/srv/irclogs.ubuntu.com/2008/07/10/#ubuntu-x.txt

sorenErm... I get a white screen when I log into GNOME in Intrepid. What am I doing wrong?09:24
sorenGDM looks fine.09:24
tjaaltondisable compiz09:25
tjaaltonit's broken on intel09:25
soren:(09:25
tjaalton965 at least09:25
sorenOk.09:25
sorenThanks.09:25
sorenThis will be fixed at some point, I presume?09:25
* soren kinds of likes compiz.09:25
tjaaltonthere are two bugs and they were marked as blockers for mesa 7.1 yesterday, so yes they'll get fixed soon :)09:26
sorenFantastic. Thanks.09:26
* soren hugs tjaalton 09:26
tjaaltonthe white screen one already has a fix, but then you'd end up with a black screen with some corrupt graphics :)09:26
* tjaalton hugs soren back09:26
* soren ponders the semantics of "fix" :)09:28
tjaaltonok, patch :)09:29
tjaaltontwo, actually09:29
brycetjaalton: I heard back from Intel on this09:36
bryceYong:09:36
bryceI just reproduced #14441 on 965G with tip of xserver1.5, mesa, drm, xf86-video-intel.  A quick way to resolve it is commenting out09:36
bryce    if (pI830->useEXA)09:36
bryce        pDRIInfo->texOffsetStart = I830TexOffsetStart;09:36
brycein xf86-video-intel/src/i830_dri.c. Then compiz works.  You can ask Ubuntu guys for more test.09:36
bryceFor #15477, It doesn't happen on G965. We may need to find a 945 box to look at it.09:36
bryce- Peng09:37
brycenight.09:37
tjaaltoncommenting that out breaks other setups09:37
tjaaltonthe workaround is already mentioned on the bug09:37
tjaalton15477 does happen on 96509:39
tjaaltonbut for that bug there already are patches09:39
tjaaltonoh, night :)09:41
pwnguinheh14:10
pwnguini filled up my / partition14:10
pwnguinoops14:10
pwnguinfun finding random crap on cleanup14:10
pwnguinXFree86.0.log14:10
=== Q-FUN1 is now known as Q-FUNK
=== Q-FUN1 is now known as Q-FUNK
=== Q-FUN1 is now known as Q-FUNK
=== Q-FUN1 is now known as Q-FUNK
tjaaltonnow, let's clear all l-r-m* from nvidia/fglrx bugs and let tseliot and superm1 have some fun :P19:58
tjaaltona great way to boost my lp karma 20:02
pwnguinman, if you just mark them all invalid20:03
pwnguinyou will earn a new top score on the "people pwnguin is angry at" table20:04
tjaaltonhah20:05
pwnguinpvalli does that and it ticks me off20:05
pwnguini just love getting told my bug is a dupe, but if i want to know WHICH one, i should just go look it up myself20:05
tseliottjaalton: I'll have a lot of fun then ;)20:09
pwnguinbtw20:10
pwnguinthis might be a good time to point out that there is an upstream bug tracker email address20:10
pwnguinand that launchpad apparently supports this20:10
tjaaltonpwnguin: what benefit does it give?20:12
pwnguina feeling that at least we tried?20:12
pwnguinplus, if we used it there might be an incentive to use something smarter on their end20:13
tjaaltonok, so does bugzilla have such a feature (email)?20:13
pwnguinI dont understand the motivation of the question20:13
tjaaltonupstreams tend to use bugzilla20:14
pwnguinnvidia is upstream20:14
tjaaltonso where to send the bug20:14
tjaaltonah20:14
pwnguinyou've got this backwards i think20:14
tjaaltonyeah, but they want that people run their script20:14
tjaaltonhehe20:14
pwnguinLP does have an email, but that wasnt the point20:14
pwnguintjaalton: the installer script or some crazy bug report script?20:15
tjaaltonsome crazy ..20:15
pwnguinfun20:15
tjaaltonnvidia-bug-report.sh20:15
tjaaltontseliot: yes, you'll have a party of the century ;)20:18
tjaaltonit never ends20:18
brycetjaalton: does the patch on 15477 look ok for inclusion in our packages?20:36
bryce(good morning btw!)20:37
tjaaltonbryce: morning bryce! yes, it does, but it's pointless without a fix for 14441 :)20:40
tseliotbryce: you're going on a vacation soon, right? May I ask how long it will last? (so that I know when I can bug you again about phase 1)20:44
brycetjaalton: ok, I highlighted the 14441 issues back to them20:46
brycetseliot: next week will be the distro sprint in London, then I'll be on vacation the week after that20:46
tseliotok, thanks20:47
bryceI can't promise I'll get anything useful done during the sprint; most of the time is meetings and talking with other devs about bugs and such20:47
tseliotbryce: no problem, in the meantime I'll do my part ;)20:48
bryceawesome20:49
brycetjaalton: btw, do you have any ideas on #246585?  I was trying to help mdz on it yesterday20:49
brycetjaalton: it appears the new -vesa doesn't report its screens correctly to gdk.  I've dug through the code from the gdk side of things yesterday, but am thinking to look at it from the -vesa side, but am not sure what angle to approach it20:50
tjaaltonbryce: the driver is very simple.. I believe the client code is doing something wrong here20:51
brycejcristau observed it seems that vesa seems to be using xinerama or thinking its using xinerama20:51
tjaaltonhmm20:51
tjaaltonI bet fedora is seeing something similar then20:51
bryceyeah I've been through the driver source; I doubt the bug is there, but I also dug down pretty deep in gdk and didn't spot it20:52
tjaaltoncompared vesa logs from hardy with the one from that bug.. nothing apparent in sight21:02
tjaaltonI'll subscribe ubuntu-x to fglrx-installer/nvidia* bugmail21:14
mario_limoncielltseliot, hum are you seeing seg faults starting X now on intrepid w/ nvidia too?21:17
mario_limonciellfglrx most definitely broke in the last day or two (bug 247376)21:18
ubottuLaunchpad bug 247376 in fglrx-installer "undefined symbols when trying to load fglrx" [Critical,Triaged] https://launchpad.net/bugs/24737621:18
tjaaltonso fglrx doesn't work with xserver 1.521:18
tjaaltoneither21:18
tjaaltonnvidia -173 and -177 work21:18
tjaaltonbut -71 and -96 don't21:19
mario_limonciellwell i wonder why i'm segfaulting then on this box w/ 17721:19
mario_limoncielli see the splash screen and then the X server dies21:19
tjaaltonsplash screen of what?21:19
tjaaltonusplash?21:19
mario_limonciellnvidia splash screen21:19
tjaaltonok, so it's not the same then21:19
tjaaltonmaybe -177 is just buggy ;)21:19
mario_limonciellbut yeah usplash is a mess on a bunch of boxes too21:19
tjaaltonright...21:20
tjaaltonlet's get plymouth!21:20
mario_limonciellthe weird thing with the nvidia segfault, there are no errors in the log except not being able to load dri221:20
tjaalton(plymouth == the new boot-candy for fedora10, using kernel modesetting)21:21
mario_limoncielli only know it's a segfault from starting X on it's own21:21
mario_limonciellah 21:21
mario_limoncielland that's what's breaking usplash?21:21
tjaaltonno, it's not in ubuntu :)21:21
mario_limoncielloh phew.21:21
tjaaltonpossibly some framebuffer madness21:22
tseliotmario_limonciell: sorry, I haven't updated my Intrepid system yet. I'll do it tomorrow21:38
mario_limoncielltseliot, yeah hopefully it's easily resolvable.  i've only got one intrepid nvidia system up and running.  unfortunately all the other ones i have still are fglrx21:39
mario_limonciellso SOL for a bit there21:39
tseliottjaalton: plymouth == works only with Intel?21:39
bryce*sigh* ETOOMANYBUGS21:40
tjaaltontseliot: yes, currently21:43
tjaaltonhttp://katzj.livejournal.com/432586.html21:43
tseliotmario_limonciell: did you try with Disable "dri2" in the Module section of your xorg.conf?21:43
mario_limoncielli didn't know disable was a valid keyword21:43
tjaaltonthe warning is harmless21:43
mario_limoncielli'll give that a shot21:43
tseliotmario_limonciell: it's in man xorg.conf21:45
tjaaltonoh, that screencast is using vesafb, not modesetting21:45
tjaaltonso it has multiple fallbacks21:45
tseliotmario_limonciell: let me know how it goes21:45
tseliottjaalton: ok, it's a sensible idea then21:46
tjaaltonbryce: i'm with you man..21:46
mario_limoncielltseliot, well no more dri2 error (duh), but still segfaulting21:47
tjaaltonmario_limonciell: have you tried -173?21:47
mario_limonciellyeah i just rolled back to it fearing something in 177 went wacky21:48
tjaaltonwhat the hell to do with nvidia/fglrx bugs that are a) old b) have no logs c) no idea what chip the user has21:49
tjaaltongoing through lrm-2.6.1521:49
mario_limonciellapport is off by default now...21:50
mario_limonciellso no logs via apport21:50
brycetjaalton: have xserver upgrades always been this rough?  I don't recall the 1.4 update triggering this many serious issues21:50
pwnguindidnt everything dangerous fall out of 1.4?21:50
tjaaltonbryce: nah, the vendors are just lazy21:51
tseliotmario_limonciell: does it help if you do startx -- -ignoreABI ?21:52
brycetjaalton: for old lrm bugs, I'd just let them know the bug's out of date and won't be further investigated here, and kindly ask that they re-test with hardy and report against $package if it still occurs21:52
tjaalton"kindly".. forget about it then :)21:52
bryce*short*21:52
bryces/h/n/21:52
tjaaltonah, monty python on the telly21:52
tjaaltonhehe21:52
bryceon the plus side, I'm finding myself getting pretty good at gdb21:52
tseliottjaalton: I receive a lot of emails which say only "I used Envy and now my card doesn't work" :-P21:53
brycetseliot: all of our Xorg bugs go that way :-)21:53
bryceexcept s/Envy/<$package>/21:53
brycefun fun21:53
mario_limoncielltseliot, well very odd, but i reinstalled 177 again (after downgrading to 173) and it appears to be working now?21:54
tseliotbryce: we should just learn how to read our users' thought, that's all...21:54
bryceI dream about making a web interface with some regex's to detect if they've included Xorg.0.log, mentioned a crash but not included a backtrace, etc. etc. and have some nice checkboxes (and 'check-all') so us triagers can process those stub bug reports faster21:55
mario_limoncielltjaalton, well i don't know that i would say they are lazy, but there was no predictable release date for xorg 1.521:55
tseliotmario_limonciell: did you do rmmod and modprobe before restarting the Xserver?21:55
mario_limoncielltseliot, yeah i did21:55
mario_limonciellso it's very hard to get your schedule together when you don't have a stable list of changes for the ABI and such21:55
tjaaltonmario_limonciell: and there still isn't but yeah21:55
tseliotmario_limonciell: ok, naive question21:55
brycebtw, I talked with AMD yesterday.  21:55
mario_limonciellabout xorg 1.5 support?21:55
tjaaltonbut fedora9 has been out two months, so either they have no fedora users or something is really wrong21:55
brycethey're working on implementing xrandr 1.2 for fgrx 8.53, which is slated to be released in Sept21:56
tjaaltonsweet21:56
brycebetas will be released to testers next month21:56
mario_limonciellbryce, er well i'm not sure you should be mentioning that with ubuntulog sitting around...21:56
tjaaltonsssh21:56
tjaalton:)21:57
* bryce shushes up21:57
bryceactually I'm not sure it's private.  maybe you know differently though21:57
mario_limonciellwell it's not been announced anywhere in the past21:57
tseliotmario_limonciell: we are the ones who signed an NDA, not Bryce21:57
mario_limoncielltseliot, it's not been announced to the beta list either21:58
mario_limonciellbut I had thought there was a 3 way NDA via AMD/Canonical/Dell as well21:59
mario_limonciellanyhow though.  did they talk about xorg 1.5 support?  randr 1.2 is kinda useless without xorg 1.5 :)21:59
bryceheh, I'm saying no more without a go-ahead ;-)22:00
tjaaltonworks just fine from 1.3 onwards :)22:00
brycebut I've requested clarification on it22:00
mario_limonciellhaha okay22:00
brycehave I ever mentioned how much I hate secrets?22:01
tseliot:-)22:02
mario_limonciellyeah i wish that they would publicly comment on this kind of stuff22:02
mario_limonciellwhat would you call the "contours" that are showing in this image: http://imagebin.org/22276 ?22:02
brycebanding22:03
mario_limonciellbecause that only happens with the 'ati' driver22:03
mario_limonciellnot with the fglrx22:03
bryceyeah we used to have that in inkscape with gradient patterns22:03
mario_limonciellwhat causes it?22:03
brycecolor interpolation bugs, 22:04
brycenumerical issues in gradient printing code22:04
brycebtw, is that background a png or a svg?22:04
mario_limonciellits the default ubuntu one22:05
bryceok so a png22:05
mario_limonciellthis was a hardy install upgraded to intrepid (before fglrx broke)22:05
brycesee if you can reproduce it in inkscape - draw a circle and color it with a gradient22:05
mario_limoncieller well i crashed inkscape trying to fill it with a gradient :)22:09
mario_limonciellill try again22:09
brycecould also try gimp22:09
bryceinkscape crashes are quite rare though; I'll bet backtracing that crash would give some handy info 22:10
mario_limonciellyeah inkscape definitely does it too22:10
mario_limoncielli did it with a big red circle gradienting to white22:10
mario_limoncielltheoretically linearly22:11
bryceok, that rules out it being just a png issue then22:11
mario_limoncielli'll put all this info into a bug then22:11
mario_limonciellits in bug 243372 now22:12
ubottumario_limonciell: Bug 243372 on http://launchpad.net/bugs/243372 is private22:12
mario_limonciellbah22:12
mario_limonciellokay it shouldnt be private :)22:12
mario_limonciellbug 24337222:13
ubottuLaunchpad bug 243372 in xserver-xorg-video-ati "RadeonHD 3670, unable to show entire color scheme" [Undecided,New] https://launchpad.net/bugs/24337222:13
tjaaltonthere, lrm-2.6.15 cleared and unsubscribed23:16
tjaalton~45 bugs less to worry about23:17
Awsoonnthought you might be in here Bryce :)23:18
Awsoonn bug #24299023:18
ubottuLaunchpad bug 242990 in xorg "xorg does not synchronize to vertical refresh" [Undecided,New] https://launchpad.net/bugs/24299023:18
AwsoonnI can't tell if it is my monitors ghosting or if I really do see what he sees, but what say you an that bug?23:19
brycelooking23:19
Awsoonndomo23:19
bryceAwsoonn: does it make a difference if compiz is enabled/disabled?23:20
Awsoonnlet me disable and see23:20
Awsoonnmore noticeable when disabled23:21
brycehmm, well without seeing a screenshot it sounds like https://bugs.launchpad.net/ubuntu/+bug/96991, but that should go away with compiz is disabled23:22
Awsoonnoh! here we go!23:22
ubottuLaunchpad bug 96991 in xorg-server "3D stuff breaks with Compiz:  Redirected Direct Rendering is needed in DRI" [Unknown,Fix released] 23:22
AwsoonnI took a screenshot of it23:22
AwsoonnI can't believe that worked. :) let me attach it23:23
Awsoonn*posted*23:29
brycehmm23:33
AwsoonnIs there somewhere that I should tell xorg to use vsync?23:34
brycewell, you can specify it in xorg.conf23:36
brycewhat you might want to do is compare your monitor's documented rates against what's listed in Xorg.0.log to see if something's misdetected23:36
bryceor try swapping monitors, if that's easier23:36
bryceI notice in the Xorg.0.log there's some warnings about pipe-A issues, so you could setting that option (google launchpad for 'Pipe-A' quirk)23:37
brycehowever that usually exhibits itself as a crash on lid close, not like this23:37
bryceI'd encourage you to post your Xorg.0.log, for comparison against the original reporter's23:37
brycealso, it couldn't hurt to test booting Intrepid alpha-1 or alpha-2 when it's out, just in case this issue's already fixed upstream23:38
bryceif it's a mis-detection of sync rate, you could also try the NoDDC option 23:39
Awsoonnmy Section "Monitor" contains nothing for a refresh rate, is that anything?23:39
brycethat just means you're letting xserver autodetect it23:39
brycewhich works 99% of the time :-)23:39
AwsoonnScreen resolution settings window shows that it is set at..... 50Hz23:40
bryceoh, also another approach if you suspect modeline issues, is to try alternate resolutions and refresh rates23:40
brycemodeline bugs tend to be specific to one particular setting23:41
Awsoonninteresting that the nvidia settings tool says I'm at 60 Hz23:41
bryceyou can also use 'ddcprobe' and 'get-edid | parse-edid' to check things23:42
bryceoh you're using nvidia?  the log shows intel...23:42
bryceyeah there's a known bug with -nvidia where it reports wrong refresh rates23:42
Awsoonnmy montiors report 60Hz as well23:42
Awsoonnthose logs are the OP's23:43
brycenVidia knows about the bug (I think it's in their release note and/or faq), but I don't know of any plans they have to fix it23:43
Awsoonnso MY issue is nvidia, but his is something else you think?23:44
bryceit's quite possible23:59
bryceit's very typical for unrelated bugs to have similar symptoms23:59
bryceand this gets confusing especially when the symptoms are described in text, rather than screenshots23:59

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