[llvm] r176768 - LoopVectorizer: Ignore dbg.value instructions

[Jakob Stoklund Olesen]

On Mar 10, 2013, at 12:21 PM, Arnold <aschwaighofer at apple.com> wrote:

> Yes, we do with any potentially vectorizable but unrecognized call. 
> 
> It is especially bad for -g. If you turn on debugging you will get a dbg.value for the induction variable turning off vectorization for at lot of loops I have seen. :(

I couldn't actually find documentation requiring this, but usually we don't want the -g option to affect optimization or code generation. The assembly code produced with and without -g should be identical except for the actual debug information.

This usually means that all optimizations must carefully ignore llvm.dbg.* intrinsics and DBG_VALUE instructions in the backends.

/jakob