[PATCH 1/2] riscv: create a custom CPU implementation for PolarFire SoC

[Jamie.Gibbons at microchip.com]

Hi Leo,

On Thu, 2025-12-04 at 15:45 +0800, Leo Liang wrote:
> EXTERNAL EMAIL: Do not click links or open attachments unless you know
> the content is safe
> 
> Hi Jamie,
> 
> On Wed, Nov 19, 2025 at 12:38:42PM +0000, Jamie Gibbons wrote:
> > [EXTERNAL MAIL]
> > 
> > From: Conor Dooley <conor.dooley at microchip.com>
> > 
> > PolarFire SoC needs a custom implementation of top_of_ram(), so stop
> > using the generic CPU & create a custom CPU instead.
> > 
> > Signed-off-by: Conor Dooley <conor.dooley at microchip.com>
> > ---
> >  arch/riscv/Kconfig                     |  1 +
> >  arch/riscv/cpu/mpfs/Kconfig            | 16 +++++++++++
> >  arch/riscv/cpu/mpfs/Makefile           |  6 ++++
> >  arch/riscv/cpu/mpfs/cpu.c              | 22 +++++++++++++++
> >  arch/riscv/cpu/mpfs/dram.c             | 38
> > ++++++++++++++++++++++++++
> >  arch/riscv/include/asm/arch-mpfs/clk.h |  8 ++++++
> >  board/microchip/mpfs_generic/Kconfig   |  4 +--
> >  7 files changed, 93 insertions(+), 2 deletions(-)
> >  create mode 100644 arch/riscv/cpu/mpfs/Kconfig
> >  create mode 100644 arch/riscv/cpu/mpfs/Makefile
> >  create mode 100644 arch/riscv/cpu/mpfs/cpu.c
> 
> The cpu.c file only contains "cleanup_before_linux" and seems
> identical
> with the one provided in arch/riscv/cpu/generic/cpu.c.
> 
> Other than that, LGTM.
> 
> If you don't mind, I could fix this on my side that you don't need to
> resend the patchset again.
> 
> Reviewed-by: Leo Yu-Chi Liang <ycliang at andestech.com>

Thank you for your response, review and feedback.

Regarding 'mpfs/cpu.c', you're corect that it is currently identical to
the generic implementation. Our intent was to provide a placeholder for
any future Polarfire-specific logic and also assumed it was manditory,
but if duplication is unneccessary, I'm happy for you to adjust as you
see fit.

In other words, if U-Boot prefers to avoid duplication and this file is
not mandatory for build or architextural reasons per CPU implementation,
than please go ahead and use the generic version.

Thanks,
Jamie.