/srv/irclogs.ubuntu.com/2009/05/15/#ubuntu-x.txt

bryceI'm pretty sure it skips directly to the LTS00:00
mnemommkay00:04
pwnguinheh00:18
pwnguin"ubuntu needs to stop shipping new kernels and xorg builds every release"00:19
pwnguinhttp://lunduke.com/?p=42900:21
brycepwnguin: heh00:39
brycepwnguin: of course, would we skip revving xorg for a release, we'd get equally skewered00:39
brycehmm, what if we hadn't revved jaunty00:41
brycetoday we'd be on xserver 1.5.2-ish, mesa 7.2, librandr 1.2, -intel 2.4.100:43
brycewe could have stuck with fglrx 9.3, which would be nice.00:44
bryceATI hardware would not be as well supported, but Intel stuff could probably have been backported00:45
pwnguinoh, the guy is full of crack00:58
LLStarks?00:58
LLStarksdid i miss something?00:58
jcristauno00:58
pwnguinLLStarks: nothing X related00:59
LLStarksi'm not sure if you guys are following the conversation in #intel-gfx but i was wondering if dri2 swapbuffers even has a shot at making karmic.00:59
LLStarks?01:12
SarvattLLStarks: I'm about to add jbarnes' cursor flicker patch to edgers in a bit after the new drm builds and everything incase you dont want to build it yourself :D01:18
LLStarksi know how to build, but i miss out on all the debian patches.01:19
LLStarksalso, i posed this question a few minutes and it went unanswered: "i've always been curious as to how good (quality and performance-wise the linux intel driver is when compared to windows or mac. i can understand that windows likes to abandon device support and forces the user to use older drivers. i like how linux drivers are furiously updated on almost a monthly basis."01:20
LLStarksany thoughts?01:20
jcristauseems pretty obvious that drivers for other platforms are more mature..01:23
Sarvatti havent compared them because i dont use intel video on anything i care about performance on01:23
jcristaualso get more development effort put into them, since they actually matter.01:23
superm1bryce, er intel 2.4 driver /does/ work on jaunty though.  siretart set up a backport on his ppa :)03:07
superm1been the only way i could confidently get decentish performance03:08
brycesuperm1: yep I know03:27
quentusrexOk, what is the best way to check the ram usage of Xorg?05:07
tjaaltonbryce: sounds like the lunduke guy didn't get the memo07:07
tjaaltonalso, he seems pissed that fglrx dropped support for the old hw, although it's not mentioned07:08
brycetjaalton: link?07:30
tjaaltonhttp://lunduke.com/?p=42907:31
tjaaltona lot of handwaving07:32
bryceah, right was looking at that earlier07:32
brycewears me out.07:33
brycefscking backseat drivers.07:36
tjaaltonyeah07:40
bryceit's scary when I can just listen to people describe their problem with no mention of the hardware and know what driver they're using07:45
tjaaltonheh07:46
bryceyeah this guy doesn't understand X.org07:51
pwnguinindeed08:00
pwnguinhe could have been much more specific with his hate08:00
bryceyep08:00
pwnguinthree letters08:00
pwnguinxkb08:00
bryceI don't agree with his assertion that "good Linux software can only be made by well paid developers"08:05
bryceI would offer Inkscape as a counter-example.08:05
tjaaltonbryce: ok to sync -keyboard and drop the hangul hack? evdev is used by default anyway08:44
brycetjaalton: yes, that sounds ok by me08:45
tjaaltongood, I'm filing the rest that are syncable08:46
bryceawesome08:46
tjaaltonaiptek, suncg3, suncg6, suntcx08:48
tjaaltonI should merge the input packages next09:00
pwnguinbryce: don't worry, inkscape is obviously crap because he says so09:05
pwnguinartists need adobe because that's the standard09:05
pwnguini dont even know why you Free Software hackers bother trying09:05
* bryce snorts09:06
pwnguinon the hubris scale, i think this guy takes second place09:06
pwnguinthe ex-microsoftie who declared ubuntu was violating the spirit of the GPL by forking debian still holds king09:07
brycehehe09:07
pwnguins/holds/is/09:07
pwnguinwho, by his own logic, debian itself largely violates the GPL spirit09:08
pwnguingrammar's not on my side tonight09:08
pwnguinhe does have a point though; the community college i work for runs a graphic design program that feeds into places like hallmark09:10
pwnguinunfortunately, he murders this point by declaring nobody anywhere can avoid adobe09:11
bryceyeah, anything that goes "Everyone does foo" ends up just being a straight man for something new.09:12
pwnguinbringing peace to the middle east by converting everyone to hinduism09:14
pwnguinthat reminds me to polish up my Linux Packaging form reply09:14
bryceright09:15
tjaaltonhmm, evdev is syncable09:58
tjaaltonsorry, isn't10:00
tjaaltonactually, it is. having an extra fdi file that loads evdev for mice & keyboards isn't harmful, and those can be then dropped from hal10:05
jcristautjaalton: it's probably doable to make debian/rules only install the fdi if(!ubuntu)11:03
tjaaltonjcristau: yeah, but it might go away later, so it's not a huge deal right now11:17
jcristauack.11:17
=== crevette_ is now known as crevette
jcristauis there any replacement for http://people.ubuntu.com/~bryce/Xorg/versions_current.html? that seems to fail.11:52
tjaaltonhmm, works here11:53
jcristauhrm11:53
jcristauf*cking firewalling.11:54
tjaaltonheh :)11:54
jcristauworks when i work around that.  thanks :)11:58
tjaaltonI'll merge mesa 7.4.1 and clean up the diff, waiting for 7.5rc2 before working on it..12:11
tjaalton+start12:12
jcristau7.4.2 looks like it should be coming soon12:33
tjaaltonheh, so it seems12:47
Sarvatt7.5rc2 was supposed to be out 2 days ago, was waiting around to update the ppa for it but gave up :D13:12
tjaalton7.4.1-1u1 uploaded13:13
tjaaltonjcristau: what was the command again you use to check the consistency in the git repo?13:13
tjaaltons/in/of/13:14
jcristaugit clean -dnx13:14
jcristauto check if there are no spurious files13:14
tjaaltonthanks, I'll add it to the wiki13:15
tjaaltonbryce: seems you have stuff to push to xorg-server git?13:35
jcristauthere.  Subject: [Mesa3d-announce] Mesa 7.4.2 released13:48
Sarvattthere's 7.4.2 like 30 minutes after you pushed it, repeat of intel 2.7.1 right after bryce pushed 2.7.0 :D13:49
tjaaltonhah13:53
jcristauxorg-server's 100_xserver_exa_force_greedy.patch looks weird.  how do you make sure EXA_MIGRATION_GREEDY doesn't conflict with EXA_HANDLES_PIXMAPS?14:26
tjaaltonsshh, it'll go away14:27
=== albert231 is now known as albert23
apwjbarnes, hiya ... wondering about KMS, i see you are poking that ship quite a bit18:01
jbarnesapw: yep, I wrote big chunks of it18:02
apwam struggling to find any up to date patches or even discussions on it and wondering what is planned to hit soon and where i might find patches18:02
apwso i can eval what might make karmic18:02
apwobviously intel mostly hit the tree in .29, wondering about ati etc.18:03
jbarnesyeah intel support is in 2.918:04
jbarnes.2918:04
jbarneswith tons of bug fixes since it first landed18:04
jbarnes2.6.x was the first 2d driver to really support it18:04
jbarnesati support should be coming in .3018:05
jbarneserr .3118:05
apwyeah we are tracking for at least .3018:05
apwso ati is likely .31 material18:05
jbarnesbut airlied has a branch he's shipping in F11 with all the ati bits18:05
jbarnesso if you ended up on .30 you could take the upstream ati development driver18:05
apwis there a a preview branch for the planned ati stuff?  against currentish heads?18:05
apwwhere might i find that ... i need to get my ducks lined up to help make some decision on what we are shooting for18:06
* jbarnes digs up the urls18:07
apwthanks ... that'll make my life a lot easier :)18:07
jbarneshttp://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=summary18:07
jbarnesthe drm-rawhide branch has the latest radeon support18:08
apwhmm i poked about in his repo and it all seems 2.6.29-rc based.  which seemed wrong18:08
jbarnesyeah he hasn't updated in awhile it looks like18:09
jbarnesbut I think all you need is the drivers/gpu/drm and include/drm stuff anyway18:09
apwyeah i am sure its not beyond me to rebase it should it be neeed18:10
jbarneslemme ping glisse, he might have a tree too18:10
apwany idea where the nv supprot is kms wise?18:10
jbarnesI know the red hat guys are working furiously on it18:11
jbarnesbut I don't know what the current status is18:11
apwthanks for your help :)18:12
jbarnesyou might try sending a note to dri-devel too, Jerome Glisse is the main radeon kms guy right now afaik, and Ben Skeggs is the nouveau guy iirc18:12
apwcool.  will do that ...18:13
jbarnesah apparently glisse has a kms tree for radeon: git://anongit.freedesktop.org/~glisse/drm-next drm-next-radeon18:13
jbarneshttp://neo-technical.wikispaces.com/kms-glisse18:14
jbarneshttp://jglisse.livejournal.com/1822.html18:16
jbarnesapw: the things you get when you ask a simple question on #dri-devel :)18:16
* apw adds that to his list of irc channels18:17
jbarnesseems like there's a lot of interest in ati kms so everyone jumped on me when I asked ;)18:20
tormodyou can find those things on planet.freedesktop.org18:20
apwheh i bet there is :)  i am sure we will be shouted at if we don't have it in kk18:58
tormodthere will be shouting whatever you do19:04
tormodI am a bit worried now with all bugs moving into the kernel, will X debugging gets a lot more difficult? will we need serial console to sort out mode problems?19:06
* tormod didn't try kms yet19:07
tormodanyone here understanding MTRR ranges? http://launchpadlibrarian.net/26765241/dmesg19:42
LLStarksbryce, you there?20:16
jbarnesbryce: btw fix for https://bugs.freedesktop.org/show_bug.cgi?id=21488 is a good one, might be the root cause of the 965 hangs20:18
ubottuFreedesktop bug 21488 in Driver/intel "[GM45] [UXA] [KMS] lockup while using opera" [Critical,New]20:18
rzrtormod: hi20:48
tormodhi rzr20:49
rzrtormod: my membership is about to expire and https://bugs.launchpad.net/bugs/189393 is still open :)20:49
ubottuUbuntu bug 189393 in xorg-server "[U1] TV not detected with Radeon IGP 320M" [Unknown,In progress]20:49
rzrtormod: i have to track this20:50
tormodrzr: membership?20:50
tormodrzr: do the Options mentioned in the upstream bug help?20:52
rzrno20:53
tormodrzr: I suggest following up upstream - they are aware of the problem. make sure you always try latest git, just in case20:58
rzri have the git changelog in my rss feed 20:59
tormodI am not there yet, but I check cgit.freedesktop.org quite often :)21:00
tormodSarvatt: did you make that mesa hook? otherwise I'll take a stab on it21:02
Sarvatttormod: sprry. no i didnt21:11
Sarvattsorry too :)21:13
brycetjaalton: pushed.21:36
brycetjaalton: just flipping the debug patch back on21:36
brycewe can probably get rid of 100_xserver_exa_force_greedy.patch; we've not been forcing greedy on intel for a while so that patch is obsolete.  besides, -> uxa.21:38
LLStarkshey bryce21:46
bryceheya LLStarks21:50
LLStarksbryce, i'm not sure if you've overheard my interest in seeing dri2 swapbuffers get backported into karmic.21:51
LLStarksbut i am of the opinion that we shouldn't have to wait until next april to have it in a release.21:52
LLStarksi can easily see the requisite code for dri2proto, mesa, xserver and xf86-video-intel making the cut21:53
LLStarksthe problem will most certainly lie within a decision to use 2.6.30 rather than 2.6.3121:54
LLStarksin that case, a backports to 2.6.30 may be in order.21:55
LLStarksbryce, what do you think?21:55
brycemm21:55
bryceI don't know enough about the patch in particular, but yes I do sense we're going to need a lot of kernel stuff backported just in general, if we're going to make kms work well21:56
LLStarksthis code will be very desirable if karmic is dri2+uxa+kms21:56
bryceLLStarks: can you file a bug with the git url for the thing that needs pulled, so we can keep track of it?21:56
LLStarksbryce, the patches in question will allow tear-free output21:56
mnemoi guess it will be decided at UDS, but is it more likely at this point that we get .30 for karmic?21:56
LLStarksbryce: http://virtuousgeek.org/blog/index.php/jbarnes/2009/05/07/pageflipping_blocking_etc21:56
LLStarksthe 3rd comment has the requisite bits21:57
bryceLLStarks: set a milestone for it and target it to the karmic release.  That will ensure it gets reviewed21:57
LLStarkswhat packages should i file against? all requisites?21:57
bryceLLStarks: mm, how about linux and -intel?21:57
LLStarksgotcha21:58
bryceeven though it sounds like there's nothing to do for -intel, it'll help us keep track of it.21:58
brycemnemo: well, I would guess .30 but it's a kernel team decision ultimately, and I think they plan to decide it at UDS22:00
brycemnemo: apw knows about the -ati/.31 stuff so I suspect they'll be factoring into the decision whether to do the backport in .30 or move ahead to .3122:00
mnemook nice22:00
brycemnemo: I'm also a bit concerned that X bugs are going to be moving into the kernel, which could make them tricky for us to debug, however we've got a good kernel team, so it may be a net win for us22:08
brycemnemo: in any case I'm hoping to chat with with them about bug process22:08
mnemoyea its going to be different for sure22:11
mnemoi guess no more EDID in xorglog with KMS, heh22:11
brycedunno, maybe; people will still want to xrandr their screen to other rez's22:16
LLStarksbryce: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/377090/22:16
ubottuUbuntu bug 377090 in xserver-xorg-video-intel "[RFC Karmic] DRI2 swapbuffers" [Undecided,New]22:16
tormodoops already a 7.4.2 regression, fdo #2175622:30
mnemosounds like the whole intel QA team was playing ut2004 ;)22:34
mnemoops no I looked at the wrong bug :)22:36
Sarvatttormod: thats also present in 7.5 and 7.6, have had that problem for awhile.22:51
Sarvattmnemo: you dont get EDID in the log? I do - http://sarvatt.com/downloads/Xorg.0.log.txt23:01
bryceSarvatt: it seems inconsistent.  Some logs have it, some don't.  I haven't figured what makes it show sometimes, and not other times23:02
bryceI'd really like to force the edid to always be printed in the log.23:03
mnemoyea that'd be nice23:03
mnemoim upgrading my intel box to karmic right now and it got hosed big time when I turned on edgers23:03
brycethere's a few things I'd like to change about the logs, if I ever get the time23:03
mnemoits a G45 though, was that a know issue?23:03
brycelike, I'd like it to consistently put the nicely formatted lspci stuff in the log, so we can stop asking people to post their lspci -vvnn23:04
brycemnemo: 'hosed'? sure23:05
mnemoyeah doesnt boot at all23:05
mnemoi get that "the greeter program crashed, trying another one"23:05
mnemobut this is after I added edgers23:05
brycewell you'd need to look at your Xorg.0.log and gdm logs to see why it failed23:06
mnemohehe... yup23:06
Sarvattwas it kubuntu?23:09
mnemonah23:09
mnemorebooted it again now with no changes and now it started... xorglog says im now 2.7.99 now23:10
mnemoI get 4:3 resolution though which is wrong23:10
Sarvattwhat type of input?23:10
Sarvattvga?23:11
mnemodvi23:11
mnemoapport found an intel oops now it seems23:12
mnemobug 37710723:14
ubottuLaunchpad bug 377107 in linux "WARNING: at /build/buildd/linux-2.6.30/drivers/gpu/drm/i915/i915_gem_tiling.c:320 i915_gem_set_tiling+0x1d6/0x1e0 [i915]()" [Undecided,New] https://launchpad.net/bugs/37710723:14
mnemonothing on fdo for that one23:15
mnemothe dmesg keeps spamming "unable to read EDID from VGA-1" ... and i just doubled checked, its connected using dvi for sure23:22
Sarvattah of course you're on x6423:34
Sarvatti was going to say you might want to try it with the latest drm-intel-next code added to the kernel and i have it here http://sarvatt.com/downloads/2.6.30-5.7/ but didnt build it for x6423:34
=== jldugger is now known as pwnguin
Sarvatthopefully that stuff gets pulled into rc6 that should be coming in the next day or two23:50
Sarvattspeak of the devil, already did23:50
Sarvattmnemo: http://lists.freedesktop.org/archives/intel-gfx/2009-May/002391.html looks sort of relevant, https://bugs.freedesktop.org/show_bug.cgi?id=2121023:55
ubottuFreedesktop bug 21210 in Driver/intel "[G43 KMS] VGA wrongly detected as connected (actually disconnected)" [Normal,New]23:55
Sarvatti can apply the patch and throw the driver up on a ppa if you want to test it23:57

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