It fuzzed on the first chunk, so I wiped the source directory, re un-tarred asterisk so it was clean, and then manually applied the changes.. doing removes and adds where necessary. So, unfortunately no, I do not have the information, though I suppose I could run it again.
Keep in mind, also, that I'm running 1.2.9.1, not svn-trunk. On 6/21/06, BJ Weschke <[EMAIL PROTECTED]> wrote:
Do you have that patch/commit message from svn-commits where this code was introduced so we can track it back? On 6/21/06, Matt <[EMAIL PROTECTED]> wrote: > On 6/21/06, BJ Weschke <[EMAIL PROTECTED]> wrote: > > On 6/21/06, Matt <[EMAIL PROTECTED]> wrote: > > > Ok, > > > So here's some information I've, to this point, left out. I applied > > > the patch to allow * to be pressed in queues to park a call (*270). > > > After reverting the patch the system seems stable. So, it almost > > > seems like the crash is directly related to the number of times > > > someone parks a call that came in a queue. > > > > > > I don't understand, though.... > > > > > > /* terminates call */ > > > ast_frfree(f); > > > f = NULL; > > > > > > How does removing those 2 lines cause asterisk to crash? Basically > > > the code looks for a *, if it sees it it runs that... so the patch > > > removes those lines.. so when asterisk sees a * it doesn't do > > > anything. The variables are so cryptic I can't exactly figure out > > > what its doing.. is ast_frfree(f) suppose to 'free' something, perhaps > > > that is not getting freed? But then what is f = NULL. Yikes, who > > > comes up with these variable names? > > > > > > > Yes, ast_frfree(..) is supposed to free any allocated memory > > associated with the pointer to the Asterisk frame structure given to > > it. If the structure has already been freed somewhere else in the code > > and you try to call this again, you will segfault Asterisk. If you > > don't call it at all, you will leak memory associated with the > > structure. The pointer is being set to NULL after the call so that > > further checks of the pointer itself will be able to determine that > > the structure has been free'd already and doesn't need to be done > > again (preventing a segfault from calling it twice). > > > > Ok... that makes sence, and is probably why asterisk was crashing then > (good news is since reversing the patch and installing clean 1.2.9.1 > we've been rock solid over night). Now.. my question is... how do > those 2 lines of code disconnect a call? > _______________________________________________ > --Bandwidth and Colocation provided by Easynews.com -- > > Asterisk-Users mailing list > To UNSUBSCRIBE or update options visit: > http://lists.digium.com/mailman/listinfo/asterisk-users > -- Bird's The Word Technologies, Inc. http://www.btwtech.com/ _______________________________________________ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
_______________________________________________ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
