[01:15] i confirmed a bug involving CONFIG_MPENTIUMM=y and maverick's toolchain producing unbootable (hanging) kernels [01:15] happens with local builds both on real hw, and in qemu [01:15] and i tested the build on a remote fresh 10.10 system, and it hangs in the same place [01:16] affects at least 2.6.35.7, 2.6.35.8, and 2.6.36 [01:16] does this belong under gcc bugs? [01:18] also, gcc testsuite produces some fun results :) === Pilif12p_ is now known as Pilif12p === manusheel is now known as manusheel_afk === fabbione|lpc is now known as fabbione === yofel_ is now known as yofel [12:34] pitti: question about apport: when using 'ubuntu-bug --package XX' is file-bug mode supposed to be implied or do those options only work now if you add -f ? --help says that file-bug mode is implied if a single argument is given but for some people '--package=bash' is a single argument too but apport silently fails currently if you don't add -f/--file-bug [12:34] (bug 665953) [12:34] Launchpad bug 665953 in apport (Ubuntu) "ubuntu-bug --pid and --package options do not work" [Medium,Confirmed] https://launchpad.net/bugs/665953 === tkamppeter_ is now known as tkamppeter === oubiwann is now known as oubiwann-away === oubiwann-away is now known as oubiwann === tlyu_ is now known as tlyu === sanchaz-away is now known as sanchaz [18:04] not sure if this is relevant here, but latest flashplayer crashes on hardy when going fullscreen on any video site. flash tries to use gtk_widget_get_window which is available since gtk-2.14 and hardy has gtk-2.12. I have reported this bug on bugs.adobe.com [18:05] i guess reporting this on launchpad is not neccessary but please tell me if otherwise [18:06] jazzdog: I think it has to be fixed at Adobe, so I think an LP bug isn't necessary. [18:07] funny thing on bugs.adobe.com i can not see my own reported issue :) [18:08] they say I reported a bug related to a non-public product or component [18:08] jazzdog: well, flash 9 is still in multiverse [18:09] It seems like having an LP bug would still make sense, especially if multiple people start noticing this [18:09] For tracking purposes more than "we need to fix this" purposes [18:10] it should affect everyone using hardy heron and watching videos on any video site in fullscreen [18:10] and a link to the upstream Adobe bug in the description would help [18:10] jazzdog: is this Flash 10 and Flash 9 or just Flash 10? [18:11] latest flash from partner repo and from adobe.com download site [18:11] so flash10 [18:11] ok, so you can file against adobe-flashplugin [18:11] but my reported bug is not public on adobe.com even I can not see it [18:12] That's fine. What matters is that we know the bug exists [18:13] jazzdog: do you know about the public bug tracker for flash? https://bugs.adobe.com/jira/secure/IssueNavigator.jspa? [18:14] that's where I've submitted the bugreport, or at least the domain is the same [18:14] jazzdog: please file a bug in launchpad against adobe-flashplugin [18:14] ok [18:22] You need to specify a package or a PID. See --help for more information. [18:22] after running ubuntu-bug adobe-flashplugin [18:24] It's in the Partner repository, not Ubuntu, so I'm not sure ubuntu-bug is intended to work for that case. [18:24] it needs -p adobe-flashplugin [18:24] -p was not on the wiki page [18:25] jazzdog: ah, that's no longer necessary in LUcid [18:25] what is not in lucid? and btw I'm on hardy :) [18:25] jazzdog: right [18:26] i see [18:30] okay I've reported it on LP [18:42] oops I've just realized this was already reported on karmic [18:42] which is strange because karmic has gtk-2.18 [18:43] only adobe can solve this :) [18:43] or lightspark or gnash === dendrobates is now known as dendro-afk === MattJ100 is now known as MattJ === tkamppeter_ is now known as tkamppeter [19:36] cjwatson, I just proposed a merge for parted and dmraid to finally drop the ubuntu patches removing the 'p' from the device name. last we talked you wanted to do that during maverick but I guess didn't have time... hopefully you can just bzr merge and commit easily [19:41] psusi: this is for things like /dev/mapper/loop0p1? [19:43] ebroder, no.. for /dev/mapper/nvidia_13412352435 [19:43] ah, ok [19:44] ebroder, upstream went to inserting a 'p' between the base raid device name and the partition number a while ago but ubuntu has been carrying patches to reverse that for a while [19:44] err, right. dmraid, not dmsetup. neverm ind [19:44] I believe it causes breakage when the base name happens to end in a digit [19:45] things like grub can't figure out where the base name ends and the partition number begins [20:13] pitti, you were trying to work out automounting of esata disks right? I've been looking into that a bit lately and it seems AHCI has the capability to tell the os a port is external, but the driver doesn't bother using that bit === hanska is now known as dapal [21:47] why does a process opening a partition device for write access generate a udev change event, and why is that a time we want to ask lvm to activate all volumes? [21:48] I don't exactly understand the first half, but the second half is easy - all of Ubuntu's device detection is event driven. So new LVM device -> bring it online [21:49] you need that sort of behavior for things like "LUKS on LVM on RAID" to come up in any sane way [21:49] right, so you want to activate when a new disk is added, but why when a partition is "changed" which apparently means anything wrote to it [21:49] No, "changed" means the device changed. The partition was resized, or something like that [21:49] Like I said, I don't know opening the device for writing generates a change event [21:50] nope, a change event is emitted simply from opening the dev node in read/write mode [21:50] A change event is *also* emitted if the partition is resized, which *is* a case where you need to bring VGs online [21:50] on the partition device that is [21:50] not the raw disk, which you would think would get a change event when you change the partition table [21:51] I would think that if the change added a partition, you would get an add event on the partition and want to activate... but right now it activates on any add or change event for a disk or partition... which seems a bit more than needed [21:52] so keybuck wrote a patch to lvm to make sure it does not open devices for write access unless absolutely needed or it caused a feedback loop where the open causes a change, which runs lvm, which opens, etc [21:52] The changelog does describe it as a "sledgehammer" [21:53] I can't quite come up with the exact scenario where you'd need it, although I'm sure cjwatson or Keybuk could explain it better [21:53] but I don't understand why the problem happens in the first place since I can't see any sense to activating every time a partition is changed, if changed means somebody wrote to it [21:53] hrm... [21:53] Running pvcreate /dev/sda2 -> activating VGs is the closest I can get [21:53] I guess vgcreate will write to the PVs, and activating the VG after that happens makes sense [21:54] But the behavior definitely makes sense to me, even if I can't explain it exactly [21:54] there's nothing in it to activate yet when you pvcreate though.. [21:56] vgcreate, though, not pvcreate [21:58] same thing... when you first create a volume group it is empty.. [21:58] Like I said, I don't know the exact scenario, but it's very easy for me to imagine that there are scenarios *like* those where the behavior is desirable [21:59] though I suppose if lvcreate does not automatically activate the new volume then you would do that with this udev rule... [22:00] This rule only triggers on PVs, not LVs, no? [22:00] the rule triggers whenever any block device is added or changed [22:00] and it has been identified by blkid as being a pv actually... [22:00] Right. "ENV{DM_UDEV_RULES}=="", GOTO="persistent_storage_dm_end"" [22:01] yea... [22:03] In general, there's more than enough subtlety in LVM that I try not to assume behavior is unnecessary just because I can't come up with a scenario that needs it [22:04] (especially if I can imagine scenarios that *almost* need it) [22:04] that's why I asked ;) [22:08] kirkland: i was thinking for powernap. Roll back to original (before adams changes) for the main trunk and create a series with his changes. So that I can first work with the main trunk by commiting stuff, because I;ve not yet dec [22:10] kirkland: haven't yet decided what to do. So I was planingn first to do everything related to pm-powersave and then work on the Monitors. My approach right now is to use the original branch... but I might decide for adam;s approach [22:12] natty http://dpaste.de/mGIK/ *** buffer overflow detected ***: tftp terminated======= Backtrace: =========/lib/libc.so.6(__fortify_fail+0x37)[0x7f29b446b527]... [22:12] what's the cli to post that to lp? [22:13] CarlFK: Do you get anything if you run apport-cli? [22:14] juser@dhcp232:~$ apport-cli [22:14] No pending crash reports. Try --help for more information. [22:17] ran the tftp command manually (instead of the python Popen([tftp shaz... - crash, but still "No pending crash reports. [22:17] ebroder: hehe "You and @ebroder both follow... Marigold Farmer" *high-five* [22:18] CarlFK: you may need to enable apport for it to catch that. edit /etc/default/apport and then sudo /etc/init.d/apport start [22:20] or sudo service apport start force_start=1 [22:20] on Lucid and later [22:20] kees: will apport catch glibc throws like that? [22:21] http://dpaste.de/wssw/ (gdb) bt full#0 0x00007ffff7a8cba5 in raise (sig=)... [22:22] enabled=0 Souldn't that be =1 for betas? [22:22] ebroder: yes [22:23] ebroder: I wrote the code to do it. :) [22:23] kirkland: /win 3 [22:24] -_- sry bout that :) [22:24] 1: Launch a browser now... grumble. I am in an ssh shell [22:24] lynx! ;) [22:25] ok, I am playing along.. "copy this URL into a browser on another computer." [22:26] times like that, you wish that it'd file the bug by email [22:26] CarlFK: which tftp is that, btw? [22:26] kees: whatever is installed by apt-get install tftp [22:26] netkit-tftp, ok [22:29] "The file "CoreDump.gz" was attached to the bug report." um... I am on a different box? [22:29] oh wow. [22:29] did it push stuff up to some secrit place? [22:30] CarlFK: in the meantime, you might try tftp-hpa [22:30] CarlFK: yeah, it send it directly to LP, and then the bug report creation uses the hash to look it up [22:31] kees: brilliant! [22:31] what is the package that contains libc-dev or -dbg or whatever debug symboles? [22:31] kees: out of curiosity, how do glibc bug reports get captured? patch to glibc? [22:31] CarlFK: what's the bug#, btw? I can add some comments [22:32] #672325 [22:32] https://bugs.launchpad.net/ubuntu/+source/netkit-tftp/+bug/672325 [22:32] 'Error: Could not parse data returned by Launchpad: HTTP Error 401: Unauthorized\nResponse headers:\n---\ncontent-length: 21\ncontent-type: text/plain\ndate: Sun, 07 Nov 2010 22:32:11 GMT\nserver: zope.server.http (HTTP)\nstatus: 401\nvary: Accept-Encoding\nvia: 1.1 wildcard.edge.launchpad.net\nx-powered-by: Zope (www.zope.org), Python (www.python.org)\n---\nResponse body:\n---\nBug 672325 is private\n---\n (https://launchpad.net/bugs/672325)' [22:32] CarlFK: look for -dbg and -dbgsym https://wiki.ubuntu.com/DebuggingProgramCrash but -dbgsym needs a separate repo [22:32] ebroder: yes. I wrote code for glibc to retain the abort message. drepper took it without crediting me. \o/ [22:32] opps, marked it private.. fixing... [22:33] Haha, awesome [22:34] at least I would like to un-private it.. how do I do that? [22:34] CarlFK: upper right corner [22:34] This bug report should be private fixed [22:34] https://bugs.launchpad.net/ubuntu/+source/netkit-tftp/+bug/672325 [22:34] Launchpad bug 672325 in netkit-tftp (Ubuntu) "tftp assert failure: *** buffer overflow detected ***: tftp terminated" [Undecided,New] [22:35] juser@dhcp232:~$ aptitude search '~i' | fgrep -e '-dbg' [22:35] i libc6-dbg - Embedded GNU C Library: detached debugging [22:35] er, nm - thougth that was installed [22:38] "libc6-dbg is already the newest version." so it is installed. anything else I can do? [22:38] installed tftp-dbgsym for more readability (check that wiki page I linked to) [22:40] 0x4013bc : callq 0x401118 <__strcpy_chk@plt> [22:41] cp = tp->th_stuff; [22:41] strcpy(cp, name); [22:41] * ebroder sighs [22:45] gah, this is a bug in gcc again! [22:46] I will take a closer look on monday [22:47] gcc has bugs? [22:47] * lifeless is shocked [22:47] it's almost as if it was software [23:22] Failed to fetch http://ddebs.ubuntu.com/dists/natty/Release Unable to find expected entry main/source/Sources in Meta-index file (malformed Release file? [23:22] Is maverick-proposed still frozen? [23:24] slangasek, lool: can I upload a fix for LP: #631980 to maverick-proposed? [23:26] mdeslaur: go ahead; the freeze doesn't block you from uploading anyway, it just means the package won't necessarily be accepted right away [23:26] mdeslaur: the mail to u-d-a mentions Nov 8th as date of -proposed thaw [23:26] slangasek: ok, thanks [23:26] the freeze isn't quite cleared yet because we've had some difficult-to-verify SRUs in the queue that I need to bypass; but things should be opened up by tomorrow [23:27] no source packages? http://ddebs.ubuntu.com/dists/natty/main/ [23:28] CarlFK: ddebs are supplementary binary packages generated at build time; look to the main archive for source packages [23:28] slangasek: k - just wanted to make sure it was expected [23:35] kees: http://dpaste.de/IarY/ [23:36] should I add that to the bug report as an attachment? [23:37] i guess it isn't that big... I was thinking it was too big for pasting into a comment