[Asterisk-code-review] sched: Don't allow ast_sched_del to deadlock ast_sched_runq from same... (...asterisk[13])

[Walter Doekes]

Walter Doekes has posted comments on this change. ( https://gerrit.asterisk.org/c/asterisk/+/10991 )

Change subject: sched: Don't allow ast_sched_del to deadlock ast_sched_runq from same thread
......................................................................


Patch Set 4:

> > Quick question: do you have any use for the other patches I attached to https://issues.asterisk.org/jira/browse/ASTERISK-28282 ?
> 
> The one for mixmonitor to reduce locking I'm not sure how useful it is.  It does introduce reentrancy problems with those new variables.

Yes, it was quickly hacked up after we determined that the deadlock occurred more often after we started using the 'b' MixMonitor flag. That tight lock seemed to be a good candidate for extra context switching.

We did take it into production; no adverse effects seen.

> ASTERISK-28282_no_sip_poke_peer_now_if_no_qualify.patch - I cannot say much about that patch as I don't know what it is trying to fix.

It is an alternate fix to this specific deadlock. Qualify was 0, so no peer poking was necessary. That patch stops it before it was scheduled.

> Those other patches are chan_sip specific sched debugging, changes to induce the deadlock, or combined in this patch.

The scheduler debug patch actually improves it for other users of the scheduler as well. But it's not ABI compatible, so not something for released branches anyway.

> So they aren't that useful.

Thanks for checking!


-- 
To view, visit https://gerrit.asterisk.org/c/asterisk/+/10991
To unsubscribe, or for help writing mail filters, visit https://gerrit.asterisk.org/settings

Gerrit-Project: asterisk
Gerrit-Branch: 13
Gerrit-Change-Id: Ic26777fa0732725e6ca7010df17af77a012aa856
Gerrit-Change-Number: 10991
Gerrit-PatchSet: 4
Gerrit-Owner: Walter Doekes <walter+asterisk at wjd.nu>
Gerrit-Reviewer: Friendly Automation
Gerrit-Reviewer: Joshua Colp <jcolp at digium.com>
Gerrit-Reviewer: Matthew Fredrickson <creslin at digium.com>
Gerrit-Reviewer: Richard Mudgett <rmudgett at digium.com>
Gerrit-Reviewer: Walter Doekes <walter+asterisk at wjd.nu>
Gerrit-Comment-Date: Thu, 18 Jul 2019 07:34:50 +0000
Gerrit-HasComments: No
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-code-review/attachments/20190718/93f9f4c7/attachment.html>