Summary:ASTERISK-03154: [patch] sip reload deadlock
Reporter:Brian West (bkw918)Labels:
Date Opened:2005-01-02 01:25:37.000-0600Date Closed:2008-01-15 15:19:23.000-0600
Versions:Frequency of
Environment:Attachments:( 0) astobj.diff.txt
Description:We finally found this with help of kpfleming.  This is an extra break so we never unlock when we should.


disclaimer on file.
Comments:By: Olle Johansson (oej) 2005-01-02 02:18:21.000-0600

Is this patch disclaimed ;-)

By: Kevin P. Fleming (kpfleming) 2005-01-02 09:39:05.000-0600

Confirmed, this is the correct fix for this problem.

I don't think we need a specific disclaimer for a one-line patch, but since I had an equal share in creating it (<G>), I'll disclaim it.

By: Brian West (bkw918) 2005-01-02 10:34:31.000-0600

oej knows I have a disclaimer :P

By: Andrew Lindh (andrew) 2005-01-02 11:46:03.000-0600

I was having the same reload deadlock problem. The patch solved the issue for me. Thanks...

By: khb (khb) 2005-01-02 11:58:43.000-0600

I don't understand why this astobj stuff is in CVS if it hasn't been tested prior

By: Kevin P. Fleming (kpfleming) 2005-01-02 12:03:28.000-0600

It was tested, but there's no way that one or two testers can provide enough coverage to thoroughly test something that is this "invasive". When I tested it, I did not run a "sip reload", so I missed this problem.

But remember that CVS HEAD is not _stable_, it's possible for it to be borked at any time. If it had to be always guaranteed to work perfectly, then those of us working on major changes would have to have yet a third tree to work in, and this tree wouldn't get as much testing.

By: khb (khb) 2005-01-02 13:30:47.000-0600

I disagree,  something this invasive should be tested much more thoroughly and the kind the problems I have seen here attribute to that, they could have easily been thought through and test over a few weeks' period on an active system; from moment one I had the impression this was very imature stuff.

By: twisted (twisted) 2005-01-02 13:55:34.000-0600

khb: restating what was previously stated - if you want a STABLE release that does not have testing/development features and code, check out the STABLE branch. HEAD is the development branch, and it can, and will, contain code that is not stable from time to time.  

FYI: cvs co -r v1-0 asterisk zaptel libpri    <-- how to check out stable branch.

By: Olle Johansson (oej) 2005-01-02 14:50:14.000-0600

The cvs head tree is our testing platform. That is why these things happen and we constantly communicate that you should expect this from CVS head :-)

khb: you know this, even though you might have a different opinion...

Personally, I feel good about having a testing platform where we can test new stuff without worrying about total stability for production systems.

By: Brian West (bkw918) 2005-01-02 16:02:16.000-0600

oej couldn't have said it better ;)

By: Mark Spencer (markster) 2005-01-02 17:46:15.000-0600

Fixed in CVS, thanks

By: Digium Subversion (svnbot) 2008-01-15 15:19:23.000-0600

Repository: asterisk
Revision: 4634

U   trunk/include/asterisk/astobj.h

r4634 | markster | 2008-01-15 15:19:23 -0600 (Tue, 15 Jan 2008) | 2 lines

Remove silly break (bug ASTERISK-3154)