Sarvattkklimonda: updated lbm-nouveau uploading now that should fix it00:00
RAOFSarvatt: Have you made it conflict/replaces nouveau-kernel-source?00:00
Sarvattnope, will do that next upload, working on libdrm/ddx now and libdrm is a pain in the butt00:01
RAOF*Another* update to nouveau_class.h?00:02
Sarvattyeah they do this all the darn time, part of the reason i was so skeptical about nouveau in lucid00:02
RAOFBut they haven't updated the DDX for those changes; does the DDX not use them at all?00:03
Sarvattthink its just mesa that needs the updated nouveau_class.h for now00:05
Sarvattthey added texture_from_pixmap to nouveau_vieux to have texture_from_pixmap and stuff a few hours ago00:06
Sarvatterr, ignore the bad english :)00:06
Sarvattthe nv50 stuff just looks like formatting changes and new defines instead of changed values? I dont know00:08
Sarvattguess we'll see if it builds00:08
Sarvattguess i should hold off just incase it needs updating huh? :D00:09
Sarvattlibdrm is *such* a pain to update, dont mind holding off :)00:10
Sarvattmaybe the kernel guys should just bring nouveau in from upstream instead of linus since its 2.6.32 based and could go right in the kernel without backporting all of drm, that way nouveau would just be the only sketchy thing :)00:14
RAOFIt's not really 2.6.32-based, though.00:24
RAOFIt's 2.6.32+drm-next00:24
RAOFIE: It's basically 2.6.33's drm stack (and, soon, is likely to be 2.6.34-rc1's drm stack).00:25
superm1hmm umm is there any known scenarios that nouveau freezes and next time you try to boot the machine is hung before the BIOS screen...?00:42
superm1just a flashing volume LED, no POST no nothing...00:44
RAOFIt would be theoretically possible for nouveau to do that, I think.  I've never experienced it, though.00:45
RAOFYou've powered-off, rather than warm-rebooted?00:45
superm1unplugged, and removed battery00:45
superm1i was just using the machine all day today with a 9.10 install no problems and was trying out the today's 10.04 daily00:45
RAOFI don't know; I've never run into anything like that.  I'd give raising this in #nouveau a whirl.00:46
superm1* #nouveau :Cannot send to channel00:48
superm1is there something special to do to be allowed to talk there?00:48
RAOFAre you registered with nickserv?00:48
superm1i must have lost my registration during a netsplit or something00:49
Sarvattwow, that would be insane00:55
Sarvattcan you blind flash whatever model you're using? alot of people had that happen with the shipping bios on these aspire ones but luckily we can blind flash these to fix it00:56
Sarvattmaybe try pulling power/battery and holding the power button for a few minutes and trying again?00:57
RAOFLooks like darktama's on it in #nouveau.00:57
superm1Sarvatt, flashing unfortunately requires it to be booted on these, and only from DOS binaries01:01
Sarvattno hidden key combo to bypass it and flash? i think its function+escape in most insyde bioses01:04
superm1if it at least POSTed maybe, but it's definitely hung on boot when trying to POST - the backlit keyboard doesn't even turn off01:04
superm1which is normally BIOS controlled01:04
Sarvattfunction+escape, press power, wait a few seconds and release function+escape and the power light blinks01:05
Sarvattwhen mine crashes i dont get POST or anything and it still has a recovery blind flash you can do at least01:05
superm1same behaviors when i'm trying that01:06
Sarvattdoes it through the insyde efi internal stuff instead of the bios emulation layer or something01:06
Sarvattoh that means it should work01:06
superm1no i mean same behaviors as what i was seeing before01:06
superm1flashing volume LEDs, frozen backlit keyboard no POST01:06
Sarvattneed a usb stick formatted fat, with the bios named whatever.FD on it01:07
Sarvattoh ok01:07
superm1well i'm not sure i want to sacrifice another laptop like this to try'n fetch the X log01:07
SarvattRAOF: forgot to attach the debdiff?03:46
RAOFTo the lbm bug?  No.  Just filing first, so I can close it in the changelog :P03:46
RAOFDo we currently have somewhere to publish that for ctxprog testing?  I don't have any cards new enough to need it.03:47
SarvattThe attached debdiff pulls the nouveau changes from 2.6.33-rc7 to 2.6.3303:47
Sarvattmaybe x-updates?03:47
RAOFIt's attached now :P03:47
Sarvattor delete all of xorg-edgers/nouveau and just put that in there03:47
Sarvattkklimonda had problems with just the ctxprogs generator :(03:48
Sarvattthere was another fix in upstream git for it03:48
RAOFWas that with the fix?03:48
RAOFRight; that's in the debdiff, too.03:48
Sarvatti'm  not sure if that fixed it for them though03:48
RAOFThat's what I was thinking, but I think bryceh has disabled xorg-edgers/nouveau?03:48
Sarvattya should be able to reenable it03:49
Sarvattonly me and tormod are admins so i think anyone can03:50
RAOFAnd are you subscribed to lbm in lucid, or something? :)03:54
Sarvattnope, I was just looking into joining bug-control and saw the bug rss feed link, yours was the first one on the list :)03:57
SarvattPosts per week: 1,537.203:59
Sarvatti thought ubuntu-x-swat was bad :)03:59
RAOFAh.  We want that for plymouth, don't we.04:03
Sarvattwe have 189_xserver_1.5.0_bg_none_root.patch in xserver but none of the ddx specific patches enabling it there04:04
Sarvattlooks like its being used too /usr/bin/X :0 -nr -verbose -auth /var/run/gdm/auth-for-gdm-j5c54J/database -nolisten tcp vt704:05
Sarvatt(the -nr option)04:07
Sarvattintel version04:08
Sarvattwell thats for 2.10, 2.9.1 version is http://cvs.fedoraproject.org/viewvc/devel/xorg-x11-drv-intel/copy-fb.patch?revision=1.9&view=markup04:10
Sarvattah we've got the intel one in the lucid package, explains why i lost the smooth transition when i switched to edgers :)04:17
Sarvattgotta add that nouveau one though04:17
Sarvattimagine theres a radeon one we need to pick up too04:18
RAOFIt looks pretty simple.04:18
Sarvattthe smooth transition  looks really nice, leaves the logo on the screen while the panel loads and stuff04:22
Sarvattnot just a jerky 1 second ubuntu splash to  a black screen like it is without it04:23
Sarvattuploading a new nouveau snapshot on edgers with that patch04:27
Sarvattthere's radeon's patch -  http://cvs.fedoraproject.org/viewvc/devel/xorg-x11-drv-ati/radeon-6.9.0-bgnr-enable.patch?view=markup04:35
Sarvattahhh looks MUCH better now04:48
Sarvatti didnt understand the point of a 1 second splash screen but its nice with these ddx patches at least04:49
RAOFI've also applied that to nouveau in pkg-xorg git.04:51
Sarvattwoot just need to see if radeon has the patch or not now04:52
Sarvattbtw found a api 0.0.16 patch for 2.6.33, might use that on top of linus' for nouveau http://cvs.fedoraproject.org/viewvc/devel/kernel/drm-nouveau-abi16.patch?view=markup04:54
Sarvattall kinds of crack in fedora cvs04:55
tjaaltonbryceh: doesn't bzr hate you just as much when you forget to do 'bzr extmerge --all' or something?-)04:57
Sarvattfbdev bgnr patch, that'd help out arm - http://cvs.fedoraproject.org/viewvc/devel/xorg-x11-drv-fbdev/BGNoneRoot.patch?view=markup05:20
brycehtjaalton, haven't found bzr hating me so far08:54
tjaaltonbryceh: just because you haven't done any merges like that with it :)08:58
tjaaltonno tool does that automatically08:58
tjaaltonbryceh: bh.org:8080 fails to connect15:42
=== yofel_ is now known as yofel
=== Sinnerman is now known as Cobalt
brycehtjaalton: yeah I'm in the middle of a hardy->lucid upgrade and it's an old machine so it's hit a variety of glitches17:58
tjaaltonbryceh: heh, ok. I'm doing karmic->lucid on my home desktop, though it still needs more space on /18:02
tjaaltonit's a 500GB disk and root has 11GB.. how wise was that?18:02
bryceh/dev/sda4              13G   11G  2.1G  83% /18:03
brycehI wasn't much smarter tjaalton18:03
brycehbut it's just a webserver so I've been able to just uninstall stuff to get it down18:04
bjsniderthere aren't any remaining issues with jockey being able to activate the nvidia driver are there?18:16
brycehbjsnider, probably will always be some issues, but at the moment I believe we believe it should work properly18:19
bjsniderthe +1 channel is choked with people complaining about it, but i think it's where they upgraded from that's the problem18:20
brycehah, are they mostly upgrading from self-installed nvidia's?18:20
bjsnidergood question. i'm guess they were using the nvidia installer18:20
brycehtseliot has been looking to these kinds of bugs so he might want to look into it18:21
superm1bryceh, tseliot perhaps a good solution is to have update-manager attempt to detect a situation that the nvidia installer has been manually ran just like the packages do, and block the upgrade until it's removed18:29
bjsnidersuperm1, what if the user does not know how to remove it?18:33
Sarvattwe should probably have nvidia-current packages conflict with -190 and -195 even though they were never in ubuntu since alot of people used PPAs18:38
Sarvatthow about a preinst script for nvidia that tries to find a nvidia-installer binary, and calls it with nvidia-installer --uninstall before?18:54
Sarvattor just wont install if it exists since it's only there if the blob has been manually installed as far as I can see18:56
mvosuperm1: if someone gives me instructions what to look for I'm happy to implement that19:09
apwRAOF, about?19:09
Sarvattjcristau: \o/ \o/ \o/ THANK YOU! :)19:18
Sarvattfor the patch on dri-devel19:18
Sarvattmoving libdrm headers to /usr/include/libdrm will be so much better19:20
superm1tseliot, can you provide mvo that information?19:29
SarvattI think you could check for /usr/bin/nvidia-bugreport.sh19:35
Sarvatti see that in the extracted .run but its not installed in the package19:35
tjaaltoninstead of nvidia-bug-report.sh, which was installed by the deb?19:36
Sarvattah darn yeah its in /usr/lib/nvidia-current/bin/ in the deb19:38
Sarvatttls-test isnt in the deb though19:39
tjaalton/usr/bin/nvidia-bug-report.sh in karmic19:39
tjaaltonbut that's 18519:39
brycehtjaalton, btw bh.org back online.  upgrade complete.19:39
Sarvatt--slave /usr/bin/nvidia-bug-report.sh nvidia_bug_report /usr/lib/nvidia-current/bin/nvidia-bug-report.sh \19:40
tjaaltonbryceh: ah, good19:41
Sarvatt/usr/bin/tls-test isnt installed in the nvidia-current package though, no clue if its installed manually but its in usr/bin/ in the extracted blob19:42
ubottuLaunchpad bug 528683 in nouveau-kernel-source "Please remove nouveau-kernel-source from the archive (lucid)" [Undecided,New]20:36
Sarvattwish i could set wontfix, closing out a crapload of nouveau-kernel-source lucid bugs20:46
johanbrsuperm1, any luck with your half-dead laptop?20:47
superm1johanbr, yes it's back to life per some comments in #nouveau, but it certainly freezes on any lucid boot now, and is broken again20:58
Sarvattthe new lbm-nouveau that just got uploaded should help you superm121:02
superm1Sarvatt, okay i'll wait until a new daily image to try again21:02
Sarvattit still might be iffy, the chipset generation you have is one of the sketchier ones with the ctxprog generator and you might need to boot with lbm-nouveau.noaccel=1 manually from tomorrows livecd21:13
Sarvattbryceh: if we're sticking to 2.9.1 shouldnt we stop passing --enable-kms-only for intel? so these 8xx people can boot with UMS21:16
Sarvattactually, i'm wondering if these people saying booting with i915.modeset=0 on 8xx are actually using vesa when they say it works that way because of that21:17
Sarvattsaying that works I mean21:17
Sarvattbecause if thats the case it changes my opinion on the 2.9.1/2.10 debate :)21:18
brycehSarvatt, yep, it's on my todo list to reverse that21:27
brycehwho is saying that modeset=0 works for them?  I agree it'd be good to look at their xorg.0.log21:28
Sarvattthat blog you linked the other day for one, i dont have any specific bug #'s handy but i'll keep an eye open when i see another next21:29
Sarvattbecause if it works with modeset=0 and they are using vesa that way because of the intel configure flag just moving them to vesa + modeset=0 or fbdev on top of KMS with intel 2.10 might be better. reverting that configure flag might break things again since they'd be using UMS intel instead21:31
brycehmm true21:33
brycehstill, I think it's a good strategy for us21:33
superm1bryceh, recently cjwatson removed xforcevesa from the daily's afaik.  if new nvidia hardware is gonna be iffy with nouveau, i think it'll be important to still have that as an option21:38
superm1although i guess the nouveau kernel module is still gonna load with that isn't it..21:38
superm1so still potential to break21:39
brycehsuperm1, yeah21:39
superm1so perhaps xforcevesa should still live on, but blacklist intel/radeon/nouveau?21:39
brycehsuperm1, on the plus side, failsafex now uses fbdev properly with kms (and I even tested it on nouveau), but I think that kicks in only for X breakage.  If something is wrong at the KMS layer, I'm not sure it'll help21:40
brycehsuperm1, sounds like something we need to chat with sconklin and apw about21:40
brycehsuperm1, along with that is the need for enabling users to override the mode detection such as if a KVM is blocking edid (as per discussion on kernel list the other day)21:41
superm1bryceh, yup21:41
superm1well once things are declared more "stable", the QA folks should try on newer hardware and see if the need is really still there too21:41
brycehsuperm1, would you mind writing this up as a bug report to the kernel?  I'll +1 it and bump up priorities for it with apw/sconklin21:42
Sarvattvideo=blah doesnt work for whoever had that problem bryceh?21:42
brycehSarvatt, I've heard it didn't; I've not tested that myself so dunno if that is a general solution to this or not21:42
brycehyou tested it yet?21:43
Sarvattyeah it works here but i only played with it with a console boot21:43
Sarvattso a kernel command line option that launches a failsafe X would work right? is that all xforcevesa did before?21:46
Sarvattoh spiffy, you got the fbdev + kms stuff working in failsafeX bryce? gotta try that out21:48
BUGabundoboas noites21:54
tjaaltonsuperm1: xforcevesa hasn't been useful since we stopped using dexconf21:58
tjaaltonie. it did nothing aiui21:58
superm1tjaalton, not true, please see casper bzr21:58
tjaaltonnot without a link :)21:58
superm1it's in casper-bottom21:59
tjaaltona browser link21:59
tjaaltonbah, don't bother :)22:00
tjaaltontrust your word on it22:00
brycehSarvatt, yeah failsafeX should be good to go with KMS now22:07
brycehSarvatt, I'm not sure that xforcevesa made failsafex launch, or just make the system boot with vesa22:07
brycehif the latter, then maybe what's needed is something to force fbdev?22:08
bryceh(I builk software too much)22:09
brycehSarvatt, it might be nice to have a way to launch into failsafex from grub or from the safeboot option maybe22:09
brycehnot sure22:09
brycehmaybe a lucid+1 thing22:10
tseliotsuperm1: ?22:19
tseliotsuperm1: the installer might break things in different ways and you need to shut down X in order to do --uninstall (if I remember correctly)22:20
superm1tseliot, well the thought process is tell the user they need to do that to upgrade, so you dont knowingly break22:24
superm1they had to stop X to do it once, they can do it again22:25
tseliotsuperm1: aah, ok22:25
bjsnidersuperm1, but do they know about the --uninstall option?22:27
bjsniderit would be better if nvidia just didn't provide an installer at all22:27
superm1bjsnider, you tell them in the popup from update-manager22:29
superm1warning: you installed using the nvidia installer, you must uninstall using blah blah command 22:29
superm1and give them directions22:29
bjsniderif i didn't already know what it is i might ask what the nvidia installer is22:31
tjaaltonheh, could reproduce the "package foo in installed and configured already" 'bug' while upgrading to lucid22:44
tjaaltonhappened because root got full, despite u-m thinking it had enough space before it started the upgrade22:45
superm1bjsnider, well hopefully they'll remember that they manually ran said command to install the nvidia driver the first time23:10
Sarvattbryceh: hmm, failsafe no workie here at all - http://pastebin.com/j9tCXSsR23:24
Sarvattat least with the old echo foobar > /etc/X11/xorg.conf method23:24
brycehSarvatt, got an Xorg.failsafe.log ?23:24
Sarvattits from feb 11th, it didnt even try to use failsafe23:25
brycehgot any xorg.conf.failsafe ?23:25
Sarvattyeah the old vesa one23:25
Sarvattfrom feb 11th too23:26
brycehdoublecheck that you're using the latest version of xorg - verify that /etc/gdm/failsafeXServer has a timestamp of yesterday or newer23:26
brycehif that's good, then check the udev rule that is supposed to trigger failsafeX23:26
Sarvatt-rwxr-xr-x 1 root root 4484 2010-02-23 01:37 /etc/gdm/failsafeXServer23:26
brycehpastebinit /etc/gdm/failsafeXServer23:27
brycehI think it should have a newer timestamp, it might be missing the fbdev bits.23:27
brycehbut that should have at least touched Xorg.failsafe.log23:28
brycehyep that's the right version of failsafeXServer23:28
Sarvattyeah it didnt even try, it just used a normal session and failed23:28
brycehcheck the udev rule23:28
Sarvatttrying to find it23:29
brycehthat's one bit I did not test23:29
Sarvattnothing in /lib/udev/rules.d/ that matches for grep -R failsafe .23:29
Sarvattits a udev rule that triggered it before? didnt know that23:30
brycehoriginally it was gdm that triggered it23:31
brycehbut they dropped the functionality from gdm that we needed 23:31
bryceh(I guess redhat didn't need it for anything *grin*)23:31
brycehso slangasek did a udev rule for it in karmic23:32
Sarvattno x11-common.failsafe* anywhere on my system23:32
bryceh        dh_installinit -px11-common --no-start --upstart-only --name failsafe-x23:33
Sarvattah init script23:33
Sarvatttheres a /etc/init.d/failsafe-x23:33
brycehthat's probably it23:34
BUGabundoguys how can I find out the EDID of my external LCDTV plugged via vga?23:34
brycehBUGabundo, xrandr --verbose or get-edid23:34
Sarvattdont be surprised when your tv doesnt expose all the resolutions i t works at over vga23:34
BUGabundoI know23:34
BUGabundovga has limits23:35
Sarvattmost 720p hdtv's i've messed with just do 800x600 :(23:35
BUGabundomine goes higher23:35
BUGabundobut its cutting of part of the screen23:35
BUGabundoor off center23:35
BUGabundodepending on res23:35
BUGabundo$ xrandr --verbose | pastebinit 23:35
Sarvattso i guess you could just do sudo /etc/init.d/failsafe-x start to start failsafe instead of the xorg.conf trick23:36
Sarvatt/etc/init.d/failsafe-x just looks like a skeleton23:39
brycehit's a symlink23:40
Sarvattcant wrap my head around how failsafe is started23:40
brycehthere must be a config file installed somewhere23:41
BUGabundoat 800x600 the pic is really nice23:41
brycehah /etc/init/failsafe-x.conf23:42
Sarvattyou have that?23:43
Sarvattthats a leftover from superm1's dkms thing i think?23:43
Sarvattits .unused here23:43
Sarvattit got marked .unused during an upgrade a few months ago23:44
brycehI'm fairly sure this machine is a fresh install within the last month or two23:44
Sarvattyou have an /etc/init/failsafe-x.conf bryceh? maybe I screwed something up23:45
brycehyeah I used the alpha-2 cd to install from23:45
Sarvattwill check my other machines when i get home23:45
tjaaltonI have that after the upgrade23:46
Sarvatti wouldnt put it past me to have renamed it screwing around with something23:46
Sarvatt.unused or .conf?23:46
Sarvattahh i screwed with it then, sorry about that23:47
bryceh$ dpkg --contents /var/cache/apt/archives/x11-common_1%3a7.5+1ubuntu6_all.deb | grep failsafe23:47
bryceh-rw-r--r-- root/root       132 2010-02-18 16:02 ./etc/gdm/failsafeBlacklist23:47
bryceh-rwxr-xr-x root/root      8610 2010-02-18 16:02 ./etc/gdm/failsafeXinit23:47
bryceh-rwxr-xr-x root/root      4595 2010-02-18 16:02 ./etc/gdm/failsafeXServer23:47
bryceh-rw-r--r-- root/root       291 2010-02-18 16:02 ./etc/init/failsafe-x.conf23:47
brycehlrwxrwxrwx root/root         0 2010-02-18 22:06 ./etc/init.d/failsafe-x -> /lib/init/upstart-job23:47
brycehSarvatt, try reinstalling x11-common23:47
Sarvatti bet it was when i was getting failsafe every time a fsck was run and it was annoying me :D23:47

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