Hi all,
A critical signal handling bug was found in Samba 2.0.1
within a few hours of release (thank goodness !) and so I've
recalled it from the site and released 2.0.2 with a fix.
The problem was a missing break statement..... (my fault
I'm afraid) in the code handling systems with slow system
calls that aren't restarted (waitpid) - a loop was being
done to restart the waitpid if it was interrupted by a
signal, but unfortunately the standard SIGCLD signal
handler was eating the child return status in the
signal handler - thus the waitpid was returning -1
with errno = ECHILD - a condition that wan't being
tested for and hence caused smbd to go into a tight
loop doing waitpid calls under certain circumstances.
I'm just sorry that none of the internal tests
caught this one before release - please accept my
apologies.
Sorry,
Jeremy Allison,
Samba TEam.
servis@purdue.edu
1999-Feb-07 15:53 UTC
Congratulations! (Re: 2.0.1 recall - 2.0.2 released.)
My congratulations to the Samba Team for such a FAST response!!!! Within hours of the first post of a problem on samba@samba.org the problem was debugged and fixed with a recall of the bad code and a new version made available. This would NEVER happen under a closed development system. Congratulations and keep up the fantastic work! -- Brian Servis --------------------------------------------------------------------- "Never criticize anybody until you have walked a mile in their shoes, because by that time you will be a mile away and have their shoes." - unknown Mechanical Engineering bservis@usa.net Purdue University http://www.ecn.purdue.edu/~servis ---------------------------------------------------------------------