Re: PCI resource problems caused by improper address rounding

[Richard Henderson <rth@twiddle.net>]

On Thu, Dec 20, 2007 at 02:24:48PM -0800, Linus Torvalds wrote:
> I'm not exactly 100% happy with it, but it does mean that if we need a big 
> area, we'll relax the suggested starting point by that amount. It's not 
> wonderful, but it essentially admits that the minimum for the allocations 
> is really just a hint, and if we need lots of space for a resource, we'll 
> relax the minimum point appropriately.

This breaks in odd cases where the amount of memory in the system
is not a nice round number.  Like throwing two 128MB sticks into
a system that already has 2gb.  A 512MB allocation will get placed
back at 2gb, on top of the end of ram.  In order to get this kind
of thing to work, you'd have to have a hard and a soft minimum.

Even then, any random large allocation is going to ignore that
buffer that you added.  It'd be better if we could still tie this
ignoring of the buffer to whether the bios placed the resource
there in the first place.

Perhaps this is one of those things that just aren't going to be
solved properly without an xserver upgrade...


r~