/srv/irclogs.ubuntu.com/2008/11/21/#ubuntu-kernel.txt

=== doko_ is now known as doko
* apw yawns08:34
amitkmorning apw 08:38
apwmorning amitk ...08:38
NCommandermorning amitk 08:38
amitkNCommander: morning08:42
NCommandermorning08:42
* NCommander wonders if we're stuck in an infinite loop or something08:43
apwheh, heres hoping not08:43
* NCommander has no desire to debug the kernel team08:44
* NCommander runs from the bad pun08:44
amitkhehe08:53
NCommanderThe question is do we need kgdb, or old fashion printk() debugging ;-)09:00
* apw strokes printk09:07
* NCommander sees apw saying "my precious"09:16
apwsomething like that09:16
apwamitk, i take it jaunty a1 will ship with your armel kernel?09:49
amitkapw: yes. 2.6.27-based kernel.09:51
amitkBenC is planning to upload a 2.6.28-based kernel today09:52
apwso that will miss a1 i assume, it sounds like the buildd's are chugging a1 as we speak10:01
rljhi all, i experienced LP#291878 and wrote a trivial patch to the kernel (uploded to LP) to add a keyboard quirk for the affected laptop model. the patch works the way it's supposed to on my laptop and solves the issue, but so far no one else affected by the bug has responded to my patch. what's the proper way to get it included into the ubuntu kernel and also getting it upstream?10:04
amitkapw: yes, 2.6.28 isn't slated for a110:05
apwrlj, the patch looks low risk.  normal process is to push the patch upstream to the maintainer in MAINTAINERS in the kernel source, that looks to be 'INPUT (KEYBOARD, MOUSE, JOYSTICK, TOUCHSCREEN) DRIVERS'10:09
apwyou would want to write up a proper description and make sure to add a Signed-off-by: to it10:10
rljah i see, the MAINTAINERS file seems helpful. didn't think to look there. :S10:11
apwprocess is always the tricky part.  also i see this patch is really about the kernel but the bug it not against the kernel10:12
rlji'd like the affected people to try out the patch to figure out if more DMI matches should be added before submitting upstream, but i can't force them to recompile... :) oh well.10:12
rljwell, the bug is about "ubuntu" currently, isn't it10:12
rljbut from my investigation, it seems the proper place to "fix" the broken hardware is with a kernel quirk10:13
apwno you never can, but its pretty limited to one laptop type and you have that and this makes it work for you10:13
rljrather than some kind of userspace quirk10:13
apwyes, and your fix looks like its of the right form.  the acid test is sending it to the maintainers and seeing if they shout at you10:13
rlj"the acid test is sending it to the maintainers and seeing if they shout at you" ?10:14
apwyeah, your patch looks reasonable to a kernel-dev who isn't particularly familiar with keyboard quirks (me), so i would say "looks good, send it to the maintainer"10:15
apwmeaning if its not right they will then tell you what you should have done and it gets fixed10:15
rljoh, misinterpreted "the acid test" for some automated software rather than a figure of speech ;)10:15
apwyeah sorry i often forget that good english does not mean 'knows all those mad engish local sayings'10:16
apwi have sorted out the bug assignement and states and importance and stuff ...10:17
apwrlj have you ever submitted a patch upstream before?10:18
rljthanks for the pointers. i'll go with the workflow in the MAINTAINERS file in the kernel distribution and try and get it upstream. this means ubuntu will include the fix indirectly after a new 2.6.27 stable kernel has been released from upstream? or should i do something else to make ubuntu get the fix before jaunty?10:18
apwwell its untlikely to make .28 now, as we are mostly in bug-fix only phase for that10:19
apwso its unlikely to make .28, and jaunty is most likely .28 based10:19
apwbut once its upstream it is much easier to justify picking it up as a backport10:19
apwactually i may as well take this bug as i know all about the bug now10:19
rljah, true10:19
apwi can help you get the thing in shape for upstream10:20
rljno, haven't submitted before. so  i email the maintainer privetely or just the lkml?10:20
rlj/s/and not the lkml/10:20
apwyou would send it to the lists as listed in the maintainers entry10:20
apwso in this case you would send it to Dmitry direct and linux-input direct, and then likely CC: linux-kernel10:21
apwas that is the 'records everything' list10:21
apwmake sure the patch has a good one liner, a full description after that10:21
rljaha10:22
rljyeah10:22
apwthen the signoff, and finally make sure the patch is clean via checkpatch10:22
apwthen you are pretty set ... you might also Cc: me as well10:22
apwquery rlj10:22
aboganismb_tp: Are you around?11:01
smb_tpabogani, Yes11:01
aboganismb_tp: If you want i push updated preempt rt support on zinc.11:02
aboganifor Hardy i meant.11:02
smb_tpabogani, Great, ok. I'll go for them as soon as I am finished fighting my build chroots11:03
aboganiThank you very much.11:04
smb_tpabogani, Thanks for updating them so quickly11:05
smb_tpabogani, mm, one question. This is all based on the current (more or less) master? 11:07
smb_tpabogani, I am asking because all the big changes for 2.6.24.7 are on the next branch of my tree http://kernel.ubuntu.com/git?p=smb/ubuntu-hardy.git;a=shortlog;h=refs/heads/stable11:10
smb_tpabogani, Unfortunately importing those changes had quite an impact on rt because there are a few central places changed (like scheduler).. :(11:11
aboganismb_tp: I failed. I have used master. I'll go to do work for stable branch. 11:13
smb_tpabogani, I sorry, this might have been unclear... ATM there is much branch juggling on some trees...11:15
aboganiNo sorry, please. It's my fault.11:15
apwsmb_tp, if we have bug which needs much more than a kernel fix, but needs a decision on direction and will likely need a coordinated change with other packages, would that be a blueprint/uds type of thing?12:07
smb_tpapw, I did not have the case myself, yet. But I think it depends on how much has to change. Minimum the affected packages have to be added to the bug and then getting the right "sponsor"12:09
apwthis is the wireless regdom thing sitting on our queue, realistically some infrastructure to tell us where we are is needed12:10
smb_tpcking, had once a case with lmsensors. I don't think it generally has to wait until UDS12:10
apwpossibly in the short term till thats available, we might place all machines in the EU or something12:10
apwas that would get us closer to no regression, a very tricky thing indeed12:11
smb_tpapw, With wireless and what you say this might not be just a little fix in another package but some more re-design. So it is possibly something that needs more discussion12:11
smb_tpapw, I would not believe placing all machines into EU will get that much liking12:12
apwthey are unrestricted prior to intrepid i believe, and i assume there is no setting for that in the new framework12:13
apwthere doesn't seem to be a good work around here as they have to be genuinly different in different places12:14
apwso i guess the question is who is the best person to ask about what to do with this problem12:15
apwdo we have anyone in the wider architecting things role12:16
smb_tpapw, Which sounds like something that not only affects one package like network manager but is a general issue probably with wireless extension? Guess Tim is more down this route than I am. But this particularly sound like an UDS topic12:16
apwgood point tim is wireless guru in our domain, will hastle him next :)12:16
smb_tpapw, As soon as he gets up which usually is quite early :)12:17
sorenDoes any of this look like a wifi nic to you guys? http://pastebin.ubuntu.com/75192/12:23
sorenI'm told that in vista, that box has wifi, but I don't see it.12:24
lagamaybe it's usb?12:24
lagaor the kill switch is acting up?12:24
sorenIt might be USB, actually.12:25
amitkNCommander: 14:18 < cjwatson> could somebody update linux-ports-meta in jaunty to the current ABI please?                                                  12:29
amitk14:18  * cjwatson <- uninstallable-packages police 12:29
smb_tpamitk, NCommander Has someone started something already? Otherwise we could use intrepid-ports-meta.git as starter...12:31
amitksmb_tp: I was hoping NCommander already started something or would be interested :)12:32
smb_tpamitk, Ah, yes. :) If he has not yet, using the other git repo as initial source and adapting it should be simple. Would do it but my network is currently sucking due to uploads...12:34
smb_tpamitk, NCommander You know whether 2.6.27-1 is current ports abi. If yes there is now a git repo at git://kernel.ubuntu.com/ubuntu/ubuntu-jaunty-ports-meta.git12:58
amitksmb_tp: I don't see such a git tree13:03
smb_tpamitk, gitweb seems not to like it :(13:03
amitksmb_tp: it takes some time to show up if you just pushed it13:06
amitkcron job every 20-30 mins13:06
smb_tpamitk, Well sometimes it seemed faster, but I guess that is rather the effect of being distracted long enough until looking again13:07
smb_tpamitk, ;-)13:07
amitkhehe13:08
BenCAnyone know if rtg has looked into the new wireless regulatory domain stuff showing up in 2.6.28?13:25
smb_tpBenC, Not sure but apw asked some questions into that direction this morning and probably will get back to him13:29
apwBenC, from the bug we have i assume that the first steps of that are at least in .27, people are talking problems for high channels in EU as a result13:30
apwi was wondering how we fixed such 'cross domain' things this morning, and we felt rtg was the right person to poke as step 113:31
BenCWell this seems to be all new stuff in 2.6.2813:31
BenCthe built-in static regulatory stuff is getting obsoleted for a userspace daemon that handles it13:31
apwheh, well thats a step forward at least13:32
rtgapw: there is a UDS session to talk about CRDA13:33
* apw wonders what that is ...13:33
rtgapw: central regulatory domain agent13:34
apwahh good ... so i assume you are taking a role in that, given your predilictions in the wifi area?13:35
rtgapw: I helped start the design, though I haven't had much of an active role in the development. That has been Luis Rodriguez who now works for Atheros.13:37
apwto expalin why i was talking about this at all, we have a bug against intrepid as clearly some changes occured in .27 on the regularty control in the ieeeNNNN layer13:38
apwand it sounds now that is changing again for .28, and we need to work out what our response is to those13:39
rtgapw: well, there is a config option in .27 to ignore regulatory controls (which we have set). iin .28 and higher we'll need to implement the crda daemon.13:40
apwrtg, oh has that changed, the config option, ie should it not longer be a problem in .2713:43
rtgapw: in Intrepid regulatory compliance is attained through eeprom settings in each adapter, or through beacon location information element detection. typically its the union of those 2 features.13:45
apwbut there is an option to ignore it, and we have it turned on?  if so what version did we do that?13:46
rtgapw: no, the config option that we have set is to ignore the _lack_ of a crda daemon (though I may be thinking about wireless stuff in LBM)13:47
matiassHello, 13:48
matiasscan i ask you about a bug that i reported but i don;t know what to do with it13:48
matiassabout the newest kernel13:48
matiass?13:49
apwahh i see13:50
apwbug# ?13:50
matiassyes13:51
matiassi reported in bugzilla 13:51
matiassalso here is a thread in forum13:51
matiassis with the newest kernel and my motherboard of INTEL13:51
matiassDP43TF13:51
matiasscan you see maybe the thread? there is picture of the bug13:52
matiasshttp://ubuntuforums.org/showthread.php?t=96366213:52
apwwhich bugzilla is it reported in, and which bug# ?13:54
matiassa second.13:54
apwthe forum thread seems to be almost unobtainable from where i am no idea  why13:55
matiasshttps://bugs.launchpad.net/bugs/29266313:55
matiassno way...13:55
matiassi;m there now13:55
matiassokey..i see they have problem with the ubuntuforum site13:57
matiassso,, there is the bug in launchpad13:57
matiassare you there?13:59
matiasshello??14:06
speakmanHi folks! I've just submitted a patch for the ATL1E driver: https://lists.ubuntu.com/archives/kernel-team/2008-November/003603.html14:18
rtgspeakman: shit, wish I'd known of this patch last night. given the age of the patch I'm surprised it isn't one of the stable updates.14:23
speakmanis it not even in linux-2.6 git?14:24
speakmanIs there a new ubuntu kernel on its way for Intrepid already?14:25
matiasshi,,,can you help me with my bug and motherboarD?14:25
matiasshere is a thread you can see the image of it14:26
matiasshttp://ubuntuforums.org/showthread.php?t=96366214:26
rtgspeakman: Intrepid 2.6.27-10.20 is on the way14:27
speakmanoooh :(14:27
speakmanis it really not possible to merge this patch before release?14:28
rtgIntrepid is already released. It'll show up in an update. Create a Launchpad bug and attach the path.14:28
rtgs/path/patch/14:29
speakmanwhich lp project?14:29
rtgspeakman: Ubuntu project, linux package14:30
speakmanhttps://bugs.launchpad.net/~ubuntu-kernel-team14:30
apwBenC, how is our jaunty kernel is your rebase up yet?14:39
BenCapw: rebase is done, doing configs now14:40
apwBenC, so wait a bit and it'll be there14:41
amitkBenC: I'm doing the armel configs too. Just be done soon14:41
BenCamitk: ok14:41
apwrtg, our intrepid tree feels like it is missing the last couple of tags?14:41
speakmanBugreport + patch submitted: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/30069814:42
rtgapw: could be. gimme  a bit to wrap up LBM/LRM packaging.14:42
BenCrtg: I was wondering if you had looked into the new userspace regulatory stuff that the 2.6.28 supports?14:43
rtgapw: try now.14:43
speakmanrtg: Can you promote it to next kernel release to 8.10 ?14:44
rtgBenC: I've beem postponing that knowledge until UDS. Luis will be there for the week.14:44
rtgspeakman: I know why this one looks familiar. I read it on LKML a few days ago.14:46
speakmanrtg: okay, it might already be in linus' git?14:47
BenCrtg: ok, I've kept the built-in compatability enabled, so it shouldn't be required yet14:47
speakmanrtg: this thread: http://lkml.org/lkml/2008/11/11/16514:48
rtgspeakman: commit 7ee0fddfe05f105d3346aa8774695e713069783614:48
rtgit'll show up with 2.6.27.8 stable14:48
speakmanoh, that's great news :)14:49
rtgBenC: right, thats the way to go until I get smarter about it.14:49
speakmanany idea when .8 will be release in public?14:49
rtgspeakman: since I just uploaded last night, it'll likely be at least 2 weeks14:50
speakmanrtg: can I download it somewhere right now?14:50
rtgspeakman: https://edge.launchpad.net/ubuntu/+source/linux14:52
matiasssomeone can have a look at my BUG  i put a photo of it in a thread  http://ubuntuforums.org/showthread.php?t=96366214:53
matiassthank you very very much,,,, i don;t want to wait maybe until april!!14:53
speakmanthis might not be a kernel issue, but how do I make Intrepid prefer my patched version of 2.6.27-7-generic instead of the intrepid version, and still reuse all restricted-* stuff plus upgrades when the new version arrives?14:57
rtgspeakman: if you are building your own kernel, then edit debian/changelog to increment the minor version, e.g. 7.16 --> 7.1714:59
speakman(if I do apt-get upgrade it will reinstall the intrepid version)14:59
speakmanrtg: oh, great thanks!14:59
apwrtg perhaps 7.16 -> 7.17~speakman15:00
rtgapw: 7.17 is safe in this instance since we'll never produce that version number15:01
apwahh fair enough15:01
rtgthough the updates kernel is gonna clobber it almost immediately, being -9.1915:01
speakmanokay, which one to prefer? will it still overwrite with new kernel even if I choose ~speakman suffix?15:02
rtgspeakman: yep - perhaps you should wait until the -updates kernel is propagated, download it and add your patch.15:03
speakmanokay, it is not possible to withdraw that kernel and attach this patch..?15:06
rtgain't gonna happen15:06
speakmanokay, just asking..15:06
speakmani think I'll just reapply the patch when that kernel is released. I need the .local domain support in my work. :)15:07
BenCamitk: I've pushed my configs15:07
amitkBenC: ok.. still working on mine.15:07
apwmatiass, can you tell me more about what CPU this thing has in it?15:08
BenCamitk: you have more, so take your time :)15:08
apwmatiass, if you have an older distro on it, then can you can /proc/cpuinfo and paste it into the launchpad bug15:09
speakmanreboot time. Great thanks to you all !15:09
rtgapw: according to the MB model, its gotta be core-duo or higher15:11
matiasssomeopne ca help mme??with this bug   http://ubuntuforums.org/showthread.php?t=96366215:11
matiassquad core INTEL Q660015:12
matiassi know someone with the same problem ...is the motherboard we think15:13
rtgmatiass: its not likely to get solved here. have you talked to Intel?15:14
matiassmmm15:15
matiassi talked to them today. "we dont support linux"15:15
apwmatiass, we really need to get all the information you have in the forum thread into the launchpad bug15:15
rtgmatiass: have you booted _any_ linux kernel on it?15:15
matiassbut listen , with Debian lenny is working, kernel 2.6.26 i think15:15
matiassokey how can i do this ?????????15:16
apwif lenny works the it would be good to get the /proc/cpuinfo off there pasted into the bug as well15:16
matiasstell me what to do i willput it in the launchpad15:16
apwwell there is an image in there that you need to register to get, you could pull that over to launchpad15:17
apwand it would help to transcribe the text off the screen to save people looking at the image15:17
matiassokey, can i upload image to launchpad?15:17
apwi think its call an attachment or something15:18
matiassokey..15:19
matiassbut /proc/cpuinfo  is needed?    q6600 is not enoguh?15:19
apwyes that has a large amount of information about how the cpu has been detected15:22
matiassokey15:23
matiassi;m not in the place with the PC ...so on sunday i will be15:23
BenCsmb_tp, rtg: That atl1e patch in kernel-team@ looks like a good SRU candidate15:23
matiasshow can  i contact you? or you enter bugzilla also?15:23
rtgBenC: yes - he's already created an LP report. I've nominated. its gonna show up in stable .815:24
matiassyou always here in this channel ?  i really sorry i don;t understand a lot about linux...just starting ..bought hardware i thought is compatible and ...have lot of problem15:25
BenCamitk: I'm rebasing on 2.6.28-rc6 since current git wont even compile because of some header weirdness15:25
BenCamitk: so no commits yet, I'll be force pushing in a minute15:25
apwputting the maximum inforamtion you can in launch pad means any of the kernel developers can look over it15:25
BenCrtg: current rebase includes two reverts of upstream patches from intrepid...I'll point them out in the summary I send to kernel-team@ so I can find out if they can be removed15:26
matiassthanks , but you think it can be solution maybe that i will compile my own new kernel?15:27
amitkBenC: ok... 15:27
rtgBenC: not surprising.15:27
apwmatiass, not sure at this point.  if you know how to build kernels then building a 2.6.27 kernel on your working lenny system would help, if that fails we know its 2.6.27 related15:28
matiassi will try...i don;t know how :)15:29
matiassalso i mean the error is when you put the disc and just boot 15:29
BenCugh15:29
BenCwhy does task_struct have to scm_work_list members :/15:29
BenC*two15:30
BenCI wonder if that's a bad merge15:30
matiassalso i had 8.04 with old kernel  and did upgrade via internet and then when it upgraded the kernel also you have the BUG after the GRUB 15:31
apwmatiass, so you had a working hardy system and it broke when upgraded15:35
apwcould you still boot the 2.6.24 kernel from hardy on the intrepid updated userspace15:36
matiassyes. i did it  but15:37
matiasslisten, 15:37
matiassi have a new very good graphic card from nvidia and you CAN't install new drivers if you don;t have the newest kernel!15:37
matiassis like all the world against :)15:37
apwwell the world is against closed source drivers, they make life hard for themselves and their users15:38
rtgBenC: i think its a bad merge, but OMG thats getting to be an ugly structure.15:38
apwthe point of trying these things it to find out what about the machine is broken so youwill be able to run the newer kernel15:38
matiassunderstand.15:39
matiassbut you mean the hardware has a problem? i think is incompatibilty15:39
matiass?15:43
matiassapw: what do you mean?15:48
apwno i mean there is something about your hardware which triggers this issue in the new kernel15:49
apwto find the bug, we need to find the difference that triggers the failure15:49
amitkBenC: do you expect to upload today?15:49
apwso  finding out a 2.6.27 virgin tree boots ok tells us its not in any of those commits15:49
apwits likely in our local modifications, a much smaller set15:50
apwbut the key is it splits the search space in two and eliminates one of them15:50
amitkBenC: build testing all the arm configs on my machine is going to take 6+ hrs. Should I push them out first?15:55
amitkBenC: configs pushed, but not yet tested. Builds ongoing16:02
rtglamont: do you know if the special compiler restrictions still apply to hppa for for the Intrepid ports build, e.g., gcc-4.1-hppa64 ?16:03
lamontrtg: that'd be a doko question16:04
rtgok, perhaps a better question is, has anyone run a .27 kernel with hppa?16:05
lamontit's gonna definitely need gcc-${mumble}-hppa6416:05
rtgwhats kyle doing with it these days?16:05
dokortg: which restrictions? it would be nice to use the 4.3 compiler if possible16:06
rtgdoko: well, thats kind of what I'm asking. does the hppa .27 kernel build work with 4.3 ?16:07
dokortg: I didn't check16:07
rtgdoko:  ok, I'll work with lamont to get the right tool set.16:08
lamontrtg: see also other window16:09
BenCamitk: is it just the configs or do you have all the arch setup in there too?16:24
Kanohi rtg , could you fix the 2.6.28 git? there are double fixes included16:36
Kanofor example acbdf3b2fb51c8d197b1cca5dd4d0f54add69d6d16:36
rtgKano: its still a work in progress. BenC and amitk are in the middle of it right now.16:37
Kanobtw. i like that you now use the complete 2.6.27.x fixes not only parts16:38
rtgyeah - it didn't make sense to me to ignore all those good bug fixes.16:42
Kanowell for me too ;)16:42
Kanoi disliked the selected fixes16:42
NCommanderdoko, I'm working at removing 4.1 from the ports kernel (I haven't done HPPA, but PowerPC and SPARC are liberated :-))16:57
ckingrtg: did you know of any Wifi issues with the Ralink RT2571WF PCI-E Mini (driver rt73usb) - I have somebody report it cannot connect reliably with WPA17:00
Kanocking: well for rt there are often 2 usb drivers available17:00
rtgcking: is it the OEM driver?17:01
Kanoalso pci-e is never rt73usb17:01
Kanonext, the official driver for ralink needs a change if you want to use wpa supplicant17:01
mjg59cking: There's a clue in the name17:01
mjg59Hm. I guess you /could/ put a USB wifi card on a mini PCI-E slot.17:02
mjg59Would seem a little strange, but still.17:02
ckingwhoaa.. hold on.. 17:02
ckingrtg: which one is the OEM version17:02
cking..this is hardy by the way17:03
rtgcking: oh17:03
rtgI can't remember if we packaged the OEM driver.17:03
ckingKano: I would agree with that initial statement, but lsmod tells me different.. hence I am mystified17:03
rtgsmb_tp worked on the rt25xx backlport17:03
Kanoit works better with ndiswrapper anyway17:04
ckingrtg: OK - thanks17:04
Kanojust not with draft-n17:04
aboganismb_tp: I did it.17:09
aboganismb_tp: Do you want a "request pull" through ml?17:10
Kanortg: do you use all bugfixes for 2.6.24 now too?17:13
rtgKano: smb_tp is working on it. I have not looked at the git repo lately17:13
smb_tprtg, That is rt286017:53
ZhenechIs there any reason, linux-libc-dev does not ship video/uvesafb.h?20:14
haydnI know this might be off topic, but could some quickly help me? I need know how to install kernel 2.6.24 in an Ibex install. I'm trapped by the busybox intramfs timeout error! Please no one else can seem to help.20:33
amitkBenC: all the arch setup is in there too22:15

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