https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50971
Jeffrey A. Law <law at redhat dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|ASSIGNED |RESOLVED
CC| |law at redhat dot com
Resolution|--- |FIXED
--- Comment #3 from Jeffrey A. Law <law at redhat dot com> ---
GCC has been finding both loops since at least gcc-4.7. The problem is passes
between their initial discovery and the loop optimizers (mergephi in
particular) would squash out one of the header blocks which creates the shared
header problem (and thus in turn the loop optimizers treat this as a single
loop).
In gcc-4.9 and beyond mergephi no longer squashes out the second header, so the
two loops are stay distinct throughout the optimizer pipeline. Given this was
never a regression, backporting those fixes to the gcc-4.8 branch isn't on the
table. I'm closing this as resolved.
