William A.Carrel wrote:
As I just added to a bug on Sourceforge, a change made in the not too
distant past broke pass, exec and possibly other things.
In agent/mibgroup/util_funcs.c:~228, cachebytes needs to be ssize_t.
Left as size_t breaks the if(cachebytes < 0) when the read() above it
returns -1, and is a type-mismatch with read() return value on all
platforms I've been able to find anyway.
As it is, the normal sleep() or usleep() waiting for data to read being
skipped and spinning (quickly) to MAXREADCOUNT. As a result, users are
experiencing missing data from "exec", and I've seen pass fail to
actually get to the input from my scripts using "pass" on faster hosts
under some load.
https://sourceforge.net/tracker/index.php?
func=detail&aid=936483&group_id=12694&atid=112694
Please back out patch #860575 to fix this issue.
I assume that you have independently verified that
pass and pass_persist work when you revert that line to ssize_t ?
After that line was changed from size_t back to ssize_t,
Builds on FreeBSD, Linux, and Win32 compile and run without problems.
I did not run pass nor pass_persist, as the problem description
indicates the error occurs under certain conditions.
The one line change was committed to V5-1-patches and main CVS branches,
and will be available in the next release.
Sorry for the inconvenience, and thank you for the increased understanding.
Sincerely,
-Mike Slifcak
-------------------------------------------------------
This SF.Net email is sponsored by: GNOME Foundation
Hackers Unite! GUADEC: The world's #1 Open Source Desktop Event.
GNOME Users and Developers European Conference, 28-30th June in Norway
http://2004/guadec.org
_______________________________________________
Net-snmp-coders mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
