[PATCH] D12036: We shouldn't need to pass -fno-strict-aliasing when building clang with clang.

Justin Bogner via cfe-commits cfe-commits at lists.llvm.org
Fri Aug 14 18:20:08 PDT 2015

Chris Bieneman <beanz at apple.com> writes:
> Ok… I wrote some CMake goop to enable sanitizers on a bootstrap build
> (I’ll clean up the patches and send those out separately). check-clang
> worked with asan and ubsan enabled.
> Is that sufficient pre-commit testing?

Yes, I'd expect that to shake out any problems.

> I still need to rework the Makefile side of the patch to match Eric’s
> suggestion.

Hal also suggested checking for the __clang__ define rather than
checking the compiler name. That seems better to me.

David Blaikie <dblaikie at gmail.com> writes:
> (does strict aliasing still have the special case for enum type
> punning?  Because I've certainly seen (& fixed) that in a few places
> in Clang/LLVM)

I'm sure it does - this pattern is used pretty pervasively.

More information about the cfe-commits mailing list