On 09/08/2015 01:19 AM, Peter Maydell wrote: > The tcg common code isn't smart enough to notice it only > needs to calculate not(t1) once ?
Correct, we do no value numbering or cse. > In the overwhelmingly common case (x86 tcg backend) > we would save an insn every time, right? Yes. It all depends on what value is used for NZCV, of course. Thankfully we *do* do dead code elimination, which is why I unconditionally compute both T1 and T2, and let them be deleted should they be unused. > I wouldn't bother to make the front-end generate different > code for the backend does/doesn't have andc situations, > certainly. I'll mock it up and see how much duplication there is. And also check it out on a Haswell host, which does have andc. r~
