Re: 2.1 kernels and dpkg.
Hi,
What does POSIX say about this (I don't have my standard
handy)? I was under the impression that ENOTDIR was required when
rmdir(2) was called on a non-dir. Also, the problem is fixed in
2.1.46, so it seems that dpkg was right in the first place.
Why fix something that is correct?
manoj
>>"Klee" == Klee Dienes <klee@mit.edu> writes:
Klee> Shaya Potter <spotter@itd.nrl.navy.mil> writes:
>> I was perusing the unofficial kernel patches at linuxhq.com, and I
>> saw that their was a patch to make the latest kernels work with
>> dpkg. This should be a big concern if dpkg is not compatable with
>> the new kernels especially if the kernels keep that behavior when
>> they turn stable.
Klee> Here's the original message posted to linux-kernel:
Klee> From: buhr@stat.wisc.edu (Kevin Buhr) Date: 20 Jul 1997 14:25:14
Klee> -0500
Klee> Don't run "dpkg" or "dselect" under 2.1.44 or 2.1.45. Both
Klee> these kernels follow symlinks on "sys_rmdir", "sys_rename", and
Klee> "sys_chown". The first breaks "dpkg" most severely, since "dpkg"
Klee> absolutely relies on a "rmdir" of a symlink returning ENOTDIR.
Klee> The second is nasty too, since "dpkg", when installing a symlink
Klee> "blahblah", creates the symlink "blahblah.dpkg-new" and then
Klee> renames it to "blahblah". The "sys_chown" problem is mostly
Klee> minor, but if "dpkg" installs a dangling symlink, it'll die when
Klee> the chown of the symlink unexpectedly fails.
Klee> I'm looking into fixing the problem now, though it may be until
Klee> the weekend before I really have time to fix it properly.
--
"It does not pay a prophet to be too specific." Sprague de Camp
Manoj Srivastava <url:mailto:srivasta@acm.org>
Mobile, Alabama USA <url:http://www.datasync.com/%7Esrivasta/>
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org .
Trouble? e-mail to templin@bucknell.edu .
Reply to: