BenC | and it uses rdtsc(), so it's x86 only | 12:04 |
---|---|---|
BenC | mjg59: any idea what a good portable replacement is for rdtsc()? | 12:04 |
BenC | and when it says "work with current 802.11 stack", it means that in the loosest way | 12:05 |
BenC | it's still using local ieee80211 headers, and abusing the real stack | 12:05 |
mjg59 | BenC: What fun | 12:06 |
BenC | I'm not sure about including it...my biggest fear is that I'll become so frustrated with it that I'll end up getting an 8180 card and rewrite the whole driver | 12:07 |
mjg59 | What's it actually using rdtsc for? | 12:07 |
mjg59 | Ha | 12:07 |
BenC | mjg59: filling in some hosttime in a struct | 12:07 |
mjg59 | Does it look important, or does it just need a monotonic timer? | 12:08 |
BenC | mjg59: don't laugh, that's how I ended up taking over the ieee1394 stack :) | 12:08 |
=== rtcm_ [n=jman@81.84.150.197] has joined #ubuntu-kernel | ||
mjg59 | Oh dear | 12:08 |
BenC | it's something to do with a prism header | 12:08 |
mjg59 | Nngh. | 12:09 |
mjg59 | BenC: It only seems to be for monitor mode | 12:10 |
mjg59 | BenC: Other code just seems to use jiffies | 12:14 |
BenC | it has it commented out to use jiffies near the top of that function | 12:14 |
BenC | guess I'll do that aswell | 12:14 |
mjg59 | I'd just go back to that, TBH | 12:14 |
mjg59 | It seems to be a fairly standard 802.11 thing | 12:14 |
BenC | wait, that's mac_time | 12:15 |
BenC | oh well, I can use it still | 12:15 |
mjg59 | If you dig for hosttime in drivers/net/wireless, there's no shortage of hits | 12:16 |
BenC | sweet | 03:39 |
=== BenC discovers git-bisect | ||
BenC | that is the most awesome tool and makes git totally worth using, even if it was the crappiest scm in the world | 03:39 |
=== BenC [n=bcollins@debian/developer/bcollins] has joined #ubuntu-kernel | ||
=== _native_ [n=alpha@cpe-66-87-4-181.ut.sprintbbd.net] has joined #ubuntu-kernel | ||
=== _native_ [n=alpha@cpe-66-87-4-181.ut.sprintbbd.net] has left #ubuntu-kernel ["cabal] | ||
Earthpig | benc: would be nice if it could be automated. | 07:52 |
=== chmj [n=chmj@wbs-146-184-221.telkomadsl.co.za] has joined #ubuntu-kernel | ||
=== CataEnry [n=cataenry@host204-30.pool8248.interbusiness.it] has joined #ubuntu-kernel | ||
CataEnry | hi all | 11:58 |
CataEnry | cya | 12:17 |
=== michel [n=michel@zux006-048-229.adsl.green.ch] has joined #ubuntu-kernel | ||
michel | i want to patch ltsp-kernel of ubuntu, howto? | 12:23 |
=== jane_ [n=JaneW@wbs-146-151-44.telkomadsl.co.za] has joined #ubuntu-kernel | ||
=== Alhica [n=Torresm@a81-84-252-5.cpe.netcabo.pt] has joined #ubuntu-kernel | ||
=== Alhica [n=Torresm@a81-84-252-5.cpe.netcabo.pt] has left #ubuntu-kernel [] | ||
fabbione | BenC: is it possible to do firewire back to back? | 01:50 |
fabbione | between 2 controllers i mean | 01:51 |
fabbione | if so do i need a special cable or something? | 01:51 |
=== chmj_ [n=chmj@wbs-146-161-75.telkomadsl.co.za] has joined #ubuntu-kernel | ||
=== CataEnry [n=cataenry@host204-30.pool8248.interbusiness.it] has joined #ubuntu-kernel | ||
CataEnry | hi all | 01:57 |
BenC | fabbione: yeah | 01:58 |
BenC | just a normal cable | 01:58 |
BenC | think of firewire as a network where any device with more than one port is like a hub :) | 01:59 |
BenC | I've had a winxp, macosx, 2 linux boxes, 3 firewire storage devices, firewire camera, all connected on a bus together | 02:00 |
BenC | using IPover1394 between the winxp/macosx/linux boxes too | 02:00 |
fabbione | ah cool | 02:00 |
fabbione | ok | 02:00 |
fabbione | i only have 2 boxes | 02:00 |
fabbione | nothing fancy | 02:00 |
fabbione | curious to try it | 02:01 |
fabbione | probably more | 02:01 |
fabbione | how fast can it go? | 02:01 |
infinity | ~400 Mbps, in theory. | 02:02 |
fabbione | that would be a cheap way to upgrade my 100Mb LAN to a ~ 400Mb | 02:02 |
fabbione | or buy a gigabit switch | 02:02 |
jbailey | fabbione: Do you actually have any disks that can push data that fast? | 02:02 |
fabbione | but the latter is hutterly expensive | 02:02 |
fabbione | jbailey: yes | 02:03 |
jbailey | Handy. =) | 02:03 |
BenC | fabbione: the linux IPover1394 isn't all that stable | 02:03 |
fabbione | BenC: well.. it's a good way to push you to fix it :P | 02:03 |
BenC | fabbione: not my driver :P | 02:03 |
fabbione | still your kernel ;) | 02:04 |
BenC | I just did the initial ethernet-over-1394, the RFC stuff came from someone else | 02:04 |
BenC | the ethernet-over-1394 was really stable, since it was just a firewire packet wrapped around a normal ethernet packet | 02:04 |
BenC | but it was linux-only | 02:04 |
fabbione | yeah i could guess so | 02:05 |
fabbione | i guess i will switch to GigaEthernet | 02:05 |
fabbione | build a 1394 network as backup | 02:05 |
fabbione | and use the 100Mb for management | 02:05 |
fabbione | or something like that | 02:05 |
BenC | get some S800 firewire cards, and you'll almost be gigabit :) | 02:05 |
fabbione | BenC: i have the cards.. i need the switch | 02:06 |
fabbione | and a good switch is $$$ | 02:06 |
kiko | hey there | 02:06 |
kiko | quick question | 02:06 |
BenC | hey kiko | 02:06 |
kiko | what's the difference between linux-image-2.6.10-6-686-smp and linux-image-2.6.10-5-686-smp? | 02:07 |
BenC | kiko: about 4 revisions :) | 02:07 |
fabbione | AHAHAHHAHA | 02:07 |
kiko | really? | 02:08 |
BenC | kiko: you'll need to be more specific than -6 and -5, since it could be -5.10 and -6.99 | 02:08 |
kiko | oh. those are just the package names | 02:08 |
BenC | kiko: the functional difference is from -5 to -6, the kernel ABI changes | 02:08 |
kiko | ii linux-image-2.6.10-5-686-smp 2.6.10-34.7 | 02:08 |
kiko | that's what we have installed | 02:09 |
BenC | oh, I don't know how the revisions worked back then, I only know the current convention for breezy/dapper | 02:09 |
kiko | hum hum | 02:10 |
BenC | Bisecting: 2 revisions left to test after this | 02:10 |
=== BenC loves git-bisect | ||
Earthpig | do you have vmware? | 02:11 |
BenC | yes, but not installed right now | 02:12 |
Earthpig | we use vmware 5's snapshots for that kind of iteration testing. 'tis wonderful. | 02:12 |
fabbione | BenC: it's the same convention | 02:13 |
fabbione | kiko: it's an ABI change | 02:13 |
fabbione | you install the new one.. boot in the new one.. you are done | 02:13 |
BenC | fabbione: I know the ABI bump is, but the 34.7 thing isn't :) | 02:13 |
fabbione | oh right | 02:13 |
fabbione | yes | 02:13 |
fabbione | we did add the ABI number to the debian version only in breezy | 02:13 |
BenC | Earthpig: for testing kernel regressions? | 02:13 |
fabbione | the old pkgs have standard numbers | 02:13 |
kiko | fabbione, should I be using the newer kernel or does it not really matter? | 02:14 |
kiko | by newer kernel I mean newer-ABI-kernel | 02:14 |
fabbione | kiko: well it's a security update that covers no less than 10 vulnerabilities | 02:14 |
fabbione | kiko: it's up to you | 02:14 |
fabbione | i suggest to use the new one | 02:14 |
kiko | oh | 02:14 |
kiko | so the -5 kernel was not updated? | 02:14 |
kiko | ah. I think I understand the problem | 02:15 |
kiko | I shouldn't have installed the -X kernel explicitly, right? | 02:15 |
Earthpig | benc: well, we don't do kernels, so no. :) | 02:15 |
BenC | kiko: the security update forced a bump to -6 | 02:15 |
kiko | I see | 02:15 |
kiko | that's okay | 02:15 |
kiko | so it won't automatically update? | 02:15 |
kiko | (i.e. JRU does an apt-get update apt-get dist-upgrade and no new kernel) | 02:15 |
BenC | not unless you have linux-image-686-smp installed | 02:15 |
BenC | the meta-packages should pull in the upgrades | 02:15 |
kiko | yeah, that's what I should have. | 02:15 |
BenC | which is the default | 02:16 |
Earthpig | benc: we do use it for automated testing of nightlies though. | 02:16 |
BenC | Earthpig: cool | 02:16 |
BenC | fabbione: I think -4.4 is going to go out today | 02:16 |
BenC | after which I will attempt to get l-r-m built against 2.6.15, and then go for linux-meta | 02:17 |
infinity | BenC : Please don't. | 02:17 |
infinity | BenC : l-r-m will be unuseably broken as soon as the new X finally builds (which will happen in the next day, it looks like), so I need to do more than just "make it build" anyway. | 02:17 |
BenC | infinity: when's your next planned upload of l-r-m? | 02:17 |
fabbione | BenC: ok.. but it's pointless to upload until the gcj thing is sorted | 02:18 |
fabbione | BenC: basically all the archive is unbuildable atm | 02:18 |
=== BenC sighs | ||
fabbione | BenC: plus infinity loves to spank l-r-m :) | 02:18 |
infinity | BenC : I'll make l-r-m a priority tomorrow, since I expect to see X build tonight. | 02:18 |
BenC | you people enjoy dashing my hopes don't you? :) | 02:18 |
BenC | infinity: ok, I'll leave it alone, and just get linux-meta ready for when things settle down | 02:19 |
infinity | I'd like to see CDs built with shiny new kernels too, so I'm on your side. Don't worry. :) | 02:20 |
infinity | Speaking of CDs, I really need to test that 640x400 vga16fb patch and get you to include it ASAP. | 02:20 |
fabbione | BenC: Kamion wants to coordinate with you some udeb reorganization stuff | 02:20 |
infinity | Certainly before we do the next flight CD anyway. | 02:21 |
BenC | fabbione: yeah, I'm in #ubuntu-boot, but so far, I don't know what coordination I need to do | 02:22 |
BenC | I was under the impression that once I got the kernel uploaded, the rest was on him :) | 02:22 |
fabbione | BenC: basically Kamion wants to revert a few changes in the amount of udebs we do build | 02:22 |
fabbione | BenC: yeah well.. he might need your help to setup git and stuff | 02:22 |
BenC | infinity: yeah, I can test that on atleast one system where I know vga16 has always been broken | 02:22 |
fabbione | he knows what needs to touch | 02:22 |
infinity | BenC : Note that Kamion/Keybuk would really rather not see linux-meta updated until we're sure the udev stuff is sorted. | 02:22 |
infinity | BenC : You have such a system?... ROCK THE FUCK ON. I've been resigned to doing this blind until now. :) | 02:23 |
BenC | infinity: it's an ATI card connected to a TV via svideo, so it's not the normal brokeness, but it is broken | 02:23 |
BenC | I've been using vesa happily | 02:23 |
BenC | fabbione: ok | 02:24 |
fabbione | translating infinity to a more common language: "Dear Ben, given your card is broken, you win! FIX IT! kthxbye!" | 02:24 |
BenC | lol | 02:25 |
infinity | BenC : Hrm, not sure if this fix will fix your case, but bonus if it does. | 02:25 |
BenC | infinity: I have a couple of vgacon "fixes" from -mm that I want to test too...not going into our kernel yet, but I'll let you know | 02:26 |
BenC | rather wait for your patch before I touch vgacon at all | 02:27 |
BenC | hrmm...not sure if I need to mark that last boot as git-bisect bad, or good | 02:27 |
BenC | it booted way past where it normally crashed, but it got a segv in the initrd, and couldn't mount the rootdev | 02:28 |
infinity | BenC : My patch should just be touching the size/timings struct, if you're mucking with stuff outside that, we won't conflict. | 02:29 |
infinity | BenC : I just need to pick the "right" timings, and test the fuck out of them before I go blowing up people's hardware. | 02:29 |
=== rtcm [n=jman@81.84.150.197] has joined #ubuntu-kernel | ||
=== sivang [i=sivan@ubuntu/member/sivang] has joined #ubuntu-kernel | ||
sivang | BenC: Hey Ben, 'sup? I installed 2.6.15 and am now getting "BUG: Soft lockup on CPU#0" | 02:48 |
sivang | BenC: anything I can help to sovle this, or do more testing for you? | 02:48 |
BenC | sivang: backtrace or oops of any kind? | 02:49 |
sivang | BenC: if logs were not rotated, I should be able to give those . let's see | 02:49 |
sivang | BenC: re: oops, machine can't boot past it. not even in "rescue" mode | 02:51 |
sivang | BenC: can't seem to find it in the syslog, where else should I be looking? | 02:53 |
BenC | dmesg | 02:53 |
BenC | when it happens, does the machine lockup? | 02:53 |
=== CataEnry [n=cataenry@host204-30.pool8248.interbusiness.it] has joined #ubuntu-kernel | ||
sivang | BenC: yes sir :) | 02:53 |
CataEnry | re | 02:54 |
=== spike [n=spike@unaffiliated/spike] has joined #ubuntu-kernel | ||
spike | hi there | 03:01 |
BenC | sivang: does Alt+SysRQ work? | 03:03 |
spike | I was trying to get myself acquainted with dapper and thinking of kernel specs for dapper and ubuntu-server | 03:03 |
spike | is selinux/grsec integration planned? | 03:04 |
spike | and what about ELSA/CSA patches? | 03:04 |
spike | or this sort of questions are welcome here? | 03:04 |
spike | arent* | 03:05 |
sivang | BenC: I will have to try. also, I can't seem to find the dmesg logs for that hangup | 03:07 |
sivang | BenC: any tips on that? | 03:07 |
sivang | BenC: (this all mean boot cycles for me) | 03:07 |
BenC | sivang: if it was a lockup, it wont be there | 03:08 |
BenC | you'll need to do a alt+sysrq, and either hand write the trace, or take a photo of it and email that too me | 03:08 |
sivang | BenC: ok, I will give it a try. | 03:09 |
sivang | I'll be bak ;-) | 03:09 |
sivang | BenC: now it locked up, on the corrupted USPlash screen , Alt+SysRq does nothing | 03:32 |
infinity | sivang : If you boot without "splash" you can avoid usplash. | 03:33 |
infinity | sivang : You might at least get some better debugging output then. | 03:33 |
infinity | sivang : And take away "quiet" from the command line too, if you want a bit more verbosity about where it might be dying. | 03:34 |
sivang | infinity: eh right, dumb me, not like I don't know that. (I used to do that when bootsplash just entered, since I am used to see text messages and bootsplash was bit uncomfortable. | 03:34 |
sivang | infinity: anyway, yet-another-boot-cycle | 03:34 |
sivang | ok, couldn't see the "Soft lock" anymore | 03:41 |
sivang | but I did see where it halted, again SysRq helped me none | 03:42 |
sivang | [17179589.808000] Intel ISA PCIC probe: not found. Before that I got "* /etc/network/options is deprecated" | 03:43 |
=== rtcm_ [n=jman@81.84.150.197] has joined #ubuntu-kernel | ||
=== JaneW [n=JaneW@wbs-146-151-44.telkomadsl.co.za] has joined #ubuntu-kernel | ||
sivang | BenC: any idea? | 03:57 |
BenC | did alt+sysrq work? | 03:57 |
BenC | ok, I see, no | 03:57 |
CataEnry | brb | 04:06 |
BenC | sivang: I'm at a loss, try booting with "noapic", "nolapic", etc, to see if that changes anything | 04:08 |
=== CataEnry [n=Enrico@host204-30.pool8248.interbusiness.it] has joined #ubuntu-kernel | ||
sivang | BenC: ok, will do. Why isn't anything on the kernel that will open and close a log file in an atomic manner, that way if something wrong happens, we would always be able to see the logs? | 04:10 |
BenC | sivang: at that point, the system dies, and likely is so trahed that opening a file would be disastrous | 04:12 |
sivang | BenC: I see. What about doing that before it dies, or is there no possible sane way to know that before it acutally dies ? :) | 04:18 |
BenC | it could die before even being able to access the disk | 04:19 |
sivang | ah right | 04:19 |
sivang | darn | 04:19 |
sivang | oh well, I will try again in a sec | 04:19 |
sivang | BenC: btw, what nolapic does? | 04:20 |
BenC | No Local APIC | 04:20 |
sivang | k, thx | 04:20 |
BenC | which kernel are you using anyway, -386, -686, -k7? | 04:20 |
sivang | BenC: 686, the smp and UP enabled | 04:22 |
BenC | try doing "smp2up=off" aswell then | 04:22 |
sivang | /boot/vmlinuz-2.6.15-3-686 | 04:22 |
sivang | and that tells it to? | 04:23 |
BenC | are you on an SMP machine? | 04:23 |
sivang | yes | 04:24 |
BenC | then it wont do anything, so don't worry about it :) | 04:24 |
sivang | BenC: there a way to know for sure if you're on an SMP machine if you're currently booted with a non smp kernel? | 04:27 |
BenC | cat /proc/cpuinfo should list total cpu's | 04:29 |
Mithrandir | it doesn't if you don't have an SMP kernel, iirc | 04:30 |
Mithrandir | yup, verified on the live cd. | 04:31 |
Mithrandir | doesn't tell me about my second core, just my first one | 04:31 |
Mithrandir | (amd64) | 04:31 |
BenC | probably doesn't parse core's, but it should parse cpu's | 04:32 |
BenC | ncpus_probes should atleast | 04:32 |
BenC | if not, it will be fixed in 2.6.15, since there were some patches for that | 04:32 |
Mithrandir | I don't have a real SMP machine here, so I can't test, but I've seen it not list all the CPUs listed before, at least. | 04:33 |
sivang | Mithrandir: I have a non real SMP machine, and when using an SMP kernel I always see two CPUs | 05:00 |
=== chuck_ [n=chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel | ||
chuck_ | heylo | 05:07 |
=== chuck_ is now known as zul | ||
mjg59 | BenC: It looks like the bcm430x people are moving to using much the same 80211 code as the rtl8180 people | 06:02 |
mjg59 | So with luck we'll have one softmac core for all of them | 06:03 |
=== siretart [i=siretart@ubuntu/member/siretart] has joined #ubuntu-kernel | ||
siretart | BenC: I noticed that iptables in breezy as well as in dapper there are some iptables modules missing, which are in debian's iptable | 06:09 |
siretart | BenC: I filed therefore bugzilla #19978 | 06:10 |
siretart | BenC: the most notably difference to debian is that debian uses an own copy of kernel headers. Do you have an idea how to fix this? | 06:11 |
BenC | siretart: doesn't sound like something I can fix from the kernel side | 07:04 |
BenC | everything in the kernel is enabled | 07:04 |
BenC | mjg59: I have a softmac patch in -4.4 that came from the bcm430x ftp site | 07:05 |
siretart | BenC: well, I suspected that iptables source isn't too happy with the headers the kernel does currently provide | 07:05 |
siretart | because the kernel headers are the only difference | 07:05 |
BenC | siretart: is that something that needs to change in the kernel, or just that iptables will need a hacked up copy of the headers for it's own build? | 07:06 |
dilinger | heh | 07:06 |
dilinger | one of my coworkers rebooted their hoary machine today. i totally forgot about the ABINAME bump | 07:06 |
dilinger | needless to say, openafs didn't work all that well for them, as there was no openafs module compiled for the new kernel | 07:06 |
siretart | BenC: debian seems to ship with a hacked up copy of the headers. fabbione changed the iptables package to use the kernel headers instead | 07:09 |
BenC | probably a good reason for that then | 07:10 |
BenC | but you'd have to ask him about what that reason was :) | 07:10 |
BenC | sounds to me like this is all squarely on the iptables package, so it's probably best to submit it as a bug against that package | 07:10 |
siretart | well, there seems to be no direct maintainer, so I wanted to try to fix it | 07:11 |
BenC | sure, if you fix it, then you can attach a patch to the bug report | 07:12 |
BenC | you can assign it to fabbione since he touched the source, and probably get an explanation from him about it | 07:13 |
BenC | or you can do the upload yourself, but I suggest atleast talking to fabbione first to find out why he switched it | 07:14 |
siretart | I uploaded a merged iptables package today which didn't FTBFS ;) | 07:16 |
siretart | I consider this as an improvement :) | 07:17 |
lamont-away | siretart: an old working version and a new ftbfs version is a completely different situation than a new compiled-but-non-working version | 07:26 |
lamont-away | since in the first case, apt-get dist-upgrade doesn't trash a working system | 07:26 |
siretart | lamont-away: yeah, right | 07:29 |
siretart | lamont-away: but a quick testrun showed my that iptables seems to be functional (I can look at my tables) | 07:30 |
lamont-away | cool | 07:33 |
siretart | but my original problem stands: There is no /lib/iptables/libipt_recent.so being built anymore. the new version from debian didn't bring any change | 07:34 |
BenC | what are the criteria for the package building that module? | 07:40 |
BenC | do you see any reference to it in the sources? | 07:40 |
BenC | also, is it possible that regardless of the kernel headers, it is checking uname and seeing that maybe the kernel doesn't support an obsolete interface? | 07:41 |
=== siretart rechecks | ||
CataEnry | bye all | 07:54 |
siretart | something is strange with linux-kernel-headers | 07:56 |
BenC | linux-kernels-headers is an oddity by design | 07:56 |
siretart | somehow all extensions of iptables get disabled when using linux-kernel-headers | 07:57 |
siretart | perhaps thats the reason why iptables ships its own kernel headers | 07:58 |
siretart | the problem is that the package FTBFS with that headers on ubuntu.. hmmm | 07:58 |
BenC | siretart: iptables is probably one of those evil programs that defines __KERNEL__ when building userspace so it can get to some kernel internals | 07:59 |
BenC | as such, linux-kernel-headers wont satisfy what it needs | 07:59 |
=== zul [n=chuck@ubuntu/member/zul] has joined #ubuntu-kernel | ||
kiko | iptables again | 08:01 |
kiko | sign | 08:01 |
siretart | I have an idea, let me try something.. | 08:02 |
infinity | iptables does Very Bad Things, yes. | 08:03 |
siretart | I think I got it | 08:04 |
siretart | debian rules sets KERNEL_DIR to /usr/include, it has to be just '/usr' instead | 08:04 |
siretart | interstingly, iptables seems to need the kernel includes only for some extensions, not for all, and not for itself.. | 08:05 |
siretart | BenC: are you okay that I upload a package with just corrected KERNEL_DIR? quick check shows iptables works, even with '-m match' parameter | 08:06 |
BenC | what do you mean "corrected KERNEL_DIR"? | 08:07 |
siretart | BenC: http://paste.ubuntulinux.nl/4924 | 08:08 |
siretart | thats the debdiff | 08:08 |
BenC | looks good to me | 08:10 |
siretart | ok. uploading | 08:10 |
siretart | this patch should imo go to breezy-updates, too | 08:11 |
siretart | because in breezy, there are no iptable extension modules atm | 08:12 |
jbailey | BenC: Yes, our lkh is designed to make those programs suck with low amounts of sympathy given. | 08:27 |
siretart | well, iptables builds nicely with lkh | 08:27 |
BenC | siretart: then iptables gets a gold star :) | 08:31 |
siretart | :) | 08:31 |
sivang | siretart: lkh are so hard to compile with for something that depends on the kernel? | 08:32 |
siretart | sivang: I didn't touch lkh at all | 08:33 |
zul | aieeeeeeee | 08:49 |
sivang | BenC: ok, I have some leads about the lockup | 08:51 |
sivang | BenC: I have no digicam so that's what I managed to take by hand | 08:53 |
sivang | BenC: pid 5570, comm modprobe | 08:53 |
sivang | BenC: EIP:0060:[<c02ec3d4>] CPU:0 | 08:53 |
sivang | BenC: EIP is at _spin_lock_irqsave+0x14/0x20 | 08:54 |
sivang | BenC: does that help? | 08:54 |
BenC | do you have the first few functions from the stack trace? | 08:54 |
BenC | sounds like it is crashing in a module init | 08:54 |
BenC | need to know which module | 08:54 |
sivang | BenC: how do I produce a stack trace there? I was virtually helpless when reaching that point | 08:55 |
BenC | it didn't spit one out? | 08:55 |
BenC | interesting | 08:55 |
BenC | any idea what the error was above the pid 5570 line? | 08:56 |
sivang | BenC: sure, the one I told you before | 08:56 |
BenC | like "BUG: ..." | 08:56 |
BenC | softlockup, that's right | 08:56 |
sivang | BenC: the first echo was that is started PCMCIA detection stuff | 08:56 |
sivang | BenC: yes | 08:57 |
sivang | BenC: exactly | 08:57 |
sivang | BenC: on CPU#0 | 08:57 |
sivang | BenC: I wonder if I can reproduce that on this machine as well | 08:57 |
sivang | I will isntall this kernel here as well | 08:57 |
BenC | ok, -386 kernel works? if so, boot, and somehow disable pcmcia stuff | 08:57 |
sivang | BenC: I can remove it's modules if they are there form /etc/modules ? | 08:57 |
BenC | probably a script in /etc/hotplug/ for it | 08:57 |
BenC | or try mv /lib/modules/(version)/kernel/drivers/pcmcia to somewhere outside of /lib/modules | 08:58 |
sivang | ok, thanks for the tips | 08:58 |
sivang | will try and let you know | 08:59 |
BenC | ok, thanks | 08:59 |
mjg59 | BenC: Yeah, I think that's being replaced with a more general solution | 09:05 |
mjg59 | BenC: But basic Broadcom functionality seems to be getting fairly close | 09:05 |
BenC | mjg59: well, they need something because ipw2200, bcm430x, prism54-softmac and rtl808x all seem to need it | 09:06 |
BenC | rtl818x | 09:07 |
BenC | and I'm getting tired if custom ieee80211 stacks in all our external wireless drivers :) | 09:07 |
mjg59 | ipw2200 should just need the ieee80211 stack | 09:07 |
BenC | it does, but it has internal softmac stuff | 09:07 |
mjg59 | bcm430x, prism54-softmac and rtl808x need an extra softmac stack | 09:07 |
mjg59 | Oh, does it? Nngh. | 09:07 |
mjg59 | I thought it did all that in firmware | 09:08 |
mjg59 | Oh well, never mind | 09:08 |
BenC | the softmac patch for ieee80211 was partly based on ipw2200 | 09:08 |
BenC | it may not need all the softmac, but it atleast has some functions that softmac needed (subset in ipw2200) | 09:09 |
siretart | not to mention madwifi... | 09:10 |
infinity | madwifi is a whole different kettle of fish. | 09:11 |
mjg59 | madwifi needs porting to Linux | 09:11 |
siretart | it has its own ieee80211 stack. Sure, it is fishy because of other reasons, too.. | 09:11 |
mjg59 | At the moment it's a *BSD driver that's shoehorned into the kernel | 09:11 |
infinity | BenC : Should I assume from your vacation announcement that I have until Monday to spit polish LRM? | 09:15 |
=== spike [n=spike@unaffiliated/spike] has left #ubuntu-kernel ["bbl"] | ||
BenC | infinity: yeah, definitely | 09:15 |
BenC | about to upload -4.4, so you can base work off of that | 09:15 |
infinity | Excellent. And we can shoot for an update -meta on Monday. | 09:15 |
infinity | (Assuming other parties agree, I'll just upload -meta right after lrm builds everywhere) | 09:16 |
infinity | I've decided that, after pulling an all-nighter, my Thursday is going to be a write-off. | 09:16 |
infinity | Go me. | 09:17 |
infinity | Anyhow. I should go catch some "oh god, the sun's already been up for two hours, eek" shuteye. | 09:17 |
jbailey | BenC: Is -4.4 one I should test for ppc64 love? | 09:18 |
BenC | jbailey: I don't expect it to work, but surely give it a try | 09:22 |
BenC | infinity: heh, get some sleep | 09:22 |
jbailey | BenC: If you're not expecting it to work, I'll just do it when I have time and not rush to it, then. | 09:27 |
jbailey | All good. =) | 09:28 |
BenC | no problem :) | 09:33 |
=== doko_ [n=doko@dslb-084-059-072-254.pools.arcor-ip.net] has joined #ubuntu-kernel | ||
fabbione | siretart: i have no issued with you working on iptables.. | 09:36 |
=== LaschW [n=LaschW@dyndsl-085-016-016-057.ewe-ip-backbone.de] has joined #ubuntu-kernel | ||
LaschW | Are there any reports of segmentation fault booting linux-image-2.6.12-10-k7 ? | 09:43 |
siretart | fabbione: a fixed version of iptables should be in dapper now on all arches | 09:44 |
fabbione | LaschW: no. works here. we did score 9/9 successes on the tests | 09:44 |
siretart | fabbione: I think we should upload a fixed version of iptables to breezy-updates | 09:44 |
fabbione | siretart: what is exaclty broken? | 09:44 |
siretart | fabbione: debian/rules sets the variable KERNEL_DIR to /usr/include | 09:45 |
siretart | fabbione: this is wrong. it must be just '/usr' | 09:45 |
fabbione | specially given that nobody did notice up till now | 09:45 |
siretart | fabbione: without the kernel headers, a lot of optional extra modules won't get built | 09:45 |
fabbione | siretart: and problems does that bring? | 09:45 |
siretart | fabbione: extra modules like recent match support and quite a few others | 09:45 |
fabbione | hmmmm | 09:46 |
siretart | I came to it because I wanted to play around with that 'recent' match target as suggested by someone on planet.debian.org | 09:46 |
siretart | and was surprised that this doesn't work on breezy | 09:47 |
fabbione | i think it is a good candidate for -updates | 09:49 |
fabbione | siretart: i am confident you did your homework | 09:49 |
LaschW | fabbione: Is there a way how I may get / collect usefull information why this happens? | 09:49 |
fabbione | please explain to mdz and coordinate with him | 09:49 |
fabbione | LaschW: it depends what happens | 09:49 |
fabbione | LaschW: segfault what? what does segfault? | 09:50 |
fabbione | you didn't tell me much | 09:50 |
LaschW | fabbione: segfault before usplash started, si IMHO there are no logs in that state of boot, isn't it? | 09:50 |
LaschW | s/si/so/ | 09:50 |
fabbione | the kernel doesn't segfault | 09:50 |
fabbione | no there are no logs.. try to boot in recovery mode without usplash | 09:51 |
fabbione | what's on the screen? | 09:51 |
fabbione | can you take a picture? | 09:51 |
=== zul__ [n=chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel | ||
siretart | mdz: would you accept a one-line patch to iptabes for breezy-updates? Rationale: due to a wrong variable, iptables fails to find its kernel headers and does not compile nearly all optional modules. | 09:52 |
LaschW | fabbione: I will reproduce it and will see what I can do, give me a 1/4hour... | 09:52 |
=== LaschW [n=LaschW@dyndsl-085-016-016-057.ewe-ip-backbone.de] has left #ubuntu-kernel ["Leaving"] | ||
fabbione | LaschW: in 1/4 hour i will be asleep | 09:53 |
fabbione | bah ok | 09:53 |
=== ispiked [n=ispiked@unaffiliated/ispiked] has joined #ubuntu-kernel | ||
=== ubuntulog [n=ubuntulo@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-kernel | ||
=== Topic for #ubuntu-kernel: Ubuntu kernel development discussion ONLY | New git tree for dapper: https://wiki.ubuntu.com/KernelGitGuide | 2.6.15-rc1-ubuntu2 uploaded (should build for x86, amd64, and ppc) | ||
=== Topic (#ubuntu-kernel): set by BenC at Tue Nov 15 17:08:39 2005 | ||
mdz | siretart: yes | 10:05 |
mdz | siretart: send me a debdiff by mail for review | 10:06 |
siretart | mdz: http://paste.ubuntulinux.nl/4924 | 10:08 |
siretart | thats what I just uploaded to dapper, I'd upload of course with adjusted version number | 10:08 |
mdz | siretart: go ahead and prepare the package and upload as 1.3.1-2ubuntu1.1 | 10:09 |
mdz | to breezy-updates | 10:09 |
siretart | on my way | 10:10 |
=== LaschW [n=LaschW@dyndsl-085-016-009-044.ewe-ip-backbone.de] has joined #ubuntu-kernel | ||
LaschW | fabbione: linux-image2.6.12-10-k7 segfault during boot: | 10:16 |
fabbione | kernel doesn't segfault... | 10:16 |
fabbione | it OOPS | 10:16 |
LaschW | fabbione: Last message isa: uncompressing Linux... OK booting.... | 10:16 |
LaschW | fabbione:then 10 lines segmentation fault | 10:17 |
LaschW | fabbione: then a line: 0 | 10:17 |
fabbione | LaschW: -> initramfs | 10:17 |
fabbione | try regenerate the initramfs | 10:17 |
LaschW | fabbione: Ahh! | 10:17 |
dilinger | mm, neat | 10:20 |
dilinger | http://www.selenic.com/linux-tiny/ | 10:20 |
LaschW | fabbione: regenerating initramfs for 2.6.12-10-k7 will affect the other kernels? I'm asking because till now I have a running system, using the older kernel images. | 10:21 |
fabbione | no | 10:21 |
fabbione | no if you do it properly | 10:21 |
LaschW | fabbione: properly means: 'sudo dpkg-reconfigure linux-image-$(uname -r)' | 10:23 |
fabbione | yes | 10:24 |
fabbione | that is one | 10:24 |
fabbione | there are other ways | 10:25 |
fabbione | but use that one | 10:25 |
LaschW | fabbione: OK, thanks a lot... | 10:25 |
=== LaschW [n=LaschW@dyndsl-085-016-009-044.ewe-ip-backbone.de] has left #ubuntu-kernel ["Leaving"] | ||
fabbione | i have the feeling i am going to get pissed soon | 10:26 |
fabbione | there is a user that keeps bombing me of emails in pvt | 10:26 |
BenC | fabbione: bounce them all back to him, maybe he'll stop | 10:27 |
BenC | fabbione: I saw one report that sound stopped working with 2.6.12-10 | 10:27 |
fabbione | BenC: i did answer him and CC kernel-team | 10:28 |
fabbione | let see if he gets the clue (again) | 10:28 |
fabbione | BenC: quite impossible.. sound didn't change | 10:28 |
fabbione | bug number? | 10:28 |
BenC | 19969 | 10:29 |
zul__ | meh...since when did kernel-team ml become support | 10:29 |
BenC | it's very vague | 10:30 |
BenC | like, I think I just gave you all the info in the bug report | 10:30 |
zul__ | shoot em | 10:31 |
zul__ | how ubuntite of me ;) | 10:34 |
zul__ | later anyways | 10:35 |
fabbione | wow | 10:35 |
fabbione | what a bug | 10:35 |
fabbione | BenC: let the bug die there for the next 2/3 weeks and close it with prejudist | 10:36 |
fabbione | bug is pointless. kthxbye | 10:36 |
fabbione | or ask Diziet to do it for you | 10:36 |
fabbione | ;) | 10:36 |
=== LaschW [n=LaschW@dyndsl-085-016-003-190.ewe-ip-backbone.de] has joined #ubuntu-kernel | ||
LaschW | fabbione: 'sudo dpkg-reconfigure linux-image-2.6.12-10-k7' didn't change anything. Still segfault messages... | 10:42 |
fabbione | LaschW: works here... ask jbailey how to debug initramfs | 10:43 |
LaschW | fabbione: dpkg-reconfigure says: Not touching initrd symlinks since we are being reinstalled (2.6.12-10.24) | 10:43 |
BenC | lol | 10:43 |
fabbione | the kernel is ok | 10:43 |
BenC | the only way to know for sure is to regen the initramfs for the old kernel and see if that works | 10:44 |
BenC | you could also try the -386 kernel | 10:44 |
fabbione | yeah that's another option too | 10:44 |
fabbione | but i am pretty sure the kernel is fine | 10:45 |
fabbione | specially because my k7 is one of the test machines | 10:45 |
fabbione | oh JEEE | 10:46 |
fabbione | this guy is persistent | 10:46 |
fabbione | CC to kernel-team | 10:46 |
fabbione | he answer back without kernel-team | 10:46 |
=== fabbione reroutes to /dev/null | ||
BenC | he's embarassed :) | 10:47 |
fabbione | I AM PISSED | 10:47 |
fabbione | why nobody cares about pissed Developers but only about embarassed (l)users? ;) | 10:48 |
=== fabbione modprobes overfiend.ko | ||
fabbione | THEY SHOULD ALL DIE A PAINFUL DEATH! | 10:48 |
=== fabbione LARTS | ||
=== fabbione LARTS | ||
=== fabbione rmmod overfiend.ko | ||
fabbione | ah | 10:48 |
fabbione | steaming down is healty | 10:48 |
LaschW | But how does it come that the 2.6.12-9-k7 works and the -10-k7 not? Just for beeing curious... | 10:49 |
fabbione | LaschW: the kernel works fine. the segfault is not coming from the kernel | 10:49 |
fabbione | the initramfs is probably corrupted or it is not generated properly | 10:50 |
fabbione | otherwise you won't see a segfault | 10:50 |
fabbione | but an OOPS | 10:50 |
fabbione | and it can be anything that breaks the initramfs | 10:50 |
fabbione | really... | 10:50 |
fabbione | LaschW: try booting 2.6.12-10-386 | 10:50 |
fabbione | and see if it works | 10:50 |
fabbione | none of the security patches to touch boot | 10:51 |
fabbione | boot process i mean | 10:51 |
fabbione | anyway i am off to bed | 10:51 |
fabbione | good night * | 10:51 |
=== mdke [n=matt@ubuntu/member/mdke] has joined #ubuntu-kernel | ||
mdke | hello | 11:14 |
mdke | i had some bugs about acx_pci closed without a comment recently. They are still reproduceable on dapper, how come they were closed, does anyone know? | 11:15 |
jbailey | mdke: Bug number? | 11:16 |
mdke | hmm | 11:17 |
mdke | lemme find them | 11:17 |
mdke | jbailey, i can't find them, i'll have to come back when I have my email with me | 11:21 |
=== mdke finds an open one | ||
mdke | perhaps it was just housecleaning, and closing bugs instead of marking them as dups | 11:24 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!