[AArch64][SVE] Implement reversal intrinsics

[c-rhodes]

Summary

Adds intrinsics for the following:

• rbit
• revb
• revh
• revw

Patterns are also defined to map the 'llvm.bswap.*' intrinsic to the SVE
revb instruction.

[sdesmalen]

LGTM, thanks @c-rhodes!

llvm/include/llvm/IR/IntrinsicsAArch64.td

 // Counting bits
 //
 
 def int_aarch64_sve_cls : AdvSIMD_Merged1VectorArg_Intrinsic;
 def int_aarch64_sve_clz : AdvSIMD_Merged1VectorArg_Intrinsic;
 def int_aarch64_sve_cnt : AdvSIMD_SVE_CNT_Intrinsic;
 
 //
+// Reversal
+//
+
+def int_aarch64_sve_rbit : AdvSIMD_Merged1VectorArg_Intrinsic;
+def int_aarch64_sve_revb : AdvSIMD_Merged1VectorArg_Intrinsic;
+def int_aarch64_sve_revh : AdvSIMD_Merged1VectorArg_Intrinsic;
+def int_aarch64_sve_revw : AdvSIMD_Merged1VectorArg_Intrinsic;
+
+//
 // Permutations and selection
 //
 
 def int_aarch64_sve_sunpkhi   : AdvSIMD_SVE_Unpack_Intrinsic;
 def int_aarch64_sve_sunpklo   : AdvSIMD_SVE_Unpack_Intrinsic;
 
 def int_aarch64_sve_uunpkhi   : AdvSIMD_SVE_Unpack_Intrinsic;
 def int_aarch64_sve_uunpklo   : AdvSIMD_SVE_Unpack_Intrinsic;

llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td

let Predicates = [HasSVE] in {
   defm SEL_ZPZZ    : sve_int_sel_vvv<"sel">;
 
   defm SPLICE_ZPZ : sve_int_perm_splice<"splice">;
   defm COMPACT_ZPZ : sve_int_perm_compact<"compact">;
   defm INSR_ZR : sve_int_perm_insrs<"insr", AArch64insr>;
   defm INSR_ZV : sve_int_perm_insrv<"insr", AArch64insr>;
   def  EXT_ZZI : sve_int_perm_extract_i<"ext">;
 
-  defm RBIT_ZPmZ : sve_int_perm_rev_rbit<"rbit">;
-  defm REVB_ZPmZ : sve_int_perm_rev_revb<"revb">;
-  defm REVH_ZPmZ : sve_int_perm_rev_revh<"revh">;
-  defm REVW_ZPmZ : sve_int_perm_rev_revw<"revw">;
+  defm RBIT_ZPmZ : sve_int_perm_rev_rbit<"rbit", int_aarch64_sve_rbit>;
+  defm REVB_ZPmZ : sve_int_perm_rev_revb<"revb", int_aarch64_sve_revb, bswap>;
+  defm REVH_ZPmZ : sve_int_perm_rev_revh<"revh", int_aarch64_sve_revh>;
+  defm REVW_ZPmZ : sve_int_perm_rev_revw<"revw", int_aarch64_sve_revw>;
 
   defm REV_PP : sve_int_perm_reverse_p<"rev">;
   defm REV_ZZ : sve_int_perm_reverse_z<"rev">;
 
   defm SUNPKLO_ZZ : sve_int_perm_unpk<0b00, "sunpklo", AArch64sunpklo>;
   defm SUNPKHI_ZZ : sve_int_perm_unpk<0b01, "sunpkhi", AArch64sunpkhi>;
   defm UUNPKLO_ZZ : sve_int_perm_unpk<0b10, "uunpklo", AArch64uunpklo>;
   defm UUNPKHI_ZZ : sve_int_perm_unpk<0b11, "uunpkhi", AArch64uunpkhi>;

llvm/lib/Target/AArch64/SVEInstrFormats.td

 class SVE_4_Op_Imm_Pat<ValueType vtd, SDPatternOperator op, ValueType vt1,
                        ValueType vt2, ValueType vt3, ValueType vt4,
                        Operand ImmTy, Instruction inst>
 : Pat<(vtd (op vt1:$Op1, vt2:$Op2, vt3:$Op3, (vt4 ImmTy:$Op4))),
       (inst $Op1, $Op2, $Op3, ImmTy:$Op4)>;
 
 def SVEDup0Undef : ComplexPattern<i64, 0, "SelectDupZeroOrUndef", []>;
 
+//
+// Common but less generic patterns.
+//
+
+class SVE_1_Op_AllActive_Pat<ValueType vtd, SDPatternOperator op, ValueType vt1,
+                             Instruction inst, Instruction ptrue>
+: Pat<(vtd (op vt1:$Op1)),
+      (inst (IMPLICIT_DEF), (ptrue 31), $Op1)>;
+
 //===----------------------------------------------------------------------===//
 // SVE Predicate Misc Group
 //===----------------------------------------------------------------------===//
 
 class sve_int_pfalse<bits<6> opc, string asm>
 : I<(outs PPR8:$Pd), (ins),
   asm, "\t$Pd",
   "",
: I<(outs zprty:$Zd), (ins zprty:$_Zd, PPR3bAny:$Pg, zprty:$Zn),
   let Inst{9-5}   = Zn;
   let Inst{4-0}   = Zd;
 
   let Constraints = "$Zd = $_Zd";
   let DestructiveInstType = Destructive;
   let ElementSize = zprty.ElementSize;
 }
 
-multiclass sve_int_perm_rev_rbit<string asm> {
+multiclass sve_int_perm_rev_rbit<string asm, SDPatternOperator op> {
   def _B : sve_int_perm_rev<0b00, 0b11, asm, ZPR8>;
   def _H : sve_int_perm_rev<0b01, 0b11, asm, ZPR16>;
   def _S : sve_int_perm_rev<0b10, 0b11, asm, ZPR32>;
   def _D : sve_int_perm_rev<0b11, 0b11, asm, ZPR64>;
+
+  def : SVE_3_Op_Pat<nxv16i8, op, nxv16i8, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
+  def : SVE_3_Op_Pat<nxv8i16, op, nxv8i16, nxv8i1,  nxv8i16, !cast<Instruction>(NAME # _H)>;
+  def : SVE_3_Op_Pat<nxv4i32, op, nxv4i32, nxv4i1,  nxv4i32, !cast<Instruction>(NAME # _S)>;
+  def : SVE_3_Op_Pat<nxv2i64, op, nxv2i64, nxv2i1,  nxv2i64, !cast<Instruction>(NAME # _D)>;
 }
 
-multiclass sve_int_perm_rev_revb<string asm> {
+multiclass sve_int_perm_rev_revb<string asm,
+                                 SDPatternOperator int_op,
+                                 SDPatternOperator ir_op> {
   def _H : sve_int_perm_rev<0b01, 0b00, asm, ZPR16>;
   def _S : sve_int_perm_rev<0b10, 0b00, asm, ZPR32>;
   def _D : sve_int_perm_rev<0b11, 0b00, asm, ZPR64>;
+
+  def : SVE_3_Op_Pat<nxv8i16, int_op, nxv8i16, nxv8i1,  nxv8i16, !cast<Instruction>(NAME # _H)>;
+  def : SVE_3_Op_Pat<nxv4i32, int_op, nxv4i32, nxv4i1,  nxv4i32, !cast<Instruction>(NAME # _S)>;
+  def : SVE_3_Op_Pat<nxv2i64, int_op, nxv2i64, nxv2i1,  nxv2i64, !cast<Instruction>(NAME # _D)>;
+
+  def : SVE_1_Op_AllActive_Pat<nxv8i16, ir_op, nxv8i16, !cast<Instruction>(NAME # _H), PTRUE_H>;
+  def : SVE_1_Op_AllActive_Pat<nxv4i32, ir_op, nxv4i32, !cast<Instruction>(NAME # _S), PTRUE_S>;
+  def : SVE_1_Op_AllActive_Pat<nxv2i64, ir_op, nxv2i64, !cast<Instruction>(NAME # _D), PTRUE_D>;
 }
 
-multiclass sve_int_perm_rev_revh<string asm> {
+multiclass sve_int_perm_rev_revh<string asm, SDPatternOperator op> {
   def _S : sve_int_perm_rev<0b10, 0b01, asm, ZPR32>;
   def _D : sve_int_perm_rev<0b11, 0b01, asm, ZPR64>;
+
+  def : SVE_3_Op_Pat<nxv4i32, op, nxv4i32, nxv4i1,  nxv4i32, !cast<Instruction>(NAME # _S)>;
+  def : SVE_3_Op_Pat<nxv2i64, op, nxv2i64, nxv2i1,  nxv2i64, !cast<Instruction>(NAME # _D)>;
 }
 
-multiclass sve_int_perm_rev_revw<string asm> {
+multiclass sve_int_perm_rev_revw<string asm, SDPatternOperator op> {
   def _D : sve_int_perm_rev<0b11, 0b10, asm, ZPR64>;
+
+  def : SVE_3_Op_Pat<nxv2i64, op, nxv2i64, nxv2i1,  nxv2i64, !cast<Instruction>(NAME # _D)>;
 }
 
 class sve_int_perm_cpy_r<bits<2> sz8_64, string asm, ZPRRegOp zprty,
                          RegisterClass srcRegType>
 : I<(outs zprty:$Zd), (ins zprty:$_Zd, PPR3bAny:$Pg, srcRegType:$Rn),
   asm, "\t$Zd, $Pg/m, $Rn",
   "",
   []>, Sched<[]> {

llvm/test/CodeGen/AArch64/sve-intrinsics-reversal.ll

+; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s
+
+;
+; RBIT
+;
+
+define <vscale x 16 x i8> @rbit_i8(<vscale x 16 x i8> %a, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %b) {
+; CHECK-LABEL: rbit_i8:
+; CHECK: rbit z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> %a,
+                                                                <vscale x 16 x i1> %pg,
+                                                                <vscale x 16 x i8> %b)
+  ret <vscale x 16 x i8> %out
+}
+
+define <vscale x 8 x i16> @rbit_i16(<vscale x 8 x i16> %a, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %b) {
+; CHECK-LABEL: rbit_i16:
+; CHECK: rbit z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> %a,
+                                                                <vscale x 8 x i1> %pg,
+                                                                <vscale x 8 x i16> %b)
+  ret <vscale x 8 x i16> %out
+}
+
+define <vscale x 4 x i32> @rbit_i32(<vscale x 4 x i32> %a, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %b) {
+; CHECK-LABEL: rbit_i32:
+; CHECK: rbit z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> %a,
+                                                                <vscale x 4 x i1> %pg,
+                                                                <vscale x 4 x i32> %b)
+  ret <vscale x 4 x i32> %out
+}
+
+define <vscale x 2 x i64> @rbit_i64(<vscale x 2 x i64> %a, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %b) {
+; CHECK-LABEL: rbit_i64:
+; CHECK: rbit z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> %a,
+                                                                <vscale x 2 x i1> %pg,
+                                                                <vscale x 2 x i64> %b)
+  ret <vscale x 2 x i64> %out
+}
+
+;
+; REVB
+;
+
+define <vscale x 8 x i16> @revb_i16(<vscale x 8 x i16> %a, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %b) {
+; CHECK-LABEL: revb_i16:
+; CHECK: revb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> %a,
+                                                                <vscale x 8 x i1> %pg,
+                                                                <vscale x 8 x i16> %b)
+  ret <vscale x 8 x i16> %out
+}
+
+define <vscale x 4 x i32> @revb_i32(<vscale x 4 x i32> %a, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %b) {
+; CHECK-LABEL: revb_i32:
+; CHECK: revb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> %a,
+                                                                <vscale x 4 x i1> %pg,
+                                                                <vscale x 4 x i32> %b)
+  ret <vscale x 4 x i32> %out
+}
+
+define <vscale x 2 x i64> @revb_i64(<vscale x 2 x i64> %a, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %b) {
+; CHECK-LABEL: revb_i64:
+; CHECK: revb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> %a,
+                                                                <vscale x 2 x i1> %pg,
+                                                                <vscale x 2 x i64> %b)
+  ret <vscale x 2 x i64> %out
+}
+
+;
+; REVB (bswap)
+;
+
+define <vscale x 8 x i16> @revb_i16_bswap(<vscale x 8 x i16> %a) {
+; CHECK-LABEL: revb_i16_bswap:
+; CHECK: ptrue [[PG:p[0-9]+]].h
+; CHECK-NEXT: revb z0.h, [[PG]]/m, z0.h
+; CHECK-NEXT: ret
+  %res = call <vscale x 8 x i16> @llvm.bswap.nxv8i16(<vscale x 8 x i16> %a)
+  ret <vscale x 8 x i16> %res
+}
+
+define <vscale x 4 x i32> @revb_i32_bswap(<vscale x 4 x i32> %a) {
+; CHECK-LABEL: revb_i32_bswap:
+; CHECK: ptrue [[PG:p[0-9]+]].s
+; CHECK-NEXT: revb z0.s, [[PG]]/m, z0.s
+; CHECK-NEXT: ret
+  %res = call <vscale x 4 x i32> @llvm.bswap.nxv4i32(<vscale x 4 x i32> %a)
+  ret <vscale x 4 x i32> %res
+}
+
+define <vscale x 2 x i64> @revb_i64_bswap(<vscale x 2 x i64> %a) {
+; CHECK-LABEL: revb_i64_bswap:
+; CHECK: ptrue [[PG:p[0-9]+]].d
+; CHECK-NEXT: revb z0.d, [[PG]]/m, z0.d
+; CHECK-NEXT: ret
+  %res = call <vscale x 2 x i64> @llvm.bswap.nxv2i64(<vscale x 2 x i64> %a)
+  ret <vscale x 2 x i64> %res
+}
+
+;
+; REVH
+;
+
+define <vscale x 4 x i32> @revh_i32(<vscale x 4 x i32> %a, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %b) {
+; CHECK-LABEL: revh_i32:
+; CHECK: revh z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> %a,
+                                                                <vscale x 4 x i1> %pg,
+                                                                <vscale x 4 x i32> %b)
+  ret <vscale x 4 x i32> %out
+}
+
+define <vscale x 2 x i64> @revh_i64(<vscale x 2 x i64> %a, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %b) {
+; CHECK-LABEL: revh_i64:
+; CHECK: revh z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> %a,
+                                                                <vscale x 2 x i1> %pg,
+                                                                <vscale x 2 x i64> %b)
+  ret <vscale x 2 x i64> %out
+}
+
+;
+; REVW
+;
+
+define <vscale x 2 x i64> @revw_i64(<vscale x 2 x i64> %a, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %b) {
+; CHECK-LABEL: revw_i64:
+; CHECK: revw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> %a,
+                                                                <vscale x 2 x i1> %pg,
+                                                                <vscale x 2 x i64> %b)
+  ret <vscale x 2 x i64> %out
+}
+
+declare <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i1>, <vscale x 16 x i8>)
+declare <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i1>, <vscale x 8 x i16>)
+declare <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i1>, <vscale x 4 x i32>)
+declare <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i1>, <vscale x 2 x i64>)
+
+declare <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i1>, <vscale x 8 x i16>)
+declare <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i1>, <vscale x 4 x i32>)
+declare <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i1>, <vscale x 2 x i64>)
+
+declare <vscale x 8 x i16> @llvm.bswap.nxv8i16(<vscale x 8 x i16>)
+declare <vscale x 4 x i32> @llvm.bswap.nxv4i32(<vscale x 4 x i32>)
+declare <vscale x 2 x i64> @llvm.bswap.nxv2i64(<vscale x 2 x i64>)
+
+declare <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i1>, <vscale x 4 x i32>)
+declare <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i1>, <vscale x 2 x i64>)
+
+declare <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i1>, <vscale x 2 x i64>)