On 9/2/21 2:40 PM, Song Gao wrote:
+TRANS(fmov_s, gen_mov, tcg_gen_mov_tl)
Hmm. The spec says only the low 32-bits are modified? But is also says that if the source is not in the single-precision format the result is uncertain?
I'm not sure how to reconcile these two statements. Ideally the language would allow the result to be nanboxed... r~
