From mboxrd@z Thu Jan  1 00:00:00 1970
From: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Date: Wed, 10 Apr 2019 21:38:30 +0300
Subject: [U-Boot] bootefi failures on armv7
In-Reply-To: <61f8820b-b7ec-8754-773a-82f8cf8fb11f@gmx.de>
References: <20190410165704.GA10242@apalos>
 <61f8820b-b7ec-8754-773a-82f8cf8fb11f@gmx.de>
Message-ID: <20190410183830.GA12913@apalos>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 8bit
To: u-boot@lists.denx.de

Hello Heinrich,
> 
> @Simon, @Alex
> Why did we ever introduce this value?
> 127 MB in baf70c02107 ("efi: Relocate FDT to 127MB instead of 128MB")
> 128 MB in ad0c1a3d2ce ("efi_loader: Put fdt into convenient location")
> 
> baf70c02107 says that the Sandbox has only 128 MiB. But why should we
> need to hard code this here. When the Sandbox has initialized the memory
> map efi_allocate_pool() will take care of finding an address in the
> available memory anyway.
> 
> ad0c1a3d2ce mentions 128 MB as recommended value but does not provide
> any reference. I think we should provide the source of this recommendation.
> 
> According to
> https://www.st.com/en/microcontrollers-microprocessors/stm32mp157.html
> STM32MP157 comes with support for TrustZone and SecureBoot.
> 
> https://www.st.com/en/microcontrollers-microprocessors/stm32mp157.html#tools-software:
> 
> "The developer package comes with
> 
>     Boot chain based on TF-A and U-Boot
>     OP-TEE secure OS running on the Arm® Cortex®-A in secure mode."
> 
> The Linux device tree does not reserve a memory region for the trusted
> firmware (TF-A) and the trusted execution environment (OP-TEE) Shouldn't
> there be a reserved memory area for both.
> 
> If TF-A or OP-TEE reside near 0xC8000000, this would explain the
> observed behavior.
> 
> @Ilias
> Do you use TF-A or OP-TEE in your setup?
No none of those are present in my tests. 

Note that in both workaround 1/2 fdt ends up at the same address (0xc7ef4000).

Thanks
/Ilias