Re: [PATCH] drm: i915: ensure objects are allocated below 4GB onPAE

[Kyle McMartin]

On Tue, May 26, 2009 at 11:41:25PM +0200, Peter Zijlstra wrote:
> On Tue, 2009-05-26 at 14:35 -0700, Eric Anholt wrote:
> > On Tue, 2009-05-26 at 21:17 +0200, Peter Zijlstra wrote:
> > > On Tue, 2009-05-26 at 12:27 -0400, Kyle McMartin wrote:
> > > > From: Kyle McMartin <kyle@xxxxxxxxxx>
> > > > 
> > > > Ensure we allocate GEM objects below 4GB on PAE machines, otherwise
> > > > misery ensues. This patch is based on a patch found on dri-devel by
> > > > Shaohua Li, but Keith P. expressed reticence that the changes unfairly
> > > > penalized other hardware.
> > > > 
> > > > (The mm/shmem.c hunk is necessary to ensure the DMA32 flag isn't used
> > > >  by the slab allocator via radix_tree_preload, which will hit a
> > > >  WARN_ON.)
> > > 
> > > Why is this, is the gart not PAE friendly?
> > > 
> > > Seems to me its a grand way of promoting 64bit hard/soft-ware.
> > 
> > No, the GART's fine.  But the APIs required to make the AGP code
> > PAE-friendly got deprecated, so the patches to fix the AGP code got
> > NAKed, and Venkatesh  never sent out his patches to undeprecate the APIs
> > and use them.
> > 
> > It's been like 6 months now, and it's absurd.  I'd like to see this
> > patch go in so people's graphics can start working again and stop
> > corrupting system memory.
> 
> For .30 yes, for .31 we need to resolve that AGP issue, 6 months does
> seem excessive to get something like that sorted.
> 

Yeah, sorry, I should have explained it in the description better, this
is just a paper-over fix for the problem on >4GB 32-bit machines (which
is why I CC'd stable@.)

Thanks,
	Kyle
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/