From: Brian Julin <bri@forcade.calyx.net>
  To  : ggi-develop@eskimo.com
  Date: Mon, 14 Sep 1998 17:39:00 -0400 (EDT)

Re: Time for a stable version release ?

On Mon, 14 Sep 1998, Marcus Sundberg wrote:
> Yes, I know that. But the question I'm asking is if there's any
> need to lock a _framebuffer_ for threadsafety. 

The framebuffer underlying the MMAP may not be linear and
may have some sort of funky handler in-kernel underneath it.
Also if any libGGI accel driver libs change the buffer layout
to do something really fast, they should do their operations
in a different virtual address space so another thread would fault 
on the linear framebuffer and get a linear layout when the fault 
is processed.  Mixing libGGI and DB access between threads 
or processes should be discouraged anyway though.

Does this require locks for threads?  Maybe not? -- but 
two threads banging on a banked framebuffer with no coordination 
as to who is drawing where would be awfully slow.

Perhaps we should put the capability in there if we can...

--
Brian S. Julin