[llvm] [DAG] Use known-bits when creating umulh/smulh. (PR #160916)

[Simon Pilgrim via llvm-commits]

RKSimon wrote:

@davemgreen Sorry for not getting back to you - I've taken a look at the x86 mulh lowering code and have a few ideas on improvements - if you want to merge this now that's fine, or do you want to wait until I push some changes?

https://github.com/llvm/llvm-project/pull/160916