Re: glibc 2.0.94 for testing
At 11:12 -0700 1998-07-29, Dan Jacobowitz wrote: On Wed, Jul 29, 1998 at 04:27:15PM +0200, Anders Hammarquist wrote: Haven't noticed that one (at least not on the alpha under 2.1.106), but I probably don't want to know... Hmm. I know it appears on intel... I saw a report of dpkg failing with gzip: stdin: not in gzip format here a few weeks ago. -- Joel Espy Klecker URL:mailto:[EMAIL PROTECTED] URL:http://web.espy.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: glibc 2.0.94 for testing
On Wed, Jul 29, 1998 at 04:27:15PM +0200, Anders Hammarquist wrote: Haven't noticed that one (at least not on the alpha under 2.1.106), but I probably don't want to know... Hmm. I know it appears on intel... Dan -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: glibc 2.0.94 for testing
At 11:41 -0700 1998-07-27, Jules Bean wrote: On Sat, 25 Jul 1998, Joel Klecker wrote: BTW, someone needs to build dpkg = 1.4.0.25 for sparc, otherwise, folks running 2.1 kernels will not have fun. What breaks? Everything involving unpacking a .deb. -- Joel Espy KleckerDebian GNU/Linux Developermailto:[EMAIL PROTECTED] http://www.espy.org/ ftp://ftp.espy.org/pub/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: glibc 2.0.94 for testing
At 11:41 -0700 1998-07-27, Jules Bean wrote: On Sat, 25 Jul 1998, Joel Klecker wrote: BTW, someone needs to build dpkg = 1.4.0.25 for sparc, otherwise, folks running 2.1 kernels will not have fun. What breaks? Everything involving unpacking a .deb. Well, that overstating it a bit I think. What breaks is that dpkg-deb[1] runs chown(2) on everything it extracts from the archive, including symlinks. In recent 2.1 kernels, there is an lchown(2) call, and chown(2) follows symlinks (unlike 2.0 kernels where chown(2) will chown the symlink). If the symlink dpkg-deb tries to chown is broken, it exits, as chown(2) returned ENOENT. Regards, /Anders [1] 1.4.0.25 that is -- -- Of course I'm crazy, but that doesn't mean I'm wrong. Anders Hammarquist | Mud at Kingdoms| [EMAIL PROTECTED] NetGuide Scandinavia | telnet kingdoms.se 1812| Fax: +46 31 50 79 39 http://www.netg.se | | Tel: +46 31 50 79 40 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: glibc 2.0.94 for testing
On Tue, Jul 28, 1998 at 10:11:40PM +0200, Anders Hammarquist wrote: Everything involving unpacking a .deb. Well, that overstating it a bit I think. What breaks is that dpkg-deb[1] runs chown(2) on everything it extracts from the archive, including symlinks. In recent 2.1 kernels, there is an lchown(2) call, and chown(2) follows symlinks (unlike 2.0 kernels where chown(2) will chown the symlink). If the symlink dpkg-deb tries to chown is broken, it exits, as chown(2) returned ENOENT. That's not the only issue. A random fflush() that probably shouldn't even be there causes dpkg to horribly die on debs on ext2 filesystems using recent dev kernels. Don't ask. Dan -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: glibc 2.0.94 for testing
Christian Meder [EMAIL PROTECTED] writes:
On Wed, Jul 22, 1998 at 06:15:25PM +0200, Anders Hammarquist wrote:
If it was still needed for it should be there - though I haven't looked
at it so I don't now for sure. IIRC it was Johnnie Ingram who compiled
the last libc6.
The glibc-pre2.1 maintainer is Juan Cespedes.
I asked him if I could go ahead and do a NMU for glibc-2.0.94. He
wrote that he's got no time right now and I should try.
So I fought two weeks with a 2.0.94 snapshot taken from Thorsten
Kukuk's rpm-src package. Finally I've got a libc which is usuable
for me (Sparc 10, sun4m). I left out all the patches from the 2.0.93
version which didn't impact my box. Note that I left out the
getpagesize fix.
Same problem as 2.0.93, bash segfaults after running any command. I
guess it just doesn't like recent kernels. (I'm running a version
2.1.103 kernel, originally from cvs.)
Does anyone know what the problem is?
The Debian tree in question is unusable (just the straight
base2_0.tgz, both with and without the above packages added).
I tried copying the new ld-linux.so.2 to the UltraPenguin installation
and running the Debian bash with LD_LIBRARY_PATH pointing to
/mnt/root/lib (the Debian /lib tree) and it won't segfault, but if I
use chroot to run bash, it does segfault.
I managed to strace the chroot `pwd` bin/bash, the end of the output
is below. (I also tried running the UltraPenguin gdb under chroot,
and the machine is no longer responding.)
wait4(
Output from ls occurs here.
-1,[WIFEXITED(s) WEXITSTATUS(s) == 0], WUNTRACED, NULL) = 740
sigprocmask(SIG_BLOCK, [TSTP CHLD TTIN TTOU], [CHLD]) = 0
ioctl(255, TIOCSPGRP, 0xefffe860) = 0
sigprocmask(SIG_SETMASK, [CHLD], NULL) = 0
ioctl(255, TCGETS, {B38400 opost isig icanon echo ...}) = 0
sigprocmask(SIG_BLOCK, [CHLD TTOU], [CHLD]) = 0
sigprocmask(SIG_SETMASK, [CHLD], NULL) = 0
sigprocmask(SIG_BLOCK, [CHLD], [CHLD]) = 0
sigprocmask(SIG_SETMASK, [CHLD], NULL) = 0
sigprocmask(SIG_SETMASK, [], NULL) = 0
--- SIGCHLD (Child exited) ---
wait4(-1, 0xefffe654, WNOHANG|WUNTRACED, NULL) = -1 ECHILD (No child processes)
sigreturn( unfinished ...
+++ killed by SIGSEGV +++
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: glibc 2.0.94 for testing
At 21:03 -0700 1998-07-24, Steve Dunham wrote: Same problem as 2.0.93, bash segfaults after running any command. I guess it just doesn't like recent kernels. (I'm running a version 2.1.103 kernel, originally from cvs.) Must be a sparc problem (on certain machines I'm guessing), 2.0.94 works just fine on powerpc (now that the dpkg problem is fixed). BTW, someone needs to build dpkg = 1.4.0.25 for sparc, otherwise, folks running 2.1 kernels will not have fun. -- Joel Espy KleckerDebian GNU/Linux Developermailto:[EMAIL PROTECTED] http://www.espy.org/ ftp://ftp.espy.org/pub/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
