From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from foss.arm.com (foss.arm.com [217.140.110.172]) by sourceware.org (Postfix) with ESMTP id B53DA3858C78 for ; Thu, 30 Mar 2023 10:24:17 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org B53DA3858C78 Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=arm.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=arm.com Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.121.207.14]) by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id DA82212FC; Thu, 30 Mar 2023 03:25:01 -0700 (PDT) Received: from e121540-lin.manchester.arm.com (e121540-lin.manchester.arm.com [10.32.110.72]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id 01F143F663; Thu, 30 Mar 2023 03:24:16 -0700 (PDT) From: Richard Sandiford To: binutils@sourceware.org Cc: Richard Sandiford Subject: [PATCH 04/43] aarch64: Make SME instructions use F_STRICT Date: Thu, 30 Mar 2023 11:23:20 +0100 Message-Id: <20230330102359.3327695-5-richard.sandiford@arm.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20230330102359.3327695-1-richard.sandiford@arm.com> References: <20230330102359.3327695-1-richard.sandiford@arm.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Spam-Status: No, score=-33.4 required=5.0 tests=BAYES_00,GIT_PATCH_0,KAM_DMARC_NONE,KAM_DMARC_STATUS,KAM_LAZY_DOMAIN_SECURITY,SPF_HELO_NONE,SPF_NONE,TXREP autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org List-Id: This patch makes all SME instructions use F_STRICT, so that qualifiers have to be provided explicitly rather than being inferred from other operands. The main change is to move the qualifier setting from the operand-level decoders to the opcode level. This is one step towards consolidating the ZA parsing code and extending it to handle SME2. --- include/opcode/aarch64.h | 2 + opcodes/aarch64-asm.c | 8 +++- opcodes/aarch64-dis.c | 93 ++++++++++++++++++++-------------------- opcodes/aarch64-tbl.h | 18 ++++---- 4 files changed, 64 insertions(+), 57 deletions(-) diff --git a/include/opcode/aarch64.h b/include/opcode/aarch64.h index d0a0b629d99..691247aa934 100644 --- a/include/opcode/aarch64.h +++ b/include/opcode/aarch64.h @@ -659,7 +659,9 @@ enum aarch64_insn_class pcreladdr, ic_system, sme_misc, + sme_mov, sme_ldr, + sme_psel, sme_str, sme_start, sme_stop, diff --git a/opcodes/aarch64-asm.c b/opcodes/aarch64-asm.c index e050197e8be..bfabcb9e3a2 100644 --- a/opcodes/aarch64-asm.c +++ b/opcodes/aarch64-asm.c @@ -1852,6 +1852,11 @@ aarch64_encode_variant_using_iclass (struct aarch64_inst *inst) int variant = 0; switch (inst->opcode->iclass) { + case sme_mov: + case sme_psel: + /* The variant is encoded as part of the immediate. */ + break; + case sve_cpy: insert_fields (&inst->value, aarch64_get_variant (inst), 0, 2, FLD_SVE_M_14, FLD_size); @@ -1872,8 +1877,9 @@ aarch64_encode_variant_using_iclass (struct aarch64_inst *inst) encoding. */ break; + case sme_misc: case sve_misc: - /* sve_misc instructions have only a single variant. */ + /* These instructions have only a single variant. */ break; case sve_movprfx: diff --git a/opcodes/aarch64-dis.c b/opcodes/aarch64-dis.c index 02ce8345979..01881ea377d 100644 --- a/opcodes/aarch64-dis.c +++ b/opcodes/aarch64-dis.c @@ -1785,44 +1785,35 @@ aarch64_ext_sme_za_hv_tiles (const aarch64_operand *self, /* Deduce qualifier encoded in size and Q fields. */ if (fld_size == 0) - info->qualifier = AARCH64_OPND_QLF_S_B; - else if (fld_size == 1) - info->qualifier = AARCH64_OPND_QLF_S_H; - else if (fld_size == 2) - info->qualifier = AARCH64_OPND_QLF_S_S; - else if (fld_size == 3 && fld_q == 0) - info->qualifier = AARCH64_OPND_QLF_S_D; - else if (fld_size == 3 && fld_q == 1) - info->qualifier = AARCH64_OPND_QLF_S_Q; - - info->za_tile_vector.index.regno = fld_rv + 12; - info->za_tile_vector.v = fld_v; - - switch (info->qualifier) { - case AARCH64_OPND_QLF_S_B: info->za_tile_vector.regno = 0; info->za_tile_vector.index.imm = fld_zan_imm; - break; - case AARCH64_OPND_QLF_S_H: + } + else if (fld_size == 1) + { info->za_tile_vector.regno = fld_zan_imm >> 3; info->za_tile_vector.index.imm = fld_zan_imm & 0x07; - break; - case AARCH64_OPND_QLF_S_S: + } + else if (fld_size == 2) + { info->za_tile_vector.regno = fld_zan_imm >> 2; info->za_tile_vector.index.imm = fld_zan_imm & 0x03; - break; - case AARCH64_OPND_QLF_S_D: + } + else if (fld_size == 3 && fld_q == 0) + { info->za_tile_vector.regno = fld_zan_imm >> 1; info->za_tile_vector.index.imm = fld_zan_imm & 0x01; - break; - case AARCH64_OPND_QLF_S_Q: + } + else if (fld_size == 3 && fld_q == 1) + { info->za_tile_vector.regno = fld_zan_imm; info->za_tile_vector.index.imm = 0; - break; - default: - return false; } + else + return false; + + info->za_tile_vector.index.regno = fld_rv + 12; + info->za_tile_vector.v = fld_v; return true; } @@ -1914,26 +1905,14 @@ aarch64_ext_sme_pred_reg_with_index (const aarch64_operand *self, info->za_tile_vector.regno = fld_pn; info->za_tile_vector.index.regno = fld_rm + 12; - if (fld_tszh == 0x1 && fld_tszl == 0x0) - { - info->qualifier = AARCH64_OPND_QLF_S_D; - imm = fld_i1; - } - else if (fld_tszl == 0x4) - { - info->qualifier = AARCH64_OPND_QLF_S_S; - imm = (fld_i1 << 1) | fld_tszh; - } - else if ((fld_tszl & 0x3) == 0x2) - { - info->qualifier = AARCH64_OPND_QLF_S_H; - imm = (fld_i1 << 2) | (fld_tszh << 1) | (fld_tszl >> 2); - } - else if (fld_tszl & 0x1) - { - info->qualifier = AARCH64_OPND_QLF_S_B; - imm = (fld_i1 << 3) | (fld_tszh << 2) | (fld_tszl >> 1); - } + if (fld_tszl & 0x1) + imm = (fld_i1 << 3) | (fld_tszh << 2) | (fld_tszl >> 1); + else if (fld_tszl & 0x2) + imm = (fld_i1 << 2) | (fld_tszh << 1) | (fld_tszl >> 2); + else if (fld_tszl & 0x4) + imm = (fld_i1 << 1) | fld_tszh; + else if (fld_tszh) + imm = fld_i1; else return false; @@ -2975,6 +2954,25 @@ aarch64_decode_variant_using_iclass (aarch64_inst *inst) variant = 0; switch (inst->opcode->iclass) { + case sme_mov: + variant = extract_fields (inst->value, 0, 2, FLD_SME_Q, FLD_SME_size_10); + if (variant >= 4 && variant < 7) + return false; + if (variant == 7) + variant = 4; + break; + + case sme_psel: + i = extract_fields (inst->value, 0, 2, FLD_SME_tszh, FLD_SME_tszl); + if (i == 0) + return false; + while ((i & 1) == 0) + { + i >>= 1; + variant += 1; + } + break; + case sve_cpy: variant = extract_fields (inst->value, 0, 2, FLD_size, FLD_SVE_M_14); break; @@ -3002,8 +3000,9 @@ aarch64_decode_variant_using_iclass (aarch64_inst *inst) variant = 3; break; + case sme_misc: case sve_misc: - /* sve_misc instructions have only a single variant. */ + /* These instructions have only a single variant. */ break; case sve_movprfx: diff --git a/opcodes/aarch64-tbl.h b/opcodes/aarch64-tbl.h index 96e6c136282..6c2862eacf3 100644 --- a/opcodes/aarch64-tbl.h +++ b/opcodes/aarch64-tbl.h @@ -2650,16 +2650,16 @@ static const aarch64_feature_set aarch64_feature_cssc = FLAGS | F_STRICT, CONSTRAINTS, TIED, NULL } #define SME_INSN(NAME,OPCODE,MASK,CLASS,OP,OPS,QUALS,FLAGS,TIED) \ { NAME, OPCODE, MASK, CLASS, OP, SME, OPS, QUALS, \ - FLAGS, 0, TIED, NULL } + F_STRICT | FLAGS, 0, TIED, NULL } #define SME_F64F64_INSN(NAME,OPCODE,MASK,CLASS,OP,OPS,QUALS,FLAGS,TIED) \ { NAME, OPCODE, MASK, CLASS, OP, SME_F64F64, OPS, QUALS, \ - FLAGS, 0, TIED, NULL } + F_STRICT | FLAGS, 0, TIED, NULL } #define SME_I16I64_INSN(NAME,OPCODE,MASK,CLASS,OP,OPS,QUALS,FLAGS,TIED) \ { NAME, OPCODE, MASK, CLASS, OP, SME_I16I64, OPS, QUALS, \ - FLAGS, 0, TIED, NULL } + F_STRICT | FLAGS, 0, TIED, NULL } #define SME_INSNC(NAME,OPCODE,MASK,CLASS,OP,OPS,QUALS,FLAGS,CONSTRAINTS,TIED) \ { NAME, OPCODE, MASK, CLASS, OP, SME, OPS, QUALS, \ - FLAGS, CONSTRAINTS, TIED, NULL } + F_STRICT | FLAGS, CONSTRAINTS, TIED, NULL } #define SVE2BITPERM_INSN(NAME,OPCODE,MASK,CLASS,OP,OPS,QUALS,FLAGS,TIED) \ { NAME, OPCODE, MASK, CLASS, OP, SVE2_BITPERM, OPS, QUALS, \ FLAGS | F_STRICT, 0, TIED, NULL } @@ -5238,10 +5238,10 @@ const struct aarch64_opcode aarch64_opcode_table[] = SME_INSN ("usmops", 0xa1800010, 0xffe0001c, sme_misc, 0, OP5 (SME_ZAda_2b, SVE_Pg3, SME_Pm, SVE_Zn, SVE_Zm_16), OP_SME_ZADA_S_PM_PM_B_B, 0, 0), SME_I16I64_INSN ("usmops", 0xa1c00010, 0xffe00018, sme_misc, 0, OP5 (SME_ZAda_3b, SVE_Pg3, SME_Pm, SVE_Zn, SVE_Zm_16), OP_SME_ZADA_D_PM_PM_H_H, 0, 0), - SME_INSN ("mov", 0xc0020000, 0xff3e0200, sme_misc, 0, OP3 (SVE_Zd, SVE_Pg3, SME_ZA_HV_idx_src), OP_SME_BHSDQ_PM_BHSDQ, 0, 0), - SME_INSN ("mov", 0xc0000000, 0xff3e0010, sme_misc, 0, OP3 (SME_ZA_HV_idx_dest, SVE_Pg3, SVE_Zn), OP_SME_BHSDQ_PM_BHSDQ, 0, 0), - SME_INSN ("mova", 0xc0020000, 0xff3e0200, sme_misc, 0, OP3 (SVE_Zd, SVE_Pg3, SME_ZA_HV_idx_src), OP_SME_BHSDQ_PM_BHSDQ, 0, 0), - SME_INSN ("mova", 0xc0000000, 0xff3e0010, sme_misc, 0, OP3 (SME_ZA_HV_idx_dest, SVE_Pg3, SVE_Zn), OP_SME_BHSDQ_PM_BHSDQ, 0, 0), + SME_INSN ("mov", 0xc0020000, 0xff3e0200, sme_mov, 0, OP3 (SVE_Zd, SVE_Pg3, SME_ZA_HV_idx_src), OP_SME_BHSDQ_PM_BHSDQ, 0, 0), + SME_INSN ("mov", 0xc0000000, 0xff3e0010, sme_mov, 0, OP3 (SME_ZA_HV_idx_dest, SVE_Pg3, SVE_Zn), OP_SME_BHSDQ_PM_BHSDQ, 0, 0), + SME_INSN ("mova", 0xc0020000, 0xff3e0200, sme_mov, 0, OP3 (SVE_Zd, SVE_Pg3, SME_ZA_HV_idx_src), OP_SME_BHSDQ_PM_BHSDQ, 0, 0), + SME_INSN ("mova", 0xc0000000, 0xff3e0010, sme_mov, 0, OP3 (SME_ZA_HV_idx_dest, SVE_Pg3, SVE_Zn), OP_SME_BHSDQ_PM_BHSDQ, 0, 0), SME_INSN ("zero", 0xc0080000, 0xffffff00, sme_misc, 0, OP1 (SME_list_of_64bit_tiles), {}, 0, 0), @@ -5275,7 +5275,7 @@ const struct aarch64_opcode aarch64_opcode_table[] = SME_INSNC ("revd", 0x52e8000, 0xffffe000, sme_misc, 0, OP3 (SVE_Zd, SVE_Pg3, SVE_Zn), OP_SVE_QMQ, 0, C_SCAN_MOVPRFX, 0), SME_INSNC ("sclamp", 0x4400c000, 0xff20fc00, sve_size_bhsd, 0, OP3 (SVE_Zd, SVE_Zn, SVE_Zm_16), OP_SVE_VVV_BHSD, 0, C_SCAN_MOVPRFX, 0), SME_INSNC ("uclamp", 0x4400c400, 0xff20fc00, sve_size_bhsd, 0, OP3 (SVE_Zd, SVE_Zn, SVE_Zm_16), OP_SVE_VVV_BHSD, 0, C_SCAN_MOVPRFX, 0), - SME_INSN ("psel", 0x25204000, 0xff20c210, sme_misc, 0, OP3 (SVE_Pd, SVE_Pg4_10, SME_PnT_Wm_imm), OP_SVE_NN_BHSD, 0, 0), + SME_INSN ("psel", 0x25204000, 0xff20c210, sme_psel, 0, OP3 (SVE_Pd, SVE_Pg4_10, SME_PnT_Wm_imm), OP_SVE_NN_BHSD, 0, 0), /* SIMD Dot Product (optional in v8.2-A). */ DOT_INSN ("udot", 0x2e009400, 0xbf20fc00, dotproduct, OP3 (Vd, Vn, Vm), QL_V3DOT, F_SIZEQ), -- 2.25.1