[U-Boot] Uboot send pull request

[Rick Chen]

Rick Chen <rickchen36 at gmail.com> 於 2018年11月21日 週三 下午4:53寫道：
>
> Bin Meng <bmeng.cn at gmail.com> 於 2018年11月21日 週三 下午3:18寫道：
> >
> > Hi Rick,
> >
> > On Wed, Nov 21, 2018 at 2:00 PM Rick Chen <rickchen36 at gmail.com> wrote:
> > >
> > > > >
> > > > > Do any of your above patches get v2 posted? At least I did not see any follow-up
> > > > > response [1] regarding to "riscv: ax25-ae350: Pass dtb address to u-boot with a1
> > > > > register"
> > >
> > > First of all you told me about that
> > > board_fdt_blob_setup() should be completely removed.
> > > Instead the simple fix should be add CONFIG_OF_PRIOR_STAGE
> > > to ax25-ae350_defconfig.
> > >
> > > Then I explain you about that is for booting from flash.
> > > And you said
> > > In this case, we need support OF_CONTROL like other arch, eg:
> > > CONFIG_OF_EMBED and CONFIG_OF_SEPARATE.
> > >
> > > In README, I don't find the limitation about when booting from flash,
> > > it can not use  CONFIG_OF_BOARD.
> > >
> >
> > Correct. However, when ax25-ae350's U-Boot boots from ROM, where does
> > prior_stage_fdt_address get assigned to the ROM address? Besides, this
> > variable is declared in fdtdec.h as follows:
> >
> > #if CONFIG_IS_ENABLED(OF_PRIOR_STAGE)
> > extern phys_addr_t prior_stage_fdt_address;
> > #endif
> >
> > It means only OF_PRIOR_STAGE will use this variable. But in your patch
>
> I have seen that declaration in fdtdec.h
> But I was referred to
> include\configs\bcmstb.h
> extern phys_addr_t prior_stage_fdt_address;
> and that is why I borrow prior_stage_fdt_address.
>
> You do not point out this consideration directly, maybe that is why
> we have some thinking is different.
>
> > this was borrowed to support OF_BOARD, which is confusing I think. If
> > U-Boot boots from reset vector, the canonical way to support DT is via
> > CONFIG_OF_EMBED or CONFIG_OF_SEPARATE.
> >
> > > That is why ax25-ae350 use
> > > CONFIG_OF_BOARD to support boot from RAM and ROM.
> > > It is a flexible approach to meet Andes' HW mechanism.
> > >
> > > I think it is a configuration in ax25-ae350_defconfig.
> > > It will not affect qemu-riscv32(64)_defconfig.
> > >
> >
> > I understand it does not affect qemu riscv board. The thing is that I
> > wasn't clear why you did that way for ax25-ae350.
>
> Actually I just want to remove the redundant code in start.S
> -       li      t0, CONFIG_SYS_SDRAM_BASE
> -       SREG    a2, 0(t0)
> so borrowed prior_stage_fdt_address in CONFIG_OF_BOARD
>

Maybe I can keep CONFIG_SYS_SDRAM_BASE
and modify it as below
       li      t0, CONFIG_SYS_SDRAM_BASE
       SREG    a1, 0(t0)

But it is a little weird here.


>
> Rick
>
>
>
> >
> > > That is why I do not prepare v2 to modify it as
> > > CONFIG_OF_EMBED or CONFIG_OF_SEPARATE
> > > as your willingness.
> > >
> >
> > Regards,
> > Bin