[asterisk-dev] Regression after the END-of-queue control frame patch

[Olle E. Johansson]

The patch for 1.4 in https://reviewboard.asterisk.org/r/1013/  might have fixed AMI redirect in a bridged call,
but did break it for a meetme call. When I issue AMI redirect on a channel in a meetme, it hangs up.

It doesn't hang up immediately - but anything that calls ast_waitfor will not work and cause a hangup at some point, due to lack of input frames. ast_waitfor terminates with an answer indicating there's a frame to read, but reading frames we get nothing on the channel.

I've tested rev 288636 - the one before, and in that revision the transfer works as expected.

I've documented my findings in issue #18585 

I've spent many hours trying to understand this and have failed miserably. If you have a moment to spend on this and know this area of Asterisk - please help.

Thanks,
/O

https://issues.asterisk.org/view.php?id=18585