[PATCH] ARM: map_init_section flushes incorrect pmd

[will.deacon@arm.com (Will Deacon)]

On Thu, May 30, 2013 at 09:15:26AM +0100, Po-Yu Chuang wrote:
> On Wed, May 29, 2013 at 5:34 PM, Po-Yu Chuang <ratbert.chuang@gmail.com> wrote:
> > Hi Will,
> >
> > On Wed, May 29, 2013 at 4:54 PM, Will Deacon <will.deacon@arm.com> wrote:
> >> On Wed, May 29, 2013 at 03:14:58AM +0100, Po-Yu Chuang wrote:
> >>> Will,
> >>> I guess nobody noticed this because the MMU of later v7 processors
> >>> fetches page table
> >>> from D-cache. It even doesn't need to clean pmd to PoU.
> >>
> >> It does if it's UP. The walker is only guaranteed to read from L1 if you
> >> have the multiprocessing extensions.
> >
> > Ya, I see.
> >>
> >> As for this function, looks like it's ok because it has precisely one
> >> caller, so it might be worth prefixing it with some underscores to make it
> >> clear that nobody else should be calling it!
> >
> > I am fine with that.
> > Should I create a new patch?
> 
> Hi,
> 
> Does anyone have more comment about this?

Sorry, I was snowed under yesterday. If you fancy adding the underscores to
__map_init_section, that would help preserve what little remains of my
sanity.

Your existing patch looks technically correct.

Will