Re: [PATCH] target/sh4: Fix TB_FLAG_UNALIGN

[Richard Henderson]

On 8/30/22 18:30, Yoshinori Sato wrote:
> On Tue, 30 Aug 2022 01:10:29 +0900,
> Richard Henderson wrote:
> > On 8/29/22 02:05, BALATON Zoltan wrote:
> > > On Sun, 28 Aug 2022, Richard Henderson wrote:
> > > > The value previously chosen overlaps GUSA_MASK.
> > > >
> > > > Cc: qemu-stable@nongnu.org
> > > > Fixes: 4da06fb3062 ("target/sh4: Implement prctl_unalign_sigbus")
> > > > Resolves: https://gitlab.com/qemu-project/qemu/-/issues/856
> > > > Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> > > > ---
> > > > target/sh4/cpu.h | 2 +-
> > > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > > >
> > > > diff --git a/target/sh4/cpu.h b/target/sh4/cpu.h
> > > > index 9f15ef913c..e79cbc59e2 100644
> > > > --- a/target/sh4/cpu.h
> > > > +++ b/target/sh4/cpu.h
> > > > @@ -84,7 +84,7 @@
> > > > #define DELAY_SLOT_RTE         (1 << 2)
> > > >
> > > > #define TB_FLAG_PENDING_MOVCA  (1 << 3)
> > > > -#define TB_FLAG_UNALIGN        (1 << 4)
> > > > +#define TB_FLAG_UNALIGN        (1 << 13)
> > >
> > > Is it worth a comment to note why that value to avoid the same
> > > problem if another flag is added in the future?
> >
> > Hmm, or perhaps move it down below, so that we see bit 3 used, then bits 4-12, 
> > then bit 13.
> >
> >
> > r~
>
> It looks like the gUSA and unalign access flags are mixed.
> I think the flags should also be separated as the two features are not related.

Well, of course.  That's what the first patch is fixing.
Balaton is merely discussing the order in which the bits
are defined.

r~