Hi Manorit,
On Tue, 18 Jul 2023 at 22:46, Manorit Chawdhry m-chawdhry@ti.com wrote:
Hi Simon,
On 16:04-20230718, Simon Glass wrote:
Hi,
On Mon, 17 Jul 2023 at 07:45, Nishanth Menon nm@ti.com wrote:
On 13:39-20230717, Manorit Chawdhry wrote:
[...]
+#define BOOT_TARGET_DEVICES(func) \
BOOT_TARGET_MMC(func) \BOOT_TARGET_USB(func) \BOOT_TARGET_PXE(func) \BOOT_TARGET_DHCP(func)+#include <config_distro_bootcmd.h>
With standard boot you should be able to drop all of the above, since the normal order is mmc, usb, pxe, dhcp by default. But you can add a "boot_targets" env var if you like.
The one exception is TI_MMC. What is that, exactly?
TI_MMC is our custom boot mechanism that we actually support as a part of our SDK, we had been using this in am62ax like the way I have done here, am not really sure why we hooked it into the distroboot if that is the question that you are asking, maybe Nishanth/Bryan can help with that as Bryan had done it for am62ax [0] but we do this boot mechanism for sure.
2 cents: There is no reason for us to have our own TI_MMC option. we should just go standard_boot. all the SDK stuff really should be fixed if they dont work with standard_boot.
If for some reason you do need something special, it could be implemented as a new bootmeth, as has been done for ChromeOS.
Can you point me to the respective implementation file so that I can maybe have a look at it and see how we can do it for our platforms?
Yes, please see boot/bootmeth_uclass.c for the uclass and boot/bootmeth_*.c for the impls.
https://u-boot.readthedocs.io/en/latest/develop/bootstd.html#bootmeth
Regards, Simon