[08:30] <fabbione> BenC: ping?
[08:32] <fabbione> BenC: if you have a chance please test .20-5 on your PB 17". It OOPS here like hell
[08:32] <fabbione> and it can't get to boot the machine
[08:33] <fabbione> best part is that it you get dropper to busybox, but usb keyboard doesn't work..
[10:09] <tepsipakki> BenC: bug 76341 is still there with 2.6.20-5
[10:14] <tepsipakki> (Intel ISA PCIC probe: not found)
[12:37] <fabbione> kylem: 
[12:37] <fabbione>   CC [M]   ubuntu/fs/gfs/glock.o
[12:37] <fabbione> ubuntu/fs/gfs/glock.c: In function 'gfs_glock_be_greedy':
[12:37] <fabbione> ubuntu/fs/gfs/glock.c:1820: warning: passing argument 1 of 'schedule_delayed_work' from incompatible pointer type
[12:37] <fabbione> this looks like a consequence of the INIT_WORK changes you did
[02:29] <BenC> fabbione: fixing it
[02:53] <fabbione> BenC: did you also look at the PPC bug?
[02:54] <BenC> ppc bug?
[02:54] <BenC> ah, I see
[02:54] <BenC> are the daily cd's using -5 yet?
[02:55] <BenC> fabbione: BTW, where is it oopsing?
[02:55] <fabbione> BenC: yes they should.. right at the beginning.. before starting init
[02:55] <fabbione> BenC: so boot without splash and quiet
[02:55] <BenC> fabbione: can you give me some info on what/where the oops is?
[02:56] <BenC> I may be able to fix it without downloading the ISO, and booting my wifes computer with it :)
[02:57] <fabbione> BenC: i can't get anywhere to save it . so if you can wait for a picture..
[02:58] <fabbione> i think the OOPS is storage related since it can't find root afterwards
[02:58] <fabbione> that's all i could see
[03:06] <zul> i so much dislike winter 
[03:07] <kylem> we haven't really had one yet...
[03:07] <zul> it was freaking cold waiting for the bus
[03:08] <kylem> hmm, -12. /me makes a note not to go outside.
[03:41] <BenC> heh, it just started feeling like winter here
[03:42] <BenC> got down to freezing last night
[03:42] <zul> distro sprint?
[03:42] <Mithrandir> BenC: it's not cold now. :-(
[03:43] <Mithrandir> around 0C here ATM.
[03:43] <Mithrandir> I'm hoping for snow, though
[03:43] <zul> thank you el nino
[03:43] <Mithrandir> it'd be nice with -5C, clear skies and crisp air.  And lots of snow.
[03:44] <BenC> Mithrandir: I don't want snow mucking with my flights, so let's save that till after I leave :)
[03:45] <Mithrandir> BenC: you forget we're used to snow here, so once it starts snowing, stuff slows down for half a day or a day, then life goes on.
[03:45] <BenC> Mithrandir: Wish it were that way here...people seem to think that snow means "end of the world"
[03:45] <Mithrandir> lack of snow means "end of the world".
[03:46] <BenC> ppl here do not know how to drive in the snow too...the first day of snow, there's cars all in the ditches and wrapped around trees
[03:47] <Mithrandir> oh, sure, first day is chaotic here too, but the next day or so, people are "oh, this snow thing, this is how it works".
[03:47] <BenC> "I have a big expensive SUV, I can handle 60Mph through the snow"
[03:47] <BenC> ok, so it's not just here :)
[03:47] <zul> BenC: i dont know how to drive in the snow :)
[03:48] <BenC> I do, 5mph and use neutral before breaking when your car is rear-wheel drive
[03:48] <zul> first winter i was driving in canada i  winded up in a ditch with my wifes car in the middle of no where
[03:48] <zul> $200 to get the car out of the ditch
[03:48] <BenC> hehe
[03:48] <Mithrandir> BenC: or just don't brake.
[03:49] <Mithrandir> it's fun going through a winter in a hilly city with just about zero brakes on the bike. :-)
[03:50] <BenC> Mithrandir: ouch...I remember trying to drive in the snow when I had a motorcycle...but that was flat...the hills must suck for a bicycle in the snow
[03:50] <zul> my wife goes "wheeee!" whenever she starts to skid and freaks me out
[03:50] <BenC> zul: lol
[03:50] <Mithrandir> haha :-)
[03:57] <grazieno> hi people, can you help me with this bug? https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.17/+bug/75935
[05:00] <Keybuk> BenC: how are you on the subject of process execution in Linux/
[05:00] <BenC> Keybuk: depends on the difficulty of your question
[05:01] <Keybuk> I want to change argv[1] 
[05:01] <BenC> so that it shows different in ps?
[05:02] <Keybuk> yes
[05:03] <Keybuk> the cmdline, fwict, is stored as a character sequence with \0 between the arguments
[05:03] <Keybuk> ps just treats \0 as a space, therefore
[05:03] <Keybuk> so if I write \0 over argv[1] , 2, etc. all I get in ps is "progname                              "
[05:03] <Keybuk> where the number of spaces afterwards is equal to the length of the previous arguments
[05:03] <Keybuk> normally this results in a blank line :p
[05:04] <Keybuk> I can't seem to find a way to _shorten_ the cmdline
[05:04] <BenC> I think sshd does this...it actually rewrites the command line
[05:04] <BenC> Have you tried changing argc as well?
[05:04] <kylem> Keybuk, try a newline?
[05:05] <BenC> hmm, argc is on the local stack, not a pointer
[05:05] <Keybuk> BenC: argc is just a local variable to the function isn't it?
[05:05] <Keybuk> kylem: shows up in ps asa a ?
[05:05] <BenC> Keybuk: Check sshd and see what it does. I remember this being a common issue for people the last time I search around for it
[05:06] <BenC> and I think I used sshd's approach to do it
[05:06] <Keybuk> doesn't just change argv[0] 
[05:06] <Keybuk> it
[05:06] <BenC> No, it changes the whole command line
[05:06] <Keybuk> changing argv[0]  and shuffling up the rest of the command line appears to be well understood
[05:06] <Keybuk> but decreasing the length of the command line seems to be a new one :p
[05:06] <BenC> could be
[05:17] <zul> cool i didnt know you could pipe in shell commands for cvs commits
[05:17] <Keybuk> 18428 pts/9    S+     0:00 test              $
[05:18] <Keybuk> ^ that's what I'm seeing at the moment :-/  the extra spaces are where I've written \0 over the remainder of argv
[05:18] <Mithrandir> setproctitle, maybe?
[05:19] <kylem> -ENOSYS
[05:19] <kylem> :)
[05:21] <thom> hrm, the setproctitle manpage says "Peter Wemm <peter@FreeBSD.org> stole the idea from the Sendmail 8.7.3 source code by Eric Allman <eric@sendmail.org>.
[05:21] <thom> "
[05:21] <thom> so there might be something useful there
[05:21] <Keybuk> thom: I would guess that has the same code as sysvinit
[05:21] <Keybuk> which seems to be all about increasing the space available by moving the environment up
[05:21] <Keybuk> I want to do the opposite, I have arguments that I want to *hide* from ps :p
[05:22] <BenC> Keybuk: Maybe move the environment down?
[05:23] <Keybuk> BenC: doesn't seem to be a way to move the actual kernel pointer
[05:23] <Keybuk> just the local stuff in the heap
[05:30] <kylem> we could add a syscall if you really want
[05:30] <Keybuk> note-to-self: ps is in procps, not psutils <g>
[05:31] <kylem> heh
[05:36] <Keybuk> HAHAHAHA
[05:36] <Keybuk> W!NN0R!!!
[05:36] <Keybuk> I used the "find some magic loop-hole/bug in the kernel source" trick
[05:37] <Keybuk> it assumes that if the last character in cmdline is not NULL, that it's been overwritten/extended into the environment, so calls strlen() on it instead
[05:38] <kylem> heh
[05:41] <Keybuk> char *cmdline_end;
[05:42] <Keybuk> cmdline_end = argv[argc-1]  + strlen (argv[argc-1] );
[05:42] <Keybuk> *cmdline_end = '!';
[05:52] <fabbione> BenC: is there any chance to get mol running on ppc before Norway? (including a kernel that can boot my PPC?)
[05:52] <fabbione> otherwise i am going to have to install some ssh on MacOS
[06:26] <BenC> fabbione: If you can get me a screen shot of the oops, I can probably fix it...and as for Mol, I can add that in now
[06:27] <BenC> fabbione: gfs1 is broken...i fixed all the warnings, but the aio_read and aio_write ops are the wrong proto, and I can't see how to easily convert them to the new proto
[06:28] <BenC> fabbione: gfs2 just uses generic ops for aio_read and aio_write, and for the read/write callbacks
[06:28] <BenC> fabbione: I'm going to commit the fixes I have now, but the remaining two warnings need fixing, else gfs1 remains broken
[06:32] <fabbione> BenC: i will take a pic tomorrow.. mol would be cool.. gfs1 i know.. it's a bitch and i am talking about it with upstream.
[06:33] <_MMA_> BenC: I get random shutdowns at boot with the feisty -generic and -lowlatency kernels. What info is need in a bug report? Where can I look?
[06:33] <BenC> _MMA_: are you passing panic= on the kernel command line?
[06:33] <_MMA_> In menu.lst?
[06:33] <BenC> _MMA_: Yes
[06:33] <_MMA_> Ill look.
[06:34] <BenC> _MMA_: Main thing is, edit the command line in grub's menu (no need to edit the menu.lst file), remove the "quiet splash" options and try to see what exactly is going on
[06:34] <_MMA_> k
[06:35] <_MMA_> It happens about 25% of the time. Ill try to notice but it happens so fast.
[06:45] <Lure> Keybuk: this might help explaining: http://lightconsulting.com/~thalakan/process-title-notes.html
[06:46] <Lure> Keybuk: hp-ux even has syscall to do it ;-)
[06:47] <Keybuk> Lure: that's all about how to *increase* the argv area, so that you can write an arbitrarily long string into argv[0]  (and overwriting the rest of the args)
[06:47] <Keybuk> I wanted to *decrease* the argv area, removing the process's arguments from the ps list
[06:49] <Keybuk> so all that trick isn't needed
[06:49] <Keybuk> writing " " or \0 over top of the arguments "works", but it means that ->cmdline is still
[06:49] <Keybuk> /sbin/init\0\0\0\0\0\0\0\0\0\0
[06:49] <Keybuk> which means ps displays that as
[06:49] <Keybuk> "/sbin/init          "
[06:50] <Keybuk> if the arguments are quite long, you can end up with two or three "blank lines" in the output of "ps ax | less"
[06:51] <Lure> Keybuk: ok, you abuse is nice ;-)
[06:51] <Keybuk> yeah, I trick the kernel into thinking that I've used the above trick by putting something not '\0' in the last char of argv
[06:51] <Keybuk> so the kernel assumes that argv now runs into envp
[06:52] <Keybuk> it deals with that by only returning argv[0]  in cmdline, and using strlen to calculate the length of it
[06:52] <Keybuk> because argv[0]  is right (I'm not changing it), that is a neat trick to hide the arguments from ps <g>
[08:06] <BenC> fabbione: I forgot that mol was in ubuntu/, but was disabled because it didn't build anymore..it builds not, but I need you to test it
[08:07] <BenC> fabbione: Pushed if you want to test it. You should be able to install linux-headers and cd ubuntu/misc/mol/; make -C /lib/modules/$(uname -r)/build SUBDIRS=`pwd`modules CONFIG_MOL=m
[09:41] <fabbione> BenC: that assumes i can boot a kernel....
[09:42] <BenC> fabbione: have you tried -5?
[09:42] <BenC> don't you still have an edgy kernel on there?
[09:42] <fabbione> BenC: the last one i can boot is .20-1 or .20-2 IIRC
[09:43] <fabbione> all the others after do oops
[09:43] <fabbione> some at randoms
[09:43] <fabbione> some constantly
[09:43] <fabbione> i will check better tomorrow and let you know
[09:43] <BenC> ok
[09:44] <fabbione> worst case scenario we will cook up something the first evening in Olso
[09:44] <fabbione> Oslo
[09:45] <fabbione> i need to run skype on ppc via mol
[09:45] <fabbione> that's all i need
[09:45] <fabbione> anyway i am off for the evening
[09:45] <fabbione> thanks dude
[09:47] <BenC> np
[10:03] <zul> later
[10:45] <macd> zul_, the bug concerning more than one domU on xen, any status on that?
[12:10] <Jerub> hi!
[12:11] <Jerub> ubuntu feisty has the 'ipset' package, which currently requires rebuilding the kernel with a netfilter patch-o-matic-ng thingy.
[12:12] <Jerub> I'm currently working on a dynamic firewalling project that requires ipset support to work efficiently, and it'd be really nice if I could just say, 'it works in feisty' instead of having to say, 'rebuild your kernel like this'.
[12:13] <lifeless> Jerub: how stable is the patch ? 
[12:15] <Jerub> lifeless: good question, I'm just going to have a look at the svn log
[12:16] <Jerub> well, we run it in production and haven't had any problems with it.
[12:17] <Jerub> svn log says there was a commit in October '06 for a bugfix, commit before that was May.
[12:17] <Jerub> so it's not a fast moving target.