> Richard L. Hamilton wrote:
> > A truly multithreaded
> > X server would help though, so that xlsfonts -lll
> wouldn't be
> > a DoS attack, but I expect that's a lot to ask.)
> 
> We'd be more likely to EOL the ancient core fonts,
> except for the
> two required by the protocol.   There are very few
> things the X
> server can do in multiple threads that take long
> enough to care
> about and don't end up all waiting on the same lock
> for access to
> the hardware.   If you really worry about having to
> protect the
> X server from yourself, you can also change your font
> path to
> redirect all legacy font accesses off to a font
> server, effectively
> putting that code into an out-of-process separate
> thread.

Like I said, "I expect that's a lot to ask."  My desire for
such a thing is maybe a bit more theoretical than anything
(wanting all the concurrency I can get), although xlsfonts -lll
with old fashioned core fonts rather than a font server or
client side rendering is a pretty dramatic _example_ of
how to make a single-threaded X server get very obviously bogged down.
I didn't literally mean that someone might use it as a DoS, so
much as the effect.  But whether there are other examples
that really demonstrate a performance hit from inability to
handle requests truly in parallel as much as possible, I don't know.
I would imagine that they'd exist, but the effect would be
nowhere near as obvious, esp. lacking something better to
compare against.

Anyway, I'm well aware I was off into pie-in-the-sky with that one.
(And I half suspected that you'd point out something along the
lines of the cost/benefit probably not being there on that.
I could swear I'd heard long ago of an experimental multi-threaded
X server, but I don't recall whatever became of it.  Clearly, it didn't
catch on, since Xorg isn't using it.)
-- 
This message posted from opensolaris.org
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to