[llvm] [AArch64][SelectionDAG] Lower multiplication by a constant to shl+add+shl+add (PR #89532)

[David Green via llvm-commits]

davemgreen wrote:

Thanks for the info, that sounds good then. I think for newer Arm cores the cost model between the two is going to be pretty close. One or the other might be better in different places.

https://github.com/llvm/llvm-project/pull/89532