From: Srinath Parvathaneni <srinath.parvathaneni@arm.com>
To: binutils@sourceware.org
Cc: richard.earnshaw@arm.com, nickc@redhat.com
Subject: [PATCH 4/6][Binutils] aarch64: Add SVE2.1 dupq, eorqv and extq instructions.
Date: Mon, 15 Jan 2024 09:37:32 +0000 [thread overview]
Message-ID: <8e052b41-815e-4c17-8e9b-631a7085be45@arm.com> (raw)
In-Reply-To: <73155200-f7c2-4226-b4be-4a320ea82044@arm.com>
[-- Attachment #1: Type: text/plain, Size: 190 bytes --]
Hi,
This patch add support for SVE2.1 instruction dupq, eorqv and extq.
Regression testing for aarch64-none-elf target and found no regressions.
Ok for binutils-master?
Regards,
Srinath.
[-- Attachment #2: 4_6.patch --]
[-- Type: text/x-patch, Size: 13581 bytes --]
diff --git a/gas/config/tc-aarch64.c b/gas/config/tc-aarch64.c
index 04dd08a6fa71b84b3e71e0ab422fb6deb9fedb38..0665732fe03cc59df4ebd36ee1afbad08c22b72e 100644
--- a/gas/config/tc-aarch64.c
+++ b/gas/config/tc-aarch64.c
@@ -6698,6 +6698,8 @@ parse_operands (char *str, const aarch64_opcode *opcode)
case AARCH64_OPND_SVE_Zm4_11_INDEX:
case AARCH64_OPND_SVE_Zm4_INDEX:
case AARCH64_OPND_SVE_Zn_INDEX:
+ case AARCH64_OPND_SVE_Zm_imm4:
+ case AARCH64_OPND_SVE_Zn_5_INDEX:
case AARCH64_OPND_SME_Zm_INDEX1:
case AARCH64_OPND_SME_Zm_INDEX2:
case AARCH64_OPND_SME_Zm_INDEX3_1:
diff --git a/gas/testsuite/gas/aarch64/sve2p1-1-bad.l b/gas/testsuite/gas/aarch64/sve2p1-1-bad.l
index 6b07eee9e94d93a9e8d6357a741d2d6ef90601e0..f5a80d26768882f2b2e16840ad587612d34ae15e 100644
--- a/gas/testsuite/gas/aarch64/sve2p1-1-bad.l
+++ b/gas/testsuite/gas/aarch64/sve2p1-1-bad.l
@@ -35,3 +35,23 @@
.*: Error: selected processor does not support `uminqv v4.2d,p3,z2.d'
.*: Error: selected processor does not support `uminqv v8.2d,p4,z1.d'
.*: Error: selected processor does not support `uminqv v16.4s,p7,z0.s'
+.*: Error: selected processor does not support `dupq z10.b,z20.b\[0\]'
+.*: Error: selected processor does not support `dupq z10.b,z20.b\[15\]'
+.*: Error: selected processor does not support `dupq z10.h,z20.h\[0\]'
+.*: Error: selected processor does not support `dupq z10.h,z20.h\[7\]'
+.*: Error: selected processor does not support `dupq z10.s,z20.s\[0\]'
+.*: Error: selected processor does not support `dupq z10.s,z20.s\[3\]'
+.*: Error: selected processor does not support `dupq z10.d,z20.d\[0\]'
+.*: Error: selected processor does not support `dupq z10.d,z20.d\[1\]'
+.*: Error: selected processor does not support `eorqv v0.16b,p0,z16.b'
+.*: Error: selected processor does not support `eorqv v1.8h,p1,z8.h'
+.*: Error: selected processor does not support `eorqv v2.4s,p2,z4.s'
+.*: Error: selected processor does not support `eorqv v4.2d,p3,z2.d'
+.*: Error: selected processor does not support `eorqv v8.2d,p4,z1.d'
+.*: Error: selected processor does not support `eorqv v16.4s,p7,z0.s'
+.*: Error: selected processor does not support `extq z0.b,z0.b,z10.b\[15\]'
+.*: Error: selected processor does not support `extq z1.b,z1.b,z15.b\[7\]'
+.*: Error: selected processor does not support `extq z2.b,z2.b,z5.b\[3\]'
+.*: Error: selected processor does not support `extq z4.b,z4.b,z12.b\[1\]'
+.*: Error: selected processor does not support `extq z8.b,z8.b,z7.b\[4\]'
+.*: Error: selected processor does not support `extq z16.b,z16.b,z1.b\[8\]'
diff --git a/gas/testsuite/gas/aarch64/sve2p1-1.d b/gas/testsuite/gas/aarch64/sve2p1-1.d
index d3d14f3c455aa352d31e01195196e03397ed4271..6d718aec7cad5511bda282865d0667a6cdaa188d 100644
--- a/gas/testsuite/gas/aarch64/sve2p1-1.d
+++ b/gas/testsuite/gas/aarch64/sve2p1-1.d
@@ -44,3 +44,23 @@
.*: 04cf2c44 uminqv v4.2d, p3, z2.d
.*: 04cf3028 uminqv v8.2d, p4, z1.d
.*: 048f3c10 uminqv v16.4s, p7, z0.s
+.*: 0530268a dupq z10.b, z20.b\[0\]
+.*: 053f268a dupq z10.b, z20.b\[15\]
+.*: 0521268a dupq z10.h, z20.h\[0\]
+.*: 052f268a dupq z10.h, z20.h\[7\]
+.*: 0522268a dupq z10.s, z20.s\[0\]
+.*: 052e268a dupq z10.s, z20.s\[3\]
+.*: 0524268a dupq z10.d, z20.d\[0\]
+.*: 052c268a dupq z10.d, z20.d\[1\]
+.*: 041d2200 eorqv v0.16b, p0, z16.b
+.*: 045d2501 eorqv v1.8h, p1, z8.h
+.*: 049d2882 eorqv v2.4s, p2, z4.s
+.*: 04dd2c44 eorqv v4.2d, p3, z2.d
+.*: 04dd3028 eorqv v8.2d, p4, z1.d
+.*: 049d3c10 eorqv v16.4s, p7, z0.s
+.*: 056a27c0 extq z0.b, z0.b, z10.b\[15\]
+.*: 056f25c1 extq z1.b, z1.b, z15.b\[7\]
+.*: 056524c2 extq z2.b, z2.b, z5.b\[3\]
+.*: 056c2444 extq z4.b, z4.b, z12.b\[1\]
+.*: 05672508 extq z8.b, z8.b, z7.b\[4\]
+.*: 05612610 extq z16.b, z16.b, z1.b\[8\]
diff --git a/gas/testsuite/gas/aarch64/sve2p1-1.s b/gas/testsuite/gas/aarch64/sve2p1-1.s
index c56ebf856ab5815efa01e06f40d04360f8afc7bc..5278dcf5e67b4cb34ab45b2b2725ab3af14c2594 100644
--- a/gas/testsuite/gas/aarch64/sve2p1-1.s
+++ b/gas/testsuite/gas/aarch64/sve2p1-1.s
@@ -39,3 +39,25 @@ uminqv v2.4s, p2, z4.s
uminqv v4.2d, p3, z2.d
uminqv v8.2d, p4, z1.d
uminqv v16.4s, p7, z0.s
+dupq z10.b, z20.b[0]
+dupq z10.b, z20.b[15]
+dupq z10.h, z20.h[0]
+dupq z10.h, z20.h[7]
+dupq z10.s, z20.s[0]
+dupq z10.s, z20.s[3]
+dupq z10.d, z20.d[0]
+dupq z10.d, z20.d[1]
+
+eorqv v0.16b, p0, z16.b
+eorqv v1.8h, p1, z8.h
+eorqv v2.4s, p2, z4.s
+eorqv v4.2d, p3, z2.d
+eorqv v8.2d, p4, z1.d
+eorqv v16.4s, p7, z0.s
+
+extq z0.b, z0.b, z10.b[15]
+extq z1.b, z1.b, z15.b[7]
+extq z2.b, z2.b, z5.b[3]
+extq z4.b, z4.b, z12.b[1]
+extq z8.b, z8.b, z7.b[4]
+extq z16.b, z16.b, z1.b[8]
diff --git a/include/opcode/aarch64.h b/include/opcode/aarch64.h
index 1af49c406e06e79ba81a1f01887d43da37d8a625..de161db75d509b0ac96c604da7bc9743193d23b2 100644
--- a/include/opcode/aarch64.h
+++ b/include/opcode/aarch64.h
@@ -727,8 +727,10 @@ enum aarch64_opnd
AARCH64_OPND_SVE_Zm3_19_INDEX, /* z0-z7[0-3] in Zm3_INDEX plus bit 19. */
AARCH64_OPND_SVE_Zm3_22_INDEX, /* z0-z7[0-7] in Zm3_INDEX plus bit 22. */
AARCH64_OPND_SVE_Zm4_11_INDEX, /* z0-z15[0-3] in Zm plus bit 11. */
+ AARCH64_OPND_SVE_Zm_imm4, /* SVE vector register with 4bit index. */
AARCH64_OPND_SVE_Zm4_INDEX, /* z0-z15[0-1] in Zm, bits [20,16]. */
AARCH64_OPND_SVE_Zn, /* SVE vector register in Zn. */
+ AARCH64_OPND_SVE_Zn_5_INDEX, /* Indexed SVE vector register, for DUPQ. */
AARCH64_OPND_SVE_Zn_INDEX, /* Indexed SVE vector register, for DUP. */
AARCH64_OPND_SVE_ZnxN, /* SVE vector register list in Zn. */
AARCH64_OPND_SVE_Zt, /* SVE vector register in Zt. */
@@ -1002,7 +1004,8 @@ enum aarch64_insn_class
cssc,
gcs,
the,
- sve2_urqvs
+ sve2_urqvs,
+ sve_index1,
};
/* Opcode enumerators. */
diff --git a/opcodes/aarch64-asm.h b/opcodes/aarch64-asm.h
index d4b6407dc5de8d6e103ee8ca5b5f2c6bb814647f..e48bf0db8a86149155e325923f6644bb90410ccb 100644
--- a/opcodes/aarch64-asm.h
+++ b/opcodes/aarch64-asm.h
@@ -93,6 +93,7 @@ AARCH64_DECL_OPD_INSERTER (ins_sve_float_half_one);
AARCH64_DECL_OPD_INSERTER (ins_sve_float_half_two);
AARCH64_DECL_OPD_INSERTER (ins_sve_float_zero_one);
AARCH64_DECL_OPD_INSERTER (ins_sve_index);
+AARCH64_DECL_OPD_INSERTER (ins_sve_index_imm);
AARCH64_DECL_OPD_INSERTER (ins_sve_limm_mov);
AARCH64_DECL_OPD_INSERTER (ins_sve_quad_index);
AARCH64_DECL_OPD_INSERTER (ins_sve_reglist);
diff --git a/opcodes/aarch64-asm.c b/opcodes/aarch64-asm.c
index 1dfd59df42dbbe5640ece7c83f43f027a8329d06..0de09f0435abc3b7761707b2c81c58f5f3b1a10e 100644
--- a/opcodes/aarch64-asm.c
+++ b/opcodes/aarch64-asm.c
@@ -1220,6 +1220,21 @@ aarch64_ins_sve_index (const aarch64_operand *self,
return true;
}
+/* Encode Zn.<T>[<imm>], where <imm> is an immediate with range of 0 to one less
+ than the number of elements in 128 bit, which can encode il:tsz. */
+bool
+aarch64_ins_sve_index_imm (const aarch64_operand *self,
+ const aarch64_opnd_info *info, aarch64_insn *code,
+ const aarch64_inst *inst ATTRIBUTE_UNUSED,
+ aarch64_operand_error *errors ATTRIBUTE_UNUSED)
+{
+ insert_field (self->fields[0], code, info->reglane.regno, 0);
+ unsigned int esize = aarch64_get_qualifier_esize (info->qualifier);
+ insert_fields (code, (info->reglane.index * 2 + 1) * esize, 0,
+ 2, self->fields[1],self->fields[2]);
+ return true;
+}
+
/* Encode a logical/bitmask immediate for the MOV alias of SVE DUPM. */
bool
aarch64_ins_sve_limm_mov (const aarch64_operand *self,
@@ -2079,6 +2094,7 @@ aarch64_encode_variant_using_iclass (struct aarch64_inst *inst)
case sme_shift:
case sve_index:
+ case sve_index1:
case sve_shift_pred:
case sve_shift_unpred:
case sve_shift_tsz_hsd:
diff --git a/opcodes/aarch64-dis.h b/opcodes/aarch64-dis.h
index 9a38c1ab50f7fdb27588c7451ade19c166e69c96..30212f2ae2c2759b5667e5a007912d22c4a702fc 100644
--- a/opcodes/aarch64-dis.h
+++ b/opcodes/aarch64-dis.h
@@ -117,6 +117,7 @@ AARCH64_DECL_OPD_EXTRACTOR (ext_sve_float_half_one);
AARCH64_DECL_OPD_EXTRACTOR (ext_sve_float_half_two);
AARCH64_DECL_OPD_EXTRACTOR (ext_sve_float_zero_one);
AARCH64_DECL_OPD_EXTRACTOR (ext_sve_index);
+AARCH64_DECL_OPD_EXTRACTOR (ext_sve_index_imm);
AARCH64_DECL_OPD_EXTRACTOR (ext_sve_limm_mov);
AARCH64_DECL_OPD_EXTRACTOR (ext_sve_quad_index);
AARCH64_DECL_OPD_EXTRACTOR (ext_sve_reglist);
diff --git a/opcodes/aarch64-dis.c b/opcodes/aarch64-dis.c
index d395438966f16d1fc0fa7117a434cff50901f96e..bffa760004a3ede5c14287ee4db09d6db371bc87 100644
--- a/opcodes/aarch64-dis.c
+++ b/opcodes/aarch64-dis.c
@@ -2097,6 +2097,26 @@ aarch64_ext_sve_index (const aarch64_operand *self,
return true;
}
+/* Decode Zn.<T>[<imm>], where <imm> is an immediate with range of 0 to one less
+ than the number of elements in 128 bit, which can encode il:tsz. */
+bool
+aarch64_ext_sve_index_imm (const aarch64_operand *self,
+ aarch64_opnd_info *info, aarch64_insn code,
+ const aarch64_inst *inst ATTRIBUTE_UNUSED,
+ aarch64_operand_error *errors ATTRIBUTE_UNUSED)
+{
+ int val;
+
+ info->reglane.regno = extract_field (self->fields[0], code, 0);
+ val = extract_fields (code, 0, 2, self->fields[2], self->fields[1]);
+ if ((val & 15) == 0)
+ return 0;
+ while ((val & 1) == 0)
+ val /= 2;
+ info->reglane.index = val / 2;
+ return true;
+}
+
/* Decode a logical immediate for the MOV alias of SVE DUPM. */
bool
aarch64_ext_sve_limm_mov (const aarch64_operand *self,
@@ -3231,6 +3251,17 @@ aarch64_decode_variant_using_iclass (aarch64_inst *inst)
}
break;
+ case sve_index1:
+ i = extract_fields (inst->value, 0, 2, FLD_SVE_tsz, FLD_SVE_i2h);
+ if ((i & 15) == 0)
+ return false;
+ while ((i & 1) == 0)
+ {
+ i >>= 1;
+ variant += 1;
+ }
+ break;
+
case sve_limm:
/* Pick the smallest applicable element size. */
if ((inst->value & 0x20600) == 0x600)
diff --git a/opcodes/aarch64-opc.c b/opcodes/aarch64-opc.c
index cf76871930f9f4e8613a977efb81464dce3d8ba7..1d8ed26c7090e4b73489b15e74a911e33b54555c 100644
--- a/opcodes/aarch64-opc.c
+++ b/opcodes/aarch64-opc.c
@@ -1794,6 +1794,18 @@ operand_general_constraint_met_p (const aarch64_opnd_info *opnds, int idx,
return 0;
break;
+ case AARCH64_OPND_SVE_Zm_imm4:
+ if (!check_reglane (opnd, mismatch_detail, idx, "z", 0, 31, 0, 15))
+ return 0;
+ break;
+
+ case AARCH64_OPND_SVE_Zn_5_INDEX:
+ size = aarch64_get_qualifier_esize (opnd->qualifier);
+ if (!check_reglane (opnd, mismatch_detail, idx, "z", 0, 31,
+ 0, 16 / size - 1))
+ return 0;
+ break;
+
case AARCH64_OPND_SME_PNn3_INDEX1:
case AARCH64_OPND_SME_PNn3_INDEX2:
size = get_operand_field_width (get_operand_from_code (type), 1);
@@ -4074,6 +4086,7 @@ aarch64_print_operand (char *buf, size_t size, bfd_vma pc,
case AARCH64_OPND_SME_Zm_INDEX3_1:
case AARCH64_OPND_SME_Zm_INDEX3_2:
case AARCH64_OPND_SME_Zm_INDEX3_10:
+ case AARCH64_OPND_SVE_Zn_5_INDEX:
case AARCH64_OPND_SME_Zm_INDEX4_1:
case AARCH64_OPND_SME_Zm_INDEX4_10:
case AARCH64_OPND_SME_Zn_INDEX1_16:
@@ -4082,6 +4095,7 @@ aarch64_print_operand (char *buf, size_t size, bfd_vma pc,
case AARCH64_OPND_SME_Zn_INDEX3_14:
case AARCH64_OPND_SME_Zn_INDEX3_15:
case AARCH64_OPND_SME_Zn_INDEX4_14:
+ case AARCH64_OPND_SVE_Zm_imm4:
snprintf (buf, size, "%s[%s]",
(opnd->qualifier == AARCH64_OPND_QLF_NIL
? style_reg (styler, "z%d", opnd->reglane.regno)
diff --git a/opcodes/aarch64-tbl.h b/opcodes/aarch64-tbl.h
index f433257634e72b6afb64d58a1f0f052164291033..07f4eb319e9be1a8150224b59aba1ab831e51b29 100644
--- a/opcodes/aarch64-tbl.h
+++ b/opcodes/aarch64-tbl.h
@@ -6337,6 +6337,10 @@ const struct aarch64_opcode aarch64_opcode_table[] =
SVE2p1_INSNC("sminqv",0x040e2000, 0xff3fe000, sve2_urqvs, 0, OP3 (Vd, SVE_Pg3, SVE_Zn), OP_SVE_vUS_BHSD_BHSD, F_OPD_SIZE, C_SCAN_MOVPRFX, 0),
SVE2p1_INSNC("umaxqv",0x040d2000, 0xff3fe000, sve2_urqvs, 0, OP3 (Vd, SVE_Pg3, SVE_Zn), OP_SVE_vUS_BHSD_BHSD, F_OPD_SIZE, C_SCAN_MOVPRFX, 0),
SVE2p1_INSNC("uminqv",0x040f2000, 0xff3fe000, sve2_urqvs, 0, OP3 (Vd, SVE_Pg3, SVE_Zn), OP_SVE_vUS_BHSD_BHSD, F_OPD_SIZE, C_SCAN_MOVPRFX, 0),
+ SVE2p1_INSNC("eorqv",0x041d2000, 0xff3fe000, sve2_urqvs, 0, OP3 (Vd, SVE_Pg3, SVE_Zn), OP_SVE_vUS_BHSD_BHSD, F_OPD_SIZE, C_SCAN_MOVPRFX, 0),
+
+ SVE2p1_INSN("dupq",0x05202400, 0xffe0fc00, sve_index1, 0, OP2 (SVE_Zd, SVE_Zn_5_INDEX), OP_SVE_VV_BHSD, 0, 0),
+ SVE2p1_INSN("extq",0x05602400, 0xfff0fc00, sve_misc, 0, OP3 (SVE_Zd, SVE_Zd, SVE_Zm_imm4), OP_SVE_BBB, 0, 0),
{0, 0, 0, 0, 0, 0, {}, {}, 0, 0, 0, NULL},
};
@@ -6816,11 +6820,17 @@ const struct aarch64_opcode aarch64_opcode_table[] =
Y(SVE_REG, sve_quad_index, "SVE_Zm4_11_INDEX", \
4 << OPD_F_OD_LSB, F(FLD_SVE_i2h, FLD_SVE_i3l, FLD_SVE_imm4), \
"an indexed SVE vector register") \
+ Y(SVE_REG, sve_quad_index, "SVE_Zm_imm4", \
+ 5 << OPD_F_OD_LSB, F(FLD_SVE_Zm_5, FLD_SVE_imm4), \
+ "an 4bit indexed SVE vector register") \
Y(SVE_REG, sve_quad_index, "SVE_Zm4_INDEX", \
4 << OPD_F_OD_LSB, F(FLD_SVE_Zm_16), \
"an indexed SVE vector register") \
Y(SVE_REG, regno, "SVE_Zn", 0, F(FLD_SVE_Zn), \
"an SVE vector register") \
+ Y(SVE_REG, sve_index_imm, "SVE_Zn_5_INDEX", 0, \
+ F(FLD_SVE_Zn, FLD_SVE_i2h, FLD_SVE_tsz), \
+ "a 5 bit idexed SVE vector register") \
Y(SVE_REG, sve_index, "SVE_Zn_INDEX", 0, F(FLD_SVE_Zn), \
"an indexed SVE vector register") \
Y(SVE_REGLIST, sve_reglist, "SVE_ZnxN", 0, F(FLD_SVE_Zn), \
next prev parent reply other threads:[~2024-01-15 9:37 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-01-15 9:28 [PATCH 1/6] [Binutils] aarch64: Add support for FEAT_B16B16 instructions Srinath Parvathaneni
2024-01-15 9:34 ` [PATCH 2/6][Binutils] aarch64: Add support for FEAT_SME2p1 instructions Srinath Parvathaneni
2024-01-15 9:35 ` [PATCH 3/6][Binutils] aarch64: Add support for FEAT_SVE2p1 Srinath Parvathaneni
2024-01-15 9:37 ` Srinath Parvathaneni [this message]
2024-01-15 9:38 ` PATCH 5/6][Binutils] aarch64: Add SVE2.1 fmin and fmax instructions Srinath Parvathaneni
2024-01-15 9:40 ` [PATCH 6/6][Binutils] aarch64: Add SVE2.1 Contiguous load/store instructions Srinath Parvathaneni
2024-01-15 11:46 ` [PATCH 1/6] [Binutils] aarch64: Add support for FEAT_B16B16 instructions Nick Clifton
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=8e052b41-815e-4c17-8e9b-631a7085be45@arm.com \
--to=srinath.parvathaneni@arm.com \
--cc=binutils@sourceware.org \
--cc=nickc@redhat.com \
--cc=richard.earnshaw@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).