[vlc-devel] [PATCH] coreaudio: fix deadlock on interruption

[Thomas Guillem]

On Tue, Dec 12, 2017, at 17:21, Rémi Denis-Courmont wrote:
> Le tiistaina 12. joulukuuta 2017, 16.18.15 EET Thomas Guillem a écrit :
> > > Also mixing atomic ops and muteces smells very bad, TBH.
> > 
> > I said to Caro to do it like that. I agree it smells bad and seems
> > complicated.
> 
> That was a polite way to say that it makes no sense.
> 
> Indeed, a quick look at the code shows that the atomic vars don't make
> sense. 

I still don't know what the problem is.

> Also the circular buffer is busted.

Maybe, that's easy to say since almost all old atomic algorithms are
busted.
We could get ride of it and use mutexes for VLC 4.0 and don't care about
the wait-less comments (and since mutexes are hybrid now...).

> 
> > I invite you to take a closer at ca_Flush (in the drain == false case)
> > and ca_SetAliveState(). For me, it's thread safe.
> 
> So you wasted my time, and now you want to review again. How about, hell
> no?

We are an open source project and reviews should be never wasted (for
example, I took your last comment about the circular buffer in
consideration).

I would never say that to someone even if the problem come from the user
(like PEBKAC issue, miss configuration of OS/VLC). Even more, I would
never say that in a public ml.

> 
> -- 
> 雷米‧德尼-库尔蒙
> https://www.remlab.net/
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel