[Buildroot] [PATCH 00/21] Add ti_j721e_sk_defconfig

[Andreas Dannenberg]

Hi Alexander,

On Sat, Feb 17, 2024 at 04:13:16AM +0100, Alexander Sverdlin wrote:
> Hello Andreas,
> 
> On Fri, 2024-02-16 at 16:20 -0600, Andreas Dannenberg wrote:
> > On Fri, Feb 16, 2024 at 10:52:21PM +0100, Romain Naour wrote:
> > > Hello Andreas,
> > > 
> > > Le 16/02/2024 à 21:19, Andreas Dannenberg a écrit :
> > > > On Tue, Feb 13, 2024 at 06:27:56PM +0100, Romain Naour wrote:
> > > > > This series aim to add the SK-TDA4VM board support based on TI K3 j721e SoC.
> > > > > 
> > > > > This series start by adding U-boot binman support already submitted in the
> > > > > preview series "Add support for AM62x-SK HS-FS devices" from Dario Binacchi
> > > > > [1], with small improvements.
> > > > > See the FOSDEM 2024 talk "Standardizing the generation and signing of boot images" [2]
> > > > 
> > > > Thanks Romain and Dario for all the work on this, and submitting this.
> > > > Will be spending some time over the next few days to see how I can help
> > > > out with this effort. I can already see many great improvements since
> > > > the last time I worked on AM62x/AM64x support last year.
> > > 
> > > Your feed back on HS-FS devices is really helpful. Alexander Sverdlin also
> > > helped on AM62x.
> > > 
> > > Maybe if you can also test on AM64x just to be sure the defconfig is working.
> > 
> > I'm planning on testing the entire series here pretty soon and will
> > provide feedback (I have access to any and all boards:)
> > 
> > One of the things I'm trying to assess is what's still missing vs. the
> > previous attempt I made upstreaming this. For example, AM62x graphics
> > isn't there yet (ti-rogue-km / ti-rogue-um).
> 
> maybe you could shed some light onto the AM62x situation. My conclusion
> is that since your initial buildroot BSP U-Boot has been switched
> to standard boot (CONFIG_BOOTSTD), refer to (U-Boot repo):
> 
> commit 355c0afcd4b7c8c442c4d03e07649c24194b8e06
> Author: Nishanth Menon <nm at ti.com>
> Date:   Fri Aug 25 13:02:52 2023 -0500
> 
>     configs: am62x_evm_a53_defconfig: Switch to bootstd
> 
> and
> 
> commit ab54d9b28d8578cc43b214d39009e7b1def797e4
> Author: Roger Quadros <rogerq at kernel.org>
> Date:   Thu Oct 5 16:06:41 2023 +0300
> 
>     board: ti: am62x: am62x.env: Fix boot_targets
>     
>     ti_mmc is not a valid boot_target for standard boot flow so
> 
> I suppose TI's BSP still uses older U-Boot and is still not affected.
> Do we need uEnv.txt now? Do you have an idea which distro Nishanth may
> have tested with BOOTSTD change?

I wasn't aware of those changes but I just looked at some of them. Looks
like most of the effort was targeted around the BeaglePlay board.

And this was tested with Debian. You can find one of Nishanth's boot
logs here [1]. It was referenced in the cover letter [2] of the U-Boot
series that introduced the 'Switch to bootstd' change you pointed out.

If this indeed causes some widespread breakage including Buildroot we
need to understand and think about how to best resolve this. Will dig
into this some more as I'm trying to get back into helping with the
Buildroot effort here.

Thanks, Andreas



[1] https://gist.github.com/nmenon/c74acb3a895053e05623e886df77c8fe
[2] https://lore.kernel.org/all/20230824031101.3460411-1-nm@ti.com/



> 
> -- 
> Alexander Sverdlin.
>