Re: [PATCH] sh: cache: Fill in invalidate_icache_all()

Hi Marek,

On Tue, Sep 10, 2024 at 7:55 PM Marek Vasut <[email protected]> wrote:
> On 9/10/24 7:21 PM, Tom Rini wrote:
> > On Tue, Sep 10, 2024 at 02:15:58AM +0200, Marek Vasut wrote:
> >
> >> Implement invalidate_icache_all() by clearing all V bits in
> >> IC and OC. This is done by setting CCR cache control register
> >> ICI and OCI bits.
> >>
> >> Signed-off-by: Marek Vasut <[email protected]>
> >> ---
> >> Cc: Ilias Apalodimas <[email protected]>
> >> Cc: Nobuhiro Iwamatsu <[email protected]>
> >> Cc: Tom Rini <[email protected]>
> >> Cc: [email protected]
> >> ---
> >>   arch/sh/cpu/sh4/cache.c | 11 +++++++----
> >>   1 file changed, 7 insertions(+), 4 deletions(-)
> >>
> >> diff --git a/arch/sh/cpu/sh4/cache.c b/arch/sh/cpu/sh4/cache.c
> >> index d3c480e79ed..99acc599965 100644
> >> --- a/arch/sh/cpu/sh4/cache.c
> >> +++ b/arch/sh/cpu/sh4/cache.c
> >> @@ -33,8 +33,9 @@ static inline void cache_wback_all(void)
> >>      }
> >>   }
> >>
> >> -#define CACHE_ENABLE      0
> >> -#define CACHE_DISABLE     1
> >> +#define CACHE_ENABLE                0
> >> +#define CACHE_DISABLE               1
> >> +#define CACHE_INVALIDATE    2
> >>
> >>   static int cache_control(unsigned int cmd)
> >>   {
> >> @@ -46,7 +47,9 @@ static int cache_control(unsigned int cmd)
> >>      if (ccr & CCR_CACHE_ENABLE)
> >>              cache_wback_all();
> >>
> >> -    if (cmd == CACHE_DISABLE)
> >> +    if (cmd == CACHE_INVALIDATE)
> >> +            outl(CCR_CACHE_ICI | ccr, CCR);
> >> +    else if (cmd == CACHE_DISABLE)
> >>              outl(CCR_CACHE_STOP, CCR);
> >>      else
> >>              outl(CCR_CACHE_INIT, CCR);
> >> @@ -103,7 +106,7 @@ void icache_disable(void)
> >>
> >>   void invalidate_icache_all(void)
> >>   {
> >> -    puts("No arch specific invalidate_icache_all available!\n");
> >> +    cache_control(CACHE_INVALIDATE);
> >>   }
> >>
> >>   int icache_status(void)
> >
> > Thanks for filling this in, I wasn't sure how the implementation would
> > look from a quick skim of the linux kernel code.
> I believe it should look like the above, but lemme CC Geert to be on the
> safe side.

I have no idea.  I removed the corresponding (but incomplete) code
from Linux in commit 25c7d77d695a4104 ("sh: boot: Remove sh5 cache
handling"), as it was a relic of sh5 cache handling.

SH4 cache handling seems to be buried deeper. At least the above
matches for invalidation:

arch/sh/mm/cache-sh4.c- /* Flush I-cache */
arch/sh/mm/cache-sh4.c: ccr = __raw_readl(SH_CCR);
arch/sh/mm/cache-sh4.c- ccr |= CCR_CACHE_ICI;
arch/sh/mm/cache-sh4.c: __raw_writel(ccr, SH_CCR);

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds