[PATCH] CBFS bounce buffer fix

[Myles Watson]

src/boot/selfboot.c: In function 'get_bounce_buffer':
src/boot/ssrc/boot/selfboot.c: In function 'get_bounce_buffer':
src/boot/selfboot.c:113: warning: declaration of 'bounce_size' shadows a
global elfboot.c:113: warning: declaration of 'bounce_size' shadows a global

The global never got initialized, so I couldn't boot a kernel.  Now I can,
but I'm not sure this is the most elegant fix, or even that it is 100%
correct.

While I was there I changed another bounce_size to size to avoid confusion
with the global, and made selfboot static to get rid of this warning:

src/boot/selfboot.c:508: warning: no previous prototype for 'selfboot'

Signed-off-by: Myles Watson

Thanks,
Myles

[ron minnich]

Acked-by: Ronald G. Minnich

[Myles Watson]

On Wed, Sep 23, 2009 at 2:28 PM, ron minnich  wrote:
> Acked-by: Ronald G. Minnich

Rev 4663.

Thanks,
Myles

[Patrick Georgi]

Am Mittwoch, den 23.09.2009, 14:23 -0600 schrieb Myles Watson:
> The global never got initialized, so I couldn't boot a kernel.  Now I
> can, but I'm not sure this is the most elegant fix, or even that it is
> 100% correct.

Looks good to me. I had similar changes here already, but messed around
with That Other Selfboot Bug[tm] and thus have nothing finished yet.
> src/boot/selfboot.c:508: warning: no previous prototype for 'selfboot'
>
> Signed-off-by: Myles Watson

Acked-by: Patrick Georgi

Regards,
Patrick

[Myles Watson]

On Wed, Sep 23, 2009 at 2:30 PM, Patrick Georgi wrote:
> Am Mittwoch, den 23.09.2009, 14:23 -0600 schrieb Myles Watson:
> > The global never got initialized, so I couldn't boot a kernel.  Now I
> > can, but I'm not sure this is the most elegant fix, or even that it is
> > 100% correct.
>
> Looks good to me.

Thanks for the review.
> I had similar changes here already, but messed around
> with That Other Selfboot Bug[tm] and thus have nothing finished yet

I hope you get the fix committed before I find it the hard way :)

Thanks,
Myles

[Patrick Georgi]

Am Mittwoch, den 23.09.2009, 14:35 -0600 schrieb Myles Watson:
>     I had similar changes here already, but messed around
>     with That Other Selfboot Bug[tm] and thus have nothing
>     finished yet
>
> I hope you get the fix committed before I find it the hard way :)

The issue is that the loader doesn't properly cope with the case that
the payload overlaps with coreboot and starts at a lower address.

That happens on some AMD boards (with RAMBASE at 2MB, for whatever
reason) and FILO (which has a huge .bss section that moves well into
coreboot's memory).

Any other configuration should not run into this issue, and for AMD, it
should be enough to move RAMBASE to somewhere else (both 1MB and 4MB
worked for me) as a workaround.

Regards,
Patrick

[Myles Watson]

On Wed, Sep 23, 2009 at 2:42 PM, Patrick Georgi wrote:
> Am Mittwoch, den 23.09.2009, 14:35 -0600 schrieb Myles Watson:
> >     I had similar changes here already, but messed around
> >     with That Other Selfboot Bug[tm] and thus have nothing
> >     finished yet
> >
> > I hope you get the fix committed before I find it the hard way :)
> Any other configuration should not run into this issue, and for AMD, it
> should be enough to move RAMBASE to somewhere else (both 1MB and 4MB
> worked for me) as a workaround.

Great.  I'd forgotten that one, probably because it won't affect me.

Thanks,
Myles