[Bug target/107199] New: AVX512 fully masked loop vectorization needs extract_last pattern for vectorization of live variables

[rguenth at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107199

            Bug ID: 107199
           Summary: AVX512 fully masked loop vectorization needs
                    extract_last pattern for vectorization of live
                    variables
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: rguenth at gcc dot gnu.org
  Target Milestone: ---

For fully masked vectorization with AVX512 we'd need to define extract_last
which given a loop mask {kN} extracts the lane of the vector corresponding
to the last set bit in {kN} (the last iteration of the loop).

I don't see a kOP for this so I suppose moving {kN} to a GPR and then
doing BSR, moving it back and then doing a compress might work.