Summary:ASTERISK-06885: [patch] "asterisk -x" does not disconnect session when built with "dont-optimize"
Reporter:Rudolf E. Steiner (res)Labels:
Date Opened:2006-05-02 15:41:52Date Closed:2006-06-18 16:43:48
Versions:Frequency of
Environment:Attachments:( 0) 7071.branch-1.2.patch
Description:After upgrading to 'Asterisk SVN-branch-1.2-r24097' the command 'asterisk -r -x <command>' does not close the session. Scripts using this feature will be blocked.

****** STEPS TO REPRODUCE ******

Use the command 'asterisk -r -x <command>'.
Comments:By: BJ Weschke (bweschke) 2006-05-02 15:50:17

Can you attach an example? asterisk -r -x "sip show peers" seems to return just fine from the session for a copy of SVN-branch-1.2-r24295 that I just tested with.

By: Rudolf E. Steiner (res) 2006-05-02 15:58:29

It's on any commands. Maybe it's corrected on 'r24295'?

By: Rudolf E. Steiner (res) 2006-05-02 16:00:03

Note: Asterisk is compiled with the "dont-optimize"-flag.

By: BJ Weschke (bweschke) 2006-05-02 16:22:00

It may be, but I don't really think there's been too many changes to 1.2 that would cause it to be fixed between then and now. I'm a little more suspect that there's something with your system that is different than mine. What version of gcc did you compile with? What kernel / OS are you running? What version of the glibc are you running?

By: Rudolf E. Steiner (res) 2006-05-02 17:24:55

'bweschke' wrote:

> What version of gcc did you compile with?


> What kernel / OS are you running?

Kernel: 2.6.16-1.2080_smp
OS: Fedora CORE 5

> What version of the glibc are you running?


Please note:

I have upgradet from 'r22866' to 'r24097'. On 'r22866' the problem was not yet.

By: Rudolf E. Steiner (res) 2006-05-03 13:49:24

I have made a few tests to this matter:

If you compile your Asterisk without the "dont-optimize"-flag, it work's correctly.

If you compile it with the "dont-optimize"-flag, you can reproduce the issue.

By: Russell Bryant (russell) 2006-05-07 08:54:02

Give this patch a try.  Thanks to bweschke for providing access to machine where I could reproduce the problem!  This patch fixed it for me.  However, it's not much more than a guess ...

By: Russell Bryant (russell) 2006-05-07 09:12:10

Alright, scratch this idea.  It breaks on mac osx...

By: Andrey S Pankov (casper) 2006-05-07 09:27:16

I think FC4+ should be added to the list of NON-supported platforms.

This is not an asterisk bug, this is a bug in the FC's gcc and should be
filed to the FC bugtracker (like any other bugs where gcc generates a
buggy code with dont-optimize or without it).

This is not an anti-RH campaign, but it turns out that FC4+ distros are not
for asterisk :(

By: Rudolf E. Steiner (res) 2006-05-08 03:51:08


On Fedora it work's fine. Thank you very much.

By: BJ Weschke (bweschke) 2006-05-08 10:49:36

need to figure out what to do about >= gcc 4.0.0 ... for open discussion on 5/9 dev conf call.

By: Russell Bryant (russell) 2006-05-19 14:05:31

I have added this change to be effective for all systems except for Mac.  I tested it on my linux machines, mac, as well as FreeBSD.

revisions 28754 for 1.2 and 28771 for trunk

By: Russell Bryant (russell) 2006-06-12 22:57:39

I have reverted the original patch I committed for this problem but I think I have fixed the real problem here.  If you have a machine that could reproduce this issue, please re-test and verify that the problem is still gone with the current code.  Thanks!

The new fix is in revisions 33753 for 1.2 and 33754 for trunk

By: Russell Bryant (russell) 2006-06-18 16:43:47

Well, I'm pretty sure this is fixed now.  If there is still a problem, feel free to re-open this bug or email me directly - russell at digium.com