Hi Jonas / Simon,
I have submitted a patch based on Jonas's suggestion. Apologies, I am new to email based patch management and it went as new thread (at least for me)
It's here - https://lore.kernel.org/u-boot/20231114203309.852289-1-i@shantur.com/
Kind regards, Shantur
On Mon, Nov 13, 2023 at 12:07 AM Shantur Rathore i@shantur.com wrote:
Hey Jonas,
This sounds like a good idea. Please implement this.
Kind regards, Shantur
On Sun, Nov 12, 2023 at 3:01 PM Jonas Karlman jonas@kwiboo.se wrote:
Hi Shantur,
On 2023-11-12 15:21, Shantur Rathore wrote:
Hey Jonas,
For normal generic use the full bootstd commands should not be needed, do you have special scripting requirements that require access to full bootstd commands?
All rockchip boards use standard boot, this only enables full commands for one particular board, why is this board special? Or
This board isn't special.
is there merit to imply enable of full commands for all rockchip boards?
Maybe not only on rockchip boards but all boards.
Do we need full commands in general ? - Maybe not
Then why did you need to enable full commands ? Bootstd works when the distro used is supporting it. In my case when I installed the distro (Armbian), EFI was installed to /boot/efi/EFI/debian by default instead of /boot/efi/EFI/boot as expected by Bootstd. Bootstd couldn't find the EFI and failed to boot. In this scenario there is no way to figure out what's going on, what bootmethods / bootflows is bootstd able to find or try to manually boot some. To be able to support the smallest bit of debugging you end up building U-Boot from scratch to enable bootstd full for bootstd info commands. All other commands in U-boot by default come with options to display more information ( for ex. nvme info, nvme detail, usb tree, usb info, usb dev, fatinfo and many more ) then why is bootstd restricted? Thus, in my opinion we should enable BOOTSTD_FULL for users to be able to see more information. Restricting might be beneficial for end user devices but atleast development devices like RockPro64 should come with BOOTSTD_FULL by default.
Thanks for clarifying the need! And I also see the benefit of having full feature commands enabled.
For RK35xx family of devices I have tried to ensure all boards have same features and sounds like something like below could benefit all rk boards, or at least for ROCKCHIP_RK3568 and ROCKCHIP_RK3588 boards.
diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig index c39ae40335a0..07e93a9c7ad5 100644 --- a/arch/arm/Kconfig +++ b/arch/arm/Kconfig @@ -1987,6 +1987,7 @@ config ARCH_ROCKCHIP imply CMD_DM imply DEBUG_UART_BOARD_INIT imply BOOTSTD_DEFAULTS
imply BOOTSTD_FULL if BOOTSTD_DEFAULTS imply FAT_WRITE imply SARADC_ROCKCHIP imply SPL_SYSRESETRegards, Jonas
Kind regards, Shantur
On Sun, Nov 12, 2023 at 12:50 PM Jonas Karlman jonas@kwiboo.se wrote:
Hi Shantur,
On 2023-11-12 13:34, Shantur Rathore wrote:
Hi Jonas,
The CMD_SPI is not used to interact with the SPI flash.
CMD_SF is already enabled and you can use "sf probe" and any other sf related action to interact with the SPI flash on this board.
You are right, this is not needed, thanks for correcting me. I will update my patch.
What is the reasoning behind enabling full bootstd commands? Especially why just this board need it enabled by default.
Standard boot is already fully functional, it just does not have full features commands enabled by default.
By default standard boot only allows you to boot from the first available boot device. This board supports SDCard, NVME (via PCIe port), USB and network boot and with BOOTSTD_FULL we can choose what to exactly boot. The boot_targets environment variable only allows you to choose which device to boot, not which boot method / flow to use. We have ample space in SPI Flash, so why not let it have the full potential of U-Boot by default.
You can control boot targets using the boot_targets env var and boot methods using the bootmeths env var.
https://docs.u-boot.org/en/latest/develop/bootstd.html#controlling-ordering
For normal generic use the full bootstd commands should not be needed, do you have special scripting requirements that require access to full bootstd commands?
All rockchip boards use standard boot, this only enables full commands for one particular board, why is this board special? Or is there merit to imply enable of full commands for all rockchip boards?
Regards, Jonas
Kind regards, Shantur
On Sun, Nov 12, 2023 at 10:39 AM Jonas Karlman jonas@kwiboo.se wrote:
On 2023-11-11 01:13, Shantur Rathore wrote: > RockPro64 has a 16MB onboard SPI chip and current u-boot takes > around 2MB, we can enable more features. > Updating config to enable SPI commands and full BootSTD support. > --- > configs/rockpro64-rk3399_defconfig | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/configs/rockpro64-rk3399_defconfig b/configs/rockpro64-rk3399_defconfig > index 4cd6b76665..cad0b8a5d7 100644 > --- a/configs/rockpro64-rk3399_defconfig > +++ b/configs/rockpro64-rk3399_defconfig > @@ -46,10 +46,12 @@ CONFIG_CMD_BOOTZ=y > CONFIG_CMD_GPT=y > CONFIG_CMD_MMC=y > CONFIG_CMD_PCI=y > +CONFIG_CMD_SPI=y
The CMD_SPI is not used to interact with the SPI flash.
CMD_SF is already enabled and you can use "sf probe" and any other sf related action to interact with the SPI flash on this board.
> CONFIG_CMD_USB=y > # CONFIG_CMD_SETEXPR is not set > CONFIG_CMD_TIME=y > CONFIG_CMD_BOOTSTAGE=y > +CONFIG_BOOTSTD_FULL=y
What is the reasoning behind enabling full bootstd commands? Especially why just this board need it enabled by default.
Standard boot is already fully functional, it just does not have full features commands enabled by default.
Regards, Jonas
> CONFIG_SPL_OF_CONTROL=y > CONFIG_OF_SPL_REMOVE_PROPS="pinctrl-0 pinctrl-names clock-names interrupt-parent assigned-clocks assigned-clock-rates assigned-clock-parents" > CONFIG_ENV_IS_IN_SPI_FLASH=y