Hi Richard,
Quick attempt to remove MO_TE uses for fixed-endianness targets.
For the other targets I started to propagate MemOp in the call
chain, but since IIUC it is constant, I wonder if it isn't simpler
to have a CPUClass::code_endian_mo field, set once with a
target_big_endian() call. I can post another RFC to show, WDYT?
Regards,
Phil.
Philippe Mathieu-Daudé (5):
accel/tcg: Add endianness variants of cpu_ld{uw,l,q}_code()
target/alpha: Use little-endian variant of cpu_ldl_code()
target/loongarch: Use little-endian variant of cpu_ldl_code()
target/sparc: Use big-endian variant of cpu_ldl_code()
target/s390x: Use big-endian variant of cpu_ld{uw,l}_code()
include/accel/tcg/cpu-ldst.h | 43 +++++++++++++++++++++++++++++-----
target/alpha/mem_helper.c | 2 +-
target/loongarch/tcg/tcg_cpu.c | 2 +-
target/s390x/tcg/mem_helper.c | 6 ++---
target/sparc/int32_helper.c | 2 +-
5 files changed, 43 insertions(+), 12 deletions(-)
--
2.51.0
