[U-Boot] [PATCH v2 4/4] arm: factorize relocate_code routine
Benoît Thébaudeau
benoit.thebaudeau at advansee.com
Thu May 16 16:28:55 CEST 2013
Hi Albert,
On Thursday, May 16, 2013 4:29:49 PM, Albert ARIBAUD wrote:
> Hi Benoît,
>
> On Wed, 15 May 2013 19:55:31 +0200, Albert ARIBAUD
> <albert.u.boot at aribaud.net> wrote:
>
> > Hi Benoît,
> >
> > On Wed, 15 May 2013 18:36:05 +0200 (CEST), Benoît Thébaudeau
> > <benoit.thebaudeau at advansee.com> wrote:
> >
> > > Hi Albert,
> > >
> > > On Wednesday, May 15, 2013 10:30:00 AM, Albert ARIBAUD wrote:
> > > > On Wed, 15 May 2013 09:31:37 +0200, Albert ARIBAUD
> > > > <albert.u.boot at aribaud.net> wrote:
> > > >
> > > > > Then I should add =relocate_code to r3, not =_start, and also -- as
> > > > > r9
> > > > > is not the right offset here -- compute r7 as the delta between the
> > > > > link-time =_start and the run-time relocate_code (r7 becomes useless
> > > > > once R10, r2 and r3 are fixed).
> > > >
> > > > I badly need a coffee break... Make this "r7 equal to =relocate_code",
> > > > simply.
> > >
> > > Or "adr r7, relocate_code" for the same result in order to avoid using
> > > the
> > > literal pool when useless.
> >
> > Correct. That'll be in v3.
>
> Sorry, I missed this change in V3 and left the ldr. Functionally it
> does not have any impact, since as already mentioned, this code always
> runs at the link-time location. You may want Want me to fix it anyway?
I have not yet looked at your V3, but yes, I'd prefer.
Best regards,
Benoît
More information about the U-Boot
mailing list