/srv/irclogs.ubuntu.com/2010/08/26/#ubuntu-kernel.txt

=== bjf is now known as bjf[afk]
=== kentb is now known as kentb[out]
amitkmorning all07:18
ikepanhcamitk: good morning07:19
jjohansenmorning amitk07:21
amitkhi ikepanhc, jjohansen 07:21
ikepanhcjj never sleep...07:21
=== ivoks_away is now known as ivoks
amitksecurity people never can...07:22
amitkikepanhc: I hope you meant vacation instead of vocation in your email. Vocation is just more work :)07:24
ikepanhceh.....07:25
ikepanhcahley07:25
ikepanhcI must need more coffee07:25
ikepanhcnext time I will write "day off"07:25
ikepanhcI sent two mail because after push the sent, I found my phone number is wrong... seems not only number is wrong07:26
amitkheh07:28
smbmorning all08:37
RAOFGood morning.08:38
AceLangood morning08:38
RAOFHm.  Has the kms-disablement sauce for i8xx chips got lost somewhere in the Maverick changes?08:41
smbGuess we would need to use the SOURCE for that. :-P08:43
RAOFPerhaps I could phrase that in a slightly less passive way: Hm.  The kms-disablement sauce appears to have got lost somewhere in the Maverick changes.  Is this intentional?08:43
smbI probably can find out who committed the change, but I would not remember (if I ever knew) why.08:45
=== ivoks is now known as ivoks_away
smbRAOF, So those rather seem to have vanished in either a rebase or rework of the tree. I think the best option here is to ask ogasawara when she comes online.09:01
RAOFWill do.  Thanks.09:01
jjohansensmb: what do we need to do to kick the mainline builds?10:23
smbjjohansen, We should not need to do anything. Do we miss one?10:24
jjohansensmb: hrmm, I'm not sure.  I am trying to figure out what is meant by "due to the kernel-package in ubuntu being broken once again, I can't build myself"10:25
jjohansenhttps://bugzilla.kernel.org/show_bug.cgi?id=1671110:25
ubot2bugzilla.kernel.org bug 16711 in ext4 "Oops with 2.6.36-rc1" [High,New]10:25
smbjjohansen, Need to look at the bug, maybe meaning the source package...10:25
jjohansenright maybe10:26
smbjjohansen, Is he really comparing 2.6.35-rc2 with 2.6.36-rc1???10:27
jjohansensmb: I believe so10:27
jjohansenhe report 36 rc1 and rc2 fail10:28
smbcannot see 36 rc2 from the upper level (if not hidden in the links)10:29
jjohansensmb: yeah mean neither I assum he did something manual for rc210:29
* jjohansen questions the rc2 bit altogether10:30
smbWell on the rc2 side of it is 35 not 3610:30
smbat least in the text10:30
smbwhat he may mean is probably different10:31
jjohansenhis last comment he mentions both rc1 and rc2 oopsing10:31
jjohansenfirst comment .35-rc2 works10:31
smbah I see in the jpeg10:31
jjohansenyeah10:31
jjohansenI'll get him to clarify what is broken with the ppa10:31
jjohansenbetter than wasting time guessing10:32
smbI would not really be worried about a -rc1 oopsing,  I rather be surprised it boots at all10:32
smbI thought apw changed it to have patches10:32
smbbut let me check10:32
smbjjohansen, mainline build -rc2 is actually there10:33
jjohansensmb: well I am not particularly worried, I pretty much ignore anything before -rc310:33
jjohansenah cool, I haven't really poked10:33
jjohansenMy first concern was potential AA bug10:33
jjohansensecond was kernel packaging being broken in some way10:34
jjohansenI would much rather him do the bisecting on something broken than us :)10:34
smbNo I guess its "how can I compile the same"10:34
smbIt should be possible with git and the patches that now are added10:34
smbAndy removed the source package as that was often useless10:34
jjohansenhrmm, /me is going to have to look into the packaging changes sometime10:36
smbjjohansen, Basically from the BUILD.LOG you get the sha1 -> 76be97c1fc945db08aae1f1b746012662d643e9710:36
smb(which is Linus -rc2 commit in this case)10:36
jjohansenah, sha1 for upstream kernel is good :)10:36
smbThen there are now 5 patches to add debian packaging and configs and whatnot.10:36
jjohansengah, /me needs to look into overheating I've had my laptop shutdown twice today because of it10:38
jjohansenand the one time I had the cpu locked to low 800 MHz10:38
smbjjohansen, nast, you may be the hero of akgraner then. She got that too 10:39
jjohansenheh, only if I figure it out10:39
jjohansenat least its easy to replicate10:39
smbDo fans at least try to cope with it or do they remain on low speed?10:40
jjohansenjust let the flash player run :(10:40
jjohansenit seems they are remaining low speed10:40
jjohansenat least the time that I had the cpus locked to their lowest freq10:40
smbIf the temp goes up I would expect them to go faster10:41
jjohansenyes, I would too10:41
smbCan be fun to track that. I think there have been cases which were a sde effect of accessing the ec the wrong way10:42
jjohansenyeah, it can be bios (unchanged), ec, acpi, ...10:42
jjohansenevery stupid machine is different10:43
smbYes, sadly10:43
jjohansenThese are the frightening messages I get10:45
jjohansenAug 25 18:49:20 ortho kernel: [ 5178.678361] Critical temperature reached (128 C10:45
jjohansen), shutting down.10:45
smbAt least thermal shutdown happens at some time. Though 128 seems a tad too high10:46
jjohansenyeah, quite hot10:46
=== yofel_ is now known as yofel
=== smb is now known as smb-afk
=== smb-afk is now known as smb
jpdsls13:32
=== diwic is now known as diwic_afk
=== kentb[out] is now known as kentb
guest1I got a problem running kernel 2.6.35 in Ubuntu 10.0414:23
smb guest1 It might help to say what the problem is...14:26
guest1PS/2 Keyboard and PS/2 Mouse dont work with X window system after I boot with 2.6.35 (custom compiled).14:26
smbHave you tried the backport 2.6.35 driver that is pre-compiled? In the kernel-ppa I believe14:27
smbhttps://launchpad.net/~kernel-ppa/+archive/ppa has a linux-maverick14:28
smbActually I meant kernel when saying driver14:29
guest1No. Actually I have an AMD Athlon II 245 system. Ubuntu work just perfect with the default kernel i.e. 2.6.32-24-generic. I downloaded kernel 2.6.35 from kernel.org and recompiled. Everything works fine in text mode. After running X, KB and mouse wont work.14:30
* smb wonders why someone would compile a kernel when the default system works14:31
smbAnd using a custom compile you may end up with options set differently which mean probably some drivers not configured...14:32
guest1But I want to try Ubuntu with 2.6.35. Plz help.14:32
smbSee ppa14:32
smbThat is a 2.6.35 kernel14:32
=== simar__mohaar is now known as simar
guest1Ok. I will check that out. Thanks for the help.14:35
=== ivoks_away is now known as ivoks
=== JFo is now known as JFo-swap
=== amitk is now known as amitk-afk
=== bjf[afk] is now known as bjf
=== diwic_afk is now known as diwic
=== sconklin-gone is now known as sconklin
=== kentb is now known as kentb_dell
jcrigbytgardner, just got your email15:46
jcrigbytgardner, when is drop dead time for sending you different update15:47
tgardnerjcrigby, that really has more to do with the beta freeze rules from Linaro15:48
jcrigbyok15:48
jcrigbylet me look and see if I can figure out what went wrong with the merge15:48
jcrigbyI'll get back to you15:49
tgardnerjcrigby, ok15:49
jcrigbywhat is lirc btw?15:49
tgardnerjcrigby, infrared controller15:50
jcrigbyok, I saw some noise about that on kernel team list15:50
ogasawaraRAOF: hi, just want to confirm with you the kms-disablement patches you're referring to are the following 6 patches - http://pastebin.ubuntu.com/484007/16:14
ogasawaraRAOF: as they do indeed appear to have been dropped from Maverick, now to investigate/remember why16:15
jcrigbytgardner, I just did a git diff of my head and don't see any missing debian stuff.  http://pastebin.ubuntu.com/484025/16:57
jcrigbytgardner, sorry diff of my HEAD and Ubuntu upstream16:58
tgardnerjcrigby, I have 2 trees; Linaro-2.6.35-1003.7 and Ubuntu-2.6.35-19.25 checked out into separate directories. diff those to see what you come up with.17:00
jcrigbyok, maybe I sent you a bad pull request the linaro version should be 1004.817:03
tgardnerjcrigby, or maybe I forgot to 'reset --hard FETCH_HEAD'. doh!17:05
tgardnerdiffing again....17:05
jcrigbyI was wondering about that.  How to format a pull request that is a complete rebase17:05
tgardnerjcrigby, ok, my bad. I totally screwed the pooch. its looks good to me.17:09
jcrigbytgardner, wonderful17:10
jcrigbytgardner, that is great news17:10
jcrigbyand yes it did boot on beagle17:10
jcrigbytgardner, btw do you guys stop rebasing now17:11
jcrigbyfor maverick that is17:11
tgardnerjcrigby, thats a bit up to ogasawara, but usually we won't reorder anything after Beta that has a tag.17:12
jcrigbyok, just hoping to avoid another mega rebase for awhile:)17:12
ogasawaratgardner, jcrigby: I actually have taken the approach to still rebase onto any upstream 2.6.35.y stable release17:13
jcrigbythats, fine.  It gets easier each time I do this17:13
tgardnerjcrigby, besides, rebases are good for the soul. it forces you to look at conflicts.17:13
jcrigbygood point17:14
jcrigbyI suppose I should try out git rerere so I can remember confict resolutions automatically17:15
bjfis there a way to look at the buildd queues, i'm curious to why it's taking so long before my source uploads are building17:21
tgardnerbjf, you're meaning the PPAs, right?17:22
smbhttos:launchpad.net/builders17:22
smberrr17:22
smbhttps://launchpad.net/builders17:22
tgardnerhmm, thats kind of a handy link17:24
bjftgardner: yes, and smb, thanks for the link 17:25
smbwelcome17:25
bjfsmb, is there a link that shows the amd64 ppa job queu (so I can see exactly where I am in the queue right now)?17:38
smbbjf, I don't know a good one, only the general build score, but that does not really mean much17:40
bjfsmb, ack17:40
smbmaybe lamont knows some magic that apprentices can use17:40
lamontbjf: the ppa queue is not exposed in the api... /ubuntu/maverick/amd64/+builds (or maybe /u/a/m/+b) is the queue for a given distroarchseries17:42
lamontand laggy in a meeting17:42
jdstrandsmb, jjohansen, lamont, elmo: fyi, I just unembargoed the fix for xen17:42
lamontjdstrand: woot.  I'll kick them ppas shortly17:43
smbjdstrand, Cool17:43
jdstrandsmb: nice work. I know that was rather painful17:44
jdstrandsmb: and thank you again :)17:44
* lamont remembers to wait for the publisher run17:44
jjohansenjdstrand: thanks, and smb yes very nice work17:45
smbSort of a Murphy issue, anything that could be wrong was wrong. o_O17:45
smbtgardner, About the packaging, maybe we should ask about that. Probably the replaces is not needed because the 35 package is not an upgrade path to the 34 package. But maybe I have not grasped the whole concept17:48
tgardnersmb AFAIK the conflicts forces you to remove the conflicting package _first_ before you are able to install the new package. thats the primary behavior I'm interested in.17:49
smbRight and the Replaces causes the old one to get actually removed when you are updated via the meta files. I just want to prevent us showing the next package and being told of again17:51
tgardnersmb, I'm reading the Debian policy manual...17:52
tgardnersmb, but there is no meta file upgrade path. you'll have to explicitly change from compat-wireless-2.6.34 to compat-wireless-2.6.3517:54
tgardnersmb, another issue with replaces is that it only replaces files, not packages.17:55
smbright, that would be point to have that requirement only for compat-wireless-2.6.34 as that will upgrade17:55
smbhm, I probably should read the manual myself17:56
ogasawaralag: I'm adding your SOB to the arm patches for 61385517:56
tgardneryes, so I think the current combination of conflicts and replaces is correct.17:56
tgardnersmb, http://www.debian.org/doc/debian-policy/ch-relationships.html#s-replaces17:56
lagogasawara: That would be fine, thank you17:56
smbSo thats the 7.6.2 usage17:57
tgardnerI think so, though I don't really understand the mail-transport-agent example17:59
smbIts a bit complicated because its for virtual packages18:00
smbBut I think it declares to provide bla and conflict and replaces it. So when you install this packages it would uninstall all other packages that provide bla18:01
tgardnerwell, I think I want the behavior to be explicit. no automatic replacement, etc. you'll have to explicitly remove the conflict.18:02
smbMost of the time yes, so the only time this should happen is when I have lbm-wireless-generic installed and the new meta package points to lbm-compat-wireless-...18:03
tgardnersmb, correct.18:04
smbWhich is what is done so you were right that only the conflicts is needed for the other package18:04
smbI got it now as well18:04
tgardneronce we're on to the next ABI series, then this is no longer a problem18:04
smbright, if people want to change, they should make that decision knowingly 18:05
smbAnd I think that would be all we were asked to. I would just take what you had pushed and make the adaption for the conflicts and buglink18:06
tgardnerdon't I have a public buglink? or did I get the wrong one?18:07
smbNo I thought you said you forgot to add a buglink on the latest patch18:07
smbI mean yes you have the public bug in the mail18:07
tgardnerI amended and re-pushed18:07
smbah ok18:07
smbThen lets do lbm try 318:08
achiangcan someone help me to understand kernel package names vs. the actual kernel binary? output from dpkg -l gives, e.g. linux-image-2.6.32-24-generic           2.6.32-24.41, 18:18
achiangso from that, i understand that 24 is the ABI version and .41 is the build number18:18
tgardnerachiang, so far, so good.18:19
achiangmy question is, in practice, do we ever increase the build number without also changing the ABI?18:19
achiange.g., would we ever ship a 2.6.32-24.42 ?18:20
tgardnerachiang, no, though the build number is completely arbitrary, the ABI number is not.18:20
tgardneroh, lemme back up a littlwe18:20
smbBut we increase the build number without abi bum18:20
tgardneryeah, what smb said18:20
smbIn fact we know should have an 2.6.32-24.42 in proposed18:21
tgardnerbump*18:21
smbyep18:21
achiangok, i think i understand. all kernel updates that do *not* require an ABI bump will simply have an increased version number, which dpkg --compare-versions would understand to be greater, right?18:22
achiangand that's how apt would know to pull down and install the new, updated package?18:22
tgardnerachiang, correct18:22
smbyep18:22
tgardnerachiang, and when the ABI bumps, then we update the meta package so that it references the _new_ package18:23
achiangis there any way to get the ubuntu version number out of vmlinuz? or is that only stored in the debian packaging?18:23
smbThe build/upload number _always_ increments, though not always by 118:23
achiangnod18:23
smbYou see it in uname -a18:23
tgardnerand in /proc/version_signature18:23
achiangsmb: no, you only see the package name, not the version string18:24
smbLinux maximegalon 2.6.32-25-generic #4318:24
achiangmaybe my terminology is messed up18:24
achiangoh!18:24
achiangduh18:24
smbSo this is 2.6.32-25.4318:24
achiangdarn, i can't pass an arbitrary kernel path to uname18:25
smberr, why would you want to do that?18:26
achiangtgardner: smb: ok, thank you for the explanations, they make sense.18:26
achiangsmb: i am doing some evil stuff for an OEM customer's customized system update tool. don't ask, it's horrible.18:27
* smb puts his fingers into his ears18:27
achiangsmb: well, conceptually, i want to know: "is this kernel newer than that kernel?"18:29
* smb still cannot hear achiang 18:29
achiangwhere "this kernel" might live somewhere in the filesystem (not /boot/) and "that kernel" is typically the one in /boot/18:30
smbBut yeah, basically you can compare version numbers, newer always inclrements. With the + and ~ magic of course18:30
achiangbut the version numbers don't really live anywhere, except in the debian packaging; or by combining some fields from uname18:31
smbso 2.6.32-24.41~pre1 comes before 2.6.32-24.41 18:31
smbWell that information is in the kernel image too. If you boot you see it. In dmesg18:32
achiangbut certainly, if you saw /boot/vmlinuz-2.6.32-24-generic you could not figure out if it was newer or older than /my/special/path/vmlinuz-2.6.32-24-generic18:32
smbnot from that filenames alone18:33
achiangbecause the kernel in /my/special/path might have a greater version number, but it's not easy to figure out. unless maybe you use strings and grep18:33
smbright strings and grep will work18:34
achiangis there a case to be made for incorporating the version number into the file name (instead of merely package name) for generic ubuntu? or is my use case truly unique?18:34
achiangobviously not for this cycle, but for future cycles18:35
smbno, as the parts that are used now are used for modules dir and we certainly don't want separate directories for each upload18:36
achiangi guess basing the filename on package name makes life a little easier in the non-ABI change case18:36
achiangah, that makes sense too, re: modules18:36
smbI guess this is just a very special usecase, but I tried strings/grep and it looks quick and simple (as long as you are not truely evil and want to do the same for old releases like Hardy ;-))18:38
achiangsmb: actually, yes, i sadly do need a solution for hardy as well18:40
* smb runs away18:41
achiangsmb: hm, strings seems to work on a hardy kernel18:44
achiangstrings vmlinuz-2.6.24-27-lpia |  grep '2\.6\.'2.6.24-27-lpia (buildd@midyim) #1 SMP Thu Apr 8 18:04:46 UTC 201018:44
achiangoops, the output is garbled18:44
achiangthere should be a newline after grep '2\.6\.'18:44
smbno it does not. its aways #118:47
achiangurk18:48
* achiang runs to wherever smb is running18:48
* tgardner lunches18:51
=== bjf is now known as bjf[afk]
* jjohansen -> lunch19:50
ogasawaratgardner: when beta freeze lifts next week, could I get you to help me upload pciutils (just need to update the pci.ids file).20:08
ogasawaratgardner: it's for bug 60107920:08
ubot2ogasawara: Bug 601079 on http://launchpad.net/bugs/601079 is private20:08
tgardnerogasawara, funny you should mention taht. I uploaded pciutils about 20 minutes ago20:09
ogasawaraheh awesome20:09
ogasawaratgardner: I'll just mark that fixed then20:09
tgardnerogasawara, yep, I think thats a safe move20:10
tgardnerogasawara, are you subscribed to maverick-changes@lists.ubuntu.com ?20:10
ogasawaratgardner: I'm not at the moment, but I'll subscribe now so I see the notices20:11
tgardnerits a good way to survey what kind of development activity is going on20:12
=== bjf[afk] is now known as bjf
* ogasawara lunch20:21
=== ivoks is now known as ivoks_away
DJAshnarAny plans to incorporate the kernel patch for the damn Toshiba Sattelite ACPI/DSDT bug causing us to be forced to turn of ACPI and then only run ONE cpu core?  http://bugzilla.kernel.org/attachment.cgi?id=2395820:56
* DJAshnar hears crickets21:00
* DJAshnar stomps on em21:00
bjfDJAshnar: it looks like that patch is already in maverick21:14
bjfDJAshnar: though not in the form of the link you gave21:14
DJAshnaron my Sattelite c655d s5057, it hangs on ACPI on boot unless I use the older patched kernel21:15
DJAshnarbooting the generi kernel in maverick just leaves me a blank screen.  I have a radeon 4250 chipset if that helps21:17
bjfDJAshnar: that's likely a different issue21:17
bjfDJAshnar: try rebooting, hold down the left shift key to get into grub and remove "quiet spash" from the boot command line21:18
DJAshnarno joy21:20
achiangDJAshnar: well, do you now get debug output on your screen?21:21
DJAshnarno21:21
achiangDJAshnar: were you able to get into grub and edit the kernel command line?21:22
DJAshnaryup21:23
achiangDJAshnar: hm, weird. try repeating, and add "ignore_loglevel earlyprintk=vga"21:25
achiangso "quiet splash" should go away and the other two should be added21:25
DJAshnarjust a black screen...21:26
achiangDJAshnar: hm, maybe the earlyprintk argument was wrong. maybe just remove it completely, actually21:30
achiangbut do keep ignore_loglevel21:30
DJAshnarno luck21:42
achiangthat seems rather bad, and i'm out of ideas, sorry21:52
keesI love the daily upstream kernel debs. very handy!21:54
DJAshnarhow can I update the kernel daily?21:55
tgardnerkees, I've opened the natty repo, but I've not really advertized it yet21:55
keestgardner: oh! very cool. I guess I should start rebasing nx-emu and yama...21:55
tgardnerkees, I dropped the nx emulation patches for now. had some conflicts, so I need to get back to it21:56
keestgardner: okay, I'll go poke at it.21:56
DJAshnarcannot reserve MMIO region... wth?21:59
\_^_\bjf: We chatted recently about bug #616501 and you told me that this will be part of some review by kernel team. Do you know what came out of this review?22:34
ubot2Launchpad bug 616501 in linux (Ubuntu) "Kernel > 2.6.32-20 doesn’t boot, /dev/disk/by-uuid/... not found (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/61650122:34
bjf\_^_\: yes, i'm embarrassed to say it didn't happen, we will do it monday a.m.22:45
bjf\_^_\: it's on our list, we just need to talk about it22:45
\_^_\bjf: OK, thanks a lot. :)22:46
bjf\_^_\: np22:46
=== sconklin is now known as sconklin-gone
=== bjf is now known as bjf[afk]
=== JanC_ is now known as JanC

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