/srv/irclogs.ubuntu.com/2005/12/17/#ubuntu-kernel.txt

=== doko_ [n=doko@dslb-084-059-073-144.pools.arcor-ip.net] has joined #ubuntu-kernel
=== czr_ [n=czr@cs183171.pp.htv.fi] has joined #ubuntu-kernel
BenCinfinity: ping01:47
infinity.gnop : CenB01:56
infinityAll reports on the vga16fb change have been positive so far.  At least, I didn't seem to break any already-working setups.01:56
infinitySo, I should push you the patch for -8.10, and update {edu,ku,u}buntu-artwork01:57
BenChey01:57
BenCgot a nice bogl assertion for you01:57
infinityAh-ha.01:57
infinityLine?01:58
BenCbogl-vga16.c:43701:58
infinityLies.  That's the one I fixed in -25.01:58
BenCI dist-upgraded to dapper to make sure it wasn't a conflict somewhere01:59
BenCmaybe I should rerun update-initramfs to be sure again01:59
infinityupdate-alternatives --display usplash-artwork.so02:00
BenCI know usplash 0.25-1 is in the initrd02:01
BenCcpio'd it just to make sure02:01
infinityThat assertion will fail in two instances:02:02
infinitya) bug in the code (fixed in 0.1-25)02:02
infinityb) artwork too big for display area.02:02
infinityThe latter could happen if your usplash-artwork.so isn't the right one, but if you're seeing mjg59's scary line art, it SHOULD be right.02:02
infinityBut, if all else fails and we dunno WTF is going on with your machine, I'll just give you a debug build with a bunch of printfs to tell us WHAT it's asserting ain't true.02:03
BenCupdate-initramfs didn't help02:08
infinityYeah, no big shock there.02:08
infinityYou should be able to reproduce the same assertion at a command line in a VC, no need to reboot and do it from the initramfs.02:08
BenCit just shows a black screen, but I can see where the scanlines end, and it looks normal02:08
BenCconsole is still normal02:08
infinityUnless your VCs are still misbehaving and staying black forever after usplash runs.02:08
BenCthey do, but switching to X and back fixes it02:09
infinityAnd then shows the assertion?02:09
BenCyeah02:09
infinitySpiff.02:09
BenCanything I can test?02:11
infinity'Sec.02:11
infinityhttp://cerberus.0c3.net/~adconrad/usplash_0.1-24.1_i386.deb02:12
infinityThere's the printf-laden build I was using to find the stupid bug in the first place.02:12
infinityShould give us xx,yy,yres,xres,pixmapx,pixmapy on your machine, so we can at least see why the assert is failing.02:12
infinityIf it's the pixamp size, that's easy.  You have cruft ans we need to clean it up.02:13
infinityIf it's yy, it's a programming bug.02:13
infinityIf it's bogl_yres, we have real issues I need to figure out. :)02:13
BenCFYI, I had edubuntu installed on here02:13
infinityAh-ha.02:14
BenCis there a pointer to the artwork?02:14
infinityupdate-alternatives --display usplash-artwork.so02:14
infinitydpkg --purge edubuntu-artwork-usplash.  Or wait an hour or two for me to get around to uploading the smaller pixmaps.02:14
BenCok, updated to usplash version02:15
BenC...02:15
BenCok, it displayed!02:16
BenChas some line artifacys, but much better than the old one02:16
BenCworking on boot too02:17
BenCit's not perfect, but it is so much better than before02:17
BenCI can live with it enough to not do vga=771 anymore02:17
BenCyou should add that to you .txt :)02:18
infinityArtifacting?.. Poop.02:19
infinityMaybe that's your tvout still hating us.02:19
infinityWoudl you test it on a normal monitor and make sure it looks right there?02:19
infinitys/Woudl/Could/02:20
infinityFirst morning smoke setting in.. Typing hard.02:20
=== dilinger .oO(artifacting poop? freak.)
infinityIf it looks okay on a real monitor, I'll stamp is "good enough", push the new artwork out for all three dists, and you can patch -8.10 in time for flight-2.02:22
BenCwhen's flight 2?02:26
BenCI was thinking -7.9 was going to be flight 202:27
BenCsend me the vga16fb patch and I'll merge it02:27
infinityWell, flight-2 hasn't happened yet, so if you upload -8.10, that'll have to be hammered into shape for it.02:28
infinityconcordia.ubuntu.com:~adconrad/vga16fb.patch02:28
infinityApplies to drivers/video/vga16fb.c02:28
BenCI've been marking all bugs about vga16fb with "vga16fb" in the status whiteboard02:29
BenCso I'll be pushing a mass "please test flight 2" to all of them in -8.10 gets in it02:30
BenCif I can close all the vga16fb bugs, you will be my bestest friend, you can come tip cows with me in my front yard and everything02:31
infinityYou have cows?02:32
BenCthe largest portion of the land my house is on is leased to some people that raise cows for beef02:32
BenCmy house is surrounded by pasture except for the driveway leading out02:32
infinityAhh, so they're not, strictly speaking, your cows.02:33
BenCnah, but they are like our pets :)02:33
infinityDoes land ownership give you the right to tip them?02:33
infinityI'm not sure how this works. :)02:33
BenCI'd have to check the county by-laws, there may be a loophole :)02:33
BenCwhere'd you get these margins from?02:34
BenCone thing I know is that lilo graphic splash screen works on my tvout02:35
infinitymodedb.c02:35
infinityI was originally fiddling with calculating them by hand, but then decided to just go with modedb.c, assuming it may be "right".02:35
infinityI may fall back to doing it by hand, if we can find something more correct.02:35
BenCone thing I thought was odd is that I am at 640x400-7002:36
BenCI was expecting 60HZ02:36
infinityThat's expected.02:36
infinityYou shouldn't drop down to 60Hz until you hit 640x480 and above.02:37
BenCless scanlines == more refresh?02:37
infinityYeah, that's the basic theory.02:38
BenCit's been a long time since I messed with video modes, back when I was working on an HD set-top-box02:38
infinityLonger for me, I'm sure.02:38
infinityIt was discouraging to discover I remember pretty much nothing about it.02:39
BenCbut all of those were 540p and greater, so I didn't get a chance to mess with low modes like this :)02:39
infinityAnyhow, this seems to work everywhere 640x480 worked (so far, no regression reports), and works a few places where it didn't, so..02:39
infinityI'm calling it "Good Enough", and if we find some spare time, we can look for better timings to make it perfect.02:40
infinityI'm going to fix up the artwork stuff right now.02:40
infinitySee if we can't get some pretty artwork back for flight-2.02:41
BenCsweet02:50
BenCcommited, and pushed02:51
infinityHrm, I wonder if I should update the two -artwork-usplash packages to use update-initrafms -u in the postinst.  Probably.02:51
=== infinity does that too.
BenCyeah02:51
infinityI wonder how many times we'll rebuild initramfs on a big upgrade.02:51
infinityI wish we had a good way to do post-hooks...02:51
infinityHrm.02:51
infinityWait.02:51
BenCgziping that is starting to get slow02:52
infinityI could do post-hooks in apt.02:52
infinityDoesn't work for dpkg, though.02:52
infinity(raw dpkg, that is)02:52
BenChow does ldconfig do it?02:52
BenCisn't there an ldconfig post-hook for libraries?02:52
infinityIf there is, it's a special-case in dpkg itself.02:52
BenCnope, don't see it02:53
infinityBut in apt, you can do DPkg::Post-Invoke stuff, that runs after the dpkg run.02:53
infinityNot really "good enough", though, I guess.  Installing kernels with raw dpkg is a pretty common occurence.02:54
infinityMeh.  I'm still seeing that failed asserion every time I do update-initramfs, too.02:54
infinity[17311616.280000]  Assertion failed! qc->n_elem > 0,drivers/scsi/libata-core.c,ata_fill_sg,line=250402:55
BenCis it causing a failure?02:56
infinityNo.  Just a fright. :)03:01
BenCyeah, it looks pretty scary :)03:01
infinityAlso, 'strace -f update-initramfs -u' is a bad idea.  Don't do it.03:01
infinityStill going on the console I ran it on.03:01
infinityI had hopes of catching the Assertion in situ, but that seems hopeless with an strace this... Long. :)03:02
BenC-o strace.out03:03
BenChelps03:03
infinitySure, but then I don't get the printk in the middle of the strace output.03:04
=== lbm [n=lbm@cpe.atm4-0-1301006.0x50a0824e.vgnxx6.customer.tele.dk] has joined #ubuntu-kernel
=== noggie- [n=noggie@p549CF120.dip.t-dialin.net] has joined #ubuntu-kernel
=== CataEnry [n=Enrico@host149-23.pool8250.interbusiness.it] has joined #ubuntu-kernel
=== noggie- [n=noggie@p549CF120.dip.t-dialin.net] has left #ubuntu-kernel []
CataEnryhi :)09:48
Mithrandirany idea why a hoary box is giving me shitty raid resync speeds?  8MB/sec, driver is AMD811101:01
fabbioneMithrandir: ide or scsi?01:05
fabbioneor sata?01:05
MithrandirIDE01:05
MithrandirPATA01:05
fabbionedo you have dma enabled?01:05
Mithrandirhdparm claims so01:05
Mithrandirhda and hdb, which sucks, I know, but still.01:05
fabbionecheck dmesg... there are some cases in which raid turns off DMA to ensure proper raid resync01:06
fabbioneand i highly suggest NOT to turn it on while it is resyncing01:06
Mithrandirit doesn't say anything about turning it off01:07
=== jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-kernel
HrdwrBoBMithrandir: hda->hdb is bad01:42
Mithrandirhd13:05 < Mithrandir> hda and hdb, which sucks, I know, but still.01:42
HrdwrBoBI'd say that's a large portion of the problem01:42
HrdwrBoByeah01:42
Mithrandirit's not.01:42
Mithrandiryou can still get quite good performance.01:42
=== zul [n=chuck@ubuntu/member/zul] has joined #ubuntu-kernel
zulheylo02:26
=== slushpupie [i=jay@slushpupie.com] has joined #ubuntu-kernel
=== mgalvin [n=mgalvin@ubuntu/member/mgalvin] has joined #ubuntu-kernel
=== Kamion [n=cjwatson@83-216-156-196.colinw664.adsl.metronet.co.uk] has joined #ubuntu-kernel
MithrandirBenC: I wonder if the devmapper snapshot target is broken (at least with cloops)03:29
MithrandirBenC: both Colin and I are seeing really weird stuff with the live cd; FS corruption, it looks like03:30
BenCMithrandir: ok, I'll check the cloop patches from breezy03:31
MithrandirBenC: please do.  Breezy works for me at least.03:31
MithrandirBenC: we're using the snapshot target, backed with a cow device in ramfs.03:31
Mithrandirit works in breezy. :-)03:32
Kamiondm linear doesn't appear to cause corruption03:34
fabbioneBenC: when you included cloop, did you take it from breezy or upstream?03:35
BenCI took a fresh upstream03:35
fabbioneBenC: ok03:35
fabbionetake it from breezy and the problem will be solved03:35
fabbionethere is an extra fix in breezy exactly for that problem03:35
fabbionethat upstream did reject as: "It works here you fuckheads"03:36
BenCwould have been easier if that patch was seperate from the main one :)03:36
fabbioneif you find our patch and upstream in the same version you should be able to get just the fix to the device mapper you need03:36
fabbioneBenC: i fully agree...03:36
BenCI'll find part it needs and just patch it in03:36
fabbioneit was at my second / third kernel upload03:36
fabbionei admit i didn't have my clear at that time yet03:37
fabbioneand forgot about it..03:37
BenChehe, no problem :)03:37
BenCtook my couple of months before I was felt comfortable with the build system and baz03:37
infinityYou got comfortable with baz?03:38
infinityNow that bzr is the way of the future, I'm allowed to talk shit about baz/tla publically, right?03:38
fabbioneoh at that time was not even in ba03:39
fabbionebaz03:39
BenCMithrandir: fortunately I am doing -8.10 today, so I'll get that patch in there03:39
infinityAnd conveniently, Flight-2 is now blocked on -8.1003:39
infinityWhich means I'm guaranteed to get my vga16fb fixes. :)03:39
infinityYay.03:39
BenCbrb03:40
=== infinity really goes ot bed now, so he can get more LRM crack done in the morning.
jbaileyBenC: Do you need a bug filed for the asm symlink on ppc in the headers package03:40
infinity'Night, guys.03:40
jbailey?03:40
BenCnah, already fixed in -8.1003:40
BenCoh, and we get correct EXTRAVERSION now too in linux-headers packages03:40
fabbionebbl03:41
jbaileyBenC: Lovely, thanks.  I'll wait for that then, before doing more on klibc.03:41
Kamiongah, jackass no longer has 2.6.9-7 source, which is where that patch came in03:42
BenCI can get it from baz03:47
Kamiondoesn't 2.6.9-7 predate baz?03:48
BenCyeah, I but I'll diff the patch against pristine and find it03:48
Kamionseems to be http://people.debian.org/~cjwatson/tmp/cloop-dm.patch if I'm reading recent cloop source package changes correctly03:51
BenCdiff between breezy and dapper cloop.c is trivial03:53
BenCyeah, that while loop was the change03:53
BenCI'm just going to copy the breezy cloop.c03:54
BenCKamion, Muthrandir: consider this fixed for -8.10, which will be uploaded in about 5 hours03:57
MithrandirBenC: ok, thanks.03:58
mjg59Someone still needs to track down what broke suspend03:58
mjg59I'm lacking the time to do a git bisect righ tnow03:59
mjg59I suspect it's somewhere in IDE - things seem more reliable if I use the libata driver04:00
KamionBenC: great, thanks04:00
BenChmm, wonder if I have any machines that support suspend04:01
BenCcan I susped a G4?04:01
jbaileyBenC: I think I suspended my pegasos.04:05
jbaileyIt might have been STD, though.04:05
BenCis there a prefered way to suspend? I've actually never done it on anything (my laptop didn't work)04:05
zulBenC: can you pull in mine stuff as well? thanks04:08
mjg59It may well be an x86-specific issue04:10
makxsuspend to ram is more likely not to work on my experience04:12
makxsuspend to disk is not that fast but already nice.04:12
Kamionmy G4 PowerBook suspends to RAM fine04:14
KamionI normally just press the power button, although closing the lid works if your lid is less wobbly than mine04:14
BenCzul: URL again?04:22
BenCis there a script I can run to try suspend?04:22
=== lamont__ [n=lamont@mib.fc.hp.com] has joined #ubuntu-kernel
BenClamont: elilo with initramfs support ready yet?04:24
BenCoh wait, I got that elilo.efi from you didn't I04:24
BenCI'll try it  when I boot -8.1004:24
mjg59makx: The specific case is that our 2.6.15 kernels have broken suspend04:25
mjg59At least on x86 with IDE04:25
mjg59Suspend to RAM works fine on most laptops with 2.6.1104:26
mjg59Uh, 2.6.1204:26
BenCwell my G4 has IDE, so I should atleast be able to tell you whether it's x86 specific or not04:26
zulBenC: http://zulinux.homelinux.net/git/ubuntu-2.6.git04:27
makxmjg59: ive an old broken gericom laptop which never suspended to ram.04:32
makxanyway quite surpring that this bad hardware still lives.04:33
BenCzul: just the efi patch?04:33
BenCzul: can you pull from me to fix merges?04:34
lamont__BenC: thanks for testing04:35
lamont__elilo was uploaded to debian, haven't checked to see where it got with ubuntu04:35
fabbionelamont__:  do you have any idea why util-linux -2 didn't get autosynced?04:37
lamont__no clue04:40
=== lamont__ pokes elmo
=== lamont__ bets it could be because -1 was ftbfs?
zulBenC: sure..i can do that tonight when i get home..04:41
zulor i can try now..04:43
BenCgot it04:43
zulnifty04:44
=== JaneW [n=JaneW@dsl-146-129-50.telkomadsl.co.za] has joined #ubuntu-kernel
=== lbm [n=lbm@x1-6-00-13-10-7a-d1-e4.k233.webspeed.dk] has joined #ubuntu-kernel
BenClamont: I installed the elilo from debian (3.4-pre5-2)05:43
BenCit booted my ia64, going to try initramfs now05:44
zulBenC: i fixed up my tree..06:03
jbaileyBenC: Apparnetly there's klibc segfaults.06:12
jbaileyI haven't tried stuff yet.06:12
jbailey(On conference call, lagging)06:12
BenCit got into the initramfs, and yes, looked like there was a segv of some kind06:18
BenCit's progress though06:19
Mithrandirhmm, is usplash broken in the current initramfs?06:21
BenCdepends, what does broken look like to you? :)06:22
Mithrandirdoesn't show anything at att, just spits out "0" (no \n, even) when run by hand in the initramfs06:23
Mithrandirthis is the live cd, so there might be a lot of other things broken, which is why I ask if it's known-broken before investigating06:24
BenCnot sure about that one, guess you need to ask infinity06:26
Mithrandirmhm.06:27
Mithrandirhe's asleep now, though. Hopefully.06:27
BenCshouldn't be broken, afaik06:27
Mithrandirok06:27
BenCbut there is the 640x400 change with vga16fb that is being coordinated with my -8.10 kernel upload06:27
BenCmake sure it has 0.25-1 usplash06:27
Mithrandirdo you mean 0.1-26?06:28
BenCyeah, that's it :)06:28
BenC-26 may be expecting new kernel, I'd have to ask infinity about that06:29
BenClame, I cannot even peruse alsa bug tracker without an account06:30
BenCnm, seems they have guest login06:30
=== BenC [n=bcollins@debian/developer/bcollins] has joined #ubuntu-kernel
BenCthis sucks07:08
BenCbecause our ide is modular, none of the ide kernel command line options are honored (ide=nodma, ide=reverse, hdc=remap, etc...)07:09
BenCthey have to be module options, but that defeats the purpose of people using them (e.g. for booting the install cd)07:09
=== BenC wonders why we can't have built-in ide-core anyway
=== Kamion looks at the comment in http://lwn.net/Articles/162853/ asking distros not to ship bcm403x yet
Kamionalthough I guess it means more "don't ship it in releases yet"07:29
BenCyeah, it's only there for testing07:35
BenCit will probably get pulled for no other reason than we have no way to include firmware yet anyway07:36
Kamionmm, I was wondering about that07:37
zulBenC: i can help out there if you want07:38
MithrandirBenC: ETA on kernel upload?09:21
BenCcouple hours09:27
mjg59zul: Around?09:32
zulmjg59: yo09:38
mjg59zul: You've got a Latitude X1, right?09:39
zulindeed09:39
zulwhat do you want me to test?09:39
mjg59Do you have backlight problems if you close it at a console?09:39
zuli havent tried or really havent noticed, i can try it tonight and let you know09:40
BenCis that the radeon backlight problem?09:40
mjg59No, intel09:42
mjg59zul: Basically, switch to a console, close the lid, open it, check if the backlight is on or not09:43
zulok cool09:47
zulill let you know when i get home09:47
BenCzul: I have something you can work on, if you want09:50
BenC#18456, creating a linux-image-dbg package with unstripped, uncompressed vmlinux in it for OProfile and such09:51
BenCmainly just for i386 and amd6409:51
zulBenC: sure..09:57
BenCwould probably be best handled in debian/post-install like the linux-headers packages are09:57
BenCzul: thanks09:57
fabbionehey Ben10:00
BenChey fabbione10:00
fabbioneBenC: we need to schedule an amd64-generic kernel built as _i386.deb10:00
fabbionedoko was asking for it and some users too10:00
BenCfor 32-bit userland and 64-bit kernel?10:01
fabbioneyes10:01
BenCeasy enough, just copy the one from amd64 config's10:01
fabbioneyeah that too10:01
BenCI can schedule that for post 8.1010:01
fabbioneafter you release 8.1010:01
fabbioneyeah10:01
fabbionebingo10:01
Kamionoh, argh, one more kernel thing10:01
fabbioneKamion: not for install10:02
fabbioneand not in main10:02
BenCnot for install or CD10:02
fabbionewe will keep for universe10:02
Kamionno, *I* have one more kernel thing I mean10:02
BenCDVD only, IMO10:02
BenCwhat?10:02
Kamionthe kernel appears to be eating arguments it doesn't recognise10:02
Kamionfor example:10:02
=== fabbione heads offline for the night
KamionUnknown boot option `base-installer/kernel/linux/extra-packages-2.6=usplash': ignoring10:03
fabbionegood night fellas10:03
=== CataEnry [n=Enrico@host149-23.pool8250.interbusiness.it] has joined #ubuntu-kernel
CataEnryhi :)10:03
Kamionalthough rescue/enable=true appears to work, since the rescue mode works fine10:03
BenCthe original untouched command line should always be in /proc/cmdline10:04
Kamionso I'm a bit confused as yet, but this is breaking d-i preseeding10:04
Kamionat the moment we fetch it from init's environment, not /proc/cmdline10:04
BenCbase- is probably matching some __setup() in the kernel10:04
Kamionok, if it's constrained to a particular pattern then I can postpone the fix10:05
BenCIMO, it should always be gotten from /proc/cmdline10:05
KamionI think we used the environment because it was already parsed into a more or less suitable form10:05
Kamionbut it can be changed10:05
Kamiondates from December 2003 if I'm reading d-i changelogs correctly10:06
Kamionright, no problem then10:06
BenCok10:07
=== fs [i=fs@213.178.77.98] has joined #ubuntu-kernel
=== slushpupie [i=jay@slushpupie.com] has left #ubuntu-kernel []
=== BenC hits the bottom of his 381 long bug list
AcidPils1BenC: omg, wanna have a beer?11:59
=== AcidPils1 is now known as AcidPils
BenCthink I'll need a case, I've spent 4 days on this buglist12:00
jbaileyBenC: And now the bugs are sitting on top of you, squishing you? =)12:00
BenCjbailey: feels like it :)12:00
BenCnow I need to reload to see how many I actually closed and/or put in pendingupload12:01
AcidPilsBenC: dont do it... i dont wanna see you cry because of the new reports ;)12:02

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