/srv/irclogs.ubuntu.com/2009/05/29/#ubuntu-x.txt

superm1apw, if you get that kernel that jbarnes was looking for yesterday built some time this morning, can you ping me?  I'd like to see that it fixes the problems that we found yesterday08:25
apwsuperm1, eep.  remind me what he was looking for08:25
apwi am building so many things for so many people ...08:27
superm1<jbarnes_BCN> 10:44:46> -http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=commit;h=07f4f3e8a24138ca2f3650723d670df25687cd0508:28
superm1<jbarnes_BCN> 10:45:06> -apw: can we get a kernel with that patch (actually a drm-intel based kernel would be even better)08:28
=== slangase` is now known as slangasek
apwjbarnes, about?08:44
* Ng shakes fist at X. Stop crashing!08:56
NgI don't sem to get any log indication that it's doing it though, I would have expected a segfault syslog08:59
apwsuperm1, 32 or 64 better for you?09:02
superm1apw, 32 09:29
superm1apw, i dont think atom's do 64..09:29
bryce_jbarnes_BCN: http://tinyurl.com/l7zw8u11:11
apwjbarnes_BCN, superm1, latest karmic kernel with the patch you requested up: http://people.ubuntu.com/~apw/jbarnes-karmic/11:38
superm1thanks apw, grabbing right now11:40
superm1jbarnes, apw unfortunately that kernel still does have the resume problem wrg to random corruption when KMS is enabled11:55
superm1adding on xorg-edgers right now to pair with it11:58
jbarnes_BCNcommit 13f4c435ebf2a7c150ffa714f3b23b8e4e8cb42f13:26
jbarnes_BCNAuthor: Eric Anholt <eric@anholt.net>13:26
jbarnes_BCNDate:   Tue May 12 15:27:36 2009 -070013:26
jbarnes_BCN    drm/i915: Don't allow binding objects into the last page of the aperture.13:26
superm1jbarnes_BCN, did you see my ping before, I sent it to jbarnes. not sure if that's a local session back in californida14:06
superm1california even14:06
jbarnes_BCNheh14:06
jbarnes_BCNno I missed it14:06
superm129-05-2009 05:55:37 > superm1: jbarnes, apw unfortunately that kernel still does have the resume problem wrg to random corruption when KMS is enabled14:06
superm129-05-2009 05:58:22 > superm1: adding on xorg-edgers right now to pair with it14:06
jbarnes_BCNbeen off & on... had to completely reschedule my travel for the next couple of weeks14:06
superm1ah14:06
jbarnes_BCNsuperm1: arg ok14:06
superm1i added xorg-edgers as well, same thing14:06
superm1so i'm on a fully updated karmic + whatever is currently on edgers for karmic's ppa14:07
jbarnes_BCNI think we need a gpu dump (did we get one yet?) and a copy of the rest of the contents of debug/dri/014:08
superm1i dont think we got a gpu dump yet14:08
superm1can you grab it from wifi again since i've not got console access while it's in crazy state?14:11
superm1although i wonder if my wifi comes up at that point post resume14:11
jbarnes_BCNyeah we can get it from ssh14:12
superm1where ya at?  i'll come to you14:12
jbarnes_BCNthe couch by sala b314:12
jbarnes_BCNat the end of the hall14:13
superm1k, ill come in a few min, just need to send an email14:13
bryce__jbarnes_BCN: http://tinyurl.com/l3netq14:16
bryce__100 papercuts:  https://wiki.ubuntu.com/LittleDetails14:49
seb128bryce__: isn't that session starting in 16 minutes ?14:58
tseliot2jbarnes: ping15:56
tseliot2or bryce15:56
SpecHeya...if I want to rebuild (with pbuilder) the xserver-xorg-core package without a specific patch, can I just apt-get source the package, delete the patch, and build it?18:03
Sarvattcomment it out from debian/patches/series18:05
Speci hope i'm not shooting myself in the foot :), thanks.18:05
Sarvattwhich patch?18:06
SpecSarvatt: 169_mipointer_nullptr_checks.patch18:09
SpecI'm wondering if the patch caused an issue I'm seeing now, or if it's just logging the issue and the "workaround" implemented with it doesn't actually solve the problem18:09
Spec(but it does solve a moar important problem....i just wanna test.)18:10
SpecI might also remove 166_nullptr_xinerama_keyrepeat.patch18:11
jcristauwhat's the problem?18:13
SpecBUG #32446518:14
Specerr, no bot?18:14
ubottuError: Could not parse data returned by Launchpad: timed out (https://launchpad.net/bugs/324465/+text)18:14
Specwait, sorry, that's a different bug that got fixed18:14
Speccurrent bug is BUG #36337518:14
ubottuError: Could not parse data returned by Launchpad: timed out (https://launchpad.net/bugs/363375/+text)18:15
Specthe mouse "loops" on my fourth monitor sometimes, and I get a lot of error messages in X.org's logfile18:15
Specso...the results of the pbuilder are a bunch of .deb files (xdmx, xdmx-tools, xnest, xserver-stuff, xvfb, etc), should I install all of them?18:43
Specand to revert dpkg --purge/reinstall w/ apt-get?18:43
jcristauSpec: just xserver-xorg-core (and -dbg if you want to debug)18:44
Specdon't know how to debug ;)18:45
Speci'll do this later tough, thanks for the help.18:45
SarvattSpec easiest way is to add all your debs to a folder, cd to it and run sudo dpkg-scanpackages . /dev/null | gzip -9c > Packages.gz  and add deb file:/opt/debs/ / to your /etc/apt/sources/list replacing /opt/debs/ with what you picked19:53
Sarvatterr sorry ignore the >/dev/null, i pasted from my crontab :D19:53
maxbThat is overkill for a single source-package build19:54
maxbMerely use "sudo debi --upgrade"19:54
Specwhat's wrong with dpkg -i? :p20:13
Nghrm, starting to see some rendering corruption after a day or so of the KMS kernel on jaunty with 2.7.120:44
Ngnow starting to see quite a lot of corruption on widgets20:47
bryce_hrm20:49
bryce_Spec: dpkg -i is fine, that's what I always use :-)20:50
bryce_Ng: hrm20:50
bryce_Ng: well, 2.7.99.1* is in xorg-edgers and a lot of people are having success running that20:51
bryce_Ng: is this with compiz?20:51
Ngyeah20:51
NgI'll switch up to edgers I guess :)20:51
Sarvattneed a kernel upgrade Ng :)20:58
Sarvatthttp://people.ubuntu.com/~apw/jbarnes-karmic/20:59
Sarvattwas just merged today but that fixes the swap related corruption21:00
Sarvatthttp://anholt.livejournal.com/41132.html21:01
hyperairooh cool21:01
Ngnice21:03
Sarvattdidnt help theres a big memory leak with 3d compositing enabled making you swap more either21:03
NgI have 3GB, I didn't notice any swapping21:07
hyperairSarvatt: i noticed that it reduced when i turned on KMS21:11
hyperairstrange eh..21:11
hyperairi've been able to last up to ~12 hours now without having to restart X21:11
Sarvattoh  ya couldn't before? i havent done UXA without KMS in a long time, it was taking about 24 hours before i had to restart when i had compiz on with 1.5gb ram but I guess it all depends on how much you actually do during that time :D21:13
Sarvatti just switched to metacity compositing for now till its fixed21:14
hyperairSarvatt: not that i couldn't, more like i didn't know how to.21:14
hyperairthen  the email came in the other day to -devel21:14
Sarvatt241MB gem obect bytes 241MB gtt bytes right now after 3 days uptime21:14
hyperaireh that's pretty awesome.21:15
hyperairso metacity's compositing is better than compiz's huh21:15
hyperairas in it doesn't leak like crazy21:15
Sarvattanholt was saying it was GL compositing doing it in that blog post i linked21:15
hyperair833M gem object bytes here, 794M gtt bytes21:16
hyperairno wait21:16
hyperair79M gtt bytes21:16
Sarvattyour aperture is that big?!21:16
hyperair1323610112 object bytes21:16
Sarvattahh21:16
hyperair79474688 gtt bytes21:16
* Ng reboots into the new kernel and the various -edgers bits21:21
Sarvatthyperair, did you try earlier drivers out to see if theres a problem there too? to  be honest i didn't notice the leak until around may 15th (i have 1.5gb ram and dont use swap so it hit hard) and i havent tried earlier drivers out yet since its easier to disable compiz instead of losing all of the bug fixes in the newer drivers21:23
Sarvatthyperair: by oh ya couldnt before I meant I didnt know it was worse without KMS, sorry21:32
hyperairSarvatt: -2.4 doesn't have it.21:39
hyperairSarvatt: i'm pretty sure i didn't see it with 2.521:39
Sarvattwonder if i should up mesa on jaunty in xorg-edgers to 7.6, the fix for 3d rendering on 8xx is only in master21:39
hyperairthough i may be wrong21:39
Sarvattoh i meant 2.7 or newer21:39
hyperairi used that very briefly21:39
hyperairoh21:40
Sarvatt2.7 or 2.7.121:40
hyperaireverything 2.7 and up has it21:40
hyperair2.7.0 had it21:40
hyperairdefinitely21:40
hyperairand 2.7.121:40
hyperairand the current master21:40
hyperairall without KMS21:40
Sarvattgotcha, i only used EXA on 2.7.0 and 2.7.1 because its so much faster21:40
hyperairby current master i meant xorg-edgers' one21:40
hyperairEXA is faster?21:40
hyperairO_o21:40
Sarvatthttp://global.phoronix-test-suite.com/index.php?k=profile&u=robert-15508-6884-1631921:41
hyperairi didn't really experiment with EXA on those, but i'm pretty sure that EXA had the memory leak too21:41
Sarvattthe 05-09 ones are with driver 2.7.1 even though it says 2.7.021:41
hyperairi see.21:42
hyperairanholt seemed willing to fix it, but only if we can find a minimal test case21:43
hyperair=\21:43
Sarvattsomeone posted one that works for me, the loop opening and killing firefox21:43
hyperairis it?21:43
hyperairhmm21:43
Sarvatton the bug report, he saw it21:43
hyperairi see21:43
hyperairi actually have a patched drm which disables BO caching21:44
hyperairtook the one from your PPA and patched the figure from 14 to 021:44
hyperairsome #define there21:44
hyperairthat one lets the memory drop a little if you close firefox21:44

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