Announce: kdb v2.3 i386 updates for kernels 2.4.18 and 2.4.19

[Keith Owens <kaos@sgi.com>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Content-Type: text/plain; charset=us-ascii

Updated files in ftp://oss.sgi.com/projects/kdb/download/v2.3/

  kdb-v2.3-2.4.18-i386-2.bz2
  kdb-v2.3-2.4.19-i386-2.bz2

I also have kdb-v2.3-2.4.18-ia64-020722-2 in progress, to add support
for zx1 serial i/o.  Not tested yet, it should be out tomorrow.

Changelog extracts between 2.4.1[89]-i386-1 and -2.

2.4.19-i386-2

2002-08-09 Keith Owens  <kaos@sgi.com>

       * Use -fno-optimize-sibling-calls for kdb if gcc supports it.
       * .text.lock does not consume an activation frame.
       * kdb v2.3-2.4.19-i386-2.

2.4.18-i386-2

2002-08-09 Keith Owens  <kaos@sgi.com>

       * Use -fno-optimize-sibling-calls for kdb if gcc supports it.
       * 2.4.18 uses _text_lock_ not .text.lock.
       * _text_lock does not consume an activation frame.
       * kdb v2.3-2.4.18-i386-2.

When gcc optimizes sibling calls (tail recursion), kdb <= v2.3 cannot
decode the backtrace.  Nor can the stack trace show what is really
going on with optimized sibling calls, the return addresses are
ambiguous.  Until I can come up with an algorithm to trace through tail
recursion, activating kdb will set -fno-optimize-sibling-calls.

kdb v2.4 is planned as a big gcc catch up, to cope with the code
generated by more recent versions of gcc.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Exmh version 2.1.1 10/15/1999

iD8DBQE9U2agi4UHNye0ZOoRArfUAJ9BXPmRQCJPenkTyBFHiV9zRB6dkACgxkPT
qeyKkLrO/ZTyQ0j3VUa92J8=
=7U08
-----END PGP SIGNATURE-----