I've reached a point where *all* targets must use the translator loop.
Changes for v2:
* Fix (drop) singlestep check for max_insns.
We already do that generically.
* Fully initialize DisasContext.
It was that last that caused the check-acceptance regression for v1.
I simply got lucky the first time (with optimization) and the
uninitialized struct members happened to have some zeros.
r~
Richard Henderson (3):
target/avr: Add DisasContextBase to DisasContext
target/avr: Change ctx to DisasContext* in gen_intermediate_code
target/avr: Convert to TranslatorOps
target/avr/translate.c | 284 ++++++++++++++++++++++-------------------
1 file changed, 154 insertions(+), 130 deletions(-)
--
2.25.1