[12:26] <mjg59> jbailey: It should still ask us in case the user doesn't have usplash
[12:26] <mjg59> Or has removed the "splash" argument
[12:28] <jbailey> Why, though?
[12:29] <jbailey> Without fbcon, it's not useful for the console.
[01:05] <BenC> I think I have the solution for our kernel post-breezy, with regards to git
[01:15] <mjg59> jbailey: If it's not loading fbcon, it ought to be
[01:36] <BenC> Subject: linux-source-2.6.12_2.6.12-9.19_source.changes ACCEPTED
[04:38] <jbailey> mjg59: Right, and it's not.  It's just vesafb and capabilities.
[04:39] <jbailey> mjg59: Fabio mentioned that he didn't know why those were in there, just that they were.  Given that usplash loads vesafb or vga16fb if it needs to, I'm inclined to just move the capabilities loading hint out of the kernel and into initramfs.
[04:40] <jbailey> I know that ppc kernels generally have the framebuffer built in.  Is that generally the right thing, or should there be some initramfs magic that loads the framebuffers right away and does something to autodetect what it can (loading all the per-arch framebuffers or something
[04:40] <jbailey> All obviously post-breezy. =)
[04:45] <desrt> BenC; nooo!!
[04:45] <desrt> BenC; first you must enable hugepages :)
[05:01] <jbailey> desrt: It's okay.  He's a *kernel hacker*
[05:01] <jbailey> Linus set a fine tradition that frozen is what you get out of a slurpy machine. ;)
[05:54] <desrt> :)
[05:54] <desrt> btw.  fix inotify.h :)
[06:04] <jbailey> yeah yeah yeah.
[06:09] <jbailey> Got a current patch for me? =)
[06:15] <desrt> ya.  it's on bugzilla
[06:15] <desrt> well... not a 'patch' per se
[06:16] <desrt> but the one-line that needs to be added
[06:16] <desrt> you actually want it in patch form?
[06:26] <desrt> jbailey; k.  updated patch uploaded here: http://bugzilla.ubuntu.com/show_bug.cgi?id=14456
[06:26] <desrt> it's perfectly in sync with the one in the kernel now with the exception that all the #ifdef KERNEL business is removed (which you did)
[06:53] <jbailey> Cool, thanks.
[06:53] <jbailey> Yeah the upstream lkh that we uses gets rid of all of that.
[06:53] <jbailey> And makes usre all the rest of the types are trivially usable from userspace.
[06:53] <jbailey> Sleep now, will hack in the morning. =)
[06:54] <jbailey> (Best to upload that lkh when I'm sober.  thankyou very much for the patch)
[06:57] <fabbione> BenC: didn't you back out the security patch???
[06:57] <fabbione> ah never mind
[08:43] <BenC> fabbione: it's removed from 00list
[07:06] <joh> Hello, I'm running ubuntu on a server which I had to restart the ugly way by pulling the plug because of a crash. Now when I boot the machine, I discover some files (about 10) located in /bin /sbin /usr/bin and /usr/sbin which all have their owner changed to uid 112 and gid 114. Now I'm unable to change their permissions back to root (Operation not permitted). Actually, I'm unable to change *any* file in those 4 directories. After running an fsck,
[07:08] <fabbione> joh: hmmmm
[07:08] <fabbione> i would try to check for chattr -i
[07:08] <mjg59> Yeah, it's likely that they've ended up immutable
[07:09] <fabbione> if you still can't change them, try to run another e2fsck on it
[07:09] <fabbione> probably there is still corruption
[07:11] <joh> fabbione: hmmmm
[07:11] <joh> fabbione: chattr -i didn't help either...
[07:12] <fabbione> joh: switch to init s and umount /usr
[07:12] <fabbione> e2fsck -fF /dev/whatever
[07:12] <fabbione> and see if it helps
[07:13] <joh> fabbione: ok, some of the files are immutable actually...
[07:13] <joh> fabbione: but removing the flag doesn't help
[07:14] <joh> fabbione: they have all the sia flags set
[07:14] <fabbione> joh: please so as i said and check for extra corruption first
[07:14] <fabbione> what fs is that?
[07:14] <joh> fabbione: ext3
[07:15] <joh> (this is the root filesystem, so I cannot umount it atm :P I can check the files in /usr though)
[07:15] <fabbione> switch to single user and mount it ro
[07:15] <fabbione> i have to go now
[07:15] <fabbione> bbl
[07:15] <joh> ok, thanks
[07:38] <joh> fabbione: thanks, it worked when I removed the sia attributes from them. both fscks run successfully (no errors as far as I could see)...
[07:38] <fabbione> joh: no problem
[07:40] <joh> fabbione: any idea what could've caused this, btw?
[07:41] <fabbione> joh: it depends from a billion things
[07:41] <fabbione> first, do you know why the machine crashed?
[07:41] <joh> fabbione: ok? :P the cause of the crash was out of memory I think...
[07:41] <fabbione> also.. what does the machine run?
[07:41] <joh> fabbione: probably a leak in hula
[07:42] <fabbione> does the machine do a lot of IO?
[07:42] <joh> fabbione: apache, hula, mysql, pgsql, ssh, php, ...
[07:42] <fabbione> so yeah
[07:43] <fabbione> it's possible that it was doing I/O while you unplugged the power
[07:43] <fabbione> and crap has been written to disk
[07:43] <fabbione> that's simplest explanation i can find
[07:43] <joh> fabbione: well, the system wasn't responsive at all when it crashed.
[07:43] <fabbione> the worst one is that you have been hacked and that the leak and corruption was caused by an attacker attempt to gain root
[07:43] <joh> fabbione: i.e. I could change VTs but not run any process...
[07:44] <fabbione> joh: that doesn't mean much.. it could have been busy doing I/O
[07:44] <fabbione> now i give you a few tips
[07:44] <joh> fabbione: ok :)
[07:44] <fabbione> when you install servers. it's always a good idea to use different partitions
[07:44] <fabbione> like /usr /var /tmp /home on separate partitions
[07:44] <fabbione> even LVM would do
[07:45] <joh> they are
[07:45] <fabbione> ok
[07:45] <fabbione> well i understood differently befor
[07:45] <fabbione> e
[07:45] <joh> well, what I meant that was that the files in /bin and /sbin were on the root filesystem.
[07:45] <fabbione> and if your server provides a lot of different services
[07:45] <joh> *root partition
[07:46] <fabbione> it's worth a try to make some partitions ro
[07:46] <fabbione> like /usr
[07:46] <fabbione> there is no need to write there
[07:46] <joh> unless I install some packages :P
[07:46] <fabbione> root is a bit too complicate...
[07:46] <joh> yeah
[07:46] <fabbione> of cours
[07:46] <fabbione> but than you can always mount/remount
[07:46] <joh> yeah
[07:46] <joh> Ok, thanks again, I have to go now :)
[07:47] <fabbione> np