Re: [PATCH 04/23] ARM: i.MX8M: add support to pass DT via imx8m{m,n,q,p}_load_and_start_image_via_tfa()

[Marco Felsch <m.felsch@pengutronix.de>]

Hi Ahmad,

thanks for your input, please see below.

On 26-01-22, Ahmad Fatoum wrote:
> Hello Marco,
> 
> On 1/18/26 8:09 PM, Marco Felsch wrote:
> > Hi Ahmad,
> > 
> > On 26-01-15, Ahmad Fatoum wrote:
> > I get your point regarding the unnecessary breakage. It would have been
> > nice if we could have a TF-A loading abstraction within the PBL since
> > loading the TF-A and (opt. OP-TEE OS) is mostly the same on each and
> > every ARMv8-A SoC and we tend to adapt barebox rather than adapting the
> > TF-A binary.
> 
> Sure, I would also love to have a generic flow for that and that's one
> reason, why I want to restructure our early PBL code and want to avoid
> extra API breakage till then.
> 
> > However I would like to mention that handover data was meant to be
> > passing data between the PBL and barebox. We would repurpose the
> > handover mechanism if we would use the handover data. In this case it
> > could be re-used in several other places. Therefore I'm not a fan of
> > this idea.
> 
> My original thought is that we could use the same DT later for passing
> to barebox proper as well, so we are really just moving the addition of
> the handoff entry earlier.
> 
> I see now that this doesn't really work as barebox can be chainloaded at
> EL2/1 and in that case, the initial setting of the oftree handoff data
> would've been lost.
> 
> > What do you think about adding the 'fdt' support only to the 'lowlevel'
> > APIs:
> >   - __imx8mp_load_and_start_image_via_tfa()
> >   - __imx8mm_load_and_start_image_via_tfa()
> >   - __imx8mn_load_and_start_image_via_tfa()
> > ?
> 
> Sure, I can live with that.

Okay, I dropped the imx8m*_load_and_start_image_via_tfa() changes but
kept the changes for __imx8m*_load_and_start_image_via_tfa().

Regards,
  Marco