Shared memory transport for X (was Re: Security question?)

[Ian Romanick]

Adam Jackson wrote:

> SGI and DEC too.  And internal to Xlib, AFAIK, still using the PF_UNIX socket 
> for synchronization.
> 
> Precision Insight did some research into adding this to XFree86 in the early 
> 4.0 days, but it was decided that it wasn't worth the added complexity since 
> shared memory transport only really helps when the graphics subsystem is much 
> more powerful than the CPU.  The original paper fell off the net, but I 
> rescued it from the wayback machine:
> 
> http://dri.sourceforge.net/cgi-bin/moin.cgi/SharedMemoryTransport
> 
> It would be interesting to see if the results in this paper can be improved 
> upon by using linux futexes rather than the Unix socket for synchronization.  
> The implementation referred to in this paper is still available on a branch 
> of the DRI xc tree, if anyone feels like some archaeology.

The performance improvement might also be a lot larger for hardware 
accelerated indirect rendering.  Typical X requests don't have a lot of 
data, but when you start throwing large vertex arrays and textures into 
the mix, things change.

Using futexes is a good idea, BTW.