/srv/irclogs.ubuntu.com/2012/05/30/#ubuntu-kernel.txt

=== smb` is now known as smb
* smb mornings07:06
ppisatimoin07:35
cooloneyppisati: moin,07:40
smbppisati, cooloney morning guys07:46
smb(or whatever time)07:46
cooloneysmb: hey, man07:47
tgardnerapw, so, are you reworking 'UBUNTU: [Config] linux-image-extras needs full postinst' ?12:21
apwtgardner, yep, have it here, forgot to send it out :/12:22
tgardnerapw, http://tech.surveypoint.com/make-grub2-boot-the-last-operating-system-you-used.html13:43
apwtgardner, thats not quite the same ... hmm13:45
tgardnerapw, not quite, but there are some clues there13:45
apwtgardner, i am sure its a doable indeed.13:49
apwogasawara, ok i've enhanced the config checker so it basically has teh same predicate language for configuration settings ... i've made a preliminary alpha1 review page for comparison: https://wiki.ubuntu.com/KernelTeam/Specs/QKernelConfigReview/Alpha114:17
ogasawaraapw: thanks, /me will review14:18
ppisatiapw: can you send the script used to generate the matrix?14:18
ppisati*send me14:18
apwppisati, its all in kteam tools, you need the before release and after release git repos, both checked out at the branch tip you want to compare14:19
ppisatiapw: ack, what's the name of the script?14:19
apwthen you do like .../kteam-tools/devel/devel-config-summary ubuntu-precise ubuntu-quantal >SUMMARY14:19
apwppisati, but let me push the latest14:20
apwppisati, ok pushed14:20
ppisatiapw: so in my case, to do a comparison between master and omap4 branchesi need two dirs with Q/master and Q/omap4, right?14:20
apwppisati, to compare a non-standard branch with a standard b14:20
apwbranch, you can checkout the say ti-omap4 at the tip you want, and then fdr genconfigs, then checkout master-next again and regenerate14:21
apwand you will get the configs mixed together14:21
ppisatiuhm, ok14:21
ppisatilet me try14:21
hallynSo I've got this lenovo s10-3.  since start of maverick, it hasn't resumed from suspend, and Len Brown basically threw his hands in the air saying the bios is doing something wrong.  Now, as of a few weeks ago, the precise kernel seems to leave the wireless NIC in a bad state (broadcom) iwth phy0 rfkill hard off14:34
hallyni have to boot into lucid, suspend, resume, and then it's back up (until i boot precise)14:35
hallynanyone heard of such a thing?14:35
hallyni'm about to resign myself to running lucid on it... (or holding down paper with it)14:35
tgardnercking, didn't you work on this one ^^14:37
ckingtgardner, I can't recall any fix I worked on for that machine14:37
apwppisati, success ??14:37
tgardnercking, the s10 sounds like a model the hwe worked on.14:37
apwvanhoof, ^^14:38
ckingbah, laptop just died14:40
apwcking, you need to fix that14:40
ckingapw, it's called get a new one14:40
ckingapw, henrix, shall I check for flights for the sprint then?14:47
henrixcking: hmm... actually, not sure yet as i may be flying from lisbon14:47
ckinghenrix, ack14:48
henrixcking: anyway, i'll let you know later today (or most probably tomorrow)14:48
henrixcking: and thanks for asking14:48
ckinghallyn, I've see another S10-3 which only  suspend/resume on worked on natty14:52
apwcking, yeah i may want to go out earlier will find out tonight14:52
hallyncking: for awhile in maverick/natty you could make suspend/resume work (not 100% reliably) by not using intel_idle.  but that stopped.14:52
hallynstill, not resuming was a pain but not a showstopper.  no wireless is more of a showstopper.14:53
ckinghallyn,  apparently using kernel parameter "nohpet" seemed to have helped that particular user14:53
hallyni just don't understand why suspend/resume unblocks wireless14:53
hallynright.  nohpet never worked for me, but i've seen that.14:53
apwhallyn, bios is run during those two14:53
hallynwhy isn't bios run during boot?  :)14:53
apwhallyn, literally _anything_ can happen14:54
hallynoh, the kernel driver must muck it up?14:54
apwhallyn, some bits of it are, but who knows if its working right14:54
apwhallyn, likely the kernle is doing what the bios asks, and that is not working14:54
ckingyou did say it was broadcom wireless- I suspect that's the binary blob wl driver 14:54
apwhallyn, which chipset is it?  perhaps brcmsmac will work better for you14:55
hallyncking: i could try blacklisting that (thought about it( but the brcmsmac driver hangs my laptop completely14:55
hallyninteresting!14:55
hallynin lucid it reports a different model # :)  Broadcom Corporation Device 4727 (rev 01)14:55
hallynpretty sure in precise kernel it said 43xx (not sure what the xx were )14:56
hallyni'm upgrading the precise partition under chroot right now, hoping for a hail mary14:56
apwhallyn, that prolly means its a pciid 4727 and lucid doesn't know it14:57
tgardnerhallyn, have you tried the Q kernel ? sforshee has done good stuff for brcmsmac. https://launchpad.net/~ubuntu-x-swat/+archive/q-lts-backport14:57
* cking notes it could be a collection of issues, wifi, firmware, etc, so teasing it out with a Q kernel may be a good starting place14:59
hallynok, i'll try the q kernel (if i can successfully upgrade under chroot)15:00
hallynthanks15:00
ckinghallyn, if you are feeling adventurous and can bear to download 650M of .ddeb, you can also try: https://wiki.ubuntu.com/Kernel/Reference/S3SystemTapDebug15:01
ppisatiapw: cut&pasting all the config diff is such a PITA...15:04
apw?15:04
apwppisati, from where to where ?15:04
hallyncking: probably worth it, will try, once wifi is up15:06
ppisatiapw: after config generation into a wiki page15:08
apwppisati, editmoin ?15:08
ppisatiapw: do you have any smarter way to get the data there?15:08
ppisatiapw: never used15:08
apwthats the smarter way 15:08
* ogasawara back in 2015:09
apwogasawara, where are we with alpha1 kernel, i am doing some config review goop, and wondering if you'd want that in or out of a115:16
tgardnerapw, I've been pushing config review related changes15:21
apwtgardner, ok ta15:23
ogasawaraapw: push it, I'm gonna upload our Alpha-1 kernel first thing Monday15:29
apwogasawara, will do, a _few_ more do to sign15:29
apwsigh15:29
apwsmb, CONFIG_XEN_ACPI_PROCESSOR ... any idea why that is on for amd64 and off i38615:47
smbapw, I think it was changed because we don't really support a dom0 on i38615:47
smbhrmpf15:48
smbwhich was true last release at least but I just did apt-cache search...15:49
tgardnerapw, smb: I don't think we want to support an i386 dom015:50
hallyncking: well i've got deb http://ppa.launchpad.net/ubuntu-x-swat/q-lts-backport/ubuntu precise main, but i'm not getting the kernel from that15:51
apwhallyn, you'd have to opt in ... tgardner whats todays meta packag for the lts kernel ?15:52
hallynd'oh15:52
tgardnerhallyn, apw: one of the packages in  linux-meta-lts-backport-quantal15:53
tgardnerlinux-image-generic-lts-backport-quantal15:53
hallynthanks.  i'd tried just 'linux-lts-backport-quantal' but tha didn't help15:53
tgardnerhallyn, it didn't help because its not a package ?15:54
tgardnerhallyn, p.s. the name of the meta package is going to change. rick doesn't like 'backport' in the name15:55
hallynright i guess it's the source pkg name?  anyway i'm now tethered over my phone to wifi so have a wider pipe. installing, thanks :)15:55
tgardnerhallyn, do you normally live at the end of a 19K baud serial line ?15:56
hallynhow i wish for 19K!  300, acoustically coupled15:57
tgardnerwhat a dinosaur15:57
smbtgardner, apw Somehow I would think at least the options should be right if there is a i386 hypervisor around. Which I thought was not but seem to be wrong15:57
hallynall right not really.15:57
tgardnersmb, -ENOPARSE, huh?15:58
apwtgardner, he is saying even if we don't support it its there and the options should be right in it15:58
smb*sigh* yeah, what he said15:59
tgardnerapw, agreed. I just don't wanna support it regardless.15:59
tgardnersome crazy outfit will bet their business on 32 bit dom0 using 10 year old Dell servers.16:01
smbzul, Can you refresh my memory why the xen source package is in main but the binaries seem all to be in universe...16:01
zulsmb: the hypervisor?16:01
smbtgardner, There probably already are them doing it with kvm and openstack16:01
smbzul, that and the utils too16:01
smbzul, at least16:01
zulsmb: because libxen is in main since its a dep for libvirt the rest isnt in main16:02
tgardnersmb, perhaps, but we don't have to give them more ways to hang themselves.16:02
smbtgardner, if I am not just going mad (more than the usual) we already have the config in a way to enable it on i386 (at least precise)16:08
tgardnersmb, I'm sure we do, but I want it dropped by the time 14.04 is released. no better time to cut those ties then right now.16:10
hallynsigh, that didn't give me a nic at all.  fraid i need to put it aside for a bit.16:11
tgardnerhallyn, wtf kind of nic do you have ?16:12
smbtgardner, apw I would not want to make that decision just factually. Right now a different setting for the acpi processor in quantal sure does not make sense with the other options being what they are16:12
apwsmb, yeah i would say i'll harmonise them to stop the errors i am seeing16:12
apwand you can argue with server and get it gone16:12
hallyntgardner: broadcom somesort16:12
tgardnerhallyn, you're gonna have to get a dmesg before we can figure out whats happening16:13
smbapw, Right, if that makes sense. Just that this is sort of a new decision/new target.16:13
hallynyeah, if i don't throw it out the window first, i'll open a bug with dmesg output and such next time i turn the thing on16:14
hallynreally would be nice to have it working, but since i still have no hope for resume anyway, it'll never be what it once was :)16:14
apwhallyn, if you are going to throw it out send it to sforshee16:14
rtg_smb, there is no CONFIG_XEN_ACPI_PROCESSOR in precise.16:15
smbrtg_, There should be. At least it is in what I got in master. And what I have been testing in -proposed16:16
apwppisati, do omap boards have 8250 serial ?16:17
hallynapw: is he building a beowulf cluster of netbooks?16:17
apwhallyn, no but he has a fine hammer16:17
smbrtg_, 15ca7f9c2a2dd8fdf263963c57504d0da6fbd84e16:17
hallynapw: it's my best laptop to use in the car, so i'm all talk - would more likely pick up a $15 wireless-n usb stick16:18
rtg_smb, ok, it would help if I was looking at the _tip_ of master-next.16:20
smbrtg_, or tip of master now16:20
apwhallyn, i was thinking he could try and fix it, is easier with the device in your hand16:20
smb(its the change that is currently under verification)16:20
* apw pokes ppisati16:21
ppisatiapw: there's an omap serial config, dunno if it exploits 825016:21
ppisatiapw: let me check16:21
rtg_smb, so I don't understand your objection to dropping this in quantal 32 bit. 16:21
smbrtg_, The objection is that it does not make sense to drop that single bit from 32bit then16:22
rtg_smb, thats circular.16:22
apwrtg_, i think he means you have to turn off all of XEN host in 32bit or this option should be harmonised even if we offer no support16:23
smbrtg_, Not when it is set y in common config16:23
smbin precise16:23
apwogasawara, do yo u have a list of options you dinged for build failures on arm back in the day ?16:24
rtg_apw, so what we _shold_ do for quantal is CONFIG_XEN_DOM0=n16:24
smbrtg_, But that is actually a new decision16:24
ogasawaraapw: I'd have to grep the logs, just a sec.  I did give them all similar commit subjects so I could find them again.16:25
rtg_smb, so? we make new decisions about config options all the time.16:25
smbWhich has not been discussed, asked for, yada, yada16:25
ogasawaraapw: 1bd2e66f5ab6063b153e7638453d0bcf719f842f UBUNTU: [Config] Temporarily disable CONFIG_AX88796 on arm16:26
ogasawaradf954481918bae6b7f93320b1be5fc0b6db42192 UBUNTU: [Config] Temporarily disable CONFIG_TI_CPSW on arm16:26
ogasawarabe637dbbc9c3cf51f91efec7236bef1f054c4a2f UBUNTU: [Config] Temporarily disable CONFIG_USB_EHCI_HCD_PLATFORM on arm16:26
ogasawara32e2f0b7eef1c46494ef4244646bccaa34d3b0e8 UBUNTU: [Config] Temporarily disable CONFIG_LIS3L02DQ on arm16:26
ogasawara37962f948cb20ee7fc1def0a3a8bff40abddc524 UBUNTU: [Config] Temporarily disable CONFIG_MFD_OMAP_USB_HOST on arm16:26
ogasawarab1dad7778d5965bd81499ad7f44214453e5a94d6 UBUNTU: [Config] Temporarily disable CONFIG_EZX_PCAP on arm16:26
ogasawara59e1b9557ddc329e6181891ef4716ec4f2c6e1b2 UBUNTU: [Config] Temporarily disable CONFIG_TOUCHSCREEN_EGALAX on arm16:26
ogasawarab45ba114bef30fbdd10c57309963f8774e8f01a9 UBUNTU: [Config] Temporarily disable CONFIG_TOUCHSCREEN_EETI on arm16:26
apwrtg_, that is cirtainly an option indeed.  for me i would just leave it on and docuemnt it unsupported ...16:26
smbrtg_, What I want to say is that we should bring this topic to next uds and then move forward and not just do it in q16:26
apwif we are going to change it, we should at least start a thread about it16:27
smbI guess at least that. Otherwise (or even with it) it ends up as the we drop non-pae16:27
ppisatiapw: totally different stuff (omap_serial vs 8250)16:28
ppisatidrivers/tty/serial/omap-serial.c16:29
* smb goes off to have a pint of fermented apple juice...16:29
apwppisati, ok so that doesn't need to be builtin for arm* then yes?  8250 serial drivers ?16:30
ppisatiapw: nope16:31
ppisatiapw: actually i'm talking about omap here16:31
apwppisati, as am i, omap only16:31
ppisatiapw: if you are doping another arm soc, it's a totally different thing16:31
ppisatiok16:31
apwomap3 only indeed16:31
apwmaster branch only16:31
ppisatiack16:31
apwppisati, thanks16:32
ogra_do we have an equivalent to the linux-version tool from debian ? (from linux-base)16:35
apwogra_, no but that one should be installable now after tgardner fixed it16:35
apwogra_, as to whether it works ... is another question, if not poke me16:35
ogra_ah, great, seems the new flash-kernel uses it 16:35
ogra_yeah, i got f-k on my TODO for tomorrow to fix the rough edges16:36
apwogra_, ok let me know, i need some u-space to play with16:36
ogra_k16:36
hallyntgardner: http://paste.ubuntu.com/1015033/ has dmesg and lspci -v output fwiw (with q-backport kernel)16:37
rtg_ogra_, all I dropped from linux-base was /usr/bin/perf. /usr/bin/linux-version should still be intact.16:37
apwrtg_, my memory is it may need ubuntuisation, but ogra_ can test that, and if it does we can fix it16:37
ogra_rtg_, awesome 16:38
ogra_yeah, i will work through the bits and pieces tomorrow and let you guys know if i stumble16:38
rtg_apw, possibly, but thats a different problem.16:38
hallynoh that's before i modprobe brcmsmac16:39
rtg_hallyn, why do you have to modprobe it? do you have black lists ?16:39
apwhallyn, do yo have wl installed, that blacklists brcmsmac by default, and bcma which it needs16:39
rtg_hallyn, what apw said. plus you may need linux-firmware-nonfree16:40
hallynoh riht that might have happened, i can't check right now bc after i tried to soft-unblock phy0 (with fn-f5) it hung16:40
hallyndo i need linux-firmware-nonfree only to use wl, or also for brcmsmac?16:41
rtg_hallyn, just for brcmsmac16:41
hallynthese names just kill me.  is brcmsmac supposed to replace b43?16:48
apwhallyn, i believe so yes16:49
hallynactually bcma-pci-bridge is what has started auto-modprobing16:49
hallynheh, but no change.  when i unblock phy0, the thing hangs16:49
hallyncking: d'oh, i assume ddebs.ubuntu.com won' thelp me when i'm running the q-backports kernel16:56
rtg_hallyn, correct16:57
ckinghallyn, yep16:57
dilekshttp://packages.ubuntu.com/precise/all/linux-firmware/filelist17:18
dilekshttp://packages.ubuntu.com/quantal/all/linux-firmware/filelist17:18
dileksare there missing fws for ath6k?17:19
bjfrtg_: when you get some free time, gomeisa could use htop17:23
rtg_bjf, done17:24
apwdileks, there appear to be fewer in quantal, presumably we don't need them there as we only have 3.4 kernels ... rtg_ ^17:25
ckingbjf just wants to see how hard he is driving it17:25
* bjf likes to make her squeal17:26
rtg_dileks, right, I removed some older firmware files. did I get overzealous ?17:26
dilekscant test as I have no ath6k wificard. I was asking bwh whuzzup on debian as ath6k was activated recently for trunk.17:28
dileksis linux-firmware the correct place or should it be in -nonefree pkg?17:29
dileksI mean whats the criteria to put fws in the one or other pkg?17:31
rtg_dileks, its must be redistributable17:32
jwirtg_: oh, that's a nice surprise for folks bisecting through older kernels...17:33
dileksif its in linux-firmware.git its "freely redistributable"?17:33
rtg_jwi, we don't support kernels older then 3.2 on precise.17:34
rtg_dileks, yep.17:34
jwirtg_: i think the ar9170* are no longer used (replaced by carl9170); ar7010 looks fairly old too (htc_7010 now?)17:36
dileksrtg_: thank you for enlightenment17:37
rtg_jwi, I mostly went by the MODULE_FIRMWARE statements in the kernel. there may well be some old cruft that I missed. 17:37
rtg_I should get back and finish up a more thorough review17:38
rtg_I was just looking to save some space on the LiveCD17:38
apwogasawara, ok i've updated the summary a few more times, added a 'BUILD FAILURES' section for the ones you t17:43
apwturned off17:43
ogasawaraapw: ah nice17:44
ogasawaraapw:  so did you just hardcode the build failure ones?  or should I stick with the same commit subject so it can be auto extracted?17:45
rtg_apw, when using quilt in a  debian package it appears you cannot patch anything under the debian directory. Is that your experience ?17:45
apwrtg_, right you can just edit those is my understanding17:45
apwogasawara, they are listed in the overrides right now17:45
rtg_apw, thats what I've done after struggling a a bit17:46
jwirtg_: dropping support is fine; but missing firmware is gonna make it a little harder to even *run* an older kernel (for testing purposes, to investigate a regression, ...).17:48
jwibut i guess those folks know where to get the files manually17:48
=== rtg_ is now known as tgardner-lunch
* ppisati is off17:56
dileksis that apt-pinning really necessary?17:56
dilekshttps://wiki.ubuntu.com/Testing/EnableProposed17:56
dileksespecially -security and -updates17:57
dileksif there is a higher ubuntu-version, this will be taken17:57
apwdileks, i beleive that lets you not get proposed items you didn't specificially ask for17:57
dileksgiving -proposed 400 will not automatically update pkgs17:58
dileksnot using a preferences file will give -proposed 50017:58
apwdileks, i believe the intent of those is to let you make -proposed optin like -backports now is, i have no idea if that will work as written17:59
dileks/etc/apt/sources.list.d/ubuntu-precise.list: http://paste.ubuntu.com/1015170/18:00
dileksI put that extra and partner repo stuff into /etc/apt/sources.list.d/ubuntu-precise-partner-and-extras.list18:02
dilekshttp://paste.ubuntu.com/1015175/18:02
dileksdunno why /etc/apt/sources.list.d/$file is not used18:03
dileksits much nicer than a big fat sources.list18:03
apwdileks, presumably cause all the code which fiddles with it from the UI would have to change18:12
apweven if ti could be cleaner and simpler now18:12
apwogasawara, am just buildtesting my config changes on x86 and then will push18:12
ogasawaraapw: ack18:12
apwogasawara, not bad for an afternoons effort, some 20 or so purples gone18:13
ogasawaraapw: nice18:13
apwthe new predicate based matching is helping get rid of 'em18:13
apwogasawara, more tommorrow :/18:14
dileksapw: what do you mean by "UI"?18:14
tgardnerogasawara, you won't send out the LTS kernel email  soon, will you ? I'm futzing with meta package names still.18:40
ogasawaratgardner: am coordinating with nick skaggs, we're thinking of sending out some sort of announcement next Thurs to align with Alpha-118:41
ogasawaratgardner: that enough time?18:41
tgardnerogasawara, thats plenty of time. thanks.18:42
apwdileks, the config manager thing for sources18:45
tgardnerogasawara, uploaded new LTS kernel and meta packages to Q-series LTS Backport with -backport removed from all names. they should appear in awhile.20:10
ogasawaratgardner: ack20:11
ogasawaraballoons: ^^20:11
tgardnerogasawara, gotta keep rick happy :)20:11
balloonsack, thanks20:11
ogasawaratgardner: heh, yah I informed nick we should strip all uses of the word "backport" from the announcement20:11
tgardnerogasawara, cool. I'll confirm everything looks good in the morning. I'm EOD.20:12
ogasawaratgardner: ack, I can test here too20:12
* tgardner -> EOD20:14
=== yofel_ is now known as yofel

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