[llvm] [RISCV] Separate doLocalPostpass into new pass and move to post vector regalloc (PR #88295)

[Luke Lau via llvm-commits]

================
@@ -288,32 +288,44 @@ define <vscale x 16 x i64> @vector_interleave_nxv16i64_nxv8i64(<vscale x 8 x i64
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    csrr a0, vlenb
-; CHECK-NEXT:    slli a0, a0, 3
+; CHECK-NEXT:    slli a0, a0, 4
 ; CHECK-NEXT:    sub sp, sp, a0
-; CHECK-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 8 * vlenb
-; CHECK-NEXT:    vmv8r.v v0, v8
+; CHECK-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x10, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 16 * vlenb
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    slli a0, a0, 3
+; CHECK-NEXT:    add a0, sp, a0
+; CHECK-NEXT:    addi a0, a0, 16
+; CHECK-NEXT:    vs8r.v v8, (a0) # Unknown-size Folded Spill
----------------
lukel97 wrote:

> Additional spill code here, likely caused by an extra vsetvli dependence during machine scheduling. There is a possibility that the scheduler itself will be able to fix this issue once we move the vsetvl insertion pass into postRA completely.

Yes, that was my thoughts as well. For what its worth there doesn't seem to be much extra spilling in practice, on llvm test suite with SPEC CPU 2017 there's actually slightly less spills overall. But the difference is so small I would just chalk it up to noise

```
      regalloc.NumSpills                          
l/r                  lhs           rhs        diff
count  2067.000000        2067.000000   278.000000
mean   88.670053          88.665215     0.000010  
std    1155.298479        1155.264766   0.000260  
min    0.000000           0.000000     -0.001185  
25%    0.000000           0.000000      0.000000  
50%    0.000000           0.000000      0.000000  
75%    0.000000           0.000000      0.000000  
max    44054.000000       44054.000000  0.004115 
```

https://github.com/llvm/llvm-project/pull/88295