Re: Race condition in rygel-changelog.vala?

[Jussi Kukkonen <jussi kukkonen intel com>]

Hi Dario,

On 25 March 2014 18:23, Russo, Dario <dario russo intel com> wrote:
> I am running CTT tests on Rygel renderer. (version 0.21.3)

I assume this is not with the gstreamer renderer, but another one that
uses librygel-renderer? If you can reproduce this with only rygel
code, that would be useful to know as well.

> Unfortunately, very often I get a fatal assert and SIGABORT in
> Rygel-changelog.vala, apparently the hash table is incurring in a race
> condition, can it be?

Without a backtrace, it's hard to tell.

> I tried protecting it locking it with a Glib.StaticMutex, and this seems to
> fix the assert/abort, but I still (very rarely) get a strange lastlog event
> with corrupted message (its in the DMR state string: for example, instead of
> "TRANSITIONING", presents random characters at the start, while the rest of
> the soap message is perfectly fine.
>
> Is that possible that the hashtable in Rygel-changelog.vala incurs in a race
> condition? If yes, should I file a bug in the bugtrack?

I don't see where the race could be but if you're seeing a crash,
please do file a bug with a back trace. Please CC me so I don't miss
it.

Jussi