15:01:20 #startmeeting F12 Alpha Blocker Bug Meeting 15:01:28 does anyone else want to run the show this week, or should I do it again? 15:01:29 #chair adamw 15:01:59 adamw: no objections from me, you kept things moving at a good pace last time 15:02:12 i was probably hungry 15:02:24 a good motivator! 15:02:27 =) 15:02:30 who's here? 15:02:38 * jlaska 15:02:41 this week i'm melting, we're in the middle of a ridiculous heatwave here 15:02:52 I'm here. 15:02:58 adamw: much cooler today 15:03:30 yeah, thank heavens 15:04:27 well, let's kick off with the coven then 15:04:35 others will chime in, if past history is any indicator :) 15:04:54 the f12alpha list seems to get longer at each meeting, which really isn't how this is supposed to be working! 15:05:10 first up: 15:05:14 #link https://bugzilla.redhat.com/show_bug.cgi?id=486284 15:05:15 Bug 486284: medium, low, ---, pjones, MODIFIED, [PATCH] RFE: ext4 support in grub 15:05:26 this is the ext4 bug, again 15:05:43 we still haven't marked it as closed. i know i haven't because i can't get far enough in anaconda to confirm whether it's actually working. same for you guys? 15:05:52 same here 15:06:12 * poelcat finally makes it to a blocker meeting 15:06:24 f13: adamw ditto! 15:06:31 heya poelcat, welcome :) 15:06:39 hope you stayed cool 15:06:54 my entire family has been camping out with me in my tiny office, which is the only room with an AC unit 15:07:13 hehe 15:07:41 so...for 486284, i'd say all we can really do is leave it as-is, and remember to check it once we can get to the frickin' partitioning stage 15:08:09 anything else on it? 15:08:18 I'd love to make a general policy that all blocker bugs must be verified ... but I realize there's much more to it than that 15:08:38 i think that's what we're shooting for - i don't think we've closed any without confirmation yet 15:09:25 #agreed no action possible on 486284 due to broken anaconda: will confirm fix when anaconda is less screwed 15:09:48 alright, next up (jlaska and I are playing with the meetbot stuff...): 15:09:54 #topic https://bugzilla.redhat.com/show_bug.cgi?id=497650 15:09:56 Bug 497650: high, low, ---, kernel-maint, NEW, mutexes are broken on PAE kernels? 15:10:09 hello 15:10:09 this is a new one since last meeting 15:10:12 hi dpravec 15:11:13 adamw: I think dpravec has some nice ascii art to share for the meeting! 15:11:48 go for it then :) 15:12:02 :) 15:12:04 is this the one that was fixed in a recent libc? 15:12:11 ok i wanted to do one change, but here it is.. 15:12:22 make sure you have monospace font 15:12:31 .--. 15:12:32 | |--.__ ,---. 15:12:34 | |--' \_ \ W / 15:12:35 | | \_ .*** . >:0OOO ' ' '. 15:12:37 | | \_*****. / M \ . 15:12:38 | | \***/ . `---' ,---. . . 15:12:40 | | >8#### . \ M / . ' ' . . ' '. 15:12:41 | | _/***\ . >:0OOO . ' 15:12:43 | | / ***** . / W \ 15:12:45 | | _ / .*** . `---' 15:12:47 | |--.__/ 15:12:51 | |--' 15:12:53 '--' /#MB@\ 15:12:55 /#MW$@E#\ 15:12:57 #@M$$#@m$#W 15:12:59 ***************************************************************************** 15:13:01 * If there's something weird * 15:13:05 * and it don't look good * 15:13:07 * Who Ya Gonna Call? * 15:13:09 * BUGZAPPERS! * 15:13:11 ***************************************************************************** 15:13:36 vote of thanks to dpravec =) 15:13:42 notting: no, that was a different one iirc 15:13:43 :) 15:13:50 notting: this was reported in april, note 15:13:59 just hasn't come up before as lennart only just added it to blockers 15:14:00 I think we have a new logo idea to run by the design folks ;) 15:14:14 i'm going to ping lennart and see if he can come in to discuss this bug... 15:14:36 adamw: is this specific to via? 15:14:39 * jlaska reading 15:14:43 *blink* 15:14:54 jlaska: i'm not sure, what i'm wondering is if it's specific to *sound* 15:15:28 if it's specific to sound i'm not sure we can consider it an alpha blocker if we're sticking strictly to our policy, as sound ain't in the critical path... 15:15:36 adamw: not sound-specific, but pa is probably the only thing using PI mutexes in a desktop install.... 15:16:34 maybe this is the same as the one notting's thinking of 15:17:25 yeah, it does kinda sound like it actually, my bad notting 15:18:05 there was one bug about PI mutexes being broken on PAE kernels and one about x86_64 15:18:19 the x86_64 one was allegedly fixed by a recent glibc update 15:18:30 ah 15:18:33 this is the other one? 15:18:44 yes 15:19:24 the fixed one is mentioned in the second-to-last comment 15:19:48 "fix memory ordering in pthread_mutex_unlock (BZ#10418)" 15:19:54 that's http://sources.redhat.com/bugzilla/show_bug.cgi?id=10418 15:19:55 Bug 10418: medium, medium, ---, johnsonm, CLOSED CURRENTRELEASE, SCSI driver modual aic7xxx crashes on boot floppy 15:20:00 which was re-opened by the reporter 15:20:05 so we know for sure that it's still busted on PAE kernels as of the rawhide-20090731 tree? 15:20:11 heh, buggbot isn't smart enough to understand sources.redhat.com :) 15:20:41 lennart isn't replying, btw 15:20:56 f13: I don't think we know that for sure 15:21:12 ok, we need to get a conformation on that. 15:21:51 ok, so we ask for test with latest rawhide 15:22:16 second question: is this an alpha blocker? 15:22:42 oh, just got a reply from lennart 15:22:45 here he is :) 15:22:46 adamw: it has the nasty side effect of locking up all your desktop if you have sound enabled... 15:22:56 mclasen: well that's what I was wondering about 15:23:05 mclasen: the version discussed on -devel-list seemed to have that impact 15:23:09 adamw: hmm, i wonder if 497650 also got fixed by that recent glibc update 15:23:30 mclasen: but none of the reporters in _this_ bug say it happens to them. they just say 'no sound'. you'd think they'd mention if their desktop had locked up. 15:23:38 mezcalero: yes, we wondered that too 15:23:45 * mclasen goes to debug why shutdown doesn't work on his system - I might drop off occasionally 15:23:49 mezcalero: so we've agreed to stick it in needinfo and ask reporters to check 15:23:54 adamw: yeah, incomplete information all around 15:24:11 mezcalero: the second question is whether it qualifies as an alpha blocker, so we're trying to pin down the impact 15:24:18 adamw: that bug is actually against f11 15:24:27 adamw: so i am not even sure if this exists on rawhide still 15:24:29 I'd say yes if it's still happening in rawhide it's a blocker 15:24:40 if the desktop locks up, you can't get to PackageKit to update out of the mess 15:24:49 but we need to know it happens on rawhide. 15:24:54 f13: as i said, we're not actually clear if this version of the bug causes lockups 15:24:59 f13: no-one in the report says it does, afaict 15:25:11 it causes asserts to be hit 15:25:16 should be pretty simple to test. 15:25:16 mezcalero: do you know if this particular issue was causing lockups? or just no sound? 15:25:24 f13: well it seems to be hardware-dependent 15:25:28 oh 15:25:48 i mean, i've been running rawhide this whole time, and i haven't seen that problem. 15:26:06 the bug is not exactly new and marked for f11, and in contrast to that other mutex issue it's relatively quiet 15:26:12 so it might be a good idea to just ignore it 15:26:39 yeah, throw it in needinfo, move on 15:26:47 yeah, my intuition is that this one isn't hitting a lot of people and doesn't seem to cause catastrophic consequences, so it's probably not an alpha blocker 15:27:01 i agree 15:27:14 ok, so - agreed to throw it in needinfo and ask for status with latest rawhide kernel and glibc, take it out of f12alpha? 15:27:26 seems fair given what we know 15:27:59 #agreed request status on 497650 in latest rawhide, drop it from f12alpha due to restricted breadth and severity of impact 15:28:22 oops, already made the change to th bug now 15:29:08 mezcalero: worked out fine :) 15:29:21 mezcalero: thanks for helping 15:29:24 ok, next up: 15:29:40 #topic https://bugzilla.redhat.com/show_bug.cgi?id=510033 15:29:41 Bug 510033: urgent, high, ---, dlehman, MODIFIED, AttributeError: 'NoneType' object has no attribute 'name' 15:29:52 I believe this, and a few other MODIFIED anaconda bugs are waiting on resolution of the NM issues 15:29:53 this is our old friend the breaks-all-installs bug from last week 15:30:13 and again we can't confirm the fix as we can't get that far in anaconda now 15:30:46 although we got one report that it was fixed from a lucky maxamillion who managed to hit the installable-tree jackpot :) 15:31:01 * f13 wonders about last night's tree 15:31:06 same as previous 15:31:36 jeremy and clumens are still working through changes introduced by the latest NM update 15:31:55 ah ok. 15:31:57 right, i think they fixed the first bug we were all seeing, but there's others behind it that have much the same impact, right? 15:32:12 that's right 15:32:31 so i guess we have the same agreement as for the first bug we discussed 15:32:37 Once they have a better handle on things, I believe a bz will be headed our way for tracking ... but right now they are knee deep in anaconda 15:32:55 #agreed no action possible on 510033 due to broken anaconda: will confirm fix when anaconda is less screwed 15:33:14 I'm just glad we're knee deep in it now, and not this time next week 15:33:36 yeah, although it's rather affecting our ability to test the later stages of anaconda, which worries me 15:33:44 not much to be done about it, though 15:33:47 hopefully we're not still knee deep (or neck deep) next week 15:34:08 ok, next up: 15:34:11 #topic https://bugzilla.redhat.com/show_bug.cgi?id=511907 15:34:12 Bug 511907: high, medium, ---, jgranado, MODIFIED, parted.Disk.duplicate exception in storage scan 15:34:31 this is another new addition 15:34:31 adamw: I believe this is falls into the same boat 15:34:45 nominated by jlaska 15:35:22 it was another ... breaks all installs bug which required a new parted iirc 15:35:31 yeah, i see from the discussion 15:35:39 so another 'it's probably fixed but we can't test it' 15:36:04 right, another data point for full disclosure from the bz ... 15:36:04 "This seems to happen during scanning of systems whose partitions were created 15:36:08 using the F11 installer." -dlehman 15:36:20 which obviously makes it quite important :) 15:36:21 I don't think this changes the Alpha blocker status, but did want to mention 15:36:39 no i don't think it does, i think it's pretty reasonable to expect people installing alpha to have an f11 system installed 15:37:04 i mean, to expect that they would have an f11 system, and so we should handle that case 15:37:12 yeah, agreed 15:38:31 #agreed no action possible on 511907 due to broken anaconda: will confirm fix when anaconda is less screwed 15:38:49 next up: 15:38:53 #topic https://bugzilla.redhat.com/show_bug.cgi?id=513629 15:38:54 Bug 513629: medium, low, ---, schwab, CLOSED RAWHIDE, every GTK+ app get stuck 15:39:06 oh, heh 15:39:13 this has been closed since the time i loaded the f12alpha page :) 15:39:59 looks like another incarnation of the mutex issue. ok. moving on 15:40:10 #topic https://bugzilla.redhat.com/show_bug.cgi?id=513879 15:40:11 Bug 513879: medium, low, ---, katzj, MODIFIED, Rawhide anaconda not using correct gtk theme 15:40:25 this got bumped to MODIFIED earlier this morning 15:40:32 adamw: no word yet from lindel 15:40:41 i find it very hard to consider this a blocker 15:40:50 I nominated this bug for a few reasons ... 15:40:55 unless the impact of 'not using correct gtk theme' is something other than 'installer looks a bit odd' 15:41:21 I think severity of this issue is low, but priority was considered higher 15:41:51 testing showed differing results on i386 and x86_64 ... so I wanted to get a better understanding of the root cause 15:42:07 in the way i define those fields? yes, i'd agree with that, it's heavily unprofessional and should block final release...but it's hard to see it blocking alpha imho 15:42:22 I agree ... and felt compelled otherwise 15:42:32 heh 15:42:32 since Alpha is really the Beta 15:42:47 no, it's really the alpha. Beta was really Alpha, remember? :D 15:42:55 uh huh ;) 15:42:59 excuse me while i screw my head back on ;) 15:43:14 so, SHOULD this bug block the Balpha release? :) 15:43:48 given that the fix is in buildinstall ... I think it's worth considering 15:44:09 well, if no-one else has strong feelings i'm happy to defer to jlaska's intuition... 15:44:15 but I agree, the impact to alpha testers is low 15:45:06 * adamw watches mclasen on his bungee 15:45:48 well, let's leave it on the list, it doesn't hurt much. i assume it's in the same state as all the other anaconda bugs - you can't test at present even if the fix gets built? 15:46:09 yeah, this will need a new build of anaconda 15:46:18 which is likely to come anyhow for the remaining NM integration 15:46:26 i mean, you can't test until we have an anaconda that gets into X 15:46:36 yeah, that or VNC 15:47:00 do you need an action item to confirm with jeremy that the potential fix makes it into the next anaconda build? 15:47:11 or will that be certain to happen since it's committed? 15:47:19 the later 15:47:19 (not sure how anaconda have their cms set up) 15:47:22 k 15:47:28 let's just slap the generic agreed on it then 15:47:35 hehe, worksforme ;) 15:47:47 dcantrel, more likely, no ? 15:47:50 #agreed no action possible on 513879 due to broken anaconda: will confirm fix when anaconda is less screwed 15:48:05 i just said jeremy as it's his fix, according to the comments 15:48:14 mclasen: yeah, dcantrell is in the loop on the NM stuff ... but I think jeremy had the fix for this other issue 15:48:37 next up! 15:48:42 #topic https://bugzilla.redhat.com/show_bug.cgi?id=514084 15:48:43 Bug 514084: medium, low, ---, jmccann, NEW, Needs to dump state database before doing callouts/sending dbus signals. 15:49:00 not an alpha blocker, imo 15:49:13 causes sound not to work reliably in multi-user scenarios 15:49:18 but doesn't lock up your desktop 15:49:51 yeah, that's my reading too 15:50:14 unless 'acls won't be setup correctly' might mean even the first user can't get sound...but then even that's not strictly an alpha blocker 15:50:51 as lennart mentions it also breaks hotplug, which is a bit of a bummer for people with laptops and USB sound adapters. and I FOR ONE EXPECT THAT TO WORK PERFECTLY IN ALPHA RELEASES! ok, not really. :) 15:51:28 heh 15:51:44 everyone agree? 15:51:52 No objections 15:51:56 do we make it an f12blocker? 15:52:04 did this work in F11? 15:52:24 hotplug works in f11, yeah. multi-user, not entirely sure 15:52:44 none here. 15:52:47 does that mean it falls into the category of a feature 15:52:54 somehow I think we're going to have to come up with some polish definition to define F12Blocker s 15:53:00 things that don't technically prevent you from doing stuff, but just reflect poorly upon the distro 15:53:25 no, multi-user has been an intended feature of pa for a while, i just don't know whether it worked properly in f11 (i've never tested). this is really just a bug in consolekit, which happens to break some features of PA. 15:53:28 adamw: feel free to move it to F12blocker 15:53:36 we're going to have it fixed for F12 anyway 15:53:44 gotcha 15:53:56 k. i'd feel happier with it on that list, just so we don't forget about it if somehow it doesn't get fixed... 15:54:09 a patch is in the works 15:54:27 #agreed drop 514084 to f12blocker, its impact does not fit the definition of an alpha blocker 15:55:31 done 15:55:50 next up: 15:55:52 #topic https://bugzilla.redhat.com/show_bug.cgi?id=514185 15:55:54 Bug 514185: medium, low, ---, katzj, MODIFIED, No module named system_config_keyboard.keyboard 15:56:35 I believe this is the issue Kamil noted last week that was blocking his virt install attempts 15:56:45 looks like another classic 'install is broken, fix is in, can't be tested' 15:56:52 yeah 15:57:56 anything else to add? 15:58:13 no, I think that captures it 15:58:24 #agreed no action possible on 514185 due to broken anaconda: will confirm fix when anaconda is less screwed 15:58:42 ok, next: 15:58:49 #topic https://bugzilla.redhat.com/show_bug.cgi?id=514501 15:58:51 Bug 514501: high, low, ---, clumens, MODIFIED, dbus.exceptions.DBusException: org.freedesktop.Hal.NoSuchDevice: No device with id /org/freedesktop/NetworkManager/Devices/0 15:59:08 adamw: this is what started the discovery around the new NM API 15:59:11 this is the initial NM bug, yep. 15:59:21 a fix is in for the reported problem ... but more lurks 15:59:32 we don't have a tracker bug yet for the 'lurk' 15:59:46 hmm, there's a fix in? the comments don't seem to indicate that 15:59:50 just chris sounding confused :) 16:00:52 adamw: well ... there is a fix for that specific traceback 16:01:10 but clumens and jeremy are still trying to get anaconda + NM to play nice again 16:01:11 oh, yeah, i see comment #4 is after the initial fix. ok 16:01:16 I think that's why harald got pulled in 16:01:35 i guess since we know the anaconda guys are active on this, we can leave the bug in the state it's in for now, though it is a bit messy. 16:02:29 yeah, that's a good point, we can move this back to ASSIGNED if they are okay with it 16:02:30 anything else to add on that? 16:02:34 until the remaining fixes are in 16:02:52 yeah, either we do that and treat it as one big bug ('port anaconda to new networkmanager interface') 16:03:03 or we close that report and file new bugs for each bump in the road 16:03:43 I was going by the later, but I'll float the former by them 16:03:52 whichever they find easiest 16:03:59 ok, we can leave that to you? 16:04:08 sure I'll take that action 16:04:11 checkin with those guys 16:04:29 #agreed no significant action on 514501 as it is under active development, but clean up bug report 16:05:07 #action jlaska work with anaconda team to decide an approach to clean up 514501 16:05:22 ok, two to go! 16:05:26 #topic https://bugzilla.redhat.com/show_bug.cgi?id=514610 16:05:27 Bug 514610: high, low, ---, kernel-maint, NEW, 0.107(x86_64) won't boot: BUG at drivers/gpu/drm/drm_gem.c:451! invalid opcode: 0000 [#1] SMP 16:05:37 another newbie 16:06:12 jlaska, did you try the kernel tom mentions in his latest comment? 16:06:45 hrmm, I did not ... I was trying this on the install path 16:06:54 I could do F-11 and upgrade to rawhide I guess right? 16:07:14 I haven't done that yet for this release, so not sure what pot holes await 16:07:20 yeah, that would be the way to get onto rawhide atm 16:07:31 I did not see the exact same error message in 107, but it wouldn't boot with some intel drm oops, and 112 works fine here 16:07:34 the major one is the rpm content compression change 16:07:48 krh said the issue was caused by the vga arbiter patches in the kernel 16:07:54 mclasen: getting better, that's good 16:07:55 update rpm from f11 updates-testing before trying to update to rawhide 16:07:58 aside from that, it should work 16:08:03 adamw: oh right, thx 16:08:46 ok, so for this one, jlaska to test with kernel 112 and confirm fixed or not 16:08:50 so this is still in NEW ... but Tom reports it's resolved with a newer kernel 16:08:59 sounds goodly! 16:09:31 #agreed jlaska to test whether 514610 is fixed for him 16:09:43 #action jlaska test whether 514610 is fixed 16:09:54 i've set the bug to needinfo jlaska 16:09:58 ok, final entry! 16:10:05 adamw: thx 16:10:05 #topic https://bugzilla.redhat.com/show_bug.cgi?id=514718 16:10:07 Bug 514718: high, low, ---, anaconda-maint-list, MODIFIED, install exited abnormally - AttributeError: 'NoneType' object has no attribute 'has_key' 16:11:00 looks like yet another untestable installer fix 16:11:06 * f13 steps out 16:11:08 or is this one early enough to test? 16:11:13 thanks for coming f13! 16:11:44 adamw: I don't know yet on this 16:11:59 "This is actually fixed in git, we just didn't do a new build as networking was 16:12:02 still falling over like it was yesterday " 16:12:12 awaiting a new build it seems 16:12:14 sounds like it comes after the other networking failures : 16:12:31 well, let's put it down as 'to be tested when we get a new anaconda build'...sound ok? 16:12:58 yeah, this was found while testing the fix for one of the first networking fixes 16:13:02 * jlaska gets lost 16:13:14 hehe 16:13:22 they're all so similar that I don't want to start marking them off until we get further into an install 16:13:29 #agreed 514718 to be tested when a new anaconda build is available 16:13:34 rock on 16:13:46 yeah, i suspect we'll get a 'breakthrough build' of anaconda which fixes all the network issues and then we can close the reports up 16:13:51 s/suspect/hope wildly/ 16:13:51 :) 16:14:07 let's see what's behind door#1 16:14:11 aaah bugs! 16:14:28 alright, let's try door#2 16:14:32 aaaaagh! more bugs, with teeth! 16:15:01 do you want to go through f12blocker too, or should we wrap things up as it's taken a while already? 16:15:08 * nirik wonders if bug 505071 / bug 514010 are worth mentioning? not that many people care about ppc32. ;) 16:15:11 Bug https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=505071 medium, low, ---, kernel-maint, NEW, ppc failed to boot 16:15:11 Bug https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=514010 medium, low, ---, kernel-maint, NEW, 2.6.31 pre rawhide kernels don't boot on ppc32. 16:15:22 always worth mentioning! 16:15:30 ppc is still officially a primary arch 16:15:42 so something which stops all ppc32 machines booting sounds fairly critical to me 16:16:01 yeah. It does here... not sure what else to try, but happy to test. 16:17:01 reading the report there seems to be some differentiation being drawn between AmigaOne and iMac hardware 16:17:22 * jlaska trying to think of ppc experts on the kernel/devel side of things ... perhaps jwb 16:17:27 if the case was that AmigaOnes don't work but iMacs do i might consider that not critical...old Macs are probably the biggest/most important PPC hardware base 16:17:28 yeah, amigaone was some ppc amiga thingie. 16:17:35 ewwww 16:17:41 isn't it the new one? 16:17:49 well, newer I guess. ;) 16:18:01 oh, no, just a bit newer. :) 16:18:05 * adamw wikipedes 16:18:27 nirik: what's the hardware on which you're seeing trouble? 16:18:52 PowerMac G4 Silver (766mhz) 16:19:00 sorry, 466mhz. ;) 16:20:00 ok. looking at 505071 in isolation, by my reading, the problem seems to be that a kernel option has been set to make AmigaOne systems work, which appears to have broken iMac systems 16:20:31 AmigaOne needs CONFIG_NOT_COHERENT_CACHE to be set, Mac hardware needs it to be not set 16:20:36 yeah, I tried disabling that option here, but it fails the same way. But that might be the related cache option that got tied to amigaone 16:21:03 well, if it's "pick one or the other", we pick imac 16:21:26 yeah, that's my feeling too 16:21:33 well, just disabling amigaone didn't get it working for me. 16:21:50 I could have failed to rebuild the kernel right tho I guess. 16:21:51 there is some discussion in the bug about fixing the kernel code to allow both with branching code, depending on what hardware's detected 16:22:06 nirik: i'm working on the assumption yours is a different problem, since you don't get the same error messages 16:22:20 adamw: yeah, it could be. I filed that other bug on it. 16:22:34 you didn't see the message "kernel coherency:off != device tree_coherency:on" , right? 16:23:07 my tentative proposal would be that we set it as a blocker and request that, for the alpha, we go with the kernel settings that make Mac hardware work 16:23:23 and then for final release, we try and help Martin get the code fix to allow both sets of hardware to work done and upstreamed 16:23:59 what do you guys think? 16:24:23 I don't think I saw that. I typed in what I saw in that comment. ;) The screen was a bit jumbled tho 16:27:55 * jlaska hits and starts the rawhide update 16:31:17 ok, discussed in -kernel, my approach was generally agreed upon 16:31:27 adamw: I've got to step away for a bit ... I'll be back shortly 16:31:32 #agreed set 505071 to block alpha release, for alpha amigaone support should be disabled 16:31:37 jlaska: we can wrap things up here i think 16:31:45 adamw: okay 16:31:52 thanks everyone who contributed! 16:32:05 thanks for running the show adamw :) 16:32:22 #endmeeting