[PATCH 2/4] X event queue mutex
keithp at keithp.com
Thu Oct 2 15:16:37 PDT 2008
On Thu, 2008-10-02 at 19:10 -0300, Tiago Vignatti wrote:
> Simon Thum escreveu:
> > I believe tiagos words are a bit misleading: The mutex makes it possible
> > to block event enqueuing, which is needed to guarantee order of events
> > enqueued on the main thread. If I got it right, the intent is to
> > 'emulate' OsBlockSignals(), though I'm missing that bit.
> Good point, Simon. I forgot about this discussion which we had in XDS.
> Given that there's only one thread to take care about all the input
> devices this would not be a problem with *devices*. Events will be
> serialized is this case. But the problems eventually can happen if the
> others guys that enqueue events (nested servers, DGA, maybe also XTest)
> try to do it at the same time with the devices. A mutex is needed in
> this situation.
Ah, yes indeed -- we need a mutex to protect writes to the tail pointer
in mieqEnqueue. But, not the tail reads from mieqProcessInputEvents.
That is what the signal blocking stuff was for, after all.
keith.packard at intel.com
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 189 bytes
Desc: This is a digitally signed message part
More information about the xorg