/srv/irclogs.ubuntu.com/2009/04/03/#ubuntu-x.txt

* Ng looks at http://blog.launchpad.net/ppa/introducing-autoppa00:03
tormodI just tried to make a new libdrm package, but now the r6xx-r7xx branch does not merge nicely with master ;(00:03
tormodwhich it should since it itself merged in master a few days ago... forgot to pull00:08
Nghmm, autoppa is kinda bzr-centric00:13
Ngtormod: wow, auto-xorg-git is pretty impressive :)00:20
tormodok uploaded 2.4.6~00:23
tormodthanks! credits go to Bryce as well00:23
Ngtormod: I'm wondering about an option which notices the git tree in the working directory is already up to date and immediately bails - that way it'd only build/upload if there are actually new changes00:24
tormodit has that option :)00:24
tormodthe ppa-update uses it00:24
* Ng sighs, sorry, I swear I did look through the options!00:24
=== jbarnes_ is now known as jbarnes
Ngtormod: so, I'm happy to do some uploading of -intel, subject to some sanity checking of what I'm doing :)00:38
NgI guess we'd need separate LP accounts for bots that were auto-uploading though00:38
NgI certainly can't leave any GPG keys associated with my main account unlocked anywhere00:39
tormodNg: cool. apply for team membership. yes you'd need some autobuilder keys00:39
tormodthe rule is kind of to make as few manual changes as possible00:40
Ngjust uploaded a test run to my own PPA00:41
tormodurl?00:42
Nghttps://edge.launchpad.net/~cmsj/+archive/ppa00:43
NgI don't think the upload has been scanned yet00:43
Ngor it has and the rejection mail hasn't hit me yet :)00:44
Ngcool, accepted. that was the result of "./auto-xorg-git -n -g -D -H hooks -w /tmp/xorg-pkg-tools/ intel"00:45
tormodthe drm-snapshot 2.4.6 is done. notice I used -t "~" because 2.4.6 is not released00:47
* Ng nods00:48
tormodif the debian patch in -intel (pci id stuff) is not needed, no manual tweaks should be necessary atm00:49
tormodI am fixing up the drm-snapshot hooks also, except merging in r6xx-r7xx the rest is automatic00:50
tormodNg: pushed the hooks cleanup to bazaar/lp01:05
tormodjbarnes: is there a txt file archive of intel-gfx ML somewhere, for snatching that patch without html?01:26
tormodnever mind, saving as text in Firefox did not screw up as much as I thought01:30
tormoduploaded a new -intel with said patch. I guess on the edge means to stay ahead of git master :)01:36
tormodjcristau: I don't know what the 2.6.28 kernel in Debian does, but I had to revert http://git.debian.org/?p=pkg-xorg/lib/drm-snapshot.git;a=commitdiff;h=c32452f1fe831fe95a197883ab6f6617cd78afb1 to build -intel in Jaunty02:10
LLStarksquick question. does kms require uxa?06:09
bryceHere's the final chart of X bugs from last quarter - http://people.ubuntu.com/~bryce/totals-proprietary-2009-Q1.svg07:08
sbeattiebryce: dumb question, but what is this 'Xlib:  extension "Generic Event Extension" missing on display ":0.0".' warning I'm seeing from jaunty clients?07:51
brycedunno, haven't seen that one before07:51
tjaaltonGE is part of XI207:52
tjaaltonAIUI07:52
bryceah07:52
brycewhy would that be showing up with clients?07:52
tjaaltonbeats me07:52
brycefwiw, I've noticed we've got a lot of weird warnings strewn about...  in fact I was just looking at the one on bug 35304907:53
ubottuLaunchpad bug 353049 in xserver-xorg-video-intel "*ERROR* Unknown parameter 6" [Low,Invalid] https://launchpad.net/bugs/35304907:53
brycewhere possible I'd like to suppress these sorts of messages to avoid confusion; lacking that, I'd like to get them documented at https://wiki.ubuntu.com/X/Glossary so we can point people to them 07:55
mnemoon every single boot I get this warning:07:55
mnemo[   24.182667] [drm:i915_setparam] *ERROR* unknown parameter 407:55
mnemoi havnt had time to file that bug yet07:55
brycemnemo: don't bother - see the link I just posted07:55
mnemoaah it covers both 4 an 607:56
mnemogood07:56
brycetjaalton: just got a reject mail on vmmouse08:00
tjaaltonbryce: heh, yeah it was my mistake08:00
bryceokie08:00
tjaaltonre-uploaded the current version08:00
tjaaltonbut then I actually applied the patch and signed & uploaded then new version08:01
tjaaltonso it should be ok now08:01
tjaaltonhmm, should sync requests be assigned to the release team now?08:02
tjaaltonsince these are new upstream versions (FFe)08:03
bryceI assume it's still just the ubuntu-archive team08:04
brycein any case, it's essentially the same group of guys anyway ;-)08:04
tjaaltonheh, right08:04
RAOFHey all.  Is there something blocking the libdrm with fixed libdrm-nouveau1.symbols file?08:06
tjaaltonRAOF: no, I've just forgot to upload it08:06
bryceI can take care of it08:06
brycesorry RAOF, it's been on my todo list, I was just trying to get an xserver update out first08:07
bryce(...which I just now uploaded!)08:07
RAOFThat's fine.  It's not as if it totally breaks nouveau or anything.08:07
RAOFHurray for xserver updates!08:07
RAOFAnyone testing nouveau should be savvy enough to ask about manually installing libdrm-nouveau1, anyway :)(08:07
bryceah just minor stuff... I've been raking through X packages looking for patches people posted, and putting them in08:08
bryceso this has some xvfb-run script fixes, and a man page for Xephyr08:09
RAOFLet's see if there's anything interesting I should be pulling for nouveau at the same time as the rebuild...08:09
brycetjaalton: btw speaking of extra patches, I saw there's a mesa patch from tormod on bug 324854 you might look at08:11
ubottuLaunchpad bug 324854 in xserver-xorg-video-intel "DRI2: (UXA) white transparency artifacts with compiz [patch]" [Undecided,Invalid] https://launchpad.net/bugs/32485408:11
tjaaltonbryce: yes, it's on my repo but not yet uploaded08:11
RAOFLooks like the Fedora nouveau testing day produced a couple of fixed bugs.  Yay!08:12
bryceRAOF: do you know how they organize that?08:14
RAOFbryce: No, not really.  There's a wiki page detailing test-cases, with a link to a rawhide ISO, but I'm not sure who writes the page, how people get notified, etc.08:15
bryceyeah, I've seen that page, very curious08:16
mnemomozilla has an awesome way of organizing test days... they have a web app which helps you go through manual steps08:16
mnemolook at it here: https://litmus.mozilla.org/08:16
mnemoso basically you log in and say what build you have, then you follow instructions and enter results08:16
mnemoyou can see which testcases havn't been checked in a long time using what browser versions08:17
bryceinteresting08:21
bryceI've wondered if we ought to do more organized test day type stuff for X08:22
mnemoI think it would be a great idea because there is so many OS/hardware/software configs etc08:22
brycewell, I think it is driven more by an ability to follow through on test results08:23
bryce(which is why I was curious how fedora had organized their test day)08:23
mnemoI agree, right now it feels like we have a shortage of bugs, heh... even upstream bug tracker seems flooded with pretty high quality bugs... i have 3-4 crash bugs at least with repro steps that isn't being fixed08:24
mnemomaybe we should start some effort to clone eric anholt instead?08:25
bryceheh, I've had eric close plenty of my bugs but not fix them (so far) ;-)08:27
brycemnemo: I've had some luck in diving in and helping fix bugs on -ati this release.  With -intel though the code seems to be getting more and more complex so it's rather intimidating to work on08:30
bryceRAOF: libdrm uploaded btw08:31
mnemobryce: cool08:31
RAOFbryce: Thanks.08:33
RAOFI'll wait till it's built everywhere, then do some DDX testing.08:34
tjaaltonsix sync requests filed for the input drivers08:40
brycexorg uploaded - fixes dexconf for kvm users08:40
tjaaltondoes dexconf run?-)08:48
brycetjaalton: :-P  of course I checked ;-)08:58
tjaalton:)08:59
tjaaltonI actually meant that "is it run on install"08:59
tjaaltonseems that it isn't at least on our preseeded installations09:00
brycehmm09:00
tjaaltonbut I haven't noticed since the xorg.conf's are distributed otherwise09:00
tjaaltonhmm wait09:00
tjaaltongot the ideapad again, so I can check09:01
tjaaltonit's empty09:03
tjaaltonfunny that no-one has complained09:03
bryceempty as in 0 bytes?09:07
tjaaltonyes09:08
Unggnuhi09:08
brycehi Unggnu09:09
brycetjaalton: that sounds undesired09:09
brycehm09:09
UnggnuI am not sure if it works in Intrepid too but Apport seems to generate very good backtraces and other usefull information if X crashes in Jaunty. Maybe we should change the Backtracing Wiki to prefer enabling Apport.09:09
bryceweird, I've not seen bug reports with xorg.conf's attached that were 0 bytes09:10
bryce(would apport exclude them?)09:10
UnggnuThings like http://launchpadlibrarian.net/24144277/Disassembly.txt and you don't need to install dbg packages or use ssh/gdb09:11
bryceUnggnu: certainly feel free to improve the Backporting page.  My thought there was that if apport works and catches the crash, the user really wouldn't need docs - if it doesn't catch it, then they're probably in a situation where they have to do it manually anyway09:11
UnggnuBut I don't know how stable the process is.09:11
UnggnuYou have a point :)09:11
UnggnuBut nearly all of the intel reports in the top ten doesn't have backtrace and apport is disabled if a stable version is used09:12
UnggnuOne of the problem is that you need ssh/a second computer. No problem for a pc geek of course but ... :)09:14
bryceyes true09:17
bryceUnggnu: if you have ideas on how we can tell users how to enable apport (post-release) to get backtraces, please do update the top of the page with that info09:18
bryceI agree the current process is a bit techie for most people09:18
brycehonestly that Backporting page probably needs a good bit of copyediting to clean it up09:18
UnggnuThe wiki is fine for me is just a lot of work09:19
methril|workmorning09:20
mnemohi09:20
methril|worki''ve a xcrash in Jauntyy beta09:20
mnemoright09:20
methril|workhi mnemo!09:20
mnemo=) =)09:20
mnemobryce might know if your bug is a known issue, thats why I wanted you to describe the bug here09:21
methril|workwell, every start of Xs give me that backtrace (at the end of Xorg.0.log)09:21
* methril|work nods09:21
Unggnumethril|work: Did apport said something?09:21
mnemothis is the backtrace methril|work is seeing:   http://pastebin.com/m2146362709:21
mnemoRhdAssertFailed() in radeon driver09:21
UnggnuWhy radeonhd?09:22
brycehmm09:22
brycecrash in -radeonhd, needs the radeonhd debug symbols09:22
brycelooks like a failed assertion09:22
methril|workhw: 01:00.0 VGA compatible controller [0300]: ATI Technologies Inc M56P [Radeon Mobility X1600] [1002:71c5]09:23
mnemomethril|work: please install the package xserver-xorg-video-radeonhd-dbg09:23
methril|workok09:23
=== seb128_ is now known as seb128
bryceX1600... is that an r6xx? I forget09:23
methril|workr5xx IIRC09:24
brycemm09:24
brycemethril|work: did you try the -ati video driver?  it should work ok for r5xx cards09:24
UnggnuI would use the radeon-Driver if it works09:24
Unggnuyes09:24
UnggnuJaunty radeon driver even works with r7xx09:25
bryce:-)09:25
methril|worki don't remember why i choose radeonhd, if you suggest others, i could change09:26
brycehmm, there's 100 calls to RhdAssertFailed in the -radeonhd codebase, so hard to guess where it's failing beyond that09:26
Unggnuradeonhd isn't the default driver. It has to be installed manually afaik09:26
brycethere's a handy chart comparing them on different chips here:  http://www.x.org/wiki/RadeonProgram09:27
methril|worknice!! thank you09:27
UnggnuWe should really start a petition to get the radeonhd and radeon devs working together09:28
bryceUnggnu: that's already in progress.  By now I think -ati has absorbed most of what's cool of -radeonhd09:28
* methril|work nods about devs working together09:28
UnggnuOh, I hope so09:28
brycesince both are opensource it's fairly easy for -ati to eat -radeonhd code ;-)09:28
methril|workhehehe09:28
brycereally, it's only 6xx/7xx that still make some sense to pick radeonhd, and even there like unggnu says the difference is going away09:29
mnemobryce: actually the xorg.log says which assert is failing, look above the backtrace.. i missed that at first ;o09:30
UnggnuNo, it doesn't. Radeonhd has no real XV support for 7xx but radeon has09:30
UnggnuThat's why I don't understand the whole separation thing09:30
mnemo../../src/rhd_mc.c:417: RHDMCSetup: Assertion 'RHDMCIdle(rhdPtr, 1)' failed.09:31
RAOFThere were some philosophical differences that got expressed in code.09:31
RAOFBut I think those actually got resolved (in favour of the -radeon approach).09:31
Unggnuradeon works great except of the missing Powerplay but it isn't their fault09:32
brycemnemo: aha good point09:33
mnemohere is the code in the driver at the assert:09:33
mnemohttp://pastebin.com/m2259f52a09:33
bryceUnggnu: oh I assumed radeonhd had Xv support.  IN that case, yeah no reason to use it over -ati09:33
UnggnuI didn't got it to work here.09:33
UnggnuXv with radeonhd even in Jaunty.09:34
brycemnemo: hmm, not sure what that code does09:34
mnemome neither09:34
brycemnemo: but this is certainly well enough triaged to send the bug upstream if you'd like09:34
UnggnuEveryone can do what they want but many resources are blown in the wind which is a shame imho. I would at first try if ati/radeon works.09:35
mnemoyup09:35
bryceUnggnu: me too09:35
methril|worki'm going to try with radeon09:36
Unggnubryce according to pitti using "sudo force_start=1 /etc/init.d/apport start" in a stable Ubuntu version should do the job. After a crash a dump should be created and the user should get a message after a restart09:37
bryceUnggnu: interesting, that'd be useful to have in the backtracing page09:38
methril|work(EE) RADEON(0): [dri] RADEONDRIGetVersion failed to open the DRM09:38
methril|workwhat does that mens?09:38
Unggnumethril|work: you need a restart I guess09:38
Unggnubecause of the kernel module09:39
methril|work:(09:39
mnemomethril|work: going with -ati is probably the best thing for you, but it would also be useful if you filed a bug report about the radeonhd bug here: https://bugs.freedesktop.org/enter_bug.cgi?product=xorg&component=Driver/radeonhd09:39
methril|workcould i force the kernel module down and up?09:39
methril|workok09:39
Unggnumethril|work: Probably09:39
methril|workid' like to help a litle bit more09:39
Unggnumethril|work: maybe also remove radeonhd first, I am not sure09:39
UnggnuIf you use current -intel you become an addicted restarter ;)09:40
UnggnuAnd removing fglrx module is also not easily possible according to my experience09:41
methril|worki see some related bug with radeon and fglrx09:41
methril|workin the launchpad09:41
UnggnuFGLRX crashes a lot at least with R77009:42
Unggnumethril|work: Does removing/adding module work?09:44
methril|worknot yet09:44
methril|worki'm going to install the dbg driver09:44
methril|work(firewall problem at work)09:45
methril|workit could be that the Kernel Modesettings are wrong?10:04
UnggnuKernel modesetting?10:06
UnggnuThis isn't a feature from Jaunty afaik?10:06
methril|workok, then forget it (next kernel release 2.6.29)10:17
RAOFThat's for intel; ATI is slated for 2.6.30 IIRC10:19
jcristauno it's not10:21
jcristaumore probably .3110:21
Unggnu:)10:23
UnggnuSome news site said that Fedora has kms for intel, radeon and the open source nvidia driver10:24
brycephoronix10:24
Unggnuyeah10:24
RAOFIndeed; they pull in the drm modules from the relevent git branches.10:25
mnemohttp://www.phoronix.com/scan.php?page=news_item&px=NzEwMA10:25
mnemo"Enabled by default on Fedora 11 will be Intel kernel mode-setting (along with the ATI kernel mode-setting that can already be found in Fedora 10)."10:25
UnggnuThey will have fun in combination with ext4 :-D10:26
methril|work:)10:26
brycenight10:26
mnemonn10:26
Unggnunight bryce10:26
methril|worknight10:27
RAOFI could concievably pull in the modesetting branch for nouveau-kernel-source; I don't think that's a wonderful idea, though.10:27
UnggnuI guess there will be enough problems if kms comes with Ubuntu 9.1010:29
RAOFNot if fedora runs into them first! :)10:31
UnggnuOf course it would me more stable than now but ...10:33
methril|worki'd like to test it :)10:37
Unggnumethril|work: use a Live CD :)10:37
UnggnuIntel driver makes so many problems lately that I think stability is more important than fancy features10:37
RAOFmethril|work: Fedora livecds should give you that fun.10:38
methril|workwell, i like ubuntu :)10:39
methril|worki was checking all the xorg drivers, no one is working radeon, radeonhd neither ati10:46
methril|worki've both, debug and not debug10:46
methril|worki've this in the gdm log: /usr/bin/X: symbol lookup error: /usr/lib/xorg/modules/extensions//libdri.so: undefined symbol: atiddxAbiDixLookupPrivate10:50
jcristaufglrx fail.10:51
methril|workit's possible that before that fail i', not able to switch to a Terminal?10:53
tjaaltoneverything is possible with fglrx10:54
tjaaltonmesa 7.4 uploaded10:55
mnemomethril|work: if you want to try KMS on ubuntu... first install 2.6.29 mainline kernel (google for "ubuntu mainline kernel") and then subscribe to the xorg-edgers PPA and then read some docs on KMS to figure out what params you need to pass at boot or whatever10:55
mnemotjaalton: ah, great work.. thanks a lot10:55
methril|workmnemo: i know, i read about it. Thank you :)10:56
Unggnumethril|work: Could you check a daily LIve CD of Jaunty?10:56
methril|worki could, where it's?10:57
Unggnumethril|work: http://cdimages.ubuntu.com/daily-live/10:57
Unggnutjaalton: indeed :)10:58
Unggnumethril|work: If you change the driver you really should restart if it has DRI support.10:58
methril|worki see10:59
methril|worki'm restarting aggain and again :D10:59
methril|worklooks like crappy windows10:59
Unggnumethril|work: At least with the Live CD you could check if the radeon works at all10:59
methril|work(it's a joke)10:59
Unggnufor you10:59
tjaaltonwell, fglrx is loaded/installed, so no wonder that DRI is not working..10:59
methril|workit's working because i restarted at the MacOS partiton and works11:00
methril|workif you didn't put fglrx qhy it has to load?11:01
tjaaltonyou have it on the xorg.conf?11:02
methril|worknot now11:02
tjaaltonso.. -ati works now?11:02
Unggnumethril|work: I guess the module is loaded before X start so it is module configuration issue.11:05
tjaaltonI don't know how fglrx works, but the nvidia module is loaded by the driver when X starts11:06
jcristauthe problem was xorg's /usr/lib/xorg/modules/extensions//libdri.so was replaced by fglrx..11:06
tjaaltonheh11:06
Unggnuoh11:06
tjaaltonok so dpkg --purge xorg-driver-fglrx11:06
methril|workwas what i was going to do now tjaalton :)11:07
tjaaltonshould've been the first thing to check ;)11:07
tjaaltonlunch ->11:07
methril|workfglrx did the trick11:10
methril|worki'm going to check the other drivers11:11
methril|workthank you11:11
tormodwe introduced an apport hook for -ati to detect fglrx loaded...11:11
methril|workcould be good to post a bug in launchpad? or put some comments?11:11
tormodwe should put more warnings in the fglrx package description that it fscks up for other drivers, but I think I already have filed a bug on this11:14
methril|worki see a bug in launchpad11:14
tormodbug #?11:14
methril|work#346803รง11:17
tormodbug 34680311:17
ubottuLaunchpad bug 346803 in fglrx-installer "Radeon XPRESS 200M 5955 not recognized by aticonfig:No supported adapters detected (dup-of: 284408)" [Undecided,New] https://launchpad.net/bugs/34680311:17
ubottuLaunchpad bug 284408 in update-manager "r3xx Hardware does not work with fglrx [EPR#257839]" [Medium,Fix released] https://launchpad.net/bugs/28440811:17
methril|workalso11:19
methril|workbug 31302711:19
ubottuLaunchpad bug 313027 in fglrx-installer "MASTER: fglrx does not support xserver 1.6" [High,Fix released] https://launchpad.net/bugs/31302711:19
Brandiey hello thare18:05
BUGabundohey everyone18:05
Brandie=D18:05
BUGabundoneed a X guru here to help out Brandie!18:06
=== Brandie is now known as everyone
=== everyone is now known as Brandie
BUGabundos/he is having trouble on jaunty with an ati card18:06
BrandieWhy didnt I just buy a nvidia? D=18:06
BUGabundoeheh18:07
BrandieI love my ubuntu ;_; <318:07
BUGabundoBrandie: calm down a notch! 18:07
BUGabundothis is a (more) serious #18:08
Brandiesorry.18:08
BUGabundobryce ping18:09
BUGabundoBrandie: seems everyone is busy18:09
BUGabundohand in here a bit longer, ok?18:09
Brandiemmm alright18:10
BrandieSo, Why does ubuntu seem to have alot of problems with ati cards?18:10
BUGabundoabi bump18:11
Brandiehttp://img8.imageshack.us/img8/2776/dscf2478.jpg I get that when i try to boot into 9.04 :\18:20
BUGabundoBrandie: you can and should start by opening a bug on LP18:22
BUGabundothat way you can get help from more devs18:23
BrandieD=18:23
BrandieI think I remember the fix.18:23
BUGabundo$ apport-cli -fp xserver-xorg18:23
BrandieI just need the name of the propietari ati driver, so I can remove it18:23
BrandieIf i remember, I remeoved that, then reinstalled it in the os. and it worked.18:24
BUGabundothat would work.. .XFIX should have fixed that too18:24
Brandiemmm, It didnt last time. I cant remember the name os the package the official ati drivers install...18:25
Brandieluckily i can still boot into 8.1018:25
BUGabundoehe18:26
Brandieor... maybe i cant... It's giving me the same error not18:26
Brandie*now18:27
BUGabundooops18:27
Brandiefglrx18:28
BUGabundobye go to go. guud lucj18:29
Brandiewhat?...18:29
jbarnesbryce: gee thanks, I was running out of high prio bugs :p20:25
brycejbarnes: no prob, I suddenly find myself with plenty of extras20:28
_MMA_So I'm running Jaunty. (Intel 855 I believe) tried to connect a external monitor to the laptop not I have a corrupted screen with a cursor that looks fine. Disconnected monitor. Rebooted several times. No dice.22:04
_MMA_Any help?22:04

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