" forum thread, and confirmed that the next releases will support 1.619:54
bryceyep, I saw a reply from him in one of the -nvidia bug reports saying similarly19:54
bryceI need to email them again... monday maybe19:54
anderskThe most recent Jaunty updates caused two problems.  First, my touchpad is moving around the screen way faster than normal. 20:36
anderskAnd second, if I drag any window to any edge of the screen, it immediately jumps to the upper-left corner. 20:37
andersk(The latter occurs with Compiz only.) 20:37
tjaaltondrag with the touchpad or same with the mouse?20:37
anderskWith the touchpad.  Let me go find a mouse... 20:38
tjaaltonthere probably are some issues with the new synaptics20:38
tjaaltonbecause some defaults have been changed20:38
anderskWith the touchpad only.  So yeah, this points to a synaptics problem. 20:39
tjaaltonok, please file a bug against it20:40
anderskWill do. 20:40
jcristaumaybe you could play with xinput to see which settings affect it20:40
tjaaltonright, there are not many changes upstream since 1.0rc320:42
anderskIt looks like most synaptics bugs are being reported against the source package xserver-xorg-input-synaptics, even though it was renamed in Hardy to xfree86-driver-synaptics. 20:47
tjaaltonyes, because it's called that in debian20:50
tjaaltonbut we "support" the old name too20:50
tjaaltonwell, the debian package really should be renamed to that20:50
tjaaltonso in future it should be called x-x-i-synaptics again20:50
jcristauyeah xfree86-driver-synaptics doesn't make much sense anymore20:55
tjaaltonespecially as the git repo is called x-x-i-s20:56
jcristauit's to make sure you get confused20:56
tjaaltonI have :)20:56
jcristauevery time20:56
tjaaltonyou surely have the power to rename it?-)20:56
anderskbug 320639 20:57
ubottuLaunchpad bug 320639 in xserver-xorg-input-synaptics "Touchpad movement problems in 0.99.3" [Undecided,New] https://launchpad.net/bugs/32063920:57
jcristaui think i'd rather rename the source package than the git repo20:57
tjaaltonjcristau: of course :)20:57
tjaaltonandersk: thanks20:57
pwnguinwhat's the xorg.conf setting i need to re-enable zapping?20:59
jcristaupwnguin: Option "NoDontZap"20:59
anderskOption "DontZap" "False" in ServerFlags. 20:59
pwnguinnouveau isn't quite to the point where i don't need it ;)20:59
jcristau(it's like Option "NoXaaNoOffscreenPixmaps" "false" all over again)21:00
pwnguinwell, i agree with the idea21:00
pwnguini just live more dangerously than is prudent21:01
jcristaui meant the fact that the option name has a negation in it21:01
_MMA_I switched video cards (both nvidia) now I cant get X to start at all. Blew out my Xorg. Tried to use vesa in xorg. No screens found. Nothing. Is there some trick to get it to build the module again or something? DKMS?21:07
tjaaltoneither specify the BusID or take one out21:12
tjaaltonI misread :)21:12
bryce_MMA_: you forgot to post a link to your Xorg.0.log ;-)21:13
_MMA_I should note this is a SLI setup so it could have to do with the busID.21:13
tjaaltonso you _do_ have two of them? yes you need the BusID with 8.10 or jaunty21:14
_MMA_tjaalton: Yeah. Went from 1 card (9 series) to 2 (both 7 series). Runnin' Intrepid.21:15
_MMA_I guess X wasn't as bullet proff as I thought. :P21:15
tjaaltonthe bug is upstream, but unfortunately not seeing much action21:16
tjaaltonalthough it's on the list of blockers for xserver 1.621:16
tjaalton(well, I added it)21:16
_MMA_I think I have the old xorg with both cards somewhere.21:16
_MMA_Thing is, the live disk gives me X. Im just trying to get, *something*21:17
* _MMA_ pokes around.21:17
tjaaltonlivecd does? strange21:18
tjaaltonit shouldn't21:18
anderskSo here's another problem.  gnome-screensaver always fails to unlock the screen now. 21:20
anderskAfter I enter the correct password, gnome-screensaver-dialog sits there displaying "Checking..." 21:20
anderskAnd top shows that the CPU is spinning, with about 65% gnome-screensaver-dialog and 35% Xorg. 21:20
tjaaltonworks here21:22
_MMA_tjaalton: It was the BusID. Thanx for the reminder.21:31
anderskI've bisected the gnome-screensaver unlock problem to the gtk+2.0 2.14.5-1ubuntu2 -> 2.15.0-0ubuntu1 upgrade. 21:50
tjaaltonI've got it too21:51
tjaaltonthat version of gtk+21:51
anderskHuh. 21:51
tjaaltonmaybe you are missing something..21:51
tjaaltonwhich mirror?21:52
anderskarchive.ubuntu.com 21:52
tjaaltonok, same21:52
anderskAha, got it.  I can only reproduce with "export GTK_IM_MODULE=xim" in ~/.gnomerc. 22:02
tseliottjaalton: did you upload a new version of the synaptics driver?22:06
tjaaltontseliot: yes22:06
tjaaltonsome defaults have chagned22:07
tseliotI have noticed these 2 bug reports:22:07
ubottuLaunchpad bug 320632 in xfree86-driver-synaptics "tap-to-click and edge-scrolling broken in Jaunty" [Undecided,New]22:07
ubottuLaunchpad bug 320639 in xserver-xorg-input-synaptics "Touchpad movement problems in 0.99.3" [Undecided,New]22:07
tseliotI'll try the new driver tomorrow and if it bugs me I'll write a patch22:07
tseliottjaalton: oh and did you read about my patch for nv22:08
tjaaltonyes, and saw that you posted it upstream22:08
tseliotwhat are we going to do about it?22:08
tseliotshall we include it as a patch in ubuntu or wait for a new release of nv?22:09
tjaaltona patch is fine22:10
tseliotand do you want me to file a bug report about it?22:10
tjaaltonpreferably yes22:10
tjaaltonmight get lost otherwise22:10
tseliotok, good22:10
tseliotyes, it better if we keep track of the patch22:11
tjaaltonwhat the heck22:12
tjaaltonDRI2 works, but I do get bigger fps than vblank with glxgears22:12
tjaaltonthis is with UXA22:12
tseliothow much is it?22:13
jcristaudoes it matter? :)22:13
tseliotor is it too fast?22:13
tjaaltonwell it should show ~5022:13
tjaaltonbut it's not22:14
tjaaltonwith EXA it does22:14
tjaaltonbut then I get no DRI222:14
jcristaui'm not sure dri2 has a mechanism for vsync yet22:14
tjaaltonprobably so22:16
tjaaltonoh well, EXA is the default anyway22:16
anderskI reported bug 320666 for gnome-screensaver and gtk+2.0. 22:18
ubottuLaunchpad bug 320666 in gtk+2.0 "gnome-screensaver fails to unlock with GTK_IM_MODULE=xim after gtk 2.15.0 upgrade" [Undecided,New] https://launchpad.net/bugs/32066622:18
brycetjaalton: http://people.ubuntu.com/~bryce/totals.svg22:18
tjaaltonbryce: wow :)22:19
tjaaltonbtw, we should probably move the -mouse/-kbd bugs over to evdev if they still apply22:20
brycehow do we determine that they still apply?  ask users to retest jaunty?22:21
tjaaltonsomething like that..22:21
tjaaltonnot that many of them22:22
tjaaltonand some are likely misfiled22:22
tseliottjaalton: here's my report with the patch: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-nv/+bug/32067122:22
ubottuLaunchpad bug 320671 in xserver-xorg-video-nv "nv driver doesn't include the pci-id of my GeForce 7300 GT" [Undecided,New]22:22
tjaaltontseliot: ok thanks22:23
brycethere might also be some -evdev bugs misfiled against xkeyboard-config22:23
bryceI took out all the hotkey bugs from there and that cut it down a lot22:23
tjaaltonso unplugging a wacom tablet really kills the xserver22:39
tjaalton..and that concludes my day, night!22:40
Alexia_Deathtjaalton: It wont if you take the 0.8.2-2 (I hope, ping put a build swithc in that one)23:01

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!