Re: [PATCH] drm/i915: Fix pte updates in ggtt clear range

[Ben Widawsky <ben@bwidawsk.net>]

On Tue, 27 Nov 2012 08:22:01 +0000
Chris Wilson <chris@chris-wilson.co.uk> wrote:

> On Mon, 26 Nov 2012 21:52:54 -0800, Ben Widawsky <ben@bwidawsk.net>
> wrote:
> > This bug was introduced by me:
> > commit e76e9aebcdbfebae8f4cd147e3c0f800d36e97f3
> > Author: Ben Widawsky <ben@bwidawsk.net>
> > Date:   Sun Nov 4 09:21:27 2012 -0800
> > 
> >     drm/i915: Stop using AGP layer for GEN6+
> > 
> > The existing code uses memset_io which follows memset semantics in
> > only guaranteeing a write of individual bytes. Since a PTE entry is
> > 4 bytes, this can only be correct if the scratch page address is 0.
> 
> Gah. Wasn't there an iowrite32_rep?

And you would hope it does what you want... but it seems like just a
memcpy of dword sized chunks.

My first thought was to write my own memset which does what you want,
but this seemed like a path of less resistance.

>  
> > This caused unsightly errors when we clear the range at load time,
> > though I'm not really sure what the heck is referencing that memory
> > anyway. I caught this is because I believe we have some other bug
> > where the display is doing reads of memory we feel should be
> > cleared (or we are relying on scratch pages to be a specific value).
> 
> That's just because we are no longer disabling outputs before updating
> the GTT and hence continue to scanout from the BIOS fb during module
> load. It's a regression that we'll be able to finally fix properly
> with fastboot - though that will not be without its downsides either.
> -Chris
>