> Maybe someone can have a look at the attribute((pointer_no_escape)) > patch I posted a while ago. With some IPA machinery we could possibly > trim down the clobber lists quite a bit. >
Well, let me confirm first that he is right. This requires a cpgram.ii that compiles (none of the attachments to the testcase do) This is rather easy, i will test the improved-aliasing-branch from right before the merge. Second, it could be one of the previously silly tests that was removed (ie the is_global_var check in is_call_clobbered, even though we have globals that are *not* clobbered). Those would now be bitmap tests. If the bitmap is too expensive for random testing, we might have to move it into a flag. > Richard. >
