=== warthylog [~warthylog@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-kernel
=== Topic for #ubuntu-kernel: Ubuntu kernel development discussion | http://www.ubuntulinux.org/wiki/KernelTeam | kernel-team@ubuntu.com--2005/kernel-debian--pre26--2.6.10 | anything else t-bone needs to get off his chest | cows don't like to do DST
=== Topic (#ubuntu-kernel): set by T-Bone at Fri Mar 4 13:06:53 2005
=== #ubuntu-kernel [freenode-info] help freenode weed out clonebots, please register your IRC nick and auto-identify: http://freenode.net/faq.shtml#nicksetup
fabbioneanybody alive?07:38
=== fabbione sighs
fabbioneguy the enable-inotify patch is kinda of horrible08:21
fabbioneit would have been enough to modify the inotify-*-optional patch08:21
fabbioneinstead of adding a 3rd one on top08:22
=== fabbione [~fabbione@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-kernel
=== calc [~ccheney@ip70-185-4-246.ma.dl.cox.net] has joined #ubuntu-kernel
=== abelli [~abelli@host-84-222-39-20.cust-adsl.tiscali.it] has joined #ubuntu-kernel
fabbioneinotify doesn't crash anymore!10:17
fabbioneit will be in 2.6.10-26 and no ABI change :P10:17
fabbioneFUCK NO10:18
fabbioneit crashed on the second removal10:18
=== abelli [~abelli@host-84-222-39-20.cust-adsl.tiscali.it] has joined #ubuntu-kernel
=== abelli_ [~abelli@host-84-222-39-20.cust-adsl.tiscali.it] has joined #ubuntu-kernel
=== abelli [~abelli@host-84-222-39-20.cust-adsl.tiscali.it] has joined #ubuntu-kernel
=== zul [~chuck@] has joined #ubuntu-kernel
fabbionehey zul02:58
zulhey fabbione how is it going?02:58
fabbionegood and bad02:58
fabbionei merged inotify 0.20 from 2.6.11-3 into 2.6.1002:58
fabbioneand i managed to unplug my usb device without crashing the kernel once02:59
fabbioneit dies the second time02:59
zuli think we should take out rml02:59
fabbionenow i am cleaning up the patch mess from inotify-*-optional and enable-inotify02:59
fabbioneso that there is one good patch02:59
fabbioneand that's it02:59
fabbioneand leave inotify disabled by default03:00
zulill have to see the patch once its done so i know what to do next time if this pops up again03:00
fabbionewell it's pretty simple03:00
zuldid you get the inode error when compiling?03:00
fabbionesee the original inotify patch03:00
fabbioneit changes the dnotify entries in fsnotify entries03:01
fabbionezul: no but that's becaused i used a more recent patch and the error is easily fixable03:01
fabbioneparse the calls to fsnotify to see what are the inotify_ calls03:01
fabbionethat leads to 5 functions03:01
fabbioneso basically inside inotify.c in these 5 functs03:02
fabbioneyou add the check03:02
fabbioneif (!inotify)03:02
fabbione return;03:02
zulah ok03:02
zulso no need to push my inotify stuff to you in the next patchset from me03:03
fabbionei am going to commit it to baz in a few minutes03:03
fabbionei only need to do a test build03:03
fabbioneand see if it works as expected03:03
zulbtw when do you want my stuff by03:04
zuli can put them up when i get home tonight am at work right now03:06
fabbionewe are not going to upload today03:06
fabbionepreview release is still ongoing03:07
zulyeah thats what i thought03:07
zulalsa is messing some people up when they upgrade because of the alsa driver for their winmodem03:08
fabbionei know...03:08
fabbionei heard03:08
zulheh its the kamion kult03:16
lamontzul: the question is, does alsa provide a working modem then??03:26
zullamont: i think i dont have a modem havent had one in the past 10 years :)03:26
lamontah sure you do... it's just not plugged into a phone line...  any semi-modern laptop will do... :-)03:27
zulif i had a phone line...i just have a cell phone03:27
=== lamont can accept that. :-)
zuli like making things dificult for you lamont :)03:28
=== lamont throws things
fabbionethere are serial cables for cell phones03:29
lamontfabbione: if you can tell me how to make the DKU-5 work with linux and my nokia phone, I will buy you _3_ beers03:29
lamont(no linux driver last I knew)03:30
fabbionei have an ericsson phone that has an internal modem03:31
fabbionei just plug the serial cable and do ATZ03:32
lamontDKU-5 is a usb cable03:32
lamontand vendor id isn't recognized.03:32
lamontOTOH, the last time I tried was 2.6.8 timeframe...03:32
fabbionei have the usb version too03:33
fabbioneand i see the cell phone as a usb-serial03:33
fabbionesame story.. minicom -> ATZ03:34
=== lamont will play with his phone today
zul*sigh* why do people have to use cheap kvm switches03:38
lamontzul: because they're _poor_, of course.03:38
zulheh...duh...it was rhetorical03:39
fabbioneit's work/wife safe :)03:44
=== lamont tries to think of something that's not canonical-work safe...
fabbionewell if you have some emploies watching your shoulders ;)03:46
lamontfabbione: heh03:51
lamontbut you notice he had it pointed in a safe direction...03:51
fabbionehe still managed to hurt himself03:52
lamontyeah - had on the slide while you pull the trigger is kinda stupid...03:52
lamontnote also that "accidents" like that always involve empty guns03:55
lamontfirst rule is: it's always loaded until I see the inside of the chamber03:55
lamontthat's like when people ask if a concealed-carry gun is loaded.  duh!03:59
=== jbailey [~jbailey@] has joined #ubuntu-kernel
zulhey jbailey 03:59
lamontmorning jbailey 03:59
fabbionehey jb03:59
jbaileyHeya Chuck04:00
jbaileyG'morning Lamont, G'd afternoon Fabio.04:00
fabbionedebian/ext3-modules-2.6.10-4-itanium-smp-di lib/modules/2.6.10-4-itanium-smp/kernel/fs/mbcache.ko04:02
fabbionedebian/ext2-modules-2.6.10-4-itanium-smp-di lib/modules/2.6.10-4-itanium-smp/kernel/fs/mbcache.ko04:02
fabbionesome modules are in more than one package04:02
fabbionecommand exited with status 25504:02
fabbionemake: *** [binary-udebs]  Error 204:02
=== lamont wonders what an mbcache is
fabbionewth is that module?04:02
fabbionethese errors are coming out after we fixed the build system04:03
fabbionei386/amd64 are ok04:03
fabbioneppc is building04:03
fabbioneia64 dies with that error04:03
lamontat least one of those builds with either ext2 or 3 =y04:03
fabbioneit means that eitehr there is an error in kernel-wedge04:03
lamontor both04:03
fabbioneor that module needs to be shared04:03
fabbionelamont: right...04:04
fabbionewhich one should we compile in? ext2 is my best bet04:05
zullowest common demoniator04:05
lamontext2 is what's on the livecd, so we know we need it. :0)04:05
lamontand it's small04:06
fabbionebut i also rememeber that changing to y has other implications04:06
lamontno clue04:07
fabbionei need to figure what04:07
fabbionelike updating some d-i/ stuff04:07
lamontthe other alternative is a shared package or something04:07
fabbioneif there is a fs-common something04:07
fabbioneanyway that's T-Bone decision04:09
zulyay perl is almost done05:01
zulcrap more changes05:09
fabbionedude.. i am not kidding05:24
fabbionethat machine is IDE05:24
fabbioneand it has been until yesterday05:24
fabbionenobody here was in my office other than me05:24
fabbioneeither i updated it in my sleeps05:24
fabbioneor something is really WEIRD05:24
zul*sigh* could it be hotplug?05:25
zulgrr...i hate it when something breaks05:28
fabbioneread #u-d05:30
fabbioneeither i misread partitioner05:30
fabbioneor it did change05:30
fabbionein the latter case it IS an issue05:30
zulok if that is the case how can we fix it mandrake has libata_enable_atapi05:31
fabbionei need to check again first05:31
zultime for lunch05:47
zulyay...i was having a little heartattack05:53
zulok lunch05:54
=== lamont [~lamont@mix.mmjgroup.com] has joined #ubuntu-kernel
=== lamont_r [~lamont@dsl-254-198.dynamic-dsl.frii.net] has joined #ubuntu-kernel
zulok what the hell is going on?06:37
fabbionezul: libata_enable_atapi is broken06:42
zulit was ripped straight from mandrake and when we tested it worked06:43
fabbioneok the problem is not the ATAPI06:47
fabbionebut the PATA change06:47
fabbionestatic struct pci_device_id piix_pci_tbl[]  = {06:47
fabbione#ifdef ATA_ENABLE_PATA06:47
fabbione        { 0x8086, 0x7111, PCI_ANY_ID, PCI_ANY_ID, 0, 0, piix4_pata },06:47
fabbione        { 0x8086, 0x24db, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich5_pata },06:47
fabbione        { 0x8086, 0x25a2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich5_pata },06:47
fabbionethis means grabbing old IDE controllers and treat them as SATA06:47
fabbionesorry PATA06:48
zuloh crap..06:48
fabbionethe fact that a patch is taken directly from Mandrake or Suse or Gentoo means nothing06:48
fabbioneif they did apply other patches on top to fix this stuff06:48
fabbioneor have older code than us06:48
fabbioneso now we have 2 options06:49
zulok im sorry it wont hapen again06:49
fabbioneeither we kill the PATA and we see how it goes06:49
fabbioneor we figure if in newer libata versions it works as it should06:50
fabbioneclearly this has very very high priority06:50
jbaileyIs this undoing the SATA fix?06:50
jbaileySATA CDrom fix, rather?06:50
fabbionezul: nothing to be sorry.. it could have happen to me06:50
zuljbailey: yes06:50
fabbionejbailey: now it's giving I/O corruption on piix chipsets06:50
jbaileyOh.  Joy.06:51
fabbioneso either we get the fix NOW06:51
fabbioneor we revert the change06:51
zulrever the change on pata06:51
jbaileyIn the face of corruption definetly revert.06:51
fabbionezul: let's divide the tasks instead06:51
jbaileyBut we'll have to reopen the bug with an explanation as to why.06:51
fabbionezul: do you have such chipset somewhere?06:51
fabbioneor jbailey ?06:51
zulfabbione: no i dont but jbailey does i think06:52
fabbione(it's 12 hours and more than i am here and i need to rest)06:52
jbaileyI do, yes.06:52
fabbionejbailey: i urge you to test and see if you can reproduce the problem06:52
fabbioneit's enough to boot the live cd06:52
jbaileyIt's in a data centre three time zones from here.06:52
fabbionethe installer will fail06:52
fabbionejbailey: well.. when i ask if you have it.. i mean handy :)06:52
jbaileysabdfl also has hardware to match it.06:53
fabbioneyeah but he quite not around06:53
fabbioneand i can't ask him to build/test kernels06:53
fabbionewhat i don't understand is why a kernel upgrade didn't show the problem06:53
zuli can build a kernel without the pata enabled06:53
jbaileyIt's handy in the sense that I can test kernels if we're somewhat confident it won't eat the system, and I can get all sorts of status information about the box.06:54
fabbionewhy on an upgrade it didn't switch to sda?06:54
fabbionejbailey: the problem seems to affect only cdrom06:54
=== T-Bone whispers 'hppa.u.c' at lamont... :)
fabbionesince i could boot from such machine06:54
zulT-Bone: not a good time06:54
T-Bonefabbione: WTF is that mbcache crap in the backlog? :P06:54
fabbioneT-Bone: not a good time06:54
T-Bonebecause there is such a thing as a "good time" ?06:55
fabbionejbailey: but i can't ask you to risk on a machine that far06:55
jbaileyfabbione: Is it any piix driver or just the ata_piix driver?06:55
jbaileyfabbione: The laptop I'm on has piix.06:55
fabbionejbailey: for what i can see is any06:56
fabbionei have 2 machine with piix and both fails06:56
fabbionejbailey: using the livecd is ok06:56
fabbioneit won't touch your hd at all06:56
fabbioneit will stop much earlier than that06:56
=== T-Bone is trying to figure out what's going on...
fabbioneT-Bone: I/O corruption from cdrom with the new libata changes06:57
jbaileyHmm, I wonder if I have a CD here.06:57
jbaileyUgh, teenagers.  Looks like the local school is having a field trip to the local coffee shop.06:57
jbailey(first time I looked around)06:57
fabbionecdrom only *APPARENTLY*06:57
T-Bonefabbione: sweet, how comes this made it in?06:57
lamont_rjbailey: I was here for one of those too.06:57
T-Bones/this/this shit/06:57
fabbioneT-Bone: the same way people are already complaining about the ppc LED turned on06:57
lamont_rwell, time to go retrieve my latin scholar06:57
T-Bonei see06:57
jbaileyfabbione: Is there anything I can test from my running system06:58
fabbionepatch -p1 < somefancystuff.diff06:58
T-Bonewe've been too laxist on the mergein policy06:58
lamont_rso who checked in the ppc changes, anyway?06:58
fabbionejbailey: no. you need to boot the livecd06:58
fabbionenot me06:58
T-Bonepart my fault06:58
fabbionei was approx 20000KM or 12000NM from any pc06:58
T-Bonebut otoh i must blame zul for that one06:58
fabbioneNO TIME TO BLAME06:59
fabbionewe need a fix06:59
T-Bonedisable it06:59
fabbioneand who can test the fix06:59
jbaileyfabbione: Lemme pull down a livecd then.  Lesse how good the net connection at this web caf really is. =)06:59
fabbionejbailey: ok.06:59
T-Bonerevert to pre-2506:59
fabbionejbailey: to verify:06:59
fabbioneboot on the liveCD06:59
fabbioneand the installer should fail with "unable to load installer components" or something like that06:59
fabbionecheck dmesg06:59
fabbioneand /proc/partition to verify that your hardware is actually recognized as scsi07:00
lamont_rT-Bone: need -25 for kickstart for kamion. Don't think we can revert07:00
fabbioneprobably checking the PCI id from above should be enough to know if it will happen or not07:00
lamont_rand I'm inclined to call this 25.1, since we already have -26 in the works...  thoughts?07:00
T-Bonelamont_r: how comes we couldn't revert the libpata diff (which is 2 lines btw), since it was working just fine before except for a few users with specific hardware??07:00
lamont_rT-Bone: that's what we need to do.07:01
fabbionefew = a lot dude07:01
fabbionepiix is quite common intel hw07:01
=== lamont_r read 'revert to pre-25' to mean 'use -24 on the preview'. which is much different than reverting that patch
T-Bonefabbione: hell, your call. If you prefer more supported user with broken implementation...07:01
T-Bonelamont_r: yeah sorry i wasn't clear enough07:01
jbaileyfabbione: array6 is fine for the livecd, right?07:01
jbaileyOr do I need something more recent?07:02
fabbionejbailey: dunno.. it needs to have -2507:02
T-Boneif we need a minute-made fix, the only solution is reverting that patch07:02
lamont_rI prefer some unusable implementations to silent data corruption anywhere07:02
jbaileyDo you have the URL for a mor recent snapshot handy?07:02
T-Bonewe wouldn't have enough time to test any other invasive change07:02
T-Boneand would introduce potential new bugs that way07:02
=== lamont_r really must run and fetch the kid - back online in about 40-45 minutes
T-Bonelamont_r: same here, hence my comment. Trying to fix both 'support more users' and 'fix data corruption' would go against our safe commit policy07:03
T-Bonebut that's just my 2 cents07:04
jbaileyfabbione: I'm getting 60k per second to cdimage.ubuntu.com  3 hours to download at this rate. =(07:07
=== T-Bone can't help. No x86 hw around
zuli have a pentium 3 where i can build a 686 kernel if you want to test07:16
jbaileyThe question got missed in #u-devel, Is this only for cases where ata-piix intead of piix gets loaded for pure pata systems?07:20
=== T-Bone needs to run out
T-Bonei'll be back in ~30'07:23
zulim building a kernel for 686 that disables undef pata again07:24
T-Gonezul: that's a no-go alas. We don't have time to rebuild a full set of kernels...07:24
=== T-Gone runs off anyway
zulim only building 68607:25
jbaileyzul: Do you know whether it's only for cases where ata-piix gets loaded instead of piix?07:25
zulim not sure07:26
jbaileyI've been running that kernel on my laptop for a while with no problem, and I've been using that kernel on a sata system that does a couple emails a second with no problem.07:26
jbaileyI'm still sitting at 40k/sec for this livecd download, though.07:27
jbaileyMy laptop does the cdrom no problem.07:29
jbaileySo I'm not sure where he's seeing the issue.07:29
zuldo a hdparm on the cdrom07:30
jbailey16 bit, a pile of offs, readahead 256, HDIO_GETGEO failed.07:32
jbaileyafk a sec.07:34
zulfabbione: is that the same  message that jbaileiy got?07:37
jbaileyzul: I think thats' pretty normal for hdparm'ing a CDROM.  I don't htin it provides the geometry info.07:37
jbaileyMy guess is that what he's seeing is that the controller is being detected as ata_piix intead of piix.07:38
jbaileyIt wouldn't affect upgrades because initrd-tools isn't that bright.07:39
=== abelli [~abelli@host-84-222-39-20.cust-adsl.tiscali.it] has joined #ubuntu-kernel
jbaileyIs that PCI ID add that you pasted here the only change with that patch?07:40
fabbioneit also enables another thing07:42
jbaileyfabbione: Is my assumptions right so far?  That you're running off of ata_piix instead of running off of piix?07:42
fabbionei need dinner07:43
jbaileyMm, lunch sounds like a good idea too.07:44
jbaileyI'll finish this first.07:45
=== lamont_r [~lamont@phantom.acmeps.com] has joined #ubuntu-kernel
=== lamont_r sits on the "neighbor"'s front porch, leeching bandwidth
lamont_r"neighbor" is about 3.5 miles away08:21
T-Bonewould that mean anything good for me? :}08:22
lamont_rfabbione: from a logistical standpoint, whatever fix we do should probably just happen on mainline, and then we can merge it over to -pre26, and then think about renaming said branch... :0)08:22
=== T-Bone ducks
zulok im building the linux headers for 2.6.10 with #undef pata08:23
T-Bonezul: what are you trying to do?08:23
zulim trying to get a test kernel going08:23
zulor...trying to help out08:24
T-Bonezul: actually, any solution involving a kernel rebuild is mostly ruled out, afaict08:24
T-Bonezul: actually it turns out that I just said crap. Forget it :P08:30
zuli have a linux-image with pata turned off again if you want to test08:30
T-Bonezul: i'd love to. Don't have the hw :P08:32
jbaileyzul: Did you revert the whole patch or just the extra PCI ids?08:33
zuljust the pci ids08:33
jbaileyI could test the kernel here just to make sure it doesn't load both anymore.08:34
zul-#undef ATA_ENABLE_ATAPI                /* define to enable ATAPI support */08:34
zul+#define ATA_ENABLE_ATAPI               /* define to enable ATAPI support */08:34
zul #undef ATA_ENABLE_PATA         /* define to enable PATA support in some08:34
zul                                 * low-level drivers */08:34
T-Bonezul: tel kamion/fabbione what you're doing on #u-devel, seems they're not aware...08:35
T-Gonedinner time, bbiab08:50
fabbionei am preparing 25.108:55
T-Bonelamont: how are you going to integrate 25.1 into the baz archive?09:28
=== T-Bone tries to return to the adequate channel
lamontT-Bone: simple.  grab source, apply diff, commit to mainline09:32
lamontthen branch it to -2,6,10-25,109:32
lamontmerge to --pre26, and commit there09:33
T-Boneseems so simple :)09:33
lamontT-Bone: that's why we don't work on the mainline branch... :-)10:02
lamontalternatively (if we had already checked in -26 on mainline), then it would go in as a patch on the -25 version branch10:02
T-Bonebtw. what about the issue that would possibly be solved today? Will it have to be postponed? (just asking, mind you :)10:05
lamontwhich issue is that?10:07
lamontoh. that.  nfc10:07
T-Bonei feared such an answer :|10:07
zulright ill be back later tonight...going to go play some soccer and ill be in pain later :)10:07
lamontstep 1: preview release ships.  step 2: figure out status of everything else, act accordingly10:08
=== T-Bone has gone for a 'nobody cares' policy on ia64
T-Bonezul: have fun10:08
zuloh i will10:08
MithrandirT-Bone: nobody (read HP) has cared enough to send me an ia64 yet, so.. ;P10:09
T-BoneMithrandir: heh. Otoh you've got full access to mine...10:09
MithrandirT-Bone: true enough.10:09
T-BoneMithrandir: more access would be "physical" at that point :)10:09
MithrandirI could use one, it's cold up here in Norway.10:10
T-BoneMithrandir: btw if you need it right wrt lib32gcc1... I can let it on 24/7 if needed :)10:11
MithrandirT-Bone: I'm going home now.  If gf is sleepy, I might fix it now so it's ready when preview freeze is over.10:11
MithrandirI'll let you know in 30-ish minutes10:11
T-Bonethx alot :)10:12
=== calc [~ccheney@ip70-185-4-246.ma.dl.cox.net] has joined #ubuntu-kernel
=== jbailey [~jbailey@CPE000ded9d787c-CM014260028338.cpe.net.cable.rogers.com] has joined #ubuntu-kernel
MithrandirT-Bone: ok, home, Karianne is already in bed, so I can hack the ia32-libs stuff.10:56
T-Bonehehe ok10:56
T-Bonehold on, booting the beast10:56
T-Bonethere it goes. Give it 2' and it'll be all yours :)10:57
T-Bonefeel free to dist-upgrade if needed10:57
Mithrandirshouldn't need to11:02
T-Bonewhatever you want :)11:03
=== lamont notes that t-bone has turned into the 'willy switch'
T-Bonelamont: huh? :)11:33
lamontI used to have willy walk over and power-cycle my A500 in my cube in 5L.11:34
lamonthence, willy switch.11:34
lamontfar better than X10. :-)(11:34
lamontmore expensive, too11:34
T-Bonedamn you :)11:34
lamontoh where oh where is fabbione...11:45
T-Boneto bed11:49
T-Bonenot necessarily sleeping, mind you ;)11:50
=== T-Bone ducks

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