On 10/16/18 3:40 AM, Peter Maydell wrote: > On 8 October 2018 at 22:21, Richard Henderson > <richard.hender...@linaro.org> wrote: >> Most of the v8 extensions are self-contained within the ISAR >> registers and are not implied by other feature bits, which >> makes them the easiest to convert. >> >> Reviewed-by: Philippe Mathieu-Daudé <phi...@redhat.com> >> Signed-off-by: Richard Henderson <richard.hender...@linaro.org> > >> diff --git a/target/arm/translate.h b/target/arm/translate.h >> index c1b65f3efb..1d60569583 100644 >> --- a/target/arm/translate.h >> +++ b/target/arm/translate.h >> @@ -7,6 +7,7 @@ >> /* internal defines */ >> typedef struct DisasContext { >> DisasContextBase base; >> + ARMCPU *cpu; /* for access to the id_* registers */ > > The translate code is not supposed to have access to either ARMCPU > or the ARMCPUState env pointer. Putting a pointer to cpu into the > DisasContext defeats this. This is why aarch64_tr_init_disas_context() > and the 32-bit equivalent extract all the info they need from > arm_cpu and env and put it into DisasContext fields.
I know that. I also know that if we don't do it this way, then we need to duplicate all of the routines that query the ID registers. I think this way is cleaner. We simply have to be vigilant about how ctx->cpu is used otherwise. r~