Re: [RFC PATCH-for-8.0] target/arm: Keep "internals.h" internal to target/arm/

[Philippe Mathieu-Daudé]

On 9/12/22 16:19, Richard Henderson wrote:
> On 12/9/22 05:17, Philippe Mathieu-Daudé wrote:
> > +++ b/target/arm/machine.c
> > @@ -6,6 +6,45 @@
> >   #include "internals.h"
> >   #include "migration/cpu.h"
> > +/* This mapping is common between ID_AA64MMFR0.PARANGE and 
> > TCR_ELx.{I}PS. */
> > +static const uint8_t pamax_map[] = {
> > +    [0] = 32,
> > +    [1] = 36,
> > +    [2] = 40,
> > +    [3] = 42,
> > +    [4] = 44,
> > +    [5] = 48,
> > +    [6] = 52,
> > +};
> ...
> > +++ b/target/arm/ptw.c
> > @@ -42,45 +42,6 @@ static bool get_phys_addr_with_struct(CPUARMState 
> > *env, S1Translate *ptw,
> >                                         ARMMMUFaultInfo *fi)
> >       __attribute__((nonnull));
> > -/* This mapping is common between ID_AA64MMFR0.PARANGE and 
> > TCR_ELx.{I}PS. */
> > -static const uint8_t pamax_map[] = {
> > -    [0] = 32,
> > -    [1] = 36,
> > -    [2] = 40,
> > -    [3] = 42,
> > -    [4] = 44,
> > -    [5] = 48,
> > -    [6] = 52,
> > -};
>
> How does this even compile with the remaining usage of pamax_map in 
> get_phys_addr_lpae?

It does not :( I rebased 62 patches then tried to extract/post unrelated
patches like this one, but didn't build at each step and failed to
realize the broken rebase, *sigh*. Sorry.