/srv/irclogs.ubuntu.com/2008/10/27/#ubuntu-kernel.txt

persiaSee, this is why I suggested you attach it to the bug when it's done :)00:01
NCommanderpersia, rofl00:01
NCommanderpersia, I had to run updateconfigs for a new configuration option. I'll have to re-test build it :-/00:11
persiaNCommander, Well, at least your processor will get some exercise :)00:14
NCommanderYeah yeah :-P00:15
NCommanderpersia, I might be wrong on needing to rebuilding it00:15
persiaHrm?00:15
NCommanderThe build finished regardless00:15
* NCommander will do a second smoke test on the build00:16
NCommanderreboot time00:17
NCommanderpersia, I can confirm the -rt kernel works00:22
NCommanderpersia, what was that bug number again?00:22
persia28968300:22
* NCommander is 99.9% sure that the second rebuild test is not necessary but I'm not willing to risk it this close to release00:22
persiaThank you for the extra care.  It's not default on the CD, but I'm fairly sure we only get one chance to upload before it becomes an SRU.00:23
NCommanderpersia, I'm having trouble now, I think the linux-rt sources tainted my /usr/src package00:24
* NCommander is getting patching failures00:24
persiaIt's designed to be built under sbuild :)00:24
NCommanderYou know00:24
NCommanderthere is an easy solution00:24
* NCommander shoots it to his PPA00:25
NCommanderIf it builds there, I'm satified00:25
persiaYep.  That's a nice clean sbuild.  Takes about an hour to build for a PPA.00:25
NCommanderpersia, I didn't think of it before cause I'm usually working on the ports kernel00:25
NCommanderand away it goes00:25
persiaI think that's 90% of the reason that -rt doesn't also target powerpc00:26
NCommanderpersia, thats going to change for jaunty00:26
* NCommander already has it on the TODO list00:26
persiaGood.  It's a frequently requested feature.00:26
NCommanderI need to get the normal ports kernel up to shape before studio can get some love00:26
NCommanderpersia, I'm also talking to jdong on the possibility of backporting the ports kernels (as a different binary package so it won't be an auto upgrade) so studio users don't have to wait for 9.0400:27
persiaYou mean ports users?00:27
NCommanderWell, linux-rt-backports :-)00:28
NCommanderand linux-ports-backports00:28
NCommanderThe powerpc varient of rt should build out of the linux-rt source package, having that changeset in the ports kernel will likely break things miserably :-/00:28
persiaDefinitely.  Unless -rt goes mainline, I think the current architecture makes sense, although it's annoying to have to upload a new -rt every time -generic is uploaded.00:29
NCommanderpersia, man, LP is lagging tonight, I still haven't got an ACCEPT or REJECT on the rt upload to PPA00:29
NCommanderpersia, agreed. Talking with the rest of the kernel team, they also agreed00:30
* NCommander brought it up at the last meeting00:30
NCommanderpersia, the only problem is the rt kernel doesn't support SMP last time I checked, so its still somewhat limited on PowerPC00:31
persiaNo, it's somewhat limited *everywhere*.  It supports SMP on 2.6.26, but 2.6.26 has all sorts of other issues.00:31
NCommanderEw00:32
NCommanderpersia, good call on my part. FTBFS :-P00:38
persiaNCommander, That you find that a good call is an indicator that your glass will always be half full :)00:40
NCommanderpersia, well, paranoia, especially with kernels and RC freeze is a healthy thing00:41
persiaIndeed :)00:42
NCommanderprobably monday is the last day we get any uploading in00:42
TheMusoConsidering how long it takes to populate dak, and test images...00:43
persiaIt's been Monday for nearly 14 hours ...00:43
persia(at least somewhere)00:43
* NCommander sends it to the PPA again00:49
NCommanderpersia, I'm also testing in pbuilder, so we should be good to go soon-ish00:51
persiaExcellent00:51
NCommanderpersia, I'm just glad to do my bit for studio00:52
persiaI know better than that : you just want to join every team on launchpad ;p00:52
TheMusoI think its called NCommander has heaps of time now, but may not in the future. :p00:56
NCommanderTheMuso, actually, I'm not taking classes for winter quarter, so I'll actually have more time in the future00:57
TheMusoCool.00:58
* NCommander HATES updateconfigs00:58
* nigel_c gets ABI differences even with a simple checkout of the right git tag. Maybe it's because I'm still using Hardy gcc.01:05
NCommandernigel_c, that would do it01:06
nigel_ck.01:08
nigel_cIt's only a few symbols - memcpy, __memcpy and something else I've forgotten, but still ....01:09
nigel_cI've nearly got my ppa sorted - some cleanups, pushing a patch upstream and remove a small feature. Still have one new symbol I can't do anything about, but that doesn't seem to kill the build.01:10
nigel_cThanks for your help the other day, NCommander01:10
NCommandernigel_c, what did I do?01:11
nigel_cI forget now, but it was helpful :)01:25
NCommanderlol01:25
NCommanderOh01:25
NCommanderI think I taught you about bumping the ABI01:26
nigel_cmm. that souns right01:26
nigel_coops one handed typing while eating lunch :)01:26
NCommanderlol01:26
NCommanderhey BenC1 03:42
=== BenC1 is now known as BenC
=== asac_ is now known as asac
=== thegodfather is now known as fabbione
=== asac_ is now known as asac
=== asac_ is now known as asac
=== Keybuk is now known as Guest49306
=== Keybuk_ is now known as Keybuk
* abogani waves09:47
aboganitseliot: Are you around?09:47
tseliotabogani: sure09:49
aboganitseliot: Sorry to bother you. Only one question about DKMS/video...09:49
tseliotok09:49
aboganitseliot: How can I simulate fglrx dkms build process (obviously i don't have that hardware)?09:50
aboganifor rt kernel09:50
tseliotabogani: you don't need a specific device to build a module09:50
tseliotabogani: you will have to install nvidia-VER-kernel-source09:51
tseliotand if you want to build the module manually (i.e. without booting into a kernel for which you want to build a module)09:52
tseliotyou can type something like:09:52
tseliotsudo dkms add -m nvidia -v 177.80 -k $(uname -r)09:52
tseliotsudo dkms build -m nvidia -v 177.80 -k $(uname -r)09:52
tseliotsudo dkms install -m nvidia -v 177.80 -k $(uname -r)09:52
aboganiPut log somewhere? 09:53
tseliotyes, /var/lib/dkms/nvidia/177.80/build/09:53
tseliot(replace 177.80 with the version of the driver)09:53
tseliotit's in the make.log09:53
aboganitseliot: Wow! Thanks a lot!09:54
tseliotabogani: you're welcome09:55
=== mdz_ is now known as mdz
apwi am trying to determine the source of broken alsa sound (crackles, OSS works) in the latest intrepid kernels, both of the reporters seem to have intel hda sound.  so am wondering if anyone has working alsa sound, and if so which sound h/w they have11:34
fqhhello,where is the definition of "get_user" in x86?12:01
ckingfqh: include/asm-x86/uaccess_*.h12:03
fqhOh, I see. But it is strange that linus-git has no a asm-x86 now.12:08
fqhhttp://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=tree;f=include;h=0fd53c2abd87c28d05796ebe9c139678917a593b;hb=f8d56f1771e4867acc461146764b4feeb524566912:08
apwhaven't they moved into arch now12:11
apwarch/x86/include/asm or something12:11
apwfqh ^^12:11
ckingfqh: arch/x86/include/asm$ vi uaccess.h12:13
ckingoops, meant: arch/x86/include/asm/uaccess.h12:13
fqhyes. I see it. :)12:13
zulfor getting fixes modules in hardy it has to go in linux-backport-modules correct?12:36
rtgzul: it depends. is it an existing module in LUM?12:37
zulits drbd 12:38
zulso im assuming no12:39
rtgzul: it doesn't sound familiar, but then there are a bunch of drivers in Hardy LUM. Is this something we _want_ to support?12:40
zulrtg: yeah we support it already, the reason Im asking is there is a push to update drbd to a newer version as a hardy SRU (not by me) and afair there is an ubuntu/drbd in the ubuntu-hardy git tree so if I get asked about Ill have the information12:41
rtgzul: ah, then you'll have to propose it on the kernel team list, and provide some corroborating evidence that it fixes real problems.12:43
zulrtg: yep ok thanks12:43
aboganitjaalton, tseliot, superm1: Could you push the attached patch for bug 286961, please?13:08
NCommanderMorning13:08
NCommanderabogani, we're in final freeze, nothing can be pushed to the archive at this point13:08
tseliotabogani: you might want to talk to superm1 about this (for an SRU) in #ubuntu-devel13:13
aboganitseliot: Ok, thanks.13:14
NCommandermorning tseliot, BenC13:14
tseliotNCommander: good morning13:14
NCommandertseliot, how goes it13:14
tseliotfine13:15
NCommanderThats a good way to start out ones morning :-)13:15
BenCNCommander: good morning13:15
tseliotNCommander: BTW my patch for abiword was accepted by upstream13:15
NCommanderBenC, morning! You'll be pleased to know AppArmor has landed for ports ;-)13:15
NCommandertseliot, woo!13:15
NCommanderBenC, so has squashFS, AuFS, a bunch of bug fixes, compcache, pretty much everything useful out of the normal ports branch13:17
NCommanders/ports/main/g13:17
BenCNCommander: nice13:17
BenCNCommander: busy weekend? :)13:17
NCommanderBenC, You could say that. I also got my first round of rt hacking (a very last minute rebase against modern kernel sources)13:18
* NCommander is also adding powerpc support to -rt :-)13:19
BenCNCommander: you're not patching ports directly with -rt are you?13:19
NCommanderNo13:19
NCommanderwith abogani's permission, I'll have it grab a ports tarball, and build against the ports kernel on powerpc (and arm/mips if we ever get those architectures in the DC)13:20
apwrtg: you made the last kernel image i believe.  i have been debugging the alsa sound issues some of us have been seeing with intel sound.  i just rebuilt the kernel and install just that one module and my sound is working now.  so i am trying to fathom how we have ended up with different modules from that process13:39
rtgapw: which one in particular? hda?13:40
apwyeah hda, snd-hda-intel13:40
rtgapw: perhaps its an ordering issue.13:41
apwi was doing a bisection on it from 2.6.2413:41
apwand ended up with all the patches installed, with the tree at our git HEAD13:41
rtgdid the bisect converge?13:41
apwnow i did build it using make13:41
rtgwith what config?13:41
apwi found sound worked at 2.6.26 and 2.6.27 and indeed then at our latest HEAD13:42
apwthe one from /boot/ for the latest intrepid kernel /boot/2.6.27-7-generic13:42
apwmy module is massive compared to yours13:43
apw3.4M against your 815k, so its clearly not compiled the same way13:43
rtgapw: I imagine its more of an ordering issue then it is a code generation problem.13:44
apwa bit of a pig to debug me thinks ...13:45
apwi guess my next step is to build this kernel the way the tools do insteal13:45
rtgapw: have you talked to the ALSA guys about it?13:45
apwnot yet, only just confirmed its a kernel issue13:46
apwwhere would you suggest i go to find them13:46
rtgapw: alsa-devel@alsa-project.org13:46
rtgapw: amd64 or i386?13:49
apwthis is amd6413:49
apwand my kernel build was on amd6413:49
* apw prays this is not a toolchain bug breaking the kernel13:49
rtgapw: describe your build process. I assume you cloned the git repo, copied the config, then 'make'13:50
apwmake oldconfig; make 13:50
apwthen i literally copied the .ko into the installed kernel13:50
rtgdpkg -l|grep gcc13:50
apwi have 4.2 and 4.3 installed it seems13:51
apwdefualt compiler  seems to be 4.313:51
apw4.3.2-1ubuntu1113:51
rtgapw: gimme a bit.13:52
apwrtg np ... will go see if it can build it your way13:52
=== Keybuk_ is now known as Keybuk
NCommanderError(/tmp/buildd/linux-ports-2.6.27//drivers/scsi/scsi_transport_fc.c:509): cannot understand prototype: 'atomic_t fc_event_seq; '14:35
NCommanderANy idea whats causing that?14:35
NCommander^- amitk, BenC, 14:35
BenCNCommander: sounds like a missing include14:39
NCommanderIt compiles fine :-/14:39
NCommanderargh14:39
BenCNCommander: what patch introduced that problem?14:39
NCommanderI dunno14:39
NCommanderIt compiles just fine14:39
NCommanderB14:52
apwrtg ok built the kernel using debian/rules binary-generic, the .ko that made seems to work too15:04
rtgapw: when you run alsamixer, are you seeing only one output control?15:05
apwyes only master15:05
apwthough in the gnome tools there are several15:05
rtgapw: well, none of those work for me.15:06
apwrtg: not sure whats not working for you there?15:14
* apw hits the build process with bigger hammers15:15
apwrtg, even the stripped version of the .ko which gets built to make the .deb's (which is more the size of yours) seems to work ok15:21
rtgapw: I think this is an app space issue. I'm having problems on laptops that used to work.15:22
NCommanderHow does docbook parse the source code15:23
apwrtg, i have working sound with the hardy kernel and not with 'your' intrepid kernel with the same userspace, and here i am only changing the module.  but i guess this doesn't rule out a userspace library and some luck making it work for me15:24
fbondHi.  I'm trying to debug a suspend issue, but I'm not sure if I'm looking at a bug or not (I think I am).15:31
fbond`echo standby >/sys/power/state` leaves the machine running with display still on after disabling devices.15:32
fbondThat is not normal, is it?15:32
fbondThe machine does seem to be "paused," in a way, and it correctly comes out of this state with normal wake interrupt activity.15:32
NCommanderBenC, I solved it, there was a comment above the struct that went /**15:36
NCommanderBenC, stupid docbook :-/15:36
=== thegodfather is now known as fabbione
jcastroLooking for kernel-team Openweek sessions: https://wiki.ubuntu.com/UbuntuOpenWeek16:22
apwrtg: ok things just got wierd here, i just got a request from update-manager to install the -7.14 kernel again so i let it17:48
apwand now sound works, even though the module is identicle to the 'non working one'17:49
apwso i suspect that adds credance to your suspicion its userland17:49
apwrtg "its back" ...18:13
apwso its actually intermittant, so all my testing is meaningless18:14
NCommanderBenC, do you know anyone with ia64 boxes? (I have the ia64 kernel more or less building, but I have no idea how to setup ski, so I'm hoping to find someone who is willing to test these :-)18:54
BenCI have one, but it is powered off atm19:02
NCommanderBenC, handy. I should see if I can pick a cheap ia-64 for myself at some point ;-)19:16
=== doko_ is now known as doko

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