/srv/irclogs.ubuntu.com/2011/08/23/#ubuntu-kernel.txt

=== BenC__ is now known as BenC
=== klebers[afk] is now known as klebers
=== klebers is now known as klebers[afk]
=== klebers[afk] is now known as klebers[AFK]
=== ivoks_ is now known as ivoks
=== kloeri_ is now known as kloeri
=== smb` is now known as smb
* smb tries to wake up07:35
RAOFThis sounds like a job for bees!07:40
smbNah, more like very strong coffee...07:40
RAOFInfused with bees?07:41
RAOFEverything is better with bees ☺07:41
* smb wonders about the bee thing that seems to go on....07:41
smbB B B07:42
smbStill not better...07:42
RAOFB-)07:43
RAOFSee?  Much better!07:43
smbHeh :)07:43
smbSeems one way to tell that apw is awake is to watch for new tracker mails pouring in ... ;)07:54
smbapw, morning07:54
apwsmb, heh tracker emails ?07:59
smbapw, modifying the cve-bzr-tracker that I am still subscribed to 07:59
apwsmb, heh yeah thats first thing normally cause it kills my machine stone dead ... as it pays the 'its 24 hours so acces timess needs to change' pain on alll my repos08:00
apwi then make some tea in parallel while it runs :)08:00
smbHeh, yeah sounds like a good way to not to actually pay that time in person... :)08:01
apwliterally takes 10 minutes of heavy crunching the first time, disk burried08:02
* apw notes that this one shows a huge heap of updates on arm like we haven't rebased arm in agaes, i guess with all the regressions on master thats not unexpected08:02
smbIf its Lucid, then sort of. At least ec2 skipped 3 or 4 versions. But only the first had a lot of changes. Still could have been one or two stable ubdates08:04
apwyeah most likley that much behind.08:07
apwsmb, though 78 CVSs moved from pending to pending (x), so something was very very behind08:08
apwkees will be pleased indeed to see them all go out08:08
* smb shouts08:14
apwsmb are you "there" or is pulse bust at my end08:14
smbyes :-P08:15
* apw cries08:15
smbNot much more joy08:15
apwsmb, can you see my lips even ?08:16
smbapw, were making fun of you08:16
smbwell we hear you08:16
ckingwe can here you08:16
smbbut you cannot hear us08:16
smbwe won't repeat that08:16
ckingsmb, you can see how many writes you did to the fs using: cat /sys/fs/ext4/sda1/lifetime_write_kbytes08:19
smbcking, Hm, I guess that would require it to pass the recovery phase which it fails because any writes seem to fail...08:30
* ppisati -> lunch11:33
zulcan we get CONFIG_BLKIO_CGROUP turned on for omap4, i already opened bug #831954 about it12:35
ubot2Launchpad bug 831954 in linux "CGROUPS_BLKIO need to be enabled on omap4" [Undecided,Incomplete] https://launchpad.net/bugs/83195412:35
apwzul on which release12:37
zuloneiric12:37
tgardnerapw, on it12:37
apwtgardner, seems you read even faster than i12:37
smbzul, Another thing for lxc?12:38
apwtgardner, we might want to see that *_CGROUP_* are the same in master and there12:38
apwtgardner, as i suspect there will be something else missing tommorrow12:38
zulsmb: yeah wheee12:38
apwzul i assume lxc works on i386/amd64 ?12:38
smbapw, Sounds quite reasonable... 12:38
bjf##12:38
bjf## Ubuntu Kernel Team Meeting - Today @ 17:00 UTC - #ubuntu-meeting12:38
bjf##      agenda: https://wiki.ubuntu.com/KernelTeam/Meeting12:38
bjf##12:38
zulapw: pandaboard12:38
=== klebers[AFK] is now known as klebers[afk]
apwzul, i know you need ti-omap4, i mean if the config is right in x86 then its worth syncing all cgroup things over at the same time12:39
smbapw, And the other answer should be yes12:39
tgardnerapw, funny, I was just thinking the same thing.12:39
zulapw: agreed12:39
apwsmb, so lxc does work yes ?12:39
smbapw, Yes, and I was thinking that it seemed to work with the last other cgroup change. But seems not all of it12:40
=== klebers[afk] is now known as klebers
* bjf -> morning walk12:56
=== bjf is now known as bjf[afk]
=== med_out is now known as medberry
=== shadeslayer_ is now known as shadeslayer
tgardnerogasawara, pushed rebase to v3.1-rc3 plus makefile patches to ubuntu-p13:44
ogasawaratgardner: sweet, was just about to start that13:44
tgardnerogasawara, now you can take an extra 20 minutes :)13:45
ogasawaraheh13:45
* apw isn't sure she can take a full 40 minutes of that13:46
* ogasawara cringes at the thought13:46
=== bjf[afk] is now known as bjf
* bjf -> back13:54
ppisatihave we ever had arm vexpress kernel built?13:57
ppisatiit seems the server/arm people need it13:57
ppisatiand they are asking to had this arm flavour13:57
tgardnerppisati, linaro agreed to supply that flavour13:58
ppisatitgardner: we supply or they supply?13:59
tgardnerppisati, tey, as in linaro13:59
tgardnerthey*13:59
ppisatiuhm13:59
* ogasawara back in 2013:59
ppisatitgardner: it seems they need it for the cloud images (that they will run on qemu in this case)14:00
ppisatianyway, i'll reroute them to linaro14:00
tgardnerbjf, are you deaf ?14:06
bjfmust be, can you hear me?14:07
smbyes14:07
bjfhuh14:07
smbkillall pulseaudio14:07
Davieyppisati: Thanks for the heads up14:12
ppisatiDaviey: ah, you are here :)14:12
Davieytgardner: Did linaro agree to provide it "in ubuntu archive" ?14:13
tgardnerDaviey, IIRC they were going to do that. contact John Rigby.14:15
Davieytgardner: ta14:16
apwDaviey, it is rather late in the day to be noticing you don't have a kernel isn't it?14:16
tgardnerDaviey, of course there might be a maintenance issue. Linaro typically abandons a release as soon as its out.14:17
amitktgardner: Daviey: John (for the config) and Ricardo Salveti (for uploads to archive if necessary)14:17
Davieyapw: who uses THAT?14:17
apwthat ?14:17
Davieya kernel.. we just ride along the userspace.14:17
smband we likely have no need for a vexpress thing... :-P14:18
* smb at least has no hw for that14:18
DavieyThe reason for wanting it, is that we were advised that the vexpress platform is better to virtualise than the ones we were using.14:18
amitksmb: very few people in the world do and most are concentrated around a place called cambridge, UK14:19
Davieythis is really purely for using full virt.14:19
Davieysmb: you do have hardware for that!  it's called qemu-system-arm :)14:19
tgardnerDaviey,  actually, I thought it was the only armel flavour that worked with qemu14:19
amitktgardner: there is now basic omap support too14:19
DavieyOther people know much more about this crap than i do :)14:20
apwi thought we dropped versatile cause qemu did omap support now14:20
tgardneramitk, Daviey: well we _do_ have omap2 support in oneiric14:20
tgardneror omap3 rather14:20
smbbjf, If that was intended to be speed it did not work14:22
ppisatiyep, qemu should grok the omap3 kernel nicely now14:22
smberr speech14:22
DavieyNCommander / ogra_ / lool: Should we use omap3 as our favoured qemu-system-arm platform?14:31
ogra_Daviey, well, that limits you to 512M14:32
apwso what have you been testing with for the last 4 months14:33
ogra_and makes your image creation more complex 14:33
Daviey   mm   mmmmm    mmm  m    m14:34
Daviey   ##   #   "# m"   " #    #14:34
Daviey  #  #  #mmmm" #   mm #mmmm#14:34
Daviey  #mm#  #   "m #    # #    #14:34
Daviey #    # #    "  "mmm" #    #14:34
ogra_mmm ?14:34
* apw wonders why your R is lifting its foot14:34
loolDaviey: There are pros and cons; vexpress has some network issues, but otherwise mostly works; OMAP supports emulation of a bunch of different boards and will actually emulate the OMAP3 boot ROM (vexpress just starts into an ELF binary from the cmdline); vexpress goes up to 1 GiB, beaglexm up to 512 MiB; network, keyboard and mouse emulation on OMAP is tricky, but you get flash15:02
Davieylool: tgardner seemed to think that linaro was working on a vexpress kernel, is this intended to be in the archive, do you know?15:03
loolDaviey: it's in the archive15:04
loolDaviey: as well as our omap kernel15:04
loolDaviey: the d-i flavor for vexpress is temporarily disabled as it lacks a couple of modules in the config which are now required15:05
* Daviey is now really quite confused.15:05
loolDaviey: rmadison linux-linaro-vexpress15:05
loolDaviey: rmadison linux-linaro-omap15:05
DavieyGah.15:05
DavieyThere seems to be pockets of unknowns floating around.15:06
loolDaviey: Note that there are 3 OMAP kernels in Ubuntu ATM; linux-ti-omap4 for OMAP4 only (not supported in qemu), linux builds an OMAP3-only flavor (should work in QEMU), and linux-linaro-omap builds and OMAP2+ flavor which works on OMAP3 + OMAP4 and in QEMU15:06
Davieylool: So you think linux-linaro-vexpress is our best bet, for most memory support?15:07
loolDaviey: that seems right, but the network issue is a big deal for a cloud usage15:07
loolDaviey: for a local run, it's fine though; but you can't test network install for the same reason15:07
tgardnerDaviey, as is the maintenance window15:08
loolone thing I didn't try is setting up the network over serial; that's really ugly, but it would be reliable and slow15:08
* Daviey sobs.15:08
jibelapw, bug 827197 is not fixed on today's desktop images.15:14
ubot2Launchpad bug 827197 in linux "/usr/bin/ecryptfs-setup-private fails with exit status 1 during install - User's home directory not encrypted" [High,In progress] https://launchpad.net/bugs/82719715:14
apwjibel, what kernel is on the CD you have15:15
ogra_lool, but the beaglexm vm needs a proper vfat, vexpress can just run with vmlinuz and .img file ... its quite some extra effort to add partitons etc15:15
jibelapw, 3.0.0-9.1315:16
jibelapw, it was fixed on .12 according to changelog15:17
apw jibel looking at it now15:17
jibelapw, thanks15:17
loologra_: I don't think it's an obvious disadvantage; for instance with vexpress you have to keep the kernel up-to-date inside the installed system and on the host, while with OMAP you just apt-get update15:18
looljust... different, both with advantages and disadvantages depending on the load15:18
ogra_sure, omap is also better supported, but you still have to do more implementation work in the beginning15:18
cr3hi folks, do some drivers take different code paths for some pci devices depending on their revision number? ie, is the revision important to know when troubleshooting a pci device?15:25
apwogasawara, about?  seems something odd has happend to the ecryptfs fix, it seems to have undone itself, even though you had to disable module checks... 15:25
ogasawaraapw: eh?15:25
apwcr3, yep, some quirks only apply to specific revisions or chip revs or any number of other things15:25
apwogasawara, seems although all the pre-requisite changes are in the commit, as of now ecryptfs itself is still =m in some cases.15:26
apwogasawara, i cannot fathom how that is possible15:26
kirklandapw: why's ecryptfs so much trouble this cycle?15:26
apwkirkland, some idiot added TPM based key support to it :/15:27
cr3apw: awesome, thanks for the confirmation! And, just to be extra sure, like wearing a belt and suspenders, the class of a device is not likely to change between revision numbers of a pci device, right?15:27
tgardnercr3, you mean like changing from a network device to a disk device ?15:27
apwcr3, i'd have not thought it was possible no15:27
apwogasawara, shall i push the fix to that onto master-next15:27
ogasawaraapw: yes please15:28
ogasawaraapw: will prep the upload asap15:28
* ogasawara tries to figure out what happened15:28
apwogasawara, done ... its utterly dumbfounding15:28
=== klebers is now known as klebers[afk]
kirklandapw: okay, how's that complicate things for us?15:28
cr3tgardner: maybe not as drastic as that but perhaps change from "RAM memory" to "FLASH memory", assuming a Memory controller15:29
apwkirkland, it added a huge dependancy pile which has to be =y to allow ecryptfs to be =y and cause we don't autoload it it has to be =y15:29
tgardnercr3, perhaps in the prototype or development phse, but certanily not after a production release.15:29
tgardnerphase*15:29
apwand it silently moved itself =m to fix the depenancy issues and we didn't notice15:29
apwogasawara, perhaps i should add erypt_fs =y to the enforcer15:30
cr3tgardner: that's a reasonable use case, thanks, exactly what I needed to know!15:30
ogasawaraapw: yep, that would be a good idea15:30
kirklandapw: ugh, that's nasty;  tyhicks will be on board next Monday;  perhaps he can help out15:30
tgardnerogasawara, don't commit the makefile changes on master just yet.15:30
ogasawaratgardner: ack15:30
ogasawaratgardner: I'm just gonna do the ecryptfs bits and upload15:31
tgardnerogasawara, good thinking15:31
apwogasawara, ok pushed the enforcer bits too, feel free to rebase them after the upload if thats easier15:32
ogasawaraapw: ack, thanks.15:32
apwogasawara, did we get the seccomp bits in time ?15:33
ogasawaraapw: I could actually sneak them in now, kees sent a pull request this morning15:33
tgardnerapw, I'm looking at them right now15:33
tgardnerogasawara, I'd say wait on seccomp. ecryptfs is more important15:34
ogasawaratgardner, apw: am thinking we should just wait till post beta-1 though15:34
apwogasawara, it feels like a first upload after beta-1 material to me, gives a few weeks to rip it out if its pants15:38
ogasawaraapw: agreed15:38
tgardnerapw, hey mr perl wizard. when you have time look at ubuntu-p.git. 'Missing argument in printf at debian/scripts/abi-check line 174.' implies that the ABI checker needs some love.15:42
apwtgardner, sure, been seeing that for some time i think15:42
tgardnerapw, ack15:43
apwtgardner, added to my todo15:43
sconklinhey apw, can you confirm that https://bugs.launchpad.net/ubuntu/+source/linux/+bug/805494 is fixed? I'll do the bug updates if you say it's good for Lucid and Maverick15:57
ubot2Ubuntu bug 805494 in linux "ubuntu/rtl8192se driver breaks build when running 3.0 and above kernels" [Low,Fix committed]15:57
bjf##16:00
bjf## Kernel team meeting in one hour16:00
bjf##16:00
tgardnersconklin, I think its pretty obviously a fix since it wouldn't build otherwise. I'll check again with a quick test build16:03
tgardnersconklin, updated to  verification-done-lucid on bug #80549416:35
ubot2Launchpad bug 805494 in linux "ubuntu/rtl8192se driver breaks build when running 3.0 and above kernels" [Low,Fix committed] https://launchpad.net/bugs/80549416:35
=== klebers[afk] is now known as klebers
=== medberry is now known as med_out
apwsconklin, indeed that is a build test if it built its ok16:56
* tgardner --> lunch17:44
bjfogasawara, we're just going with the moin text that the bot pukes out?17:45
bjfogasawara, for the meeting notes17:45
ogasawarabjf: that's what I did last week, seemed easier17:45
bjfogasawara, not sure what that buys us may as well just point at the irc log17:46
ogasawarabjf: I added a blurb to the meeting README in kt-tools about it as an alternative to running the minutes through the irc2moin script17:46
ogasawarabjf: true17:46
hertontgardner, apw, sconklin: bug 805494 needs to be marked verified for maverick as well. I did build tests under 3.x kernel with maverick sources just in case, will mark verified for it as well.17:51
ubot2Launchpad bug 805494 in linux "ubuntu/rtl8192se driver breaks build when running 3.0 and above kernels" [Low,Fix committed] https://launchpad.net/bugs/80549417:51
tgardnerherton, ack18:18
hertonsforshee: we got positive feedback from testing bug 828550, can you send your patch to the mailing list to be reviewed/applied?18:19
ubot2Launchpad bug 828550 in linux "kernel BUG at /build/buildd/linux-2.6.32/drivers/gpu/drm/i915/i915_gem_evict.c:183!" [High,Incomplete] https://launchpad.net/bugs/82855018:19
herton*on bug18:21
sforsheeherton, ack18:27
=== bjf changed the topic of #ubuntu-kernel to: Home: https://wiki.ubuntu.com/Kernel/ || Oneiric Kernel Version: 3.0 || Ubuntu Kernel Team Meeting - August - 30 - 17:00 UTC || If you have a question just ask, and do wait around for an answer!
ckingyawn19:39
ogasawaratgardner: just fyi, i ran updateconfigs for the seccomp config changes and pushed over the top of master-next21:01
tgardnerogasawara, did it just resort ?21:01
ogasawaratgardner: yep, just moved em to ubuntu.common21:01
tgardnerah, thanks.21:02
jjohansenrebooting21:04
bdmurrayThis is indicative of something bad right?21:34
bdmurray[  423.586548] aufs au_xino_do_write:378:python[4582]: I/O Error, write failed (4294967268)21:34
sorenbdmurray: You're out of space.21:51
sorenbdmurray: 4294967268 == -2821:51
=== klebers is now known as klebers[AFK]
bdmurraysoren: not me but thanks!21:52
sorenENOSPC = 28.21:52
sorenbdmurray: It's guesswork, but it sounds plausible to me.21:52
bdmurraysoren: aufs                   3097704   2522256    415552  86% /21:53
sorenbdmurray: Hm.21:55
sorenbdmurray: pass21:55
sorenbdmurray: All I know is that 4294967268 is -28 written as an unsigned int.21:55
sorenbdmurray: and -28 is what you'd get if made a system call that failed due to not enough space.21:55
sorenbdmurray: ..and then I put two and two together.21:56
=== klebers[AFK] is now known as klebers
gswallow1Hiya, dumb question I think. First, I asked #ubuntu and got crickets.  Is there a better place between there and here?  And second, I have been trying to install 10.04.2/3 on an HP BL460g7, with some cursed ServerEngines CNA.  So far the only thing I can get to bring up the network on this card is Maverick.  I have read on teh Googlez that there is a way to choose a "Maverick backports" kernel or something and still install Lucid.  Do22:10
gswallow1oh, hey. https://answers.launchpad.net/ubuntu/+source/ubiquity/+question/16453622:14

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