From nobody Wed Nov 27 13:56:18 2024 Delivered-To: importer@patchew.org Authentication-Results: mx.zohomail.com; dkim=pass; spf=pass (zohomail.com: domain of gnu.org designates 209.51.188.17 as permitted sender) smtp.mailfrom=qemu-devel-bounces+importer=patchew.org@nongnu.org; dmarc=pass(p=none dis=none) header.from=linaro.org ARC-Seal: i=1; a=rsa-sha256; t=1698522453; cv=none; d=zohomail.com; s=zohoarc; b=W7lAYJAscBvPdLexCO2tBnHIeUv86n7bMqAmL2NJ1UAZKd1Xk9VgOkk942Zds/zZmf2Uc4L4+ixGqINBhPVrswmYnTWqHneZZEA0rXcfSOzKJiHQ8rhsPUyNP1YtSi0THQ+Ikqe2MtyfPncSolI85nQWZ9NPgNYsR+DrTh2jnoY= ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=zohomail.com; s=zohoarc; t=1698522453; h=Content-Transfer-Encoding:Cc:Cc:Date:Date:From:From:In-Reply-To:List-Subscribe:List-Post:List-Id:List-Archive:List-Help:List-Unsubscribe:MIME-Version:Message-ID:References:Sender:Subject:Subject:To:To:Message-Id:Reply-To; bh=k5lsHDsasK4zIrruyQZlXl4wlS973HJoMzMc8iwOmM4=; b=nLfStLCc2Jla4w9IGgK2RynpcM2uf8dcKArSCwsZdnYa4pi+CZ4CegesMpyZ1wbLGKYRvDDE0jmMSkdKt91JcD9hF2Z4HSEVO1GxBcJGVNCQnCiP29DoftUJoLCnl+1thvOvbkfrypLmrYwDEoYkwaQH/CuKGpHOD9/suvS5tNc= ARC-Authentication-Results: i=1; mx.zohomail.com; dkim=pass; spf=pass (zohomail.com: domain of gnu.org designates 209.51.188.17 as permitted sender) smtp.mailfrom=qemu-devel-bounces+importer=patchew.org@nongnu.org; dmarc=pass header.from= (p=none dis=none) Return-Path: Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) by mx.zohomail.com with SMTPS id 1698522453582803.0534305164147; Sat, 28 Oct 2023 12:47:33 -0700 (PDT) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1qwpFr-0004nz-PD; Sat, 28 Oct 2023 15:45:55 -0400 Received: from eggs.gnu.org ([2001:470:142:3::10]) by lists.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1qwpFX-0004ie-J7 for qemu-devel@nongnu.org; Sat, 28 Oct 2023 15:45:35 -0400 Received: from mail-io1-xd30.google.com ([2607:f8b0:4864:20::d30]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_128_GCM_SHA256:128) (Exim 4.90_1) (envelope-from ) id 1qwpFU-00046S-Ir for qemu-devel@nongnu.org; Sat, 28 Oct 2023 15:45:35 -0400 Received: by mail-io1-xd30.google.com with SMTP id ca18e2360f4ac-7a68b87b265so97169139f.2 for ; Sat, 28 Oct 2023 12:45:32 -0700 (PDT) Received: from stoup.. ([71.212.149.95]) by smtp.gmail.com with ESMTPSA id h9-20020a170902f7c900b001b8a3e2c241sm3600096plw.14.2023.10.28.12.45.29 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Sat, 28 Oct 2023 12:45:29 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linaro.org; s=google; t=1698522330; x=1699127130; darn=nongnu.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:from:to:cc:subject:date :message-id:reply-to; bh=k5lsHDsasK4zIrruyQZlXl4wlS973HJoMzMc8iwOmM4=; b=yzpQrP4wpnU9pgkOoqER6fXyN3hbtTHTtczyGf/R0X/XF4KWYJo8y1QeJPm92d4F/8 oGcspEW9Q9MKqPW/6EGG8u1MgJqaw/GIlvCtaN3CM0HNyDewVYJoSDZidmbpPfNJIGMb sfU1vgB6J66ToQGnXVfFJRwNxcICTc+rpxJxhZ0rc1UoAvTIITgUP6eT4H2Wqh4OVjlb T7j1ZDoHTodeS0FrUwjnKizgkoIhO7kqiUaEe4sFLFwy2HQbA32m/gyNsOr2h4BU68ry ZaTrOPJIj/z8ShXI03lp/1Z5I5xYT3wD5HLOsXCFSpxMV+mYt83F6x0q41W/Md+COocT 6meA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1698522330; x=1699127130; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=k5lsHDsasK4zIrruyQZlXl4wlS973HJoMzMc8iwOmM4=; b=sAegQKBwVxcXkAaldvoPK+MH8tLUpR+6/NBM8O3qFsDB7K5RapVbkMTo3XkBIpXhln wy2Xg1ASxxYJ6/UUpyQo/GQ+3pejqS5i520an3KcLK+33MRVFVYNcBTE+uFKQI+LGHEZ a+RPdC1BPna25TFQoTrnMooSYh+NCKdg33dAYovxnccUrKFomnnDLUZozUoux/PySRu3 OBhHwnldcNgUvqw9hm/PMIXyv7Bjjmd9Q204rxCkiY6Aydb33NWUPVmt4M9POKeC5lMV n5BYfG7ckV66YX83ozK8vB3u1522ngUnSJzQiKqCSbdl0fI0/DL7azg/iWcvgWJTW4Ha GjIQ== X-Gm-Message-State: AOJu0YxTXjp5wdkV1wDwwtQCaOT7FdwoUw6MFWfqruyEHHQRNVQuZ/s+ vTZeNc8tLciVPSSGkLMP4eEs3MqthrqQH2cTXaM= X-Google-Smtp-Source: AGHT+IGY/WpQVHw0Ya+CVSbs96C4B7wXcWmYdOdGNhUu6oTyE3DXpwSwQanwZFdg/Cl9Hm/wZcagGg== X-Received: by 2002:a05:6e02:1be7:b0:357:a25a:572a with SMTP id y7-20020a056e021be700b00357a25a572amr9930623ilv.21.1698522330043; Sat, 28 Oct 2023 12:45:30 -0700 (PDT) From: Richard Henderson To: qemu-devel@nongnu.org Cc: philmd@linaro.org, pbonzini@redhat.com Subject: [PATCH v2 06/35] tcg/optimize: Handle TCG_COND_TST{EQ,NE} Date: Sat, 28 Oct 2023 12:44:53 -0700 Message-Id: <20231028194522.245170-7-richard.henderson@linaro.org> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20231028194522.245170-1-richard.henderson@linaro.org> References: <20231028194522.245170-1-richard.henderson@linaro.org> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Received-SPF: pass (zohomail.com: domain of gnu.org designates 209.51.188.17 as permitted sender) client-ip=209.51.188.17; envelope-from=qemu-devel-bounces+importer=patchew.org@nongnu.org; helo=lists.gnu.org; Received-SPF: pass client-ip=2607:f8b0:4864:20::d30; envelope-from=richard.henderson@linaro.org; helo=mail-io1-xd30.google.com X-Spam_score_int: -20 X-Spam_score: -2.1 X-Spam_bar: -- X-Spam_report: (-2.1 / 5.0 requ) BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1, RCVD_IN_DNSWL_NONE=-0.0001, SPF_HELO_NONE=0.001, SPF_PASS=-0.001 autolearn=ham autolearn_force=no X-Spam_action: no action X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+importer=patchew.org@nongnu.org Sender: qemu-devel-bounces+importer=patchew.org@nongnu.org X-ZohoMail-DKIM: pass (identity @linaro.org) X-ZM-MESSAGEID: 1698522454790100003 Content-Type: text/plain; charset="utf-8" Fold constant comparisons. Canonicalize "tst x,x" to equality vs zero. Canonicalize "tst x,sign" to sign test vs zero. Fold double-word comparisons with zero parts. Fold setcond of "tst x,pow2" to a bit extract. Signed-off-by: Richard Henderson Reviewed-by: Philippe Mathieu-Daud=C3=A9 --- tcg/optimize.c | 245 ++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 223 insertions(+), 22 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index 2f2d1c3001..891c28acef 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -508,9 +508,15 @@ static bool do_constant_folding_cond_32(uint32_t x, ui= nt32_t y, TCGCond c) return x <=3D y; case TCG_COND_GTU: return x > y; - default: - g_assert_not_reached(); + case TCG_COND_TSTEQ: + return (x & y) =3D=3D 0; + case TCG_COND_TSTNE: + return (x & y) !=3D 0; + case TCG_COND_ALWAYS: + case TCG_COND_NEVER: + break; } + g_assert_not_reached(); } =20 static bool do_constant_folding_cond_64(uint64_t x, uint64_t y, TCGCond c) @@ -536,12 +542,18 @@ static bool do_constant_folding_cond_64(uint64_t x, u= int64_t y, TCGCond c) return x <=3D y; case TCG_COND_GTU: return x > y; - default: - g_assert_not_reached(); + case TCG_COND_TSTEQ: + return (x & y) =3D=3D 0; + case TCG_COND_TSTNE: + return (x & y) !=3D 0; + case TCG_COND_ALWAYS: + case TCG_COND_NEVER: + break; } + g_assert_not_reached(); } =20 -static bool do_constant_folding_cond_eq(TCGCond c) +static int do_constant_folding_cond_eq(TCGCond c) { switch (c) { case TCG_COND_GT: @@ -556,9 +568,14 @@ static bool do_constant_folding_cond_eq(TCGCond c) case TCG_COND_LEU: case TCG_COND_EQ: return 1; - default: - g_assert_not_reached(); + case TCG_COND_TSTEQ: + case TCG_COND_TSTNE: + return -1; + case TCG_COND_ALWAYS: + case TCG_COND_NEVER: + break; } + g_assert_not_reached(); } =20 /* @@ -586,8 +603,10 @@ static int do_constant_folding_cond(TCGType type, TCGA= rg x, } else if (arg_is_const_val(y, 0)) { switch (c) { case TCG_COND_LTU: + case TCG_COND_TSTNE: return 0; case TCG_COND_GEU: + case TCG_COND_TSTEQ: return 1; default: return -1; @@ -660,7 +679,30 @@ static int do_constant_folding_cond1(OptContext *ctx, = TCGArg dest, } =20 r =3D do_constant_folding_cond(ctx->type, *p1, *p2, cond); - return r; + if (r >=3D 0) { + return r; + } + if (!is_tst_cond(cond)) { + return -1; + } + + /* + * TSTNE x,x -> NE x,0 + * TSTNE x,-1 -> NE x,0 + */ + if (args_are_copies(*p1, *p2) || arg_is_const_val(*p2, -1)) { + *p2 =3D arg_new_constant(ctx, 0); + *pcond =3D tcg_tst_eqne_cond(cond); + return -1; + } + + /* TSTNE x,sign -> LT x,0 */ + if (arg_is_const_val(*p2, (ctx->type =3D=3D TCG_TYPE_I32 + ? INT32_MIN : INT64_MIN))) { + *p2 =3D arg_new_constant(ctx, 0); + *pcond =3D tcg_tst_ltge_cond(cond); + } + return -1; } =20 static int do_constant_folding_cond2(OptContext *ctx, TCGArg *args) @@ -668,6 +710,7 @@ static int do_constant_folding_cond2(OptContext *ctx, T= CGArg *args) TCGArg al, ah, bl, bh; TCGCond c; bool swap; + int r; =20 swap =3D swap_commutative2(args, args + 2); c =3D args[4]; @@ -689,21 +732,54 @@ static int do_constant_folding_cond2(OptContext *ctx,= TCGArg *args) tcg_target_ulong alv =3D arg_info(al)->val; tcg_target_ulong ahv =3D arg_info(ah)->val; uint64_t a =3D deposit64(alv, 32, 32, ahv); - return do_constant_folding_cond_64(a, b, c); + + r =3D do_constant_folding_cond_64(a, b, c); + if (r >=3D 0) { + return r; + } } + if (b =3D=3D 0) { switch (c) { case TCG_COND_LTU: + case TCG_COND_TSTNE: return 0; case TCG_COND_GEU: + case TCG_COND_TSTEQ: return 1; default: break; } } + + /* TSTNE x,-1 -> NE x,0 */ + if (b =3D=3D -1 && is_tst_cond(c)) { + args[3] =3D args[2] =3D arg_new_constant(ctx, 0); + args[4] =3D tcg_tst_eqne_cond(c); + return -1; + } + + /* TSTNE x,sign -> LT x,0 */ + if (b =3D=3D INT64_MIN && is_tst_cond(c)) { + /* bl must be 0, so copy that to bh */ + args[3] =3D bl; + args[4] =3D tcg_tst_ltge_cond(c); + return -1; + } } + if (args_are_copies(al, bl) && args_are_copies(ah, bh)) { - return do_constant_folding_cond_eq(c); + r =3D do_constant_folding_cond_eq(c); + if (r >=3D 0) { + return r; + } + + /* TSTNE x,x -> NE x,0 */ + if (is_tst_cond(c)) { + args[3] =3D args[2] =3D arg_new_constant(ctx, 0); + args[4] =3D tcg_tst_eqne_cond(c); + return -1; + } } return -1; } @@ -1151,24 +1227,37 @@ static bool fold_brcond2(OptContext *ctx, TCGOp *op) case 0: goto do_brcond_const; case 1: - op->opc =3D INDEX_op_brcond_i32; - op->args[1] =3D op->args[2]; - op->args[2] =3D cond; - op->args[3] =3D label; - break; + goto do_brcond_low; + } + break; + + case TCG_COND_TSTEQ: + case TCG_COND_TSTNE: + if (arg_is_const_val(op->args[2], 0)) { + goto do_brcond_high; + } + if (arg_is_const_val(op->args[3], 0)) { + goto do_brcond_low; } break; =20 default: break; =20 + do_brcond_low: + op->opc =3D INDEX_op_brcond_i32; + op->args[1] =3D op->args[2]; + op->args[2] =3D cond; + op->args[3] =3D label; + return fold_brcond(ctx, op); + do_brcond_high: op->opc =3D INDEX_op_brcond_i32; op->args[0] =3D op->args[1]; op->args[1] =3D op->args[3]; op->args[2] =3D cond; op->args[3] =3D label; - break; + return fold_brcond(ctx, op); =20 do_brcond_const: if (i =3D=3D 0) { @@ -1829,6 +1918,104 @@ static bool fold_remainder(OptContext *ctx, TCGOp *= op) return false; } =20 +static void fold_setcond_tst_pow2(OptContext *ctx, TCGOp *op, bool neg) +{ + TCGOpcode and_opc, sub_opc, xor_opc, neg_opc, shr_opc, uext_opc, sext_= opc; + TCGCond cond =3D op->args[3]; + TCGArg ret, src1, src2; + TCGOp *op2; + uint64_t val; + int sh; + bool inv; + + if (!is_tst_cond(cond) || !arg_is_const(op->args[2])) { + return; + } + + src2 =3D op->args[2]; + val =3D arg_info(src2)->val; + if (!is_power_of_2(val)) { + return; + } + sh =3D ctz64(val); + + switch (ctx->type) { + case TCG_TYPE_I32: + and_opc =3D INDEX_op_and_i32; + sub_opc =3D INDEX_op_sub_i32; + xor_opc =3D INDEX_op_xor_i32; + shr_opc =3D INDEX_op_shr_i32; + neg_opc =3D TCG_TARGET_HAS_neg_i32 ? INDEX_op_neg_i32 : 0; + if (TCG_TARGET_extract_i32_valid(sh, 1)) { + uext_opc =3D TCG_TARGET_HAS_extract_i32 ? INDEX_op_extract_i32= : 0; + sext_opc =3D TCG_TARGET_HAS_sextract_i32 ? INDEX_op_sextract_i= 32 : 0; + } + break; + case TCG_TYPE_I64: + and_opc =3D INDEX_op_and_i64; + sub_opc =3D INDEX_op_sub_i64; + xor_opc =3D INDEX_op_xor_i64; + shr_opc =3D INDEX_op_shr_i64; + neg_opc =3D TCG_TARGET_HAS_neg_i64 ? INDEX_op_neg_i64 : 0; + if (TCG_TARGET_extract_i64_valid(sh, 1)) { + uext_opc =3D TCG_TARGET_HAS_extract_i64 ? INDEX_op_extract_i64= : 0; + sext_opc =3D TCG_TARGET_HAS_sextract_i64 ? INDEX_op_sextract_i= 64 : 0; + } + break; + default: + g_assert_not_reached(); + } + + ret =3D op->args[0]; + src1 =3D op->args[1]; + inv =3D cond =3D=3D TCG_COND_TSTEQ; + + if (sh && sext_opc && neg && !inv) { + op->opc =3D sext_opc; + op->args[1] =3D src1; + op->args[2] =3D sh; + op->args[3] =3D 1; + return; + } else if (sh && uext_opc) { + op->opc =3D uext_opc; + op->args[1] =3D src1; + op->args[2] =3D sh; + op->args[3] =3D 1; + } else { + if (sh) { + op2 =3D tcg_op_insert_before(ctx->tcg, op, shr_opc, 3); + op2->args[0] =3D ret; + op2->args[1] =3D src1; + op2->args[2] =3D arg_new_constant(ctx, sh); + src1 =3D ret; + } + op->opc =3D and_opc; + op->args[1] =3D src1; + op->args[2] =3D arg_new_constant(ctx, 1); + } + + if (neg && inv) { + op2 =3D tcg_op_insert_after(ctx->tcg, op, sub_opc, 3); + op2->args[0] =3D ret; + op2->args[1] =3D ret; + op2->args[2] =3D arg_new_constant(ctx, 1); + } else if (inv) { + op2 =3D tcg_op_insert_after(ctx->tcg, op, xor_opc, 3); + op2->args[0] =3D ret; + op2->args[1] =3D ret; + op2->args[2] =3D arg_new_constant(ctx, 1); + } else if (neg && neg_opc) { + op2 =3D tcg_op_insert_after(ctx->tcg, op, neg_opc, 2); + op2->args[0] =3D ret; + op2->args[1] =3D ret; + } else if (neg) { + op2 =3D tcg_op_insert_after(ctx->tcg, op, sub_opc, 3); + op2->args[0] =3D ret; + op2->args[1] =3D arg_new_constant(ctx, 0); + op2->args[2] =3D ret; + } +} + static bool fold_setcond(OptContext *ctx, TCGOp *op) { int i =3D do_constant_folding_cond1(ctx, op->args[0], &op->args[1], @@ -1836,6 +2023,7 @@ static bool fold_setcond(OptContext *ctx, TCGOp *op) if (i >=3D 0) { return tcg_opt_gen_movi(ctx, op, op->args[0], i); } + fold_setcond_tst_pow2(ctx, op, false); =20 ctx->z_mask =3D 1; ctx->s_mask =3D smask_from_zmask(1); @@ -1849,13 +2037,13 @@ static bool fold_negsetcond(OptContext *ctx, TCGOp = *op) if (i >=3D 0) { return tcg_opt_gen_movi(ctx, op, op->args[0], -i); } + fold_setcond_tst_pow2(ctx, op, true); =20 /* Value is {0,-1} so all bits are repetitions of the sign. */ ctx->s_mask =3D -1; return false; } =20 - static bool fold_setcond2(OptContext *ctx, TCGOp *op) { TCGCond cond; @@ -1903,22 +2091,35 @@ static bool fold_setcond2(OptContext *ctx, TCGOp *o= p) case 0: goto do_setcond_const; case 1: - op->args[2] =3D op->args[3]; - op->args[3] =3D cond; - op->opc =3D INDEX_op_setcond_i32; - break; + goto do_setcond_low; + } + break; + + case TCG_COND_TSTEQ: + case TCG_COND_TSTNE: + if (arg_is_const_val(op->args[2], 0)) { + goto do_setcond_high; + } + if (arg_is_const_val(op->args[4], 0)) { + goto do_setcond_low; } break; =20 default: break; =20 + do_setcond_low: + op->args[2] =3D op->args[3]; + op->args[3] =3D cond; + op->opc =3D INDEX_op_setcond_i32; + return fold_setcond(ctx, op); + do_setcond_high: op->args[1] =3D op->args[2]; op->args[2] =3D op->args[4]; op->args[3] =3D cond; op->opc =3D INDEX_op_setcond_i32; - break; + return fold_setcond(ctx, op); } =20 ctx->z_mask =3D 1; --=20 2.34.1