On Fri, Nov 10, 2006 at 05:13:24PM +, Athanasius wrote:
This would seem to be something that's at least triggered by the
optimisation the debian build uses (-O3). I just did an apt-get source
proftpd, which fetched proftpd_1.2.10-15sarge1.dsc and friends. Then I
edited debian/rules therein changing the -O3 to -ggdb3, and commenting
out the dh_strip -a line so the symbols would stay put. The aim of this
was to get a decent backtrace in gdb upon SEGV.
However now I can't get it to crash, when before simply logging in,
issuing an ls and then logging out again was sufficient to cause the
SEGV, buried somewhere under pam_delay().
Of course now I say that, after reinstalling the latest packages I had
before this, via:
dpkg -i proftpd-common_1.2.10-15sarge1.0.1_i386.deb
proftpd-doc_1.2.10-15sarge1_all.deb proftpd_1.2.10-15sarge1.0.1_i386.deb
in /var/cache/apt/archives, I can't now get it to crash on quit.
Anyway, it's either a bug triggered by the optimisation, or of course
possibly a compile bug with that optimisation.
Additionally, if this is to do with the delay engine stuff, then see
bug #325304 and #308313 as mentioned in there.
I suspect here an Heisenbug principle should apply. Basically mod_delay
has been fixed after sarge release with respect to proper resource sharing.
And unfortunately a good deal of cvs patches have been applied into
debian package, so problems are quite difficult to be fixed on the basis
of upstream fixes only...
--
Francesco P. Lovergine
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
