On Mon, 2022-05-16 at 12:00 +0200, Michael Nazzareno Trimarchi wrote:
Hi Alban
On Mon, May 16, 2022 at 10:23 AM Alban Bedel alban.bedel@aerq.com wrote:
If optee is running add the firmware node to the FDT to allow the kernel to use a more generic device tree.
Signed-off-by: Alban Bedel alban.bedel@aerq.com
arch/arm/mach-imx/imx8m/soc.c | 9 +++++++++ 1 file changed, 9 insertions(+)
diff --git a/arch/arm/mach-imx/imx8m/soc.c b/arch/arm/mach- imx/imx8m/soc.c index 8e23e6da326f..2a78cb6a0952 100644 --- a/arch/arm/mach-imx/imx8m/soc.c +++ b/arch/arm/mach-imx/imx8m/soc.c @@ -30,6 +30,7 @@ #include <fsl_wdog.h> #include <imx_sip.h> #include <linux/bitops.h> +#include <tee/optee.h>
DECLARE_GLOBAL_DATA_PTR;
@@ -1347,6 +1348,14 @@ usb_modify_speed: #endif
cleanup_nodes_for_efi(blob);
+ if (rom_pointer[1]) { + int err = optee_add_firmware_node(blob, "linaro,optee-tz", + "smc"); + if (err) + return err; + }
return 0; } #endif --
Adding the node is not sufficient. I think that we need reserved memory node on top of it.
My board use an imx8m and there the SoC code already carve out the optee memory (arch/arm/mach-imx/imx8m/soc.c dram_init_banksize()). In that case it is enough to just add the firmware node as the memory node created in the DTB already exclude the optee area. I expect it was done like this to also prevent using this memory area in u-boot.
This is a first step we can later add another function to also create the reserved memory nodes when needed.
The nice scenario will be only to define the size of such area and use always the last part of the memory to reserve it.
The optee area might be anywhere, the size won't be enough.
Alban