On domenica 29 aprile 2007, Antoine Martin wrote:
> Blaisorblade wrote:
> > On venerd=C3=AC 27 aprile 2007, Antoine Martin wrote:
> >> Hi all,
> >> These images aren't linked from the home page yet. Each page contains a
> >> link to the script/commands used to create the image.
> >> Here they are:
> >> http://uml.nagafix.co.uk/Ubuntu-Feisty/
> >> http://uml.nagafix.co.uk/Ubuntu-Dapper/
> >> http://uml.nagafix.co.uk/Debian-Etch/
> >> http://uml.nagafix.co.uk/BusyBox-1.5.0
> >> Some notes/issues - I think the images are usable as-is, but it would =
> >> nice to get rid of some of those messages:
> >> 1) Dapper was made from the Breezy image with a "dist-upgrade" rather
> >> than from scratch, which may trigger some issues. ie:
> >> * a boot message about /sbin/ifrename which has been replaced by udev
> >> but still referenced in a script...)
> > "dpkg --purge ifrename" should help (I had this too with Ubuntu on host=
> Ah, thx for that. I've updated the image.
> >> 5) Also in feisty:
> >> mount: none already mounted or /dev/pts busy
> >> mount: according to mtab, devpts is already mounted on /dev/pts
> >> (tried removing it from mtab to no avail)
> > Removing from /etc/fstab (if it's there) helps?
> It is there, but why should it be taken off?
Mounted from boot scripts like=20
mount devpts /dev/pts -t ... -o ...=20
Every distribution has a certain combination of whether they're mounted lik=
this or by mount -a. On Gentoo the approach above is used, but the user can=
ask to put its own item in /etc/fstab.
> >> And some new kernel issues I just discovered:
> >> * a 32-bit kernel compiled on 64-bit (using SUBARCH=3Di386) does not w=
> >> on a 32-bit host. It fails with the usual cryptic message:
> >> Kernel panic - not syncing: No init found. Try passing init=3D option=
> >> kernel.
> >> (but init is there, as the same command line works just fine with a
> >> different kernel!)
> > I got this when testing my first TLS-64bit patch (i.e. 2.6.18-bb1, IIRC=
> > Removing it made this disappear. I was always puzzled by this.
> Hmm, that was quite a long time ago!
> What do you suggest I do?
No idea right now, sorry. Consider my note on this random chatting ;-)
> >> * a 32-bit kernel compiled on 32-bit does not boot on 64-bit, all I get
> >> is: #> ./vmlinux
> >> "Checking that ptrace can change system call numbers...check_ptrace:
> >> PTRACE_OLDSETOPTIONS failed: Invalid argument"
> >> Versions are always the latest for the guest (2.6.21) and the hosts
> >> varies (184.108.40.206 for amd64 and 2.6.15 for i386) + skas
> > Applies to any (recent) 32bit UML kernel.
> > You have a bad 64bit host - the bug is the OLDSETOPTIONS support, and I
> > fixed it. 2.6.18 has this bug, but 220.127.116.11 includes my fix. It needs
> > however the attached patch, too (or it'll give another error with TLS).
> > Also, 18.104.22.168 is perfectly good (includes the attached patch).
> Thanks for the info, I always wondered when all these issues got fixed -
> I've added this info to the top of the page where I mention this issue.
Inform me of my mistakes, so I can add them to my list!
Paolo Giarrusso, aka Blaisorblade