Created attachment 279125 [details, diff] abovesubscript.diff There's a long history of GCC outputting bogus "array subscript is above array bounds" warnings: http://gcc.gnu.org/PR45978 http://gcc.gnu.org/PR43949 http://gcc.gnu.org/PR47418 http://gcc.gnu.org/PR43833 http://gcc.gnu.org/PR45180 etc. In fact building gcc-4.6.1 itself with -O2 triggers a dozen or so. I'm filing a PR for those but I think based on this history we should drop the QA warning as it's likely to just be noise. The related "below array bounds" warning seems to work correctly.
most of the ones i looked at were actual bugs in the code. minor, but PR45180 does seem like a bug in the proposed test case as well (if the func doesnt return, then use __attribute__((noreturn))). i dont feel strongly about disabling it for now.
Heh, looks like the warnings I got building gcc are valid too. Good enough to change my mind.
