[PATCH] D37579: [InstCombine] Fix PR21780 Expansion of 256 bit vector loads fails to fold into shuffles

[Sanjoy Das via Phabricator via llvm-commits]

sanjoy added a comment.

Why not leave the original load in place, and tag the load metadata like `!speculation_marker` ?  In either case, this needs a LangRef entry.


https://reviews.llvm.org/D37579