/srv/irclogs.ubuntu.com/2009/02/10/#ubuntu-x.txt

maxbapt-get source xserver-xorg-video-intel00:02
maxbapt-get build-dep xserver-xorg-video-intel00:02
maxbcd xserver-xorg-video-intel-VERSION00:02
maxbapply patch00:02
maxbdch -l zmiq "My local package to do foo"00:03
maxb(^That adds an entry to the debian/changelog, thus giving the package a distinct version number)00:03
maxbdpkg-buildpackage -b00:03
zmiq_many thanks; I'll try asap!!00:32
zmiq_the truth is I have been waiting for two months for a patch, which is now developed and tested as working, and don't want to wait anymore; it's regarding Option "SDVOBOutput"00:36
zmiq_developed by Wang Zhenyu00:36
zmiq_more at: http://bugs.freedesktop.org/show_bug.cgi?id=1782300:40
ubottuFreedesktop bug 17823 in Driver/intel "[945GM] Unable to switch to VGA with 2.4.2, VGA-1 replaced by TV-1" [Critical,New]00:40
zmiq_maxb: i've completed all steps; will dpkg-buildpackage -b also install the new module? in which direcotry?01:43
stgraberbryce: are your aware that geode GX2 are broken in Jaunty (more than they were in Intrepid) ?02:51
stgraberback in Intrepid, I just had to mention the Driver to use (geode) and force XAA to have them working, now I get a X seg fault in all cases02:51
* stgraber goes looking for a bug on LP02:52
stgraberbryce: bug 32748403:04
ubottuLaunchpad bug 327484 in xserver-xorg-video-geode "X crashing with Geode GX2 on Jaunty" [Undecided,New] https://launchpad.net/bugs/32748403:04
brycestgraber: no I hadn't heard03:40
stgraberthese things are usually extremely buggy and hard to get to work but there always were workarounds :) For jaunty the only one I found so far is using vesa which kind of sucks ...03:42
brycestgraber: unfortunately I think AMD laid off their geode support people03:43
bryceand I gather they will no longer be producing geode03:43
stgraberindeed, unfortuanately we have a few hundreds of these things out there :)03:44
brycestgraber: if you can locate a patch, or a new release which fixes the issue, we can upload it03:44
bryceelse, at least get a full backtrace, in case it's just something simple like a null pointer03:44
brycehowever my guess is that it is missing some support03:44
brycestgraber: also check the geode mailing list03:45
brycemorning everyone17:18
brycetjaalton, jcristau: does Option "ModeDebug" impact performance?  if not, would there be any reason not to switch it on while we're in development?  (up to beta perhaps?)18:02
jcristaubryce: sounds like a good idea18:03
crevettegood evening gentlemen18:35
bryceheya crevette18:37
crevetteI have a bug recently with jaunty where I can see what it is displayed after usplash, removing usplash to the "kernel " line in grub fixes the issue18:37
crevettes/can/can't/18:37
crevetteare you aware of such issue ? I remember having the same bug in intrepid :)18:37
brycecrevette: yes18:37
crevetteokay so I don't need to spam you more18:38
brycecrevette: bug in usplash; you can sub to it18:38
bryce32723018:38
brycecrevette: kees is looking at it currently - you could help him with some testing if you'd like18:39
crevettewonderful18:39
crevettekees: don't hesitate if I can help you18:39
keescrevette: okay, I've got 8 builds of usplash I need tested.  :)18:39
crevettehehe18:40
keescrevette: basically, to isolate the change that broke for some people18:40
crevettekees: do you need I enumerate my hardware ? 18:40
keescrevette: nope18:40
keescrevette: okay, ready?18:44
keescrevette: here's the list: http://people.ubuntu.com/~kees/usplash-testing/18:44
crevetteI've few minutes free18:44
keesin order, test 242 242.1.1 242.1.2 242.1.3 242.1.4 243 and 24418:44
crevettedo you have i386 deb ? 18:45
keeseach directory has 2 deb, both are needed.  at some point, after a reboot, it'll fail.  242 should work, for example.18:45
keescrevette: give me a moment, I can build those...18:45
crevetteokay thanks18:45
crevettetoo bad I don't have a VM of jaunty18:46
* crevette opens another bug in the meanwhile18:47
keescrevette: okay, i386 debs are up.18:53
crevettewonderfull18:54
crevettefirst reboot18:57
crevette242 works19:01
keesokay, good.  that's expected (242 is basically 0.5.27).19:01
* crevette will try the dichotomy way19:01
crevettelet pick on in the middle :)19:01
* kees nods19:01
crevette242.1.4 is affected19:08
keesokay.  /me suspects 242.1.319:09
keesthough why it would cause this, I have no idea.19:09
keesif 242.1.2 works and 242.1.3 is busted, then I'll revert the signal handler changes.19:10
crevettewhich one do you want me to test ?19:10
keestry 242.1.2 next.19:10
crevette242.1.2 works19:14
crevettelet's got for 242.1.319:16
keesokay, ... 242.1.3 next, I guess.  :)  Thanks so much for doing this testing19:16
crevetteno problemo, always happy to help19:17
crevette242.1.3 works too19:19
crevetteso 242.1.4 is guilty 19:20
crevette:)19:20
* kees ponders19:20
keesuhm19:20
* kees scatches his head19:20
keesseems the diff between 242.1.3 and 242.1.4 is only in the changelog...19:21
crevetteshould I try again to reboot to see if it works again19:21
tjaaltonbryce: i thought upstream turned it on already, but maybe only for master19:21
keesyeah, can you?  maybe this problem isn't always present on every boot?  geh, I wish I could reproduce this locally19:22
crevettelet do it again19:22
crevette242.1.3 didn't work the 2nd time19:30
keesokay, that does imply a race with the alarm handler... copying up another test version...19:30
keescrevette: http://people.ubuntu.com/~kees/usplash-testing/ 0.5.29~kees119:31
keesthat's 244 with the changes between 242.1.2 and 242.1.3 removed19:31
crevetteokay19:32
* kees hugs crevette19:33
brycethanks crevette :-)19:33
* crevette hugs kees & bryce19:34
crevette:)19:34
crevetteSee you soon19:35
crevetteso it seems to work19:39
crevetteI admit I didn't tested twice :)19:39
keesheh.19:39
keesokay, well, maybe I can get bryce to test the ~kees1 build too?19:40
keesbryce: do you need i386 or amd64 debs?19:40
brycesure thing19:40
brycei38619:40
keescool, they're up there for ya19:40
keeshttp://people.ubuntu.com/~kees/usplash-testing/19:40
keesthe ~kees1 build reverts a signal handler clean up.19:40
bryceinstalled; rebooting19:41
keescool, thanks19:41
bryceok, came right up19:43
keeswhee19:43
bryceso, I can confirm the fix19:43
keesokay, I'll push this version, though it's not clear to me _why_ it broke.  :P19:44
* crevette goes to eat its soup bowl19:46
keescrevette: thanks again.  I've uploaded an official 0.5.29 which reverts the signal handler changes, so hopefully this bug should stay dead.  :)19:51
crevettekewl, thanks a lot19:52
keesnp20:00
tjaaltonbryce: btw, the compiz-freeze with intel still happens for me, but it's harder to trigger with the latest patch :/20:32
brycetjaalton, ok20:33
tjaaltonso it's only partially fixed20:35
mnemotjaalton: same here but here it seems to repro more and more often the longer you run X though20:38
tjaaltonmnemo: there's another patch to try21:02
tormodtjaalton: I still have /etc/X11/Xsession.d/65mesa-check-x86-64 here, is this a conf file that needs to be deleted even if it's not shipped any longer, or is it just me?21:35
jcristauyes, the postinst should take care of that21:36
tormodjcristau: thanks. just rm it no question asked?21:45
jcristautormod: it's a bit more complicated than that21:45
tormodI was afraid so :)21:45
brycetjaalton: so do you think slipping ModeDebug into dexconf would be ok for pre-beta?  or is there a better way we could do it?21:45
jcristaubryce: don't do that. swap the default in the code21:46
jcristautormod: xsfbs.sh has remove_conffile_lookup, remove_conffile_commit, remove_conffile_rollback. there's another example on the debian wiki somewhere iirc21:47
brycejcristau: why not?  would be easier to swap than patching the xserver21:47
jcristautormod: http://wiki.debian.org/DpkgConffileHandling21:48
tormodbryce: it's easier to revert it in the server than to try to rewrite xorg.conf's later21:49
jcristaubryce: because dexconf only runs on initial install, and is the wrong place to set defaults anyway21:49
jcristaualso, what tormod says21:49
tormodjcristau: but but this is not really a configuration file... nobody should have changed if not only to disable it21:51
jcristauit's a conffile...21:51
jcristauthough i can understand if you don't care enough and want to just remove it :)21:52
jcristautormod: the xsfbs.sh stuff also makes sure the conffile is restored if for some reason the upgrade gets aborted (the rollback function)21:52
tormodwell that's almost useful - maybe reason enough21:53
tjaaltontormod: what shipped it?21:53
tjaaltonbryce: by patching the server, yes21:54
tormodtjaalton: libgl1-mesa-dri21:54
tjaaltontormod: ok.. nothing in the changelog21:55
tormodI hope jcristau looks another way :) I don't think Ubuntu supports broken release upgrades anyway.21:57
jcristauheh21:58
tjaaltontormod: oh, I dropped it22:04
tjaaltonmeh22:05
tjaaltonthere should be a /usr/share/X11/Xsession.d22:06
tjaaltonor similar22:06
tjaaltonand /etc/X11/Xsession.d left for the local admin22:06
jcristau(we had the same problem with xfree86-common -> x11-common, i think the files in /etc/X11/Xsession.d/ were left around on sarge->etch upgrades :/)22:07
jcristau(so you'd end up with two copies until you purged xfree86-common. which means two ssh-agents starting. which means fail.)22:08
maxbIt's a shame dpkg doesn't make the corner-cases of conffile handling easier22:09
tormodI thought you had to declare what's a conffile? Or is everything under /etc conffiles?22:09
maxbdebhelper declares everything under /etc for you22:10
maxbThough I believe it would be frowned upon to put something in /etc/ that wasn't a conffile22:10
tormodthere is certainly a good bunch of stuff under /etc that is not meant to be edited, no more than you would recompile something under /usr/bin22:11
maxbI guess it would be ok if you had a comment informing the local admin that the file is liable to have local changes destroyed on upgrade22:14
maxbthough still not ideal22:15
tormodfor instance there is gdm.conf, with gdm.conf-custom for making changes22:15
tormodoh, gdm.conf has such a warning :)22:16
maxbthough is still a conffile22:17

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