On 11/30/2017 03:56 PM, David Hildenbrand wrote: > On 30.11.2017 10:10, Thomas Huth wrote: >> On 29.11.2017 21:26, David Hildenbrand wrote: >>> Let's use s390_program_interrupt() instead. >>> >>> Reviewed-by: Richard Henderson <richard.hender...@linaro.org> >>> Signed-off-by: David Hildenbrand <da...@redhat.com> >>> --- >>> target/s390x/fpu_helper.c | 2 +- >>> target/s390x/int_helper.c | 14 +++++++------- >>> target/s390x/internal.h | 2 -- >>> target/s390x/misc_helper.c | 16 ---------------- >>> 4 files changed, 8 insertions(+), 26 deletions(-) >> >> Is it a disadvantage that runtime_exception() was declared as >> QEMU_NORETURN, and s390_program_interrupt() is not declared as >> QEMU_NORETURN? > > We could add that to trigger_pgm_exception instead. But if > cpu_loop_exit() would really return, we would be in more trouble AKA > nothing would work.
s390_program_interrupt and trigger_pgm_exception do return (kvm). We could export tcg_s390_program_interrupt as NORETURN, and use it directly in places that are tcg-only. That might provide some minor advantages to the compiler, but I don't think it's critical. r~
