=== smb` is now known as smb [07:54] infinity: we have still to do linux-ec2/lucid and will try to do asap. sorry about the delay === henrix_ is now known as henrix === Guest80225 is now known as panda|x201 [08:36] lp 1164074 [08:36] Launchpad bug 1164074 in linux (Ubuntu Raring) "[Highbank] Quantal to Raring upgrade issues" [High,Confirmed] https://launchpad.net/bugs/1164074 [09:01] brb [09:19] diwic, i have no sound on todays updates, on an all intel dell, 3 years old. any suggestions as to diagnostics? everything alsamixer shows for output is unmuted [09:20] apw, 13.04 ? did you have kernel and/or pulseaudio updates ? [09:20] i had kernel at least, i had 450M so probabally most things [09:20] for raring indeed [09:21] apw, let's assume kernel - last PA update was 3 weeks ago [09:21] yay [09:21] quality in action [09:22] apw, for initial investigation "ubuntu-bug audio" [09:27] diwic, https://bugs.launchpad.net/ubuntu/+source/alsa-driver/+bug/1166097 [09:27] Launchpad bug 1166097 in alsa-driver "[HDA-Intel - HDA Intel, playback] No sound at all" [Undecided,New] [09:28] apw, did you test speaker only or headphones too? [09:29] diwic, i have personally tested all outputs, nothing [09:31] infinity, yep [09:33] apw, but USB headset (if you're having any) is still working as usual I assume? [09:33] diwic, don't think we have a usb headset here... [09:33] apw, ok. [09:34] diwic, i do have some usb speakers, but they are not physically proximate [09:34] apw, let me look through recent patches to the driver and see if I find something [09:36] apw, argh, nothing there either [09:36] apw, I guess you're down to bisection then [09:37] apw, i e boot an older kernel to verify the kernel is the fault [09:37] diwic, yeep [09:37] diwic, daily quality guarenteed ... sigh [09:38] apw, how many days of update was it? [09:41] diwic, a lot ... i have had no network for weeks, i was last 'here' with b/w about 2 weeks ago [09:42] diwic, rebooting into a 3.5 kernel and i have sound [09:49] apw, you must have had sound with an earlier 3.8 kernel too? [09:51] diwic, so ... 3.8-8 is ok, 3.8-9 is bad, i must have only used my usb output since then i assume to have not noticed for so long [09:51] apw, what dates are that approximately? [09:52] commit 31b67ffaf4f9e72dce22d028180fc10a66dfec77 [09:52] Author: Tim Gardner [09:52] Date: Tue Feb 26 12:15:23 2013 -0700 [09:52] UBUNTU: Ubuntu-3.8.0-8.17 [09:52] commit f7c0639bf5538a557a1f05ea6a81fb237776e2fa [09:52] Author: Tim Gardner [09:52] Date: Thu Feb 28 08:47:51 2013 -0700 [09:52] UBUNTU: Ubuntu-3.8.0-9.18 [09:55] apw, what's the command for finding out the commits in between, for the sound/pci/hda directory? [09:55] diwic, tricky cause it is a stable update ... ick [09:55] git log 6071a7768562b970491c66127779de1f5b1250a3..Ubuntu-3.8.0-9.18 --oneline [09:56] is the only changes we made ... so it must be a mainline change, working out what that is now [09:58] diwic, it also includes 3.8 -> 3.8.1 stable [09:59] apw, hrm, I'm looking at gomeisa's ubuntu-raring tree and the only v3.8 tag it has is v3.8-rc5 [10:00] could be i have them locally separatly [10:00] apw, yeah, just haven't synced raring because I haven't got around to buy a bigger ssd :-) [10:01] diwic, looking what was in the .1 updae [10:02] diwic, [10:02] ca36807 ALSA: hda - hdmi: ELD shouldn't be valid after unplug [10:02] 97ec241 ALSA: hda - Fix broken workaround for HDMI/SPDIF conflicts [10:02] 13ba2f7 ALSA: hda - Workaround for silent output on Sony Vaio VGC-LN51JGB with ALC889 [10:02] 6a9c847 ALSA: hda - Fix default multichannel HDMI mapping regression [10:02] 41e19bd ALSA: hda - Release assigned pin/cvt at error path of hdmi_pcm_open() [10:02] 45d13ae ALSA: hda - Disable runtime PM for Intel 5 Series/3400 [10:02] looks to be the delta for hda [10:03] apw, I looked through all those already [10:03] apw, none of those looks applicable to your problme [10:03] i e your hardware [10:03] what does linux-image-extra contain (in the mainline builds) ? [10:03] * apw goes test that it _is_ 3.8->3.8.1 [10:04] amitk, 'the other half of your modules', linux-image is 'virtual' linux-image+l-i-extras is 'generic' [10:04] apw: so probably useful on a desktop [10:05] amitk, indeed yes [10:06] apw: thx [10:06] amitk, np [10:06] diwic, identifying whether it is stable or our changes next [10:06] will let you know [10:07] apw, thanks. Yeah, I think we're down to a real bisect at this point no matter what [10:07] * apw secretly blames diwic :) [10:13] diwic, ok 3.8 mainline is good, 3.8.1 is bad [10:14] diwic, what was my bug number? i've lost it already [10:15] apw, bug 1166097 [10:15] Launchpad bug 1166097 in alsa-driver (Ubuntu) "[HDA-Intel - HDA Intel, playback] No sound at all" [Undecided,New] https://launchpad.net/bugs/1166097 [10:22] diwic, ta [10:32] diwic, so it is deffo one of those 6, just reverted just those off -9 and its working [10:33] apw, the annoying thing is that most of those are HDMI related [10:33] diwic, yes [10:34] apw, hmm, you do have an HDMI codec on the same sound card though [10:34] diwic, it does have hdmi as well [10:34] apw, btw, I assume you have not tested HDMI audio? [10:34] diwic, i have not [10:34] it is not something i use regularly [10:37] apw, ok. well, continue searching. My guess is "Fix broken workaround for.." [10:38] * diwic wonders if we'll need a "Fix for fix for workaround for" patch... [10:41] * diwic goes for lunch, bbl [11:21] back [11:25] diwic, ok ... bisect says: [11:26] commit ca368073ffcbdd25486d65c150039423f677f821 [11:26] Author: David Henningsson [11:26] Date: Tue Feb 19 16:11:22 2013 +0100 [11:26] ALSA: hda - hdmi: ELD shouldn't be valid after unplug [11:26] is at fault. [11:29] apw, eh [11:29] apw, ouch ? [11:29] apw, probably it triggers some other bug in a strange relationship [11:39] diwic, what now :) [11:42] apw, I'm still in denial ;-) [11:43] apw, it's not a dangerous patch to revert [11:43] diwic, i'll reset to latest kernel and confirm it reverting it there solves it too and its not just luck [11:44] apw, could you also attach the contents of /var/lib/alsa/asound.state to the bug? [11:46] diwic, attached [11:47] apw, can't see it, are you sure the attachment uploaded correctly? [11:48] diwic, try that (i hate LP) [11:48] now it's there [12:11] diwic, ok confirmed that reverting that patch against the 3.8.0-16 base also fixes my issues. am trying to debug now, as your patch seems reasonable assuming no other bugs [12:16] apw, it should only affect the output of a file under proc. [13:05] diwic, ok i have just added some prints (no functional change) on top of your not-working patch and it starts working ... this is well peculiar [13:11] apw, oh, a heisenbug [13:11] apw, out of curiousity what prints did you add? [13:12] brb [13:12] http://paste.ubuntu.com/5689367/ [13:12] diwic, ^^ [13:12] apw, also, I saw something in your dmesg; "alsa restore exited with status 99" or something similar. [13:12] apw, do you know if this is related, i e, if this message only comes when it's working, non-working, or unrelated? [13:12] diwic, interesting .. [13:13] will check [13:13] [ 12.683939] init: alsa-restore main process (1436) terminated with status 99 [13:13] diwic, present when working as well [13:13] apw, okay [13:13] apw, then it's not that [13:20] can i poke someone about https://bugs.launchpad.net/bugs/1164739 ? i'm not sure if its regression or not, but at very least i'd expect a kernel package to not have missing dependencies inside it. [13:20] Launchpad bug 1164739 in linux "Can not mount cephfs in VM from cloud image" [Medium,Confirmed] [13:24] smoser, Could require the extras package. Or someone asking to move certain modules to the main package... [13:24] smb, right. read my comment at the bottom. [13:24] i dont have strong feeling either way on that. [13:24] but ceph.ko should not be in a package if libceph.ko is not (as it former depends on latter) [13:25] i'd assert thats a packaging bug if significant priority. i'm surprised if humans are maintaining the kernel module list without automatic dependency resolution [13:26] smoser, Right probably one or the other. I think there was at some point a request to have ceph in virt so if it is only 200k more I guess you or Ben would not mind the difference [13:26] smb, well, i'd like to see the justification. [13:27] 200k is not insignificant at all. installed size is ~ 12M, so 200k is 2% growth. [13:28] basically, i dont want to willy nilly be adding things there. [13:28] smoser, That would be coming from the bug report, right? But yeah, it has not been a problem yet to maintain the list by hand. [13:28] so it is maintained by hand, ok. [13:29] smb, ok. i'll open another bug that you can triage as you like for "module dependencies should not be managed by hand" [13:30] and then i'll ask the user here for some use case as to why they want ceph.ko in -virtual [13:31] smoser, Yes, it is. Basically whenever someone tells us something needs to be supported in the cloud image we add it. [13:31] we should have some sort of policy to guide us in such decisions [13:32] smoser, Hm, I am not sure this really qualifies (should not be done by hand). Whoever wants something now should test what is required. I wonder whether there was an actual request for ceph... [13:35] smb, well this user wanted ceph, but really only because they wanted to use ceph and didn't know about the -extra (and i told him). [13:35] clearly if you're given the choice between "do you want to have 1 extra step to use something" or "do you want to just use it", you'd pick the second. [13:36] but we have the size concern to consider. [13:36] smoser, Its just that I *think* there has been someone asking for it before. I just cannot remember [13:36] smb, right. [13:36] smoser, can you git blame that list ? [13:36] and see? [13:36] at least who did it [13:36] diwic, ok confirmed removing those printks also makes it break again [13:36] smoser, thats what I currently try [13:37] smb, reguarding the dependency resolution, i'd suggest just failing a build if you're missing dependencies. [13:37] there are no cases where that is not a bug. [13:38] apw, I assume the "restring cap" printk was never printed? [13:38] diwic, correct [13:39] smoser, Not sure this is that simple. At least now, the build is done completely and just the packaging step splits the modules up. [13:40] hm.. i dont know. but personally i'd rather have a developers build fail then have 'modprobe' fail on a user system in this way. [13:40] smoser, http://bugs.launchpad.net/bugs/1063784 [13:40] Launchpad bug 1063784 in linux "Ceph module not installed" [Medium,Fix released] [13:41] So added around quantal [13:41] But might be the other module did not exist then [13:41] smb, and then broken in raring to a kernel change probably. [13:41] smoser, right [13:41] (which gives more credence to my argument :) [13:42] (the argument that there should be some test to stop this from being released) [13:43] smoser, Hm, there is a libceph already (in q) [13:43] smb, the reporter claims this is broken in 12.10 [13:44] that is a 3.5 kernel in the report. [13:45] smoser, That would be Q, so someone did not test things after asking for it... :-P [13:45] apw, the thing is, eld->eld_valid should have been false always anyway, since you don't have an hdmi monitor plugged in. [13:47] smoser, But ok, if it is possible to run a dep check on a package subdir that would catch those things [13:49] smb, its definitely doable. it might take some re-work of the kernel build though. i dont know. [13:49] but anyway. [13:49] i guess the right thing to do at the moment is to include libceph.ko [13:49] and probably sru that. [13:51] smoser, Yeah, in theory it looks like one can point depmod at a different base dir and a system map file... so it should be possible. So let me know the bug for that. And for libceph, yes, sounds reasonable as having ceph.ko was already done, too. [13:52] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1166197 [13:52] Launchpad bug 1166197 in linux "included kernel modules may miss dependencies (-virtual)" [Undecided,Confirmed] [13:52] smb, you definitely can do that. i use depmod that way in cirros [13:53] smoser, Ok, I will try to add it to the build process then. [14:05] infinity, bjf: we are almost through with the pile of SRUs from last week, there's a precise one that psivaa is working on now, and I'm about to start the ec2 one [14:14] plars, ack, looks good [14:46] * ogasawara back in 20 [14:55] infinity: i know you took care of the transition from Q/omap to Q/generic [14:55] infinity: we need to the same for Q/highbank to Q/generic [14:55] infinity: and i didn't find any patches in our kernel tree about it [14:56] ppisati, meta should handle that [15:00] ogra_: i know, but since he already handled that [15:01] ogra_: he knows what to do better than me [15:01] yep, i just mean it should be visible in your package somewhere :) [15:06] actually was rtg who handled that [15:06] http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-raring-meta.git;a=commit;h=289b3dbee84086f2e323a1da7ccf3ea410b3ae59 [15:08] and i *think* we need a transitional pkg like we did for omap [15:08] yup [15:09] the right conflicts/replaces/breaks and a transitional package shoudl do [15:21] ogra_: ok, i'll hassle rtg tomorrow about it [15:21] :) === stgraber_ is now known as stgraber [15:46] [16:01] ogra_: No need for conflicts, breaks, or replaces, it's just a straight up depends. [16:02] ah, k [16:02] you dont want the old package to be removed ? [16:02] even though its likely impossible to roll back [16:03] When the "old package" happens to be your running kernel, forcing it off the system would be a silly idea. [16:03] ah, indeed [16:15] * ppisati -> gym [16:15] later === danjared_ is now known as danjared [16:32] psivaa: precise/omap4 seems to also be stuck in regression-testing. [16:34] infinity: it's progressing now, started a little late on that and i should be able to finish it this eve. roughly in about 4 hrs [16:34] psivaa: Cool, thanks. [21:00] infinity: just completed the regression testing task for linux-ti-omap4: 3.2.0-1429.38 (1160194) [21:01] psivaa: Danke.