[RFC PATCH 1/4] riscv: Optimize satp_mode data type

[Guo Ren]

On Wed, Jul 6, 2022 at 1:26 AM Christoph Hellwig <hch at infradead.org> wrote:
>
> On Tue, Jul 05, 2022 at 06:05:20AM -0400, guoren at kernel.org wrote:
> > From: Guo Ren <guoren at linux.alibaba.com>
> >
> > Fixup satp_mode data type. Use ulong instead of u64 for rv32
> > compatibility. Because the u64 type didn't cause any real problem, make
> > it as optimized.
>
> The changelog loooks odd, but given that CSR are Xlen sized this
> is the right thing to do even without further justification.
The last sentence is the reason not to add Fixes: tag.

>
> > -extern u64 satp_mode;
> > +extern ulong satp_mode;
>
> .. but please spell out unsigned long.
>
> >  #ifdef CONFIG_64BIT
> > -u64 satp_mode __ro_after_init = !IS_ENABLED(CONFIG_XIP_KERNEL) ? SATP_MODE_57 : SATP_MODE_39;
> > +ulong satp_mode __ro_after_init = !IS_ENABLED(CONFIG_XIP_KERNEL) ? SATP_MODE_57 : SATP_MODE_39;
> >  #else
> > -u64 satp_mode __ro_after_init = SATP_MODE_32;
> > +ulong satp_mode __ro_after_init = SATP_MODE_32;
> >  #endif
>
> And maybe make this less of a mess while we're at it:
>
> #ifdef CONFIG_32BIT
> #define SATP_DEFAULT    SATP_MODE_32
> #elif defined(CONFIG_XIP_KERNEL)
> #define SATP_DEFAULT    SATP_MODE_39
> #else
> #define SATP_DEFAULT    SATP_MODE_57
> #endif
>
> unsigned long satp_mode __ro_after_init = SATP_DEFAULT;
It a little involves the other coding convention work.

ulong satp_mode __ro_after_init = !IS_ENABLED(CONFIG_XIP_KERNEL) ?
SATP_MODE_57 : SATP_MODE_39;B
The above seems readable & clear to me.


-- 
Best Regards
 Guo Ren