From patchwork Fri Aug 4 05:28:31 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Richard Henderson X-Patchwork-Id: 9880507 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 76B4360360 for ; Fri, 4 Aug 2017 05:32:57 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 71A6328765 for ; Fri, 4 Aug 2017 05:32:57 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 6682C289A3; Fri, 4 Aug 2017 05:32:57 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.8 required=2.0 tests=BAYES_00,DKIM_SIGNED, RCVD_IN_DNSWL_HI,T_DKIM_INVALID autolearn=ham version=3.3.1 Received: from lists.gnu.org (lists.gnu.org [208.118.235.17]) (using TLSv1 with cipher AES256-SHA (256/256 bits)) (No client certificate requested) by mail.wl.linuxfoundation.org (Postfix) with ESMTPS id 8979828765 for ; Fri, 4 Aug 2017 05:32:56 +0000 (UTC) Received: from localhost ([::1]:35742 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ddVEF-00024Y-Mn for patchwork-qemu-devel@patchwork.kernel.org; Fri, 04 Aug 2017 01:32:55 -0400 Received: from eggs.gnu.org ([2001:4830:134:3::10]:36470) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ddVAG-0007f7-CG for qemu-devel@nongnu.org; Fri, 04 Aug 2017 01:28:51 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ddVAE-0001I8-QF for qemu-devel@nongnu.org; Fri, 04 Aug 2017 01:28:48 -0400 Received: from mail-pg0-x242.google.com ([2607:f8b0:400e:c05::242]:33910) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1ddVAE-0001HQ-Hk for qemu-devel@nongnu.org; Fri, 04 Aug 2017 01:28:46 -0400 Received: by mail-pg0-x242.google.com with SMTP id y192so803236pgd.1 for ; Thu, 03 Aug 2017 22:28:46 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=sender:from:to:cc:subject:date:message-id:in-reply-to:references; bh=ZzOQcxP9s6jKeIJ4rIh+gzYTzAmYoTWmnfWvZFU4YGg=; b=nbHZYYabQDFOczttL+XSwV9N7MQVNM8MCyOyWPHNt6R1SopMGYf4d5gJwK0DeKX10u WGheFzsoQYdpJAUOib3snu9IfyS67ocx3TooOifptRW4fgvbwh2Nvsm/5M9y3xq8NBLV pVsl8wSeshUXZbEV/lJ+rAOjkQvT9QhjfhbDpGE3xpv1okG05J82P/n3halmT+sysh44 03eHhZPQHZfCcfXufa61Tt44SiQB1JvBgincjZ98hYJJMpPg/fbfhxQpmvWNtrJBccb3 zKFywt4Fb5nW+rvbr0V3wpz22mvb4iD9n+ZzDctNSmwTBmjWKTCCwU0zkxkT1Z7R0yzJ BsTQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:sender:from:to:cc:subject:date:message-id :in-reply-to:references; bh=ZzOQcxP9s6jKeIJ4rIh+gzYTzAmYoTWmnfWvZFU4YGg=; b=B1wzQ7aL/waDDnYVhAQAyfiZGF+Um7rjHYgEIYKrrFYhoNqlW3QdbsBBVQoZZidg6F YBuPMxWAj0leSE6BTJQEHxlpjN5oqhnvjWcE6fXHYIF3d4G9u2FvaQXoqr1vyMUQTYqa Frm9p77q13zGL0OcuwZ9BC+5qzVTZ+QBvTbRUMEd+B+PVh0XvzJrB+JrtP7/ekywquiI 0Ly8QSnMbjobl/7UAOTwi9c2dlGPoScOz/pw0e2Mf+KCnH3y/SOblZ9UGJ7P3Ko+Jq+F Y+y7QRnzBQAgnU33xQb3KS6QnxgjzL4RuD6KYDDjKj8J3m6Bv6XTrVugynWjFsprqHF7 cvYQ== X-Gm-Message-State: AIVw113VCVLqIbNF0BRrop8Fe5JztevkRhpGSwt8jvZLNB4T6kU/VxgP w0B2IBefbE4s2243OVg= X-Received: by 10.98.206.194 with SMTP id y185mr1209420pfg.312.1501824525358; Thu, 03 Aug 2017 22:28:45 -0700 (PDT) Received: from bigtime.twiddle.net (97-126-108-236.tukw.qwest.net. [97.126.108.236]) by smtp.gmail.com with ESMTPSA id m5sm1053099pfg.22.2017.08.03.22.28.44 (version=TLS1_2 cipher=ECDHE-RSA-CHACHA20-POLY1305 bits=256/256); Thu, 03 Aug 2017 22:28:44 -0700 (PDT) From: Richard Henderson To: qemu-devel@nongnu.org Date: Thu, 3 Aug 2017 22:28:31 -0700 Message-Id: <20170804052833.10187-10-rth@twiddle.net> X-Mailer: git-send-email 2.13.3 In-Reply-To: <20170804052833.10187-1-rth@twiddle.net> References: <20170804052833.10187-1-rth@twiddle.net> X-detected-operating-system: by eggs.gnu.org: Genre and OS details not recognized. X-Received-From: 2607:f8b0:400e:c05::242 Subject: [Qemu-devel] [PATCH 6/8] tcg/s390: Use distinct-operands facility X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.21 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: borntraeger@de.ibm.com, cohuck@redhat.com, agraf@suse.de Errors-To: qemu-devel-bounces+patchwork-qemu-devel=patchwork.kernel.org@nongnu.org Sender: "Qemu-devel" X-Virus-Scanned: ClamAV using ClamSMTP This allows using a 3-operand insn form for some arithmetic, logicals and shifts. Signed-off-by: Richard Henderson --- tcg/s390/tcg-target.h | 1 + tcg/s390/tcg-target.inc.c | 118 +++++++++++++++++++++++++++++++++++----------- 2 files changed, 91 insertions(+), 28 deletions(-) diff --git a/tcg/s390/tcg-target.h b/tcg/s390/tcg-target.h index 957f0c0afe..1b5eb22c26 100644 --- a/tcg/s390/tcg-target.h +++ b/tcg/s390/tcg-target.h @@ -58,6 +58,7 @@ typedef enum TCGReg { #define FACILITY_GEN_INST_EXT (1ULL << (63 - 34)) #define FACILITY_LOAD_ON_COND (1ULL << (63 - 45)) #define FACILITY_FAST_BCR_SER FACILITY_LOAD_ON_COND +#define FACILITY_DISTINCT_OPS FACILITY_LOAD_ON_COND extern uint64_t s390_facilities; diff --git a/tcg/s390/tcg-target.inc.c b/tcg/s390/tcg-target.inc.c index 5414c9d879..a80b07db65 100644 --- a/tcg/s390/tcg-target.inc.c +++ b/tcg/s390/tcg-target.inc.c @@ -159,6 +159,16 @@ typedef enum S390Opcode { RRF_LOCR = 0xb9f2, RRF_LOCGR = 0xb9e2, + RRF_NRK = 0xb9f4, + RRF_NGRK = 0xb9e4, + RRF_ORK = 0xb9f6, + RRF_OGRK = 0xb9e6, + RRF_SRK = 0xb9f9, + RRF_SGRK = 0xb9e9, + RRF_SLRK = 0xb9fb, + RRF_SLGRK = 0xb9eb, + RRF_XRK = 0xb9f7, + RRF_XGRK = 0xb9e7, RR_AR = 0x1a, RR_ALR = 0x1e, @@ -179,8 +189,11 @@ typedef enum S390Opcode { RSY_RLL = 0xeb1d, RSY_RLLG = 0xeb1c, RSY_SLLG = 0xeb0d, + RSY_SLLK = 0xebdf, RSY_SRAG = 0xeb0a, + RSY_SRAK = 0xebdc, RSY_SRLG = 0xeb0c, + RSY_SRLK = 0xebde, RS_SLL = 0x89, RS_SRA = 0x8a, @@ -1065,23 +1078,29 @@ static void tgen_setcond(TCGContext *s, TCGType type, TCGCond cond, case TCG_COND_GEU: do_geu: /* We need "real" carry semantics, so use SUBTRACT LOGICAL - instead of COMPARE LOGICAL. This needs an extra move. */ - tcg_out_mov(s, type, TCG_TMP0, c1); + instead of COMPARE LOGICAL. This may need an extra move. */ if (c2const) { - tcg_out_movi(s, TCG_TYPE_I64, dest, 0); + tcg_out_mov(s, type, TCG_TMP0, c1); if (type == TCG_TYPE_I32) { tcg_out_insn(s, RIL, SLFI, TCG_TMP0, c2); } else { tcg_out_insn(s, RIL, SLGFI, TCG_TMP0, c2); } + } else if (s390_facilities & FACILITY_DISTINCT_OPS) { + if (type == TCG_TYPE_I32) { + tcg_out_insn(s, RRF, SLRK, TCG_TMP0, c1, c2); + } else { + tcg_out_insn(s, RRF, SLGRK, TCG_TMP0, c1, c2); + } } else { + tcg_out_mov(s, type, TCG_TMP0, c1); if (type == TCG_TYPE_I32) { tcg_out_insn(s, RR, SLR, TCG_TMP0, c2); } else { tcg_out_insn(s, RRE, SLGR, TCG_TMP0, c2); } - tcg_out_movi(s, TCG_TYPE_I64, dest, 0); } + tcg_out_movi(s, TCG_TYPE_I64, dest, 0); tcg_out_insn(s, RRE, ALCGR, dest, dest); return; @@ -1648,7 +1667,7 @@ static void tcg_out_qemu_st(TCGContext* s, TCGReg data_reg, TCGReg addr_reg, static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, const int *const_args) { - S390Opcode op; + S390Opcode op, op2; TCGArg a0, a1, a2; switch (opc) { @@ -1753,29 +1772,44 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, if (const_args[2]) { a2 = -a2; goto do_addi_32; + } else if (a0 == a1) { + tcg_out_insn(s, RR, SR, a0, a2); + } else { + tcg_out_insn(s, RRF, SRK, a0, a1, a2); } - tcg_out_insn(s, RR, SR, args[0], args[2]); break; case INDEX_op_and_i32: + a0 = args[0], a1 = args[1], a2 = (uint32_t)args[2]; if (const_args[2]) { - tgen_andi(s, TCG_TYPE_I32, args[0], args[2]); + tcg_out_mov(s, TCG_TYPE_I32, a0, a1); + tgen_andi(s, TCG_TYPE_I32, a0, a2); + } else if (a0 == a1) { + tcg_out_insn(s, RR, NR, a0, a2); } else { - tcg_out_insn(s, RR, NR, args[0], args[2]); + tcg_out_insn(s, RRF, NRK, a0, a1, a2); } break; case INDEX_op_or_i32: + a0 = args[0], a1 = args[1], a2 = (uint32_t)args[2]; if (const_args[2]) { - tgen64_ori(s, args[0], args[2] & 0xffffffff); + tcg_out_mov(s, TCG_TYPE_I32, a0, a1); + tgen64_ori(s, a0, a2); + } else if (a0 == a1) { + tcg_out_insn(s, RR, OR, a0, a2); } else { - tcg_out_insn(s, RR, OR, args[0], args[2]); + tcg_out_insn(s, RRF, ORK, a0, a1, a2); } break; case INDEX_op_xor_i32: + a0 = args[0], a1 = args[1], a2 = (uint32_t)args[2]; if (const_args[2]) { - tgen64_xori(s, args[0], args[2] & 0xffffffff); - } else { + tcg_out_mov(s, TCG_TYPE_I32, a0, a1); + tgen64_xori(s, a0, a2); + } else if (a0 == a1) { tcg_out_insn(s, RR, XR, args[0], args[2]); + } else { + tcg_out_insn(s, RRF, XRK, a0, a1, a2); } break; @@ -1804,18 +1838,31 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_shl_i32: op = RS_SLL; + op2 = RSY_SLLK; do_shift32: - if (const_args[2]) { - tcg_out_sh32(s, op, args[0], TCG_REG_NONE, args[2]); + a0 = args[0], a1 = args[1], a2 = (int32_t)args[2]; + if (a0 == a1) { + if (const_args[2]) { + tcg_out_sh32(s, op, a0, TCG_REG_NONE, a2); + } else { + tcg_out_sh32(s, op, a0, a2, 0); + } } else { - tcg_out_sh32(s, op, args[0], args[2], 0); + /* Using tcg_out_sh64 here for the format; it is a 32-bit shift. */ + if (const_args[2]) { + tcg_out_sh64(s, op2, a0, a1, TCG_REG_NONE, a2); + } else { + tcg_out_sh64(s, op2, a0, a1, a2, 0); + } } break; case INDEX_op_shr_i32: op = RS_SRL; + op2 = RSY_SRLK; goto do_shift32; case INDEX_op_sar_i32: op = RS_SRA; + op2 = RSY_SRAK; goto do_shift32; case INDEX_op_rotl_i32: @@ -1957,30 +2004,44 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, if (const_args[2]) { a2 = -a2; goto do_addi_64; + } else if (a0 == a1) { + tcg_out_insn(s, RRE, SGR, a0, a2); } else { - tcg_out_insn(s, RRE, SGR, args[0], args[2]); + tcg_out_insn(s, RRF, SGRK, a0, a1, a2); } break; case INDEX_op_and_i64: + a0 = args[0], a1 = args[1], a2 = args[2]; if (const_args[2]) { + tcg_out_mov(s, TCG_TYPE_I64, a0, a1); tgen_andi(s, TCG_TYPE_I64, args[0], args[2]); - } else { + } else if (a0 == a1) { tcg_out_insn(s, RRE, NGR, args[0], args[2]); + } else { + tcg_out_insn(s, RRF, NGRK, a0, a1, a2); } break; case INDEX_op_or_i64: + a0 = args[0], a1 = args[1], a2 = args[2]; if (const_args[2]) { - tgen64_ori(s, args[0], args[2]); + tcg_out_mov(s, TCG_TYPE_I64, a0, a1); + tgen64_ori(s, a0, a2); + } else if (a0 == a1) { + tcg_out_insn(s, RRE, OGR, a0, a2); } else { - tcg_out_insn(s, RRE, OGR, args[0], args[2]); + tcg_out_insn(s, RRF, OGRK, a0, a1, a2); } break; case INDEX_op_xor_i64: + a0 = args[0], a1 = args[1], a2 = args[2]; if (const_args[2]) { - tgen64_xori(s, args[0], args[2]); + tcg_out_mov(s, TCG_TYPE_I64, a0, a1); + tgen64_xori(s, a0, a2); + } else if (a0 == a1) { + tcg_out_insn(s, RRE, XGR, a0, a2); } else { - tcg_out_insn(s, RRE, XGR, args[0], args[2]); + tcg_out_insn(s, RRF, XGRK, a0, a1, a2); } break; @@ -2168,6 +2229,7 @@ static const TCGTargetOpDef *tcg_target_op_def(TCGOpcode op) static const TCGTargetOpDef r_rC = { .args_ct_str = { "r", "rC" } }; static const TCGTargetOpDef r_rZ = { .args_ct_str = { "r", "rZ" } }; static const TCGTargetOpDef r_r_ri = { .args_ct_str = { "r", "r", "ri" } }; + static const TCGTargetOpDef r_r_rM = { .args_ct_str = { "r", "r", "rM" } }; static const TCGTargetOpDef r_0_r = { .args_ct_str = { "r", "0", "r" } }; static const TCGTargetOpDef r_0_ri = { .args_ct_str = { "r", "0", "ri" } }; static const TCGTargetOpDef r_0_rI = { .args_ct_str = { "r", "0", "rI" } }; @@ -2211,7 +2273,7 @@ static const TCGTargetOpDef *tcg_target_op_def(TCGOpcode op) return &r_r_ri; case INDEX_op_sub_i32: case INDEX_op_sub_i64: - return &r_0_ri; + return (s390_facilities & FACILITY_DISTINCT_OPS ? &r_r_ri : &r_0_ri); case INDEX_op_mul_i32: /* If we have the general-instruction-extensions, then we have @@ -2227,32 +2289,32 @@ static const TCGTargetOpDef *tcg_target_op_def(TCGOpcode op) OI[LH][LH] instructions. By rejecting certain negative ranges, the immediate load plus the reg-reg OR is smaller. */ return (s390_facilities & FACILITY_EXT_IMM - ? &r_0_ri + ? (s390_facilities & FACILITY_DISTINCT_OPS ? &r_r_ri : &r_0_ri) : &r_0_rN); case INDEX_op_or_i64: return (s390_facilities & FACILITY_EXT_IMM - ? &r_0_rM + ? (s390_facilities & FACILITY_DISTINCT_OPS ? &r_r_rM : &r_0_rM) : &r_0_rN); case INDEX_op_xor_i32: /* Without EXT_IMM, no immediates are supported. Otherwise, rejecting certain negative ranges leads to smaller code. */ return (s390_facilities & FACILITY_EXT_IMM - ? &r_0_ri + ? (s390_facilities & FACILITY_DISTINCT_OPS ? &r_r_ri : &r_0_ri) : &r_0_r); case INDEX_op_xor_i64: return (s390_facilities & FACILITY_EXT_IMM - ? &r_0_rM + ? (s390_facilities & FACILITY_DISTINCT_OPS ? &r_r_rM : &r_0_rM) : &r_0_r); case INDEX_op_and_i32: case INDEX_op_and_i64: - return &r_0_ri; + return (s390_facilities & FACILITY_DISTINCT_OPS ? &r_r_ri : &r_0_ri); case INDEX_op_shl_i32: case INDEX_op_shr_i32: case INDEX_op_sar_i32: - return &r_0_ri; + return (s390_facilities & FACILITY_DISTINCT_OPS ? &r_r_ri : &r_0_ri); case INDEX_op_shl_i64: case INDEX_op_shr_i64: