[02:03] <ceekay> trying to understand how kexec/kdump works and hoping someone might be able to help me out... when i try to kexec -p /vmlinuz --initrd=/initrd --append=root=/dev/sda1, i get "locate_hold failed" anyone able to point me in a direction? cat /proc/iomem shows that i have an entry for "Crash Kernel" (i booted with the crashkernel=64M@16M option)
[02:03] <ceekay> sorry "locate_hole_failed", not "hold"
[02:04] <mkrufky> BenC: fyi, hermann tends to speak to the list in his emails moreso than to those in the to: field  ;-)
[02:12] <BenC> mkrufky: I gathered :)
[02:12] <mkrufky> hehe
[02:13] <BenC> ceekay: on intrepid, just install linux-crashdump-generic
[02:13] <BenC> ceekay: reboot, and all is good
[02:13] <ceekay> benc: I'm trying to backport to 7.04 if that is at all possible :\
[02:13] <ceekay> rather, trying to understand if it is possible
[02:14] <ceekay> just pulled the source pkg kexec-tools (20070330-4ubuntu3) from hardy
[02:14] <BenC> ceekay: IIRC, 7.04 kexec is broken :)
[02:14] <ceekay> it compiles and installs... but i guess the question i'd need to ask here, is does the kernel support it or how can i find out
[02:14] <BenC> kexec and/or the kernel
[02:15] <ceekay> eek
[02:15] <ceekay> there a way i can check?
[02:15] <BenC> the kernel supports it, but it's not working
[02:15] <BenC> ceekay: if you get the above error, then it doesn't work
[02:15] <BenC> ceekay: the kernel supports it, else you wouldn't see the entry in /proc/iomem
[02:16] <ceekay> so is a newer kexec-tools not going to resolve this issue? it's broken in 2.6.20-16?
[02:16] <ceekay> rather in the kernel itself, not just the kexec tools
[02:17] <BenC> right
[02:18] <ceekay> quick way to find out i guess... i just ran kexec -p..etcetc... using the new kexec tools package, i didn't get the "locate_hole_failed" error... hit alt+sysrq+c and it appears to have just re-execed the kernel
[02:19] <ceekay> would seem to me that it worked
[02:20] <ceekay> and i now have an 80MB living in /proc/kcore
[02:21] <ceekay> BenC: did i just prove that it works? (i don't want to get too excited here)
[02:44] <ceekay> probably a stupid question- how can i get the vmlinux file for a released kernel package (i.e. 2.6.20-16)? do i have to download+extract the kernel source package, copy my /boot/config-2.6.20-16 to /usr/src/linux, then make?
[02:45] <ceekay> linux-image-debug seems to have the uncompressed vmlinux but it doesn't appear to match the regular kernel
[03:03] <rtg> kirkland: everything is built in my PPA
[03:24] <BenC> ceekay: sounds good to me
[03:24] <BenC> ceekay: apt-get install linux-image-debug-generic
[03:24] <BenC> ceekay: installs the vmlinux with debug symbols that matches the kernel
[03:34] <ceekay> BenC: hm... the program "crash" didn't seem to like it...something about mismatched banners
[03:36] <ceekay> interesting... using the /etc/init.d/kdump defaults, kexec isn't happy... i think it's the irqpoll option on my system that causes it to get stuck in an infinite loop
[03:39] <ceekay> sorry the warning was about linux_banner, but crash seems to error out saying that /boot/vmlinux-2.6.20-16... and /var/crash/kernel.0 (dump that i copied from /dev/kcore after causing a dump)
[03:41] <ceekay> don't match
[03:46] <BenC> ceekay: probably your kernel is not the latest one, run apt-get update
[03:47] <ceekay> BenC: both packages say 2.6.20-16.35 :\
[03:48] <ceekay> i know 2.6.20-17 is out, but as long as these two packages match versions seems like it should work
[03:49] <BenC> ceekay: sounds like you have something askew on your system
[03:49] <ceekay> quite probable
[03:49] <BenC> ceekay: the -debug package is the vmlinux that was used to build the standard vmlinuz kernel
[03:49] <BenC> it's not even a separate build, it's an exact match
[03:50] <ceekay> BenC: one last question for you.. looking at https://launchpad.net/ubuntu/+source/kexec-tools it appears that you did the scripts to copy the dumpfile somewhere... if i'm looking to backport this to 7.04, is this something that's only in the 8.10 initrd's?
[03:51] <BenC> ceekay: you will need kexec-tools, grub and makedumpfile from intrepid
[03:51] <ceekay> noticing the boot flag kdump_needed that gets added by /etc/init.d/kdump to the kexec on panic APPEND line
[03:52] <ceekay> ah, hadn't realized makedumpfile is a package
[03:52] <BenC> it's in universe in intrepid only
[03:52] <ceekay> gotcha
[03:52] <ceekay> is the new grub really necessary?
[03:52] <BenC> yeah
[03:52] <ceekay> 7.04 seems to do the crashdump option correctly
[03:53] <ceekay> BenC: any specifics on what the new grub does differently?
[03:55] <ceekay> last changelog entry i see seems to have been in 7.04 timeframe
[03:55] <ceekay> by you :)
[03:55] <BenC> ceekay: the crashdump= is handled differently
[03:55] <BenC> automatically added
[03:58] <ceekay> BenC: crashdump=1 seems to be handled the way i'd expect it to be currently... i set it to 1, run update-grub and then i get a new alt boot option that has the crashkernel64M@16M option appended to it
[03:59] <BenC> ceekay: that is removed
[04:00] <BenC> but you're free to figure it out on your own :)
[04:00] <ceekay> gotcha :)
[04:00] <ceekay> thanks so much for you help on this... looks like i have a working backport of current kexec-tools to 7.04...
[04:01] <ceekay> only thing i really had to change was removing "irqpoll" from the append list in /etc/init.d/kdump
[04:05] <ceekay> BenC: oh and btw... my debug image does seem to work... i think i had a funky crash dump... just did a clean reboot + sysrq/alt/c, then cp /proc/vmcore /var/crash/kernel.0, now crash reads the dump file just fine :)
[10:09] <Kano> hi
[10:09] <Kano> did somebody notice that the latest 2.6.26 snapshot (and rc6 i tested before) have got problems with dell vostro 1000?
[10:10] <Kano> it only boots when acpi=off - a bit brutal for a laptop
[10:11] <Kano> amd chipset with integrated Xpress 1150 gpu
[14:09] <penfold_99> is anyone here working on kvm?
[14:16] <zul> penfold_99: #ubuntu-virt
[14:43] <rtg> kirkland: have you talked to upstream about ecryptfs and the fact that the 'eCryptfs: remove unnecessary page decrypt call' patch did not fix the issue?
[14:43] <kirkland> rtg: i have a 10am meeting with them today
[14:43] <kirkland> rtg: that's tops on the list ;-)
[14:44] <rtg> kirkland: cool. I looked through subsequent commits for 2.6.25 and 2.6.26, but didn't see anything that might make a difference.
[14:44] <kirkland> rtg: yeah, this is bothering me
[16:00] <Kano> rtg: could you add gspca to 2.6.26
[16:47] <rtg> mdomsch: so how long should a BIOS update take on an Inspiron 1420? A00 --> A08. Its been at the 'Performing BIOS Update...' screen for over 10 minutes.
[16:48] <Kano> rtg: also uvcvideo
[16:48] <Kano> rtg: dos mode?
[16:49] <rtg> Kano: why are you asking me? BenC is the Intrepid dud right now. I'm working on Hardy stability issues.
[16:49] <rtg> s/dud/dude/
[16:50] <Kano> but why do you change it then?
[16:50] <Kano> when you look at changelog
[16:50] <rtg> Kano: because we have changed roles.  Look at who has made the last dozen Intrepid commits.
[16:51] <Kano> the last was you
[16:51] <Kano> when you really look there..
[16:51] <Kano> ok before it was benc
[16:51] <rtg> Kano: forward porting a Hardy fix doesn't count.
[16:51] <Kano> ;)
[16:52] <mkrufky> Kano: can you please lobby the gspca and uvcvideo folks for an upstream merge?
[16:52] <rtg> Kano: besides, it would be a hell of a lot easier to keep track of what is missing if you sent an email, or better yet a web page somewhere. IRC logs just get lost.
[16:53] <mkrufky> Kano: if they finally merge into upstream, this issue will be solved for _everybody_
[16:53] <Kano> sure, just kanotix users already test you kernels ;) and they missed em *g*
[16:53] <mkrufky> Kano: especially us v4l/dvb folks, who are constantly (incorrectly) getting support requests for those two drivers
[16:53] <Kano> one system - dell vostro 1000 - does not boot it without acpi=off, any idea
[16:54] <rtg> not off-hand. turn off the splash screen during boot and see where it croaks.
[16:54] <mdomsch> rtg, not sure, I haven't tried on one of those
[16:55] <Kano> well you dont see much
[16:55] <mdomsch> rtg, generally not quite that long though; several minutes, but not 10+
[16:55] <rtg> mdomsch: its still flashing. drat. I hope its not toasterized.
[16:55] <Kano> i guess it has already problems to access hd
[16:56] <Kano> due to missing irqs
[16:56] <johanbr> I updated the bios on my 1420 and it didn't take long at all. I didn't time it, but it didn't seem to be longer than a minute.
[16:56] <rtg> mdomsch, johanbr - this is one of the original prototypes, so maybe it has other issues.
[16:59] <rtg> well, it power cycled OK, but its still at A00.
[16:59] <rtg> guess I'll leave it be.
[17:00] <BenC> Kano: already said those drivers will get into 2.6.26, then the priority is there
[17:00] <Kano> well whats your priority?
[17:00] <BenC> Kano: bugging us about it doesn't make it happen any faster
[17:00] <BenC> Kano: my priority is getting the tree stable...and I can't even make an upload till after alpha 1, so there's no rush
[17:01] <Kano> when is alapha1 planned?
[17:01] <BenC> Kano: priority right now is stability and required drivers for booting and testing
[17:02] <BenC> Kano: was supposed to be by today, but I haven't been updated on the progress in a few days
[17:02] <Kano> i already made a live cd with 2.6.26, no problem booting
[17:02] <BenC> Kano: as we've explained before, our criteria for a milestone CD != the same priority for yours
[17:03] <BenC> we have milestoned bugs that need to get fixed first, and we have way more testing to do than "boots for Kano"
[17:03] <mkrufky> BenC: when should I plan to have my LUM stuff for intrepid ready to pass on to you?
[17:03] <BenC> mkrufky: if by lum you mean ubuntu/ in the main tree, then give me something to pull from :)
[17:03] <mkrufky> i'm imagining i have a month yet before it matters at all
[17:04] <Kano> mkrufky: i thought it is just patched now in like in 2.6.22?
[17:04] <mkrufky> Kano: correct, i said LUM by accident
[17:04] <BenC> mkrufky: just be aware of the milestones and freezes
[17:04] <BenC> mkrufky: initial code drop should be soon, easier to justify patches late in the cycle than full drivers
[17:05] <mkrufky> BenC: ok, in that case, is it OK if I wait for 2.6.26 upstream release first?  If my hvr950q performance tweaks get in to 2.6.26, then it will be less changes that I'll have for you
[17:05] <BenC> mkrufky: Sure, I would think 2.6.26 release would be within a couple weeks timeframe
[17:06] <tseliot> ﻿Kano: why don't you use DKMS with gspca and ﻿uvcvideo (with the sources from upstream) until the kernel team updates the drivers? It should be trivial to do so.
[17:06] <mkrufky> yes... hopefully within the week.  Linus took a vacation (i think) and he hopes to not push out another -rc
[17:06] <mdomsch> rtg, you might try using 'biosdisk' from http://linux.dell.com/projects.shtml#biosdisk, and an EXE from support.dell.com
[17:07] <rtg> mdomsch: k.
[17:07] <Kano> tseliot: i use something else, just mentioned that these driver are not there where i expected em
[17:08] <tseliot> Kano: ah, ok
[17:08] <Kano> but as they easyly compile they should be simple to add, not really much time consuming i think
[17:09] <Kano> you discuss longer then it actually takes time
[17:21] <BenC> Kano: it's not the time, it's working on something that isn't as important as the other things I'm working on
[17:21]  * BenC stops explaining priorities to Kano
[17:22] <BenC> Let's make it simple, if I worked on everything that everyone pointed me to that was "simple and quick" it would probably take me 3 weeks to complete all those "simple and quick" tasks
[17:23] <BenC> all the while, the difficult, and more important things wouldn't get done
[17:23] <Kano> when i do the patch,will you add it?
[17:23] <BenC> Kano: it will take me the same amount of time to import your patch as it would for me to cp -a the driver over and add it to the build, so no
[18:38] <UlfJack> hi
[18:47] <UlfJack> what's the right way to debug a freeze? i've searched on google, but havn't come up with anything definitive. kern.log is empty and the crashdump kernels don't seem to do anything either...
[18:50] <rtg> UlfJack: perhaps minimize the number of devices that you have active?
[18:52] <UlfJack> memtest passes, and xp seems to run stable as well...
[18:53] <rtg> UlfJack: since your IRC sessions comes and goes, I suspect you have a wireless device.
[18:53] <UlfJack> i've had to reboot because it freezes
[18:53] <rtg> UlfJack: 'it' being a wireless device?
[18:54] <UlfJack> no, i'm connected via cable
[18:55] <UlfJack> the laptop has a built-in wireless card, but that never worked, so it's switched off in the bios
[19:00] <UlfJack> no overclocking, no restricted drivers, i'm trying to narrow it down, but i'm totally stumped
[19:02] <UlfJack> i'm not doing anything specific, it just freezes randomly - it doesn't seem to make a difference if i'm using firefox, openoffice, freecell, or just a console
[19:03] <BenC> UlfJack: running what version of Ubuntu?
[19:03] <UlfJack> 8.04
[19:04] <UlfJack> it happens with the -17, the -19, and the -rt kernels
[19:05] <UlfJack> hmm, maybe i should reinstall -18 to see if it freezes too
[19:05] <BenC> UlfJack: likely wont help
[19:06] <UlfJack> compositing is disabled, too
[19:06] <BenC> UlfJack: try running with no network for a bit
[19:10] <UlfJack> it also crashes when the network cable is disconnected (in my experience, i didn't just pull the cable ;)
[19:11] <UlfJack> i had it crash on me spectacularly when i tried to give a presentation the other day :(
[19:13] <BenC> odd
[19:13] <BenC> is it a solid lock, or do the caps/num lock leds blink?
[19:14] <BenC> BTW, anyone interested, 2.6.26-3.7 is about to be uploaded
[19:15] <UlfJack> i'm sorry for the load of join/quit messages, but i have to reboot every 5 min it seems :(
[19:16] <johanbr> What's your laptop model?
[19:17] <BenC> UlfJack: ﻿is it a solid lock, or do the caps/num lock leds blink?
[19:17] <UlfJack> it's an old lifebook s60, 1 GHz, 512 MB
[19:18] <UlfJack> they don't blink, it just stops (no change on the screen) and ignores alls keyboard/mouse input
[19:20] <johanbr> Have you found *any* kernel that's stable?
[19:20] <UlfJack> iirc 7.10 was running stable
[19:21] <BenC> UlfJack: have you tried switching to console and waiting to see if the hang still occurs? Maybe get some output that way
[19:21] <BenC> make sure to boot without the quiet kernel option
[19:25] <UlfJack> does quiet make a difference for console output on crash/freeze?
[20:28] <UlfJack> of course now it doesn't freeze :(
[20:28] <UlfJack> oh well, off to bed
[22:52] <Kano> hi rtg , BenC , do you want to see a pic with the error of the vostro 1000?
[22:54] <ceekay> BenC: after all that advice yesterday (and my excitement that it was working), i can't seem to get kdump working on another one of my systems running (the one i _need_ it to work on)... seems to kernel panic when loading the capture kernel... i see "unknown_bootoption" in the stack trace and everything else above that seems to sound like initial filesystem sync/load. guess i may have to take back my "kdump works on 7.04" statement... any thoug