On Fri, Feb 17, 2023 at 05:14:11PM +0100, Francesco Dolcini wrote:
On Fri, Feb 17, 2023 at 09:03:07AM -0500, Tom Rini wrote:
On Fri, Feb 17, 2023 at 01:18:34PM +0000, Marcel Ziswiler wrote:
Hi Tom
On Thu, 2023-02-16 at 13:37 +0100, Marcel Ziswiler wrote:
On Wed, 2023-02-15 at 22:36 -0500, Tom Rini wrote:
On this platform SPL_BOARD_INIT is used to call arch_misc_init which is in turn empty, drop the call.
Cc: Marcel Ziswiler marcel.ziswiler@toradex.com Signed-off-by: Tom Rini trini@konsulko.com
Acked-by: Marcel Ziswiler marcel.ziswiler@toradex.com
Sorry, I have to take my ack back. Marek (and Francesco) are, of course, right and this is indeed needed for CAAM. Please see also [1] where it got changed/introduced in/to the current form.
[1] https://lore.kernel.org/all/20220919194118.105820-1-marex@denx.de/
Alright, yeah. I've posted v2 which preserves the existing behavior, but it could be improved further still by someone with the hardware to confirm behavior on.
[PATCHv2 12/13] global: Add a select for ARCH_MISC_INIT where used -- ?
I would say that even with that patch in, this specific one should be dropped.
Yes, with v2 of 12/13, patches 10 and 11 from v1 of the series are dropped, to be clear.
At least this is what I understand from your comment here:
Yes, it's still on the board to call arch_misc_init() inside spl_board_init().
We can test your changes on the Verdin iMX8MM, even Marek should have one to my understanding, in case we are somehow late.
What I mean is that Marek has suggested a more specific fix to this problem (it should not be per-board to get the CAAM working for HAB here, it should be in the SoC code and hooks), and I would quite welcome the general clean-up but can't test it, so won't write it, right now.