[llvm] [RISCV] Reduce the VL of both operands in VMERGE_VVM (PR #144759)

[Philip Reames via llvm-commits]

https://github.com/preames approved this pull request.

LGTM

I suspect we could handle this in VLOptimizer, but I have no problem with this as an incremental improvement.

For the VLOptimizer case, v8 should the passthru operand, and the load should have a "tail undefined" property.  Given that, we should be able to reduce the VL of vmv.v.i that way as well.  

https://github.com/llvm/llvm-project/pull/144759