/srv/irclogs.ubuntu.com/2011/06/13/#ubuntu-kernel.txt

quentusrex_Has anyone run into the bug with realtech nic cards where they lose inbound traffic when under load?00:04
quentusrex_I have reproduced the issue with 3 different realtech nics and when the CPU has less than 50% idle I get over 10% loss of all udp traffic.00:06
=== jk__ is now known as jk-
=== panda is now known as Guest38046
* apw yawns07:54
jk-hey apw07:54
apwjk-, morning07:55
jk-apw: fairly early there, no?07:55
apwjk-, just  before 8 yeah07:55
micahghi, is natty not supposed to provide an automatic upgrade form 2.6.39 to 3.0?07:56
micahgoops, I meant oneiric07:56
apwmicahg, yes it will once we are sure we've gotten all the issues sorted out07:57
apwlike it renedering your whole machine unupgradable07:57
micahgah, ok :), just wanted to make sure it was known and not an oversight, thanks!07:58
apwmicahg, yes its known, the kernel packages are in but not the meta for the moment, while we confirm that the required fixes are out07:58
micahgk, great, will leave you people to work now :)07:59
apwmorning ppisati 08:48
ppisatiapw: morning08:53
apwppisati, how goes the CVE for arm slog?  08:55
apw(slog == slow never ending horrible)08:56
ppisatiapw: i took a break, i think we have enough kernels to test for now :)08:59
apwthat is for sure08:59
apwthey must hate you in #u-arm08:59
ppisatiapw: never said it, but i think so :)08:59
apwcking, morning09:14
ckingapw, hiya09:36
apwcking, how ya doing09:36
ckingnot so bad -  however I've just had to pick myself up from the floor after hearing the cost of fixing my poorly car09:38
apwwoh, how much09:38
cking..replacement gearbox and possibly replacement clutch = 4 figure sum09:39
=== cjwatson_ is now known as cjwatson
apwcking, oww09:40
ckingthe car is worth about the same amount, but to replace it will cost a load more, so I think it's worth it for another 3-4 years on the road09:41
apwcking, hard decision indeed09:41
apwsell one of your kids, that'll offset it09:41
ckingheh, I have to pay for somebody to take them off me ;-).  Considering how much one pays for tax and insurance, it's not that much over 3 years.09:42
* cking punches mumble into submission09:43
bullgard4System Monitor > Processes > tab "Waiting Channel": What is a »waiting channel«? I found: "The address of an event on which the process is waiting”. This definition seems to be poor as I hardly would call for example a poll_schedule-timeout an "address". Can you give a better definition?09:50
apwbullgard4, well that string is the name in the symbol table for that address, so it is literally an address, sometimes symbolically represented09:56
bullgard4apw: Ah, understood. --  Thank you very much for your help.09:59
nigelbwho posts to the kernel team blog?10:43
nigelbor rather who among the people posting to kernel team blog are ubuntu members :-)10:43
apwnigelb, i would think that everyone who posts to the kernel team blog (on voices yes) is an ubuntu member, why ?11:04
nigelbapw: http://nigelb.me/ubuntu/2011/06/10/cleaning-up-the-planet.html11:05
apwnigelb, are we needing to have an owner for it then?11:06
nigelbapw: yes11:07
nigelbapw: Just an Ubuntu member to own the feed11:08
apwnigelb, well i guess either pgraner, tim or myself are the obvious choices11:08
nigelbapw: whats your Launchpad ID? I'll put your ID in :-)11:08
apwnigelb, apw11:09
nigelbapw: heh, after you guys kicked pgraner to QA, he's not obvious choice anymore :-P11:09
* nigelb hides11:09
* apw gets hit boxing gloves on11:10
jwiyay, regression in lucid-proposed kills intel wifi :/. bug 79633611:14
ubot2Launchpad bug 796336 in linux "2.6.32-33-generic iwlagn firmware broken" [Undecided,Confirmed] https://launchpad.net/bugs/79633611:14
jwiand it seems to be a stupid one-liner that tgardner had doubts about months ago11:14
apwjwi, could be, i've retagged it appropriatly11:16
jwii'm pretty sure it is - from reading the code, lucid just doesn't have the code to handle the new firmware file format11:17
jwiso commit "iwlagn: Support new 5000 microcode." turns into "break 5000 stuff in funny ways"11:17
apwand what was the one liner that rtg wasn't happy with ?11:18
jwithat commit11:18
jwisee https://lkml.org/lkml/2011/4/27/41611:19
jwi(and yes, it's in maverick-proposed too. that shouldn't break though, the maverick stack seems to be capable of handling the new format)11:21
apwjwi, having read all that i am unsure why this would even affect lucid11:24
apwjwi, we have the old firmware still11:24
apwand the old driver shouldn't even see the new one11:24
jwiapw: ah, thats because iwlagn is broken in multiple ways :)11:24
apwjwi, and they are11:24
jwiit reads the api version from the firmware header - and with new firmware format, that version is 0 (which is < API_MIN = 1). now, instead of trying the other available firmware files, it just quits.11:25
jwiso thats why falling back to the older firmware files doesnt work - this should be fixed for maverick+, too11:26
apwok so could you document that in the bug, that'd make things easier11:26
apwsounds like the new firmware is a mistake in lucid given all this, hmm11:28
jwiright, i was hoping to get some confirmation first from intel on whether that firmware actually *has* the new format11:28
jwinope, the firmware is fine. it's the driver that shouldnt claim to support it - because it doesnt11:28
apwjwi, but then as the driver doesn't support it one could argue that its pointless having it on the machine11:29
jwicompat-wireless...11:29
apwwhich if it needs it should carry the firmware there and avoid the mai11:29
apwmain kernel modules seeing it11:29
jwidefinitely - but i don't think that's the standard procedure for compat-wireless?11:30
apwhmm pretty sure its meant to carry its own firmware, and even has hacks to rename them to ensure non-overlap (the ubuntu packaging thereof)11:31
jwiapw: hm, see tgardner's #2 in bug 653854 (which is currently sitting in lucid-proposed).11:33
ubot2Launchpad bug 653854 in linux-firmware "Oct 3 06:01:05 bt kernel: usb 1-7: ath9k_htc: Firmware - ar9271.fw not found Oct 3 06:01:05 bt kernel: ath9k_hif_usb: probe of 1-7:1.0 failed with error -22" [Low,Fix committed] https://launchpad.net/bugs/65385411:33
apwjwi, i think we've tended to do that as now we have more than one lbm and we'd have to put the firmware in them all if its not compatible11:34
apwbe good to confirm it is a format issue, we may be able to short circuit the load in lucid kernel11:35
apwas an expedient fix11:35
jwiapw: alright, poured my guesstimates into the report. i'll follow up if there's any further info from intel, but i'm pretty sure that this is what's going on11:55
apwjwi, sounds good thanks, will slap people when they wake up11:55
apwcjwatson, i am thinking that the 3.0 kernel should have an install dependancy on the libc changes you made, given how hard it was for me to get out of the mess, i therefore plan to add those and re-upload before shipping linux-meta ... am i making sense12:12
cjwatsonBreaks on older versions would be more appropriate.12:13
cjwatsonalthough, actually, I don't see why you need to bother12:14
cjwatsonit only breaks when you try to upgrade libc to a newer version after rebooting to the new kernel12:14
cjwatsonanyone upgrading libc to a newer version at this point is almost certainly going to be upgrading to a version with my fix anyway12:14
cjwatsonso I would just ignore it if I were you12:15
* apw thinks about that ... thats a fair point, so perhaps i will ignore it, thanks12:15
cjwatsonand it won't affect dist-upgrades from older releases since they're unlikely to reboot between upgrading the kernel and upgrading libc (even if the upgrade were somehow to an unfixed version)12:15
cjwatsonyou'll need to think about it for backports to lucid, but maybe by then it will be 3.0.x anyway12:15
apwyeah i suspect so12:16
ppisatidunno if it's more pain to compile a kernel on a panda board, or roll your own cross toolchain12:24
amitkppisati: why do you need to do either?12:25
ppisatiamitk: trying to see if i can reproduce a problem we have with latest linaro toolchain12:26
amitkppisati: aah, in that case it is probably easier to native compile the kernel on panda if you have a usb disk connected12:27
ppisatiamitk: doing both of them12:28
ppisatiamitk: and somehow my panda doesn't like my usb disk12:28
ppisatiamitk: i hopne it's not the disk that is failing12:28
diwicapw, do you have a moment?12:55
apwdiwic, yep, wassup13:02
diwicapw, I'm trying to follow rtg's instructions but I fail. There is an email ~ 2 weeks back about that I should use "git cherrypick -s -x" but it seems to fail13:03
apwok how does it fail ?13:03
apwcherry-pick yes 13:03
apw-s to sign, and -x for the upstream reference in the pick13:04
diwiclet me pastebin it for you13:04
apwack13:04
diwichttp://pastebin.canonical.com/4840513:04
apwdiwic, so there are conflicts it seems?13:05
apwdiwic, which tree are you in ?13:06
apwthat by its name would be an upstream tree no ?13:06
diwicapw, linux-2.613:06
diwicapw, but feel free to suggest another one if you think that's more appropriate13:06
apwi'd be expecting you to do the cherry-pick in the one you want to apply it to ?13:06
diwicapw, for reference, here's his message: https://lists.ubuntu.com/archives/kernel-team/2011-May/015710.html13:07
apwso i assume you made all three originally with a git format-patch -3 sort of thing13:08
diwicapw, sort of13:08
apwhe is suggestng you are missing the s-o-b and (cherry picked from xxxxx) lines in the first two13:08
apwand the easiest way to get those is just to cherry-pick the patch13:09
apwwhich you do in the destination branch13:09
diwicso I must first apply them there and *then* cherry-pick them, right?13:09
diwicand I apply them with git am?13:10
apwnormally these are upstream commits we want to cherry-pick and as long as you have some remote with them applied in the current repo13:11
apwyou can simply checkout the destaination brnach and git cherry-pick them into it13:11
apwthe cherry-pick finds extracts and applied them in one motion13:11
diwicso cherry-pick moves them between two branches in a repo13:12
diwicbut isn't this two different repos, one linux-2.6 and one ubuntu-natty repo?13:12
apwwell normally for me at least the former is the --reference'd repo in the latter, and so all of its sha1s are available in ubuntu-natty13:13
diwichrm13:16
diwicbut I don't want to change my ubuntu-natty (I want it to track the one on kernel.ubuntu.com), so I start off by cloning it13:17
diwicor should I clone with reference there as well?13:17
diwicI have the same reference btw13:18
diwicand suppose I actually succeed with making these cherry-picks, should I then use format-patch to be able to send them to the list?13:23
apwdiwic, you can safely clone with reference therre as well13:32
apwand yep after the cherry-pick you would format-patch as normal13:32
apwsforshee, hey ... morning13:59
sforsheeapw, hi13:59
apwsforshee, i think you were involved in the linux-firmware updates for 'all releases' ?13:59
apwseems we have a regression due to the iwlagn firmware on lucid-proposed13:59
apwhttps://launchpad.net/bugs/79633614:00
ubot2Ubuntu bug 796336 in linux "2.6.32-33-generic iwlagn firmware broken" [Undecided,Triaged]14:00
sforsheeapw, I thought the only thing that was updated was ralink firmware14:00
apwsforshee, the rumour is that it contains also iwlagn abi#5 firmware ... perhaps you could investiate and work out what the reality it ... believe nothing i have said as its all rumours14:01
apwbut that bug is enough to prevent linux-firmware being promoted to -updates till we have it resolved one way or the other14:01
ppisatihow do i request a pocket copy?14:06
ppisatilinux-fsl-imx51 2.6.31-609.26 from c-k-t ppa14:07
ppisatii guess an email it's enough14:07
jwiapw: its about the kernel thats in -proposed, not linux-firmware. the api-5 ucode was moved to -updates back in april ...14:12
apwsforshee, down't worry about that issue, i'll look at it, seems its not at all what i thought originally14:13
sforsheeapw, ack14:14
chadhoggin hopes that there are more eyes here on a Monday morning than a Sunday afternoon I'll repeat my question from yesterday once, but then not bother you again:14:20
chadhoggI apologize if this is the wrong venue, but I was hoping to get some real-time feedback on what I could do to continue debugging Launchpad bug 79343714:20
ubot2Launchpad bug 793437 in linux "Unusable Slowness In 2.6.38-8" [Undecided,Confirmed] https://launchpad.net/bugs/79343714:21
diwicchadhogg, my best guess would be https://wiki.ubuntu.com/Kernel/KernelBisection14:31
diwicchadhogg, using maverick and a Natty kernel on top of that could also be interesting b14:31
diwicperhaps14:31
chadhoggdiwic: thanks; that looks complicated, but I'll give it a try14:33
* diwic just sent his activity report to the public kernel-team ML...fortunately there was nothing NDA in it14:38
apwogasawara, i am liking [Acked]15:07
ogasawaraapw: I am too15:07
apwi shall take that on15:07
ogasawarabjf: you're in london tomorrow right?15:31
ogasawarabjf: I was gonna get the IRC meeting notes prepped15:32
bjfogasawara: i'm there now15:33
bjfogasawara: i could probably run it from here15:33
bjfogasawara: you want a copy of my runes?15:34
ogasawarabjf: sure, just in case.15:34
ogasawarabjf: so I'll let you handle it tomorrow but I'll be on standby15:34
bjfogasawara: wfm, runes sent15:37
ogasawarabjf: thanks15:37
apwsconklin1, looks like we have an iwl5000 regression in your -updates kernel, see bug #796336, it appears very much that the patch 'iwlagn: Support new 5000 microcode' which we got from stable upstream is completely inappropriate without the new firmware loader support15:58
ubot2Launchpad bug 796336 in linux "2.6.32-33-generic iwlagn firmware broken" [Undecided,Triaged] https://launchpad.net/bugs/79633615:58
sconklin1apw: thanks. I'm getting used to this, unfortunately. We have never had so many things broken from stable updates in as long as I can remember . . .15:59
apwsconklin1, updates kernel for lucid15:59
sconklin1yep, got it16:00
apwwell we've never tested so well either16:00
apwso ... its hard to be sure how much is new and how much is normal16:00
sconklin1fair point16:00
apwbe good if you could confirm that lucid cannot real those files ... i think not given the error and the code, so i think reverting that commit is appropriate and prolly report it to stable too16:01
apws/real/read16:01
sconklin1I'll investigate16:02
=== sconklin1 is now known as sconklin
* ogasawara back in 2016:04
* ppisati -> gym17:29
* jjohansen -> lunch19:50
=== yofel_ is now known as yofel
=== PascalFR is now known as boxershort
sconklinGrueMaster: I'm setting up the workflow tools to assign ARM testing to your new team. Should this happen for all of the following packages? linux-mvl-dove, linux-fsl-imx51, linux-ti-omap422:04
=== bryceh is now known as bryce
=== bryce is now known as bryceh
=== bryceh is now known as bryce

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