On Mon, May 05 2014 at 3:34:31 pm BST, Jon Loeliger <loeli...@gmail.com> wrote:
> Mark,
>
> Not sure on which patch of your series to base the following comment,
> but thought it was worth bringing up for discussion nevertheless.
>
> So, over in U-Boot ARMv7's arch/arm/cpu/armv7/start.S code, there
> is a bit of code under the enty-point named cpu_init_cp15 that does
> some I-Cache/MMU/state and Errata fi-xup for the boot CPU. As it should.
> That's all splendid.
>
> However, any secondary CPU coming out of reset *after* the U-Boot is
> gone and only the secure monitor remains will still need to execute code
> that is substantially similar if not exactly the same.
>
> That means the same code block (that cpu_init_cp15 code) needs to be
> assembled into the original out-of-reset sequence for the boot CPU, and
> also needs to be assembled into the secure monitor code for secondary
> CPU out-of-reset sequences.
>
> We clearly can not directly refactor that function, extracting it and
> reusing it wiith the same entry-point as it needs to be linked into
> two different sections. We could put it in some header file and
> #include it into both places. We could define a macro and instantiate
> it in each place. We could do some hacky assemble it twice with
> different names thing.
>
> Thoughts?
Yes, that makes perfect sense. Can we put this into a separate object
file and play some link-time hackery instead? Otherwise, the include
thingy seems fine.
Cheers,
M.
--
Without deviation from the norm, progress is not possible.
_______________________________________________
U-Boot mailing list
U-Boot@lists.denx.de
http://lists.denx.de/mailman/listinfo/u-boot
