aarch64: Add support for FEAT_SME2p1 instructions.

Hi,

This patch add support for FEAT_SME2p1 and "movaz" instructions
along with the optional flag +sme2p1.

Following "movaz" instructions are add:
Move and zero two ZA tile slices to vector registers.
Move and zero four ZA tile slices to vector registers.

Regression testing for aarch64-none-elf target and found no regressions.

Ok for binutils-master?

Regards,
Srinath.

7 files changed, 274 insertions, 0 deletions

diff --git a/opcodes/aarch64-asm.c b/opcodes/aarch64-asm.c
index 1db290eea7e..3fac127a589 100644
--- a/opcodes/aarch64-asm.c
+++ b/opcodes/aarch64-asm.c
@@ -1375,6 +1375,76 @@ aarch64_ins_sve_float_zero_one (const aarch64_operand *self,
   return true;
 }
 
+bool
+aarch64_ins_sme_za_vrs1 (const aarch64_operand *self,
+			     const aarch64_opnd_info *info,
+			     aarch64_insn *code,
+			     const aarch64_inst *inst ATTRIBUTE_UNUSED,
+			     aarch64_operand_error *errors ATTRIBUTE_UNUSED)
+{
+  int za_reg = info->indexed_za.regno;
+  int regno = info->indexed_za.index.regno & 3;
+  int imm = info->indexed_za.index.imm;
+  int v =  info->indexed_za.v;
+  int countm1 = info->indexed_za.index.countm1;
+
+  insert_field (self->fields[0], code, v, 0);
+  insert_field (self->fields[1], code, regno, 0);
+  switch (info->qualifier)
+    {
+    case AARCH64_OPND_QLF_S_B:
+      insert_field (self->fields[2], code, imm / (countm1 + 1), 0);
+      break;
+    case AARCH64_OPND_QLF_S_H:
+    case AARCH64_OPND_QLF_S_S:
+      insert_field (self->fields[2], code, za_reg, 0);
+      insert_field (self->fields[3], code, imm / (countm1 + 1), 0);
+      break;
+    case AARCH64_OPND_QLF_S_D:
+      insert_field (self->fields[2], code, za_reg, 0);
+      break;
+    default:
+      return false;
+    }
+
+  return true;
+}
+
+bool
+aarch64_ins_sme_za_vrs2 (const aarch64_operand *self,
+			     const aarch64_opnd_info *info,
+			     aarch64_insn *code,
+			     const aarch64_inst *inst ATTRIBUTE_UNUSED,
+			     aarch64_operand_error *errors ATTRIBUTE_UNUSED)
+{
+  int za_reg = info->indexed_za.regno;
+  int regno = info->indexed_za.index.regno & 3;
+  int imm = info->indexed_za.index.imm;
+  int v =  info->indexed_za.v;
+  int countm1 = info->indexed_za.index.countm1;
+
+  insert_field (self->fields[0], code, v, 0);
+  insert_field (self->fields[1], code, regno, 0);
+  switch (info->qualifier)
+    {
+    case AARCH64_OPND_QLF_S_B:
+      insert_field (self->fields[2], code, imm / (countm1 + 1), 0);
+      break;
+    case AARCH64_OPND_QLF_S_H:
+      insert_field (self->fields[2], code, za_reg, 0);
+      insert_field (self->fields[3], code, imm / (countm1 + 1), 0);
+      break;
+    case AARCH64_OPND_QLF_S_S:
+    case AARCH64_OPND_QLF_S_D:
+      insert_field (self->fields[2], code, za_reg, 0);
+      break;
+    default:
+      return false;
+    }
+
+  return true;
+}
+
 /* Encode in SME instruction such as MOVA ZA tile vector register number,
    vector indicator, vector selector and immediate.  */
 bool
@@ -2011,6 +2081,7 @@ aarch64_encode_variant_using_iclass (struct aarch64_inst *inst)
       break;
 
     case sme_misc:
+    case sme2_movaz:
     case sve_misc:
       /* These instructions have only a single variant.  */
       break;
diff --git a/opcodes/aarch64-asm.h b/opcodes/aarch64-asm.h
index a3bf7bda013..d4b6407dc5d 100644
--- a/opcodes/aarch64-asm.h
+++ b/opcodes/aarch64-asm.h
@@ -100,6 +100,8 @@ AARCH64_DECL_OPD_INSERTER (ins_sve_strided_reglist);
 AARCH64_DECL_OPD_INSERTER (ins_sve_scale);
 AARCH64_DECL_OPD_INSERTER (ins_sve_shlimm);
 AARCH64_DECL_OPD_INSERTER (ins_sve_shrimm);
+AARCH64_DECL_OPD_INSERTER (ins_sme_za_vrs1);
+AARCH64_DECL_OPD_INSERTER (ins_sme_za_vrs2);
 AARCH64_DECL_OPD_INSERTER (ins_sme_za_hv_tiles);
 AARCH64_DECL_OPD_INSERTER (ins_sme_za_hv_tiles_range);
 AARCH64_DECL_OPD_INSERTER (ins_sme_za_list);
diff --git a/opcodes/aarch64-dis.c b/opcodes/aarch64-dis.c
index 7e088a93c10..a14b2ca02d2 100644
--- a/opcodes/aarch64-dis.c
+++ b/opcodes/aarch64-dis.c
@@ -1929,6 +1929,84 @@ aarch64_ext_sme_za_array (const aarch64_operand *self,
   return true;
 }
 
+/* Decode two ZA tile slice (V, Rv, off3| ZAn ,off2 | ZAn, ol| ZAn) feilds.  */
+bool
+aarch64_ext_sme_za_vrs1 (const aarch64_operand *self,
+			  aarch64_opnd_info *info, aarch64_insn code,
+			  const aarch64_inst *inst,
+			  aarch64_operand_error *errors ATTRIBUTE_UNUSED)
+{
+  int v = extract_field (self->fields[0], code, 0);
+  int regno = 12 + extract_field (self->fields[1], code, 0);
+  int imm, za_reg, num_offset = 2;
+
+  switch (info->qualifier)
+    {
+    case AARCH64_OPND_QLF_S_B:
+      imm = extract_field (self->fields[2], code, 0);
+      info->indexed_za.index.imm = imm * num_offset;
+      break;
+    case AARCH64_OPND_QLF_S_H:
+    case AARCH64_OPND_QLF_S_S:
+      za_reg = extract_field (self->fields[2], code, 0);
+      imm = extract_field (self->fields[3], code, 0);
+      info->indexed_za.index.imm = imm * num_offset;
+      info->indexed_za.regno = za_reg;
+      break;
+    case AARCH64_OPND_QLF_S_D:
+      za_reg = extract_field (self->fields[2], code, 0);
+      info->indexed_za.regno = za_reg;
+      break;
+    default:
+      return false;
+    }
+
+  info->indexed_za.index.regno = regno;
+  info->indexed_za.index.countm1 = num_offset - 1;
+  info->indexed_za.v = v;
+  info->indexed_za.group_size = get_opcode_dependent_value (inst->opcode);
+  return true;
+}
+
+/* Decode four ZA tile slice (V, Rv, off3| ZAn ,off2 | ZAn, ol| ZAn) feilds.  */
+bool
+aarch64_ext_sme_za_vrs2 (const aarch64_operand *self,
+			  aarch64_opnd_info *info, aarch64_insn code,
+			  const aarch64_inst *inst,
+			  aarch64_operand_error *errors ATTRIBUTE_UNUSED)
+{
+  int v = extract_field (self->fields[0], code, 0);
+  int regno = 12 + extract_field (self->fields[1], code, 0);
+  int imm, za_reg, num_offset =4;
+
+  switch (info->qualifier)
+    {
+    case AARCH64_OPND_QLF_S_B:
+      imm = extract_field (self->fields[2], code, 0);
+      info->indexed_za.index.imm = imm * num_offset;
+      break;
+    case AARCH64_OPND_QLF_S_H:
+      za_reg = extract_field (self->fields[2], code, 0);
+      imm = extract_field (self->fields[3], code, 0);
+      info->indexed_za.index.imm = imm * num_offset;
+      info->indexed_za.regno = za_reg;
+      break;
+    case AARCH64_OPND_QLF_S_S:
+    case AARCH64_OPND_QLF_S_D:
+      za_reg = extract_field (self->fields[2], code, 0);
+      info->indexed_za.regno = za_reg;
+      break;
+    default:
+      return false;
+    }
+
+  info->indexed_za.index.regno = regno;
+  info->indexed_za.index.countm1 = num_offset - 1;
+  info->indexed_za.v = v;
+  info->indexed_za.group_size = get_opcode_dependent_value (inst->opcode);
+  return true;
+}
+
 bool
 aarch64_ext_sme_addr_ri_u4xvl (const aarch64_operand *self,
                                aarch64_opnd_info *info, aarch64_insn code,
@@ -3160,6 +3238,7 @@ aarch64_decode_variant_using_iclass (aarch64_inst *inst)
       variant = 3;
       break;
 
+    case sme2_movaz:
     case sme_misc:
     case sve_misc:
       /* These instructions have only a single variant.  */
diff --git a/opcodes/aarch64-dis.h b/opcodes/aarch64-dis.h
index 20387db7b39..9a38c1ab50f 100644
--- a/opcodes/aarch64-dis.h
+++ b/opcodes/aarch64-dis.h
@@ -124,6 +124,8 @@ AARCH64_DECL_OPD_EXTRACTOR (ext_sve_strided_reglist);
 AARCH64_DECL_OPD_EXTRACTOR (ext_sve_scale);
 AARCH64_DECL_OPD_EXTRACTOR (ext_sve_shlimm);
 AARCH64_DECL_OPD_EXTRACTOR (ext_sve_shrimm);
+AARCH64_DECL_OPD_EXTRACTOR (ext_sme_za_vrs1);
+AARCH64_DECL_OPD_EXTRACTOR (ext_sme_za_vrs2);
 AARCH64_DECL_OPD_EXTRACTOR (ext_sme_za_hv_tiles);
 AARCH64_DECL_OPD_EXTRACTOR (ext_sme_za_hv_tiles_range);
 AARCH64_DECL_OPD_EXTRACTOR (ext_sme_za_list);
diff --git a/opcodes/aarch64-opc.c b/opcodes/aarch64-opc.c
index e3ad32f5a1e..cf76871930f 100644
--- a/opcodes/aarch64-opc.c
+++ b/opcodes/aarch64-opc.c
@@ -400,6 +400,16 @@ const aarch64_field fields[] =
     { 22,  1 }, /* sz: 1-bit element size select.  */
     { 22,  2 },	/* type: floating point type field in fp data inst.  */
     { 10,  2 },	/* vldst_size: size field in the AdvSIMD load/store inst.  */
+    {  5,  3 }, /* off3: immediate offset used to calculate slice number in a
+		   ZA tile.  */
+    {  5,  2 }, /* off2: immediate offset used to calculate slice number in
+		   a ZA tile.  */
+    {  7,  1 }, /* ZAn_1: name of the 1bit encoded ZA tile.  */
+    {  5,  1 }, /* ol: immediate offset used to calculate slice number in a ZA
+		   tile.  */
+    {  6,  2 }, /* ZAn_2: name of the 2bit encoded ZA tile.  */
+    {  5,  3 }, /* ZAn_3: name of the 3bit encoded ZA tile.  */
+    {  6,  1 }, /* ZAn: name of the bit encoded ZA tile.  */
 };
 
 enum aarch64_operand_class
@@ -1938,6 +1948,49 @@ operand_general_constraint_met_p (const aarch64_opnd_info *opnds, int idx,
 	    return 0;
 	  break;
 
+	case AARCH64_OPND_SME_ZA_array_vrsb_1:
+	  if (!check_za_access (opnd, mismatch_detail, idx, 12, 7, 2,
+				get_opcode_dependent_value (opcode)))
+	    return 0;
+	  break;
+
+	case AARCH64_OPND_SME_ZA_array_vrsh_1:
+	  if (!check_za_access (opnd, mismatch_detail, idx, 12, 3, 2,
+				get_opcode_dependent_value (opcode)))
+	    return 0;
+	  break;
+
+	case AARCH64_OPND_SME_ZA_array_vrss_1:
+	  if (!check_za_access (opnd, mismatch_detail, idx, 12, 1, 2,
+				get_opcode_dependent_value (opcode)))
+	    return 0;
+	  break;
+
+	case AARCH64_OPND_SME_ZA_array_vrsd_1:
+	  if (!check_za_access (opnd, mismatch_detail, idx, 12, 0, 2,
+				get_opcode_dependent_value (opcode)))
+	    return 0;
+	  break;
+
+	case AARCH64_OPND_SME_ZA_array_vrsb_2:
+	  if (!check_za_access (opnd, mismatch_detail, idx, 12, 3, 4,
+				get_opcode_dependent_value (opcode)))
+	    return 0;
+	  break;
+
+	case AARCH64_OPND_SME_ZA_array_vrsh_2:
+	  if (!check_za_access (opnd, mismatch_detail, idx, 12, 1, 4,
+				get_opcode_dependent_value (opcode)))
+	    return 0;
+	  break;
+
+	case AARCH64_OPND_SME_ZA_array_vrss_2:
+	case AARCH64_OPND_SME_ZA_array_vrsd_2:
+	  if (!check_za_access (opnd, mismatch_detail, idx, 12, 0, 4,
+				get_opcode_dependent_value (opcode)))
+	    return 0;
+	  break;
+
 	case AARCH64_OPND_SME_ZA_HV_idx_srcxN:
 	case AARCH64_OPND_SME_ZA_HV_idx_destxN:
 	  size = aarch64_get_qualifier_esize (opnd->qualifier);
@@ -4103,6 +4156,30 @@ aarch64_print_operand (char *buf, size_t size, bfd_vma pc,
 		? style_sub_mnem (styler, "vgx4") : "");
       break;
 
+    case AARCH64_OPND_SME_ZA_array_vrsb_1:
+    case AARCH64_OPND_SME_ZA_array_vrsh_1:
+    case AARCH64_OPND_SME_ZA_array_vrss_1:
+    case AARCH64_OPND_SME_ZA_array_vrsd_1:
+    case AARCH64_OPND_SME_ZA_array_vrsb_2:
+    case AARCH64_OPND_SME_ZA_array_vrsh_2:
+    case AARCH64_OPND_SME_ZA_array_vrss_2:
+    case AARCH64_OPND_SME_ZA_array_vrsd_2:
+      snprintf (buf, size, "%s [%s, %s%s%s]",
+		style_reg (styler, "za%d%c%s%s",
+			   opnd->indexed_za.regno,
+			   opnd->indexed_za.v ? 'v': 'h',
+			   opnd->qualifier == AARCH64_OPND_QLF_NIL ? "" : ".",
+			   (opnd->qualifier == AARCH64_OPND_QLF_NIL
+			    ? ""
+			    : aarch64_get_qualifier_name (opnd->qualifier))),
+		style_reg (styler, "w%d", opnd->indexed_za.index.regno),
+		style_imm (styler, "%" PRIi64, opnd->indexed_za.index.imm),
+		opnd->indexed_za.index.countm1 ? ":" : "",
+		opnd->indexed_za.index.countm1  ? style_imm (styler, "%d",
+		opnd->indexed_za.index.imm
+		+ opnd->indexed_za.index.countm1):"");
+      break;
+
     case AARCH64_OPND_SME_SM_ZA:
       snprintf (buf, size, "%s",
 		style_reg (styler, opnd->reg.regno == 's' ? "sm" : "za"));
diff --git a/opcodes/aarch64-opc.h b/opcodes/aarch64-opc.h
index f193a90ecc5..587775152e3 100644
--- a/opcodes/aarch64-opc.h
+++ b/opcodes/aarch64-opc.h
@@ -210,6 +210,13 @@ enum aarch64_field_kind
   FLD_sz,
   FLD_type,
   FLD_vldst_size,
+  FLD_off3,
+  FLD_off2,
+  FLD_ZAn_1,
+  FLD_ol,
+  FLD_ZAn_2,
+  FLD_ZAn_3,
+  FLD_ZAn
 };
 
 /* Field description.  */
diff --git a/opcodes/aarch64-tbl.h b/opcodes/aarch64-tbl.h
index a8ccdafd044..9c7648b0a6d 100644
--- a/opcodes/aarch64-tbl.h
+++ b/opcodes/aarch64-tbl.h
@@ -1497,6 +1497,10 @@
 {                                                       \
   QLF2(S_B,S_B),                                        \
 }
+#define OP_SVE_HH					\
+{							\
+  QLF2(S_H,S_H),					\
+}
 #define OP_SVE_BBU                                      \
 {                                                       \
   QLF3(S_B,S_B,NIL),                                \
@@ -2614,6 +2618,8 @@ static const aarch64_feature_set aarch64_feature_d128_the =
   AARCH64_FEATURES (2, D128, THE);
 static const aarch64_feature_set aarch64_feature_b16b16 =
   AARCH64_FEATURE (B16B16);
+static const aarch64_feature_set aarch64_feature_sme2p1 =
+  AARCH64_FEATURE (SME2p1);
 
 #define CORE		&aarch64_feature_v8
 #define FP		&aarch64_feature_fp
@@ -2677,6 +2683,7 @@ static const aarch64_feature_set aarch64_feature_b16b16 =
 #define THE	  &aarch64_feature_the
 #define D128_THE  &aarch64_feature_d128_the
 #define B16B16  &aarch64_feature_b16b16
+#define SME2p1  &aarch64_feature_sme2p1
 
 #define CORE_INSN(NAME,OPCODE,MASK,CLASS,OP,OPS,QUALS,FLAGS) \
   { NAME, OPCODE, MASK, CLASS, OP, CORE, OPS, QUALS, FLAGS, 0, 0, NULL }
@@ -2743,6 +2750,9 @@ static const aarch64_feature_set aarch64_feature_b16b16 =
 #define SVE2_INSN(NAME,OPCODE,MASK,CLASS,OP,OPS,QUALS,FLAGS,TIED) \
   { NAME, OPCODE, MASK, CLASS, OP, SVE2, OPS, QUALS, \
     FLAGS | F_STRICT, 0, TIED, NULL }
+#define SME2p1_INSN(NAME,OPCODE,MASK,CLASS,OP,OPS,QUALS,FLAGS,TIED) \
+  { NAME, OPCODE, MASK, CLASS, OP, SME2p1, OPS, QUALS, \
+    FLAGS | F_STRICT, 0, TIED, NULL }
 #define SVE2_INSNC(NAME,OPCODE,MASK,CLASS,OP,OPS,QUALS,FLAGS,CONSTRAINTS,TIED) \
   { NAME, OPCODE, MASK, CLASS, OP, SVE2, OPS, QUALS, \
     FLAGS | F_STRICT, CONSTRAINTS, TIED, NULL }
@@ -6289,6 +6299,16 @@ const struct aarch64_opcode aarch64_opcode_table[] =
   B16B16_INSN("bfmls", 0x64200c00, 0xffa0fc00, sve_misc, 0, OP3 (SVE_Zd, SVE_Zn, SVE_Zm3_11_INDEX), OP_SVE_VVV_H, 0, 0),
   B16B16_INSN("bfmul", 0x64202800, 0xffa0fc00, sve_misc, 0, OP3 (SVE_Zd, SVE_Zn, SVE_Zm3_11_INDEX), OP_SVE_VVV_H, 0, 0),
 
+/* SME2.1 movaz instructions.  */
+  SME2p1_INSN ("movaz", 0xc0060600, 0xffff1f83, sme2_movaz, 0, OP2 (SME_Zdnx4, SME_ZA_array_vrsb_2), OP_SVE_BB, 0, 0),
+  SME2p1_INSN ("movaz", 0xc0460600, 0xffff1f83, sme2_movaz, 0, OP2 (SME_Zdnx4, SME_ZA_array_vrsh_2), OP_SVE_HH, 0, 0),
+  SME2p1_INSN ("movaz", 0xc0860600, 0xffff1f83, sme2_movaz, 0, OP2 (SME_Zdnx4, SME_ZA_array_vrss_2), OP_SVE_SS, 0, 0),
+  SME2p1_INSN ("movaz", 0xc0c60600, 0xffff1f03, sme2_movaz, 0, OP2 (SME_Zdnx4, SME_ZA_array_vrsd_2), OP_SVE_DD, 0, 0),
+
+  SME2p1_INSN ("movaz", 0xc0060200, 0xffff1f01, sme2_movaz, 0, OP2 (SME_Zdnx2, SME_ZA_array_vrsb_1), OP_SVE_BB, 0, 0),
+  SME2p1_INSN ("movaz", 0xc0460200, 0xffff1f01, sme2_movaz, 0, OP2 (SME_Zdnx2, SME_ZA_array_vrsh_1), OP_SVE_HH, 0, 0),
+  SME2p1_INSN ("movaz", 0xc0860200, 0xffff1f01, sme2_movaz, 0, OP2 (SME_Zdnx2, SME_ZA_array_vrss_1), OP_SVE_SS, 0, 0),
+  SME2p1_INSN ("movaz", 0xc0c60200, 0xffff1f01, sme2_movaz, 0, OP2 (SME_Zdnx2, SME_ZA_array_vrsd_1), OP_SVE_DD, 0, 0),
   {0, 0, 0, 0, 0, 0, {}, {}, 0, 0, 0, NULL},
 };
 
@@ -6726,6 +6746,22 @@ const struct aarch64_opcode aarch64_opcode_table[] =
     Y(SIMD_REG, regno, "SVE_Vd", 0, F(FLD_SVE_Vd), "a SIMD register")	\
     Y(SIMD_REG, regno, "SVE_Vm", 0, F(FLD_SVE_Vm), "a SIMD register")	\
     Y(SIMD_REG, regno, "SVE_Vn", 0, F(FLD_SVE_Vn), "a SIMD register")	\
+    Y(ZA_ACCESS, sme_za_vrs1, "SME_ZA_array_vrsb_1", 0,			\
+      F(FLD_SME_V,FLD_SME_Rv,FLD_off3), "ZA0 tile")			\
+    Y(ZA_ACCESS, sme_za_vrs1, "SME_ZA_array_vrsh_1", 0,			\
+      F(FLD_SME_V,FLD_SME_Rv,FLD_ZAn_1,FLD_off2), "1 bit ZA tile")	\
+    Y(ZA_ACCESS, sme_za_vrs1, "SME_ZA_array_vrss_1", 0,			\
+      F(FLD_SME_V,FLD_SME_Rv,FLD_ZAn_2,FLD_ol), "2 ZA tile")		\
+    Y(ZA_ACCESS, sme_za_vrs1, "SME_ZA_array_vrsd_1", 0,			\
+      F(FLD_SME_V,FLD_SME_Rv,FLD_ZAn_3), "3 ZA tile")			\
+    Y(ZA_ACCESS, sme_za_vrs2, "SME_ZA_array_vrsb_2", 0,			\
+      F(FLD_SME_V,FLD_SME_Rv,FLD_off2), "ZA0 tile")			\
+    Y(ZA_ACCESS, sme_za_vrs2, "SME_ZA_array_vrsh_2", 0,			\
+      F(FLD_SME_V,FLD_SME_Rv,FLD_ZAn,FLD_ol), "1 bit ZA tile")		\
+    Y(ZA_ACCESS, sme_za_vrs2, "SME_ZA_array_vrss_2", 0,			\
+      F(FLD_SME_V,FLD_SME_Rv,FLD_off2), "2 bit ZA tile")		\
+    Y(ZA_ACCESS, sme_za_vrs2, "SME_ZA_array_vrsd_2", 0,			\
+      F(FLD_SME_V,FLD_SME_Rv,FLD_ZAn_3), "3 bit ZA tile")		\
     Y(SVE_REG, regno, "SVE_Za_5", 0, F(FLD_SVE_Za_5),			\
       "an SVE vector register")						\
     Y(SVE_REG, regno, "SVE_Za_16", 0, F(FLD_SVE_Za_16),			\