[RISCV] Use v8-v23 as argument registers to conform to the proposal.

The maximum LMUL is 8. We need 16 vector registers for two LMUL-8
arguments. The modification follows the proposal of psABI in
riscv-non-isa/riscv-elf-psabi-doc#171

Differential Revision: https://reviews.llvm.org/D95134

Author: Hsiangkai

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

@@ -2565,14 +2565,15 @@ static const MCPhysReg ArgFPR64s[] = {
 };
 // This is an interim calling convention and it may be changed in the future.
 static const MCPhysReg ArgVRs[] = {
-  RISCV::V16, RISCV::V17, RISCV::V18, RISCV::V19, RISCV::V20,
-  RISCV::V21, RISCV::V22, RISCV::V23
-};
-static const MCPhysReg ArgVRM2s[] = {
-  RISCV::V16M2, RISCV::V18M2, RISCV::V20M2, RISCV::V22M2
-};
-static const MCPhysReg ArgVRM4s[] = {RISCV::V16M4, RISCV::V20M4};
-static const MCPhysReg ArgVRM8s[] = {RISCV::V16M8};
+    RISCV::V8,  RISCV::V9,  RISCV::V10, RISCV::V11, RISCV::V12, RISCV::V13,
+    RISCV::V14, RISCV::V15, RISCV::V16, RISCV::V17, RISCV::V18, RISCV::V19,
+    RISCV::V20, RISCV::V21, RISCV::V22, RISCV::V23};
+static const MCPhysReg ArgVRM2s[] = {RISCV::V8M2,  RISCV::V10M2, RISCV::V12M2,
+                                     RISCV::V14M2, RISCV::V16M2, RISCV::V18M2,
+                                     RISCV::V20M2, RISCV::V22M2};
+static const MCPhysReg ArgVRM4s[] = {RISCV::V8M4, RISCV::V12M4, RISCV::V16M4,
+                                     RISCV::V20M4};
+static const MCPhysReg ArgVRM8s[] = {RISCV::V8M8, RISCV::V16M8};
 
 // Pass a 2*XLEN argument that has been split into two XLEN values through
 // registers or the stack as necessary.

llvm/test/CodeGen/RISCV/rvv/mask-exts-truncs-rv32.ll

@@ -7,13 +7,13 @@ define <vscale x 2 x i32> @saddo_nvx2i32(<vscale x 2 x i32> %x, <vscale x 2 x i3
 ; CHECK-LABEL: saddo_nvx2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
-; CHECK-NEXT:    vmslt.vx v25, v17, zero
-; CHECK-NEXT:    vadd.vv v26, v16, v17
-; CHECK-NEXT:    vmslt.vv v27, v26, v16
+; CHECK-NEXT:    vmslt.vx v25, v9, zero
+; CHECK-NEXT:    vadd.vv v26, v8, v9
+; CHECK-NEXT:    vmslt.vv v27, v26, v8
 ; CHECK-NEXT:    vsetvli a0, zero, e8,mf4,ta,mu
 ; CHECK-NEXT:    vmxor.mm v0, v25, v27
 ; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
-; CHECK-NEXT:    vmerge.vim v16, v26, 0, v0
+; CHECK-NEXT:    vmerge.vim v8, v26, 0, v0
 ; CHECK-NEXT:    ret
   %a = call { <vscale x 2 x i32>, <vscale x 2 x i1> } @llvm.sadd.with.overflow.nxv2i32(<vscale x 2 x i32> %x, <vscale x 2 x i32> %y)
   %b = extractvalue { <vscale x 2 x i32>, <vscale x 2 x i1> } %a, 0