/srv/irclogs.ubuntu.com/2010/12/10/#ubuntu-kernel.txt

=== smb` is now known as smb
* cking waves to manjo10:20
manjocking, hello10:20
ckinghow was the party?10:20
manjoit was a blast10:21
manjowe drove under the london bridge a couple of times so got to see the bridge from a diff angle10:21
ckingcool10:21
manjoalso some of the buildings were lit 10:21
ckinggood old london town10:22
manjosaw the london eye up close 10:22
manjolit bright blue10:22
manjokept my drink count to 4... so not feeling terrible this morning 10:22
manjoalthough I was super super hungry at the end coz the food was not all  that edible10:23
ckingwas it made of rubber?10:23
manjoheh some fish and duck10:23
manjoit was ok10:23
ckingnice10:23
ckingpity I was stuck at home fixing bugs10:24
manjocking, you cannot make us feel guilty... sorry :) 10:29
manjocking, but on the downside ... I think its going to be sandwiches again for lunch ... 10:35
ckingleast you get a free lunch. 10:36
ckingI didn't get one yesterday, sorting out S3 issues10:36
* apw wonders where the morning has gone already so soon10:47
manjoapw, cking sounds a little sad that he could not join us for the parteeee11:01
* manjo going down for a reboot after upgrade11:02
manjoapw, hmmm sandwiches again ... 12:15
manjoyummy12:15
apwnever, really?12:16
manjoapw, I am going to Tamarind (curry place) this evening with my cousin 12:16
apwnice you managed to overlap with him12:17
manjoapw, my cus says this place is good ... http://www.tamarindrestaurant.com/ but it looks too classy 12:18
nirvanaolder 2.6 kernels link the asm -> asm_<arch>12:27
nirvanaI dont see that happening with a new 2.6.31+ kernels12:27
nirvanawhy is that12:27
nirvanawhere I am supported to look for headers 12:27
nirvanaWhy is it not happening now???12:27
apwsmb, i have just dropped that update for the meeting into your inbox, also copied to kate for completeness12:32
* apw runs, smb call me on my mobile if there are issues12:46
=== emma is now known as em
=== doko_ is now known as doko
=== yofel_ is now known as yofel
jMCgHello happy people!14:25
jMCgI was forwarded here from #ubuntu+114:25
jMCgI would like to know if there will be utrace support in the Natty kernels.14:26
JFobrb, rebooting from updates14:48
manjojMCg, any reason why? can't you use ptrace ?14:50
jMCg15:56 < fche> utrace & ptrace are not substitutes for each other.  utrace is an intra-kernel API to make things like ptrace and systemtap and other future debugging technologies share control over processes14:59
jMCgI couldn't have put it better, I guess.14:59
=== bjf[afk] is now known as bjf
jeremyAtgardner:  I saw your note, and wanted to make sure I understand the issue.  If I'm in an NFS-mounted directory, I can't copy a file to that same directory?15:34
jeremyAtgardner:  bug 688068, by the way :)15:35
ubot2Launchpad bug 688068 in linux (Ubuntu) "lucid system randomly locks up, does not recover (affects: 1) (heat: 8)" [Undecided,New] https://launchpad.net/bugs/68806815:35
tgardnerjeremyA, is this on the same machine as the NFS server?15:35
jeremyAit was a KVM image doing the work15:36
tgardnerbut its on the same physical HW, right?15:36
jeremyAyep, KVM guest on a physical host which is also sharing NFS15:36
tgardnerthat is the configuration that upstream says is not supported. it causes memory exhaustion because there is no throttle (like a network connection) in between the consumer and the producer15:37
jeremyAoh, interesting.15:37
jeremyAwell, that's an edge case, for sure15:38
jeremyAso 2.6.32-26 may be working great with pci=nomsi15:38
tgardnerjeremyA, you should try rsync using ssh instead15:38
jeremyAhrm.  15:38
jeremyAI'll look into that... do you have any good docs to recommend off the top of your head?15:38
tgardnerdocs for what?15:38
jeremyA(may I add our IRC conversation to the bug report, for the next sod who does this?)15:38
tgardnerof course15:39
jeremyAwell, I'd like a remotely-mounted filesystem....15:39
jeremyAIs samba safer?15:39
tgardnersamba might suffer from the same issue, but I don't know for sure15:39
jeremyAIs there a preferred solution for doing this sort of thing?  I can de-virtualize my pytivo instance, which is the primary user of this, but I'd rather not :)15:40
tgardnerjeremyA, I'd try rsync through an ssh session.15:40
jeremyAokay, I can do that for some of my cases...a read-only NFS filesystem is still safe?15:42
tgardnerjeremyA, likely. you'll only have problems when your platform is under extreme memory pressure. Hence the network suggestion since the socket interface actsas a throttle.15:43
jeremyAtgardner that makes sense, definitely.  I would imagine sshfs would work similarly.15:43
=== sconklin is now known as sconklin-afk
alex_mayorgaHi! I'm getting plenty of kernel panics on 2.6.37.8 and 2.6.37.7 can someone help me?15:55
jeremyAAlternate solutions would be to move the NFS server off to another machine, or to move the kvm guest to another piece of hardware.  Mr. Gardner wasn't sure if Samba would also be affected by this, so I have settled moving the one guest which needs r/w NFS access to a separate physical server.15:58
jeremyAoops15:58
alex_mayorgahow do I report the problems?15:59
jeremyAalex_mayorga:  try running "ubuntu-bug kernel" at the command line.  I believe you have to be root16:00
jeremyAalex_mayorga:  that will let you enter a bug report initially, then you can finish filling it out via a browser :)16:00
alex_mayorgajeremyA: do I have to be running the kernel that's panicking?16:01
jeremyAI honestly do not know.16:01
alex_mayorga"package kernel does not exist"16:01
jeremyAsorry16:02
jeremyAubuntu-bug linux16:02
jeremyA=)16:02
dokoapw: do you what kind of build failure rtg fixed with the last upload?16:03
dokoknow even16:03
alex_mayorga"The problem cannot be reported: This is not a genuine Ubuntu package"16:04
alex_mayorga:S16:04
alex_mayorgaI guess natty is not genuine Ubuntu16:04
tgardnerdoko, are you referring to ti-omap4 and binutils ?16:05
dokotgardner: yes (was looking for your rtg nick ;)16:06
tgardnerdoko, I got it fixed with some help from the Linaro guys.16:07
dokook16:07
jeremyAalex_mayorga:  how odd.  I'm afraid I'm just here getting help with my own issue, so I've exhausted my knowledge on the subject.  Perhaps one of the regulars can help...16:19
=== sconklin-afk is now known as sconklin
alex_mayorgajeremyA: no worries, thanks on the help anyway16:20
JFo<-lunch17:17
JFook NOW, I am finally going to lunch :)18:07
JFobbiab18:07
tgardnerjjohansen, how is ecryptfs coming? bug #344878 has an A2 delivery, and I'm not seeing much activity on it.18:22
ubot2Launchpad bug 344878 in linux (Ubuntu) (and 2 other projects) "file name to long when creating new file (ecryptfs_lookup: lookup_one_len() returned [-36] on lower_dentry) (affects: 48) (dups: 5) (heat: 282)" [Undecided,In progress] https://launchpad.net/bugs/34487818:22
=== cmagina is now known as cmagina-meeting
jjohansentgardner: I am working on the first pass of it18:28
tgardnerjjohansen, how about some updates in the bug so the rest of us can enjoy your progress?18:30
jjohansentgardner: sure18:32
* tgardner wonders why the tracking bug for 2.6.32.27+drm33.12 created by bjf mentions 'The following 126 patches are in the 2.6.32.19 stable release'18:45
bjftgardner, copy-paste error18:45
tgardneryeah, I made one of those once :)18:46
=== sconklin is now known as sconklin-kiddoc
bjftgardner, heh, fixed, thanks for the catch18:46
bjftgardner, didn't think anyone was actually anal enough to read those18:47
tgardneryou've already forgotten my middle name?18:47
bjfLOL18:47
* tgardner --> lunch18:52
* cking ---> calling it a day19:19
* vanhoof waves to cking 19:20
cking:-)19:20
tgardnercking, have a good weekend19:20
vanhoofcking: oh before you leave19:20
cking..?19:21
vanhoofcking: what shipping company do you prefer for international next-day type stuff19:21
vanhoofcking: does one suck less?19:21
ckingFedEx or DHL, they all seem  the same to me19:21
* vanhoof has UPS and FedEx in close proximity19:21
vanhoofok19:21
tgardnervanhoof, just get Blythe to make the arrangements19:21
vanhooftgardner: not that easy19:21
tgardnervanhoof, she makes it erally easy19:22
tgardnerreally*19:22
ckingit's never easy when one requires a special bespoke delivery to the UK late in the day.19:22
* bjf is heading out to luch today19:23
=== bjf is now known as bjf[afk]
pmatulisany reason why a network driver supported by the Lucid 10.04 ISO is not recognized during a netboot install? see [1]19:24
pmatulis[1] https://bugs.launchpad.net/ubuntu/+source/pxe/+bug/60741119:24
ubot2Launchpad bug 607411 in pxe (Ubuntu) "10.04 Atheros AR8131 network install no network device found (affects: 1) (heat: 25)" [Undecided,New]19:24
ckingvanhoof, I sent you an email of my home address, just in case you don't have it already.19:24
vanhoofcking: yeah i have it memorized19:25
ckingheh :-)19:25
vanhoof:)19:25
* cking seriously has to sort out the kids now.19:26
tgardnerpmatulis, missing firmware?19:26
tgardneror its not in a udeb19:26
pmatulistgardner: no idea19:26
tgardnerpmatulis, you need to start a bug report and attach the usual stuff.19:27
pmatulistgardner: well the bug is above.  will ask for apport info19:27
tgardnerpmatulis, doh! so it is19:28
tgardnerpmatulis, yeah, looks like its a udeb issue. it would still help to have the apport info19:30
pmatulistgardner: fixable for .2 release?19:30
tgardnerpmatulis, the timing will be tight, but perhaps19:30
* pmatulis leans a little closer to his computer19:31
tgardnerpmatulis, actually, since its a netboot problem it'll get fixed sooner then the point release19:31
tgardnerthe CDROM already supports the driver, which is why I'm sure its a udeb issue19:31
pmatulistgardner: re apport, if it won't install via PXE then i must install via media and run apport or easier route?19:33
vanhooftgardner: hey you're right ;)19:34
vanhooftgardner: blythe did make it easier19:34
* vanhoof lulz19:34
tgardnerpmatulis, if you'vve a hard ware network you can just boot the Live CD and run apport from there19:34
tgardnervanhoof, I love that girl :)19:35
pmatulistgardner: gotcha19:35
pmatulistgardner: curious, is it possible to "insert" a working driver during install time in such a case?19:44
tgardnerpmatulis, you'll have to consult someone more PXE savvy then me. ogra or cjwatson could likely tell you.19:45
pmatulistgardner: thx19:46
tgardnerpmatulis, also you could try the guys in cert. ara, cr3, etc19:47
ograpmatulis, could be that nobody from the kernel team added the driver to the inird 19:51
ogra*initrd19:51
ograwhen it was enabled19:51
ograNIC drivers need to be added manually19:51
pmatulisogra: interesting, sounds like a manual process19:52
ograyes19:52
tgardnerogra, but they generally go into a NIC udeb19:52
pmatulisogra: how to add one during PXE install time?19:52
ogratgardner, ah, right, i missed the netinstall bit above19:52
tgardnerogra, in any event, the kernel team neglected to add it :)19:53
ograi'm just checking hook-functions in initramfs-tools, seems the bit where you needed to manually add NIC drivers is gone in maverick19:54
ograso that probably changed19:54
pmatulisogra: k, using lucid here19:54
ogralucid still has it http://paste.ubuntu.com/542012/19:55
tgardnerogra, NICs aren't generally boot essential which is why (I think) that stuff was removed from the initrd. Especially since they can be added to a udeb for little cost.19:56
ogra(in /usr/share/initramfs-tools/hook-functions)19:56
ogratgardner, ugh, that would break netbooting 19:56
tgardnerogra, I thought udebs were pulled across the net during boot19:56
ogra       net)19:56
ogra                copy_modules_dir kernel/drivers/net \19:56
ogramaverick just copies the dir19:57
ograinstead of looping over a manually maintained list of modules19:57
ogratgardner, not on an installed system19:57
ograudebs are only used by the installer19:57
tgardnerogra, how are PXE images prepared?19:57
ografor an installed system the NIC drivers need to be in the initrd for netboot19:57
ograPXE pulls kernel and initrd and executes them19:58
ograthe initrd reads the server data from DHCP and mounts the rootfs after bringing up the NIC19:58
tgardnerogra, ah. how can we prepare PXE initrd's without polluting the normal initrd ?19:58
ograwe cant19:59
ograyou can either netboot without initrd and need to have the driver compiled in monolithic or use an initrd and have the driver in the initrd19:59
tgardneradding drivers impacts boot times for the general user.19:59
ograyou could make adding the kernel/drivers/net subdir optional and only copy it into netboot initrds20:00
ograseems since maverick we copy the whole dir20:00
ograup to lucid the lisz from http://paste.ubuntu.com/542012/ was used20:00
ogra*list20:00
ogramaintaining the list was massively error prone20:01
tgardnerogra, indeed.20:01
ograso i think that was the reason for just copying the dir20:01
* jjohansen lunches20:01
tgardnerpmatulis, adding to that list is probably the way to go for Lucid.20:01
ograbut that will likely add a lot of unused additional cruft and bloat your initrd20:01
pmatulistgardner: so basically i edit the file and rebuild the initrd?20:02
ograyes20:02
tgardnerpmatulis, yep20:02
pmatulistgardner, ogra: thanks.  i guess the module itself is just 'ar8131'?20:05
* ogra doesnt know20:05
tgardnerpmatulis, I don't think so. I thought it was ath1c or something. whats the PCI ID ?20:06
ograwell, for the installer you will need it in a udeb too20:06
ograthe initrd change only helps for installed systems20:07
pmatulisogra: whoops20:07
tgardnerogra, if its atl1c then its already in a udeb20:07
ogra(and for the live image, but you would need to recreate the casper initrd for that)20:08
ograthe bugs isnt really clear what the user did20:08
ograone comment talks about a PXE netinstall, the other about booting a live image from USB20:08
ografor alternate you definitely need a udeb, for live and for installed netbooting systems you need the hook-functions change and a new initrd20:09
pmatulisogra: just pxe is what i'm interested in20:10
ograPXE is the underlying boot protocol ;)20:11
ograit loads an initrd in all cases, in the case of the alternate installer the initrd needs to contain the right udeb20:11
ograin case of an installed netbooting system the driver needs to be listed in the "net" section of the auto_add_modules function in hook-functions20:12
pmatulisogra: not sure why you're talking about 'installed netbooting'.  i'm just trying to install *by* netboot.  i'm prolly missing something20:13
ograpmatulis, then you want the udeb 20:14
ograif you want to install a system that uses an nfsroot you want the other solution20:14
pmatulisogra: where can i find it and what do i do with it?  :)20:14
ograPXE is just used to boot either of them ;) 20:14
ogratalk to 20:15
ogra#ubuntu-installer 20:15
ogratgardner said the modules is already in an udeb so it might be that this udeb isnt included in the installer initrd20:15
tgardnerwell, we haven't established which driver its actually using.20:16
pmatulisogra: ok, so a udeb and the change to initrd (the net list)?20:17
ografor a proper fix we should do both, yeah20:18
pmatulisogra: just looking the necessary bit for my case20:18
ografor your installer fix you only need the udeb and need to make sure the udeb is included20:18
pmatulisogra: alright20:19
* pmatulis goes to hunt for a udeb and what to do with the bugger when he finds it20:19
ogramake sure to know the actual .ko name first 20:19
ograthen ask in the installer channel20:20
* ogra is officially on vacation, you just catched me while checking mail ;)20:20
=== tgardner is now known as tgardner-afk
=== bjf[afk] is now known as bjf
=== sconklin-kiddoc is now known as sconklin
=== cmagina-meeting is now known as cmagina
=== ogra is now known as Guest62388
JFowell, have a great weekend folks22:30

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