Re: [PATCH meta-rockchip] bsp: fix upstream TF-A deploy desync

[Trevor Woerner <twoerner@gmail.com>]

On Wed 2026-01-21 @ 06:11:26 PM, Quentin Schulz wrote:
> Hi Trevor,
> 
> On 1/21/26 6:06 PM, Quentin Schulz wrote:
> > From: Quentin Schulz <quentin.schulz@cherry.de>
> > 
> > Since commit 7bce36a2c64b ("arm/trusted-firmware-a: Use
> > firmware.bbclass") in meta-arm, the BL31 firmware is deployed in a
> > subdirectory named trusted-firmware-a and the binary is simply called
> > bl31.elf without any other string.
> > 
> > This differs from what it used to be by the use of the subdirectory and
> > the name of the file itself. This is an issue for us because the U-Boot
> > recipe is given the path to the BL31 binary in deploy directory and that
> > now was wrong.
> > 
> > The second issue is that we want rockchip-rkbin-tf-a to be a drop-in
> > replacement for the upstream recipe and have nothing to change in other
> > recipes dependent on *an* implementation of TF-A (such as the U-Boot
> > recipe).
> > 
> > We can either fixup the path like was done in 2a13a49da4af ("fix u-boot
> > for tf-a firmware") but we'll need to do that for all current and future
> > SoCs (and also wasn't a complete fix as for example px30 didn't have the
> > fixup applied). Or we can simply adapt the rockchip-rkbin-tf-a recipe to
> > deploy the same way that the upstream trusted-firmware-a recipe does by
> > installing in a trusted-firmware-a subdirectory, simply name the binary
> > bl31.elf and update the BL31 environment variable in the U-Boot recipe
> > to use the new path.
> > 
> > Because RKBIN_DEPLOY_FILENAME now contains a directory in it, we need to
> > make sure the directory is created (with install -D).
> > 
> > This was tested with upstream+downstream TF-A for RK3588, upstream TF-A
> > for RK3399 and PX30.
> > 
> > Signed-off-by: Quentin Schulz <quentin.schulz@cherry.de>
> > ---
> > This is a follow-up to
> > https://lore.kernel.org/yocto-meta-arm/28377373-a3e8-4416-aa7f-488cdfdbcf55@cherry.de/T/#m76f7d4b56849cc28fe80a76953f4e6e61ff0f7cc
> > and
> > https://lore.kernel.org/yocto-patches/dd990ccd-a418-45bc-b708-d3b1fc763ea5@cherry.de/
> 
> Forgot to say, but I think you want to branch off a whinlatter branch
> *before* this patch as the meta-arm commit that broke us is only in master
> and not whinlatter.

I broke off a whinlatter branch back in June. I try to create the
release branches once I'm aware of their existence in oe-core rather
than trying to figure out the last possible moment before they need to
be created.