From nobody Sun Apr 12 00:57:43 2026 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=1772500167; cv=none; d=zohomail.com; s=zohoarc; b=gWTC3Fs8rAhZfXB5sB3Vb5XYaVYoQ8QM82NAjGXGTNeV2vsHaqNf77IYrhUrE0TP+1fh/iJ+AIYLaqwF24Z7tA43uyGz4aAaHpUzgu8/sKlN6nP+lrN6ll+SsarM9YNIn9DVaEyvRbe0pUQc9aFb631GWP0UE6uieE9DrhLQ7/s= ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=zohomail.com; s=zohoarc; t=1772500167; 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=Ep4G4BLNwlEA/4u+aLTe3Bc1f2kKgbSUaJgl+b2hJIs=; b=F85QpkdCHFInfvsIcGVImaRfVp0wQAP/W2976vWNFyoOzkPahup9QSvVWeZn8G8KIP+vrxnztsq4+hgno8vvykV9+3HLDYBAQy+DjUKzIVJCvh0flBxcN6xDc//uoRVueUMoQ8MLeq1xeEA2KCfkD0ClF2sAVXKX/djjN6DJpyE= 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 1772500167301609.2722070059139; Mon, 2 Mar 2026 17:09:27 -0800 (PST) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1vxEFk-0000EE-77; Mon, 02 Mar 2026 20:08:48 -0500 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 1vxEFj-0000E3-6y for qemu-devel@nongnu.org; Mon, 02 Mar 2026 20:08:47 -0500 Received: from mail-pf1-x42f.google.com ([2607:f8b0:4864:20::42f]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_128_GCM_SHA256:128) (Exim 4.90_1) (envelope-from ) id 1vxEFh-00071O-0z for qemu-devel@nongnu.org; Mon, 02 Mar 2026 20:08:46 -0500 Received: by mail-pf1-x42f.google.com with SMTP id d2e1a72fcca58-827390e8a3cso3289336b3a.0 for ; Mon, 02 Mar 2026 17:08:44 -0800 (PST) Received: from stoup.. ([180.233.125.15]) by smtp.gmail.com with ESMTPSA id d2e1a72fcca58-82739d94de6sm14484988b3a.24.2026.03.02.17.08.41 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 02 Mar 2026 17:08:43 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linaro.org; s=google; t=1772500124; x=1773104924; 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=Ep4G4BLNwlEA/4u+aLTe3Bc1f2kKgbSUaJgl+b2hJIs=; b=sVT8ukjXBuMayAwz6nsSVhmN1j5pWGf+ojXF1VsCVhWywjTI1Dkk5BfXnpKyTtd3vr QeuEnsSncITE85/IaFsbrxStXiHo+EVV/KmWeQwKKaXSG8eN/ApkalYjjI/a9gKiBIja YeH99jb6LSDSUzCEZyPRkqe5X/H5OB/nKRnkbJza6iJiRRFADteAdBiIO3iptI9I6GBh voXl0UL5f6ZsB0/im7C8cgjUx0NGVFLztmxpdgAtUMHmF9VA/giD2gdHn9dH3YcgjQw7 Jn3TCwKFgN9rx35ecHUaT7bZgjmfNzZcRI41zVoNI1J1xKXD+iXRKU/KzPyHFJHTYoh1 vacg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1772500124; x=1773104924; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-gg:x-gm-message-state:from :to:cc:subject:date:message-id:reply-to; bh=Ep4G4BLNwlEA/4u+aLTe3Bc1f2kKgbSUaJgl+b2hJIs=; b=aXv7skouCA3rrFbYo+4HBHTDRx4pkFUxYFTmMhWZ3dDZFiM6nByBdVrFCHKmREl72R xBFUURjGqksEDx0xzJRe9xQC+XqRGOQK77aMTJm5FJ0Mll1QonGzCj01GspRYMea2qil O6eFxp8Vg0kFz8sGqWg/4Iz03TM0Ob5Nl13odGSntKpgtoiVLYvirIA5fxs0tF/Wi92H CEct7265XaXe7/lN2TbB8VRs2twqrDAcJwA2Dyl4FZcjh9tCnVRuzipwhoBk4Sy9HU/h daV34EYAQw7EE6EVG+nZmM3cp4LjKADhEXABvSzNjRyDOvGKoFcZ/0GMNYekojt5E1bC RL9w== X-Gm-Message-State: AOJu0YygSJdoSJ4wCPhfhJpCay2AzisJ1ccQIwbxkoiNFxBTU1zIO0TY Zy5Z6JiSPjJBu/4oJQJ03dsLYpE+4xS2O2wV8e16Ld0l+CvnBgChjJxC9m9tWu6LIUJ+zjJjfXW ZEGZIRw8= X-Gm-Gg: ATEYQzywu6oCf634ZZX22eEW1ks7SgUjRkS7K1rkpvb3WpvqLuupZhGzUCqxL4AklVo ouPqyUc3uUqybjzm0UfbKF2a7QwRnVQKzy/zRrC9e/LlP1pr+IEGsUwOXYJ3TrYwVmuKih2RxMV sxVW7DlINTu2sxKgJ+sNzcNJCfiCC0J3pZFvCr5o2gPXlPV5vT2502vQVMD78lZ5OZIT+g03ob/ W/eaBBm/2j7WV1FgeHqYGFD6YhOPEkQPP/Cuj8FXWRFfSEY4G0UXY5HEL4Rz2VlLwitBk5mchfd +1K4HzU4GeUNwk57xvT+M/jPNXOMcTwNU66BkOE+ugUy62hqphzOrteAUvaBvJ/oBHnSVbHWWXI E4If+pl0fuOIFigU1cPXyTud6dJBfQ+QILBMmBT8LfTDH/GGXomCfNoJF7gr6w8kUb+V3ve6cjk JZjbdLr/+nXzzBOwfp0lLfTZMRm2TWBXF8V+D9jlbv X-Received: by 2002:a05:6a00:9083:b0:824:3ef6:a815 with SMTP id d2e1a72fcca58-8295d87b143mr377487b3a.8.1772500123378; Mon, 02 Mar 2026 17:08:43 -0800 (PST) From: Richard Henderson To: qemu-devel@nongnu.org Cc: pbonzini@redhat.com, jim.macarthur@linaro.org Subject: [PATCH v3 2/7] tcg/optimize: Lower unsupported deposit during optimize Date: Tue, 3 Mar 2026 12:08:27 +1100 Message-ID: <20260303010833.1115741-3-richard.henderson@linaro.org> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20260303010833.1115741-1-richard.henderson@linaro.org> References: <20260303010833.1115741-1-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::42f; envelope-from=richard.henderson@linaro.org; helo=mail-pf1-x42f.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: qemu development 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: 1772500168501139100 Content-Type: text/plain; charset="utf-8" The expansions that we chose in tcg-op.c may be less than optimial. Delay lowering until optimize, so that we have propagated constants and have computed known zero/one masks. Reviewed-by: Jim MacArthur Signed-off-by: Richard Henderson --- tcg/optimize.c | 179 +++++++++++++++++++++++++++++++++++++++++++------ tcg/tcg-op.c | 83 ++--------------------- 2 files changed, 163 insertions(+), 99 deletions(-) diff --git a/tcg/optimize.c b/tcg/optimize.c index 801a0a2c68..42637c12fa 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -1652,12 +1652,17 @@ static bool fold_ctpop(OptContext *ctx, TCGOp *op) =20 static bool fold_deposit(OptContext *ctx, TCGOp *op) { - TempOptInfo *t1 =3D arg_info(op->args[1]); - TempOptInfo *t2 =3D arg_info(op->args[2]); + TCGArg ret =3D op->args[0]; + TCGArg arg1 =3D op->args[1]; + TCGArg arg2 =3D op->args[2]; int ofs =3D op->args[3]; int len =3D op->args[4]; - int width =3D 8 * tcg_type_size(ctx->type); - uint64_t z_mask, o_mask, s_mask; + TempOptInfo *t1 =3D arg_info(arg1); + TempOptInfo *t2 =3D arg_info(arg2); + int width; + uint64_t z_mask, o_mask, s_mask, type_mask, len_mask; + TCGOp *op2; + bool valid; =20 if (ti_is_const(t1) && ti_is_const(t2)) { return tcg_opt_gen_movi(ctx, op, op->args[0], @@ -1665,35 +1670,167 @@ static bool fold_deposit(OptContext *ctx, TCGOp *o= p) ti_const_val(t2))); } =20 - /* Inserting a value into zero at offset 0. */ - if (ti_is_const_val(t1, 0) && ofs =3D=3D 0) { - uint64_t mask =3D MAKE_64BIT_MASK(0, len); + width =3D 8 * tcg_type_size(ctx->type); + type_mask =3D MAKE_64BIT_MASK(0, width); + len_mask =3D MAKE_64BIT_MASK(0, len); =20 + /* Inserting all-zero into a value. */ + if ((t2->z_mask & len_mask) =3D=3D 0) { op->opc =3D INDEX_op_and; - op->args[1] =3D op->args[2]; - op->args[2] =3D arg_new_constant(ctx, mask); + op->args[2] =3D arg_new_constant(ctx, ~(len_mask << ofs)); return fold_and(ctx, op); } =20 - /* Inserting zero into a value. */ - if (ti_is_const_val(t2, 0)) { - uint64_t mask =3D deposit64(-1, ofs, len, 0); - - op->opc =3D INDEX_op_and; - op->args[2] =3D arg_new_constant(ctx, mask); - return fold_and(ctx, op); + /* Inserting all-one into a value. */ + if ((t2->o_mask & len_mask) =3D=3D len_mask) { + op->opc =3D INDEX_op_or; + op->args[2] =3D arg_new_constant(ctx, len_mask << ofs); + return fold_or(ctx, op); } =20 - /* The s_mask from the top portion of the deposit is still valid. */ - if (ofs + len =3D=3D width) { - s_mask =3D t2->s_mask << ofs; - } else { - s_mask =3D t1->s_mask & ~MAKE_64BIT_MASK(0, ofs + len); + valid =3D TCG_TARGET_deposit_valid(ctx->type, ofs, len); + + /* Lower invalid deposit of constant as AND + OR. */ + if (!valid && ti_is_const(t2)) { + uint64_t ins_val =3D (ti_const_val(t2) & len_mask) << ofs; + + op2 =3D opt_insert_before(ctx, op, INDEX_op_and, 3); + op2->args[0] =3D ret; + op2->args[1] =3D arg1; + op2->args[2] =3D arg_new_constant(ctx, ~(len_mask << ofs)); + fold_and(ctx, op2); + + op->opc =3D INDEX_op_or; + op->args[1] =3D ret; + op->args[2] =3D arg_new_constant(ctx, ins_val); + return fold_or(ctx, op); } =20 + /* + * Compute result masks before calling other fold_* subroutines + * which could modify the masks of our inputs. + */ z_mask =3D deposit64(t1->z_mask, ofs, len, t2->z_mask); o_mask =3D deposit64(t1->o_mask, ofs, len, t2->o_mask); + if (ofs + len < width) { + s_mask =3D t1->s_mask & ~MAKE_64BIT_MASK(0, ofs + len); + } else { + s_mask =3D t2->s_mask << ofs; + } =20 + /* Inserting a value into zero. */ + if (ti_is_const_val(t1, 0)) { + uint64_t need_mask; + + /* Always lower deposit into zero at 0 as AND. */ + if (ofs =3D=3D 0) { + op->opc =3D INDEX_op_and; + op->args[1] =3D arg2; + op->args[2] =3D arg_new_constant(ctx, len_mask); + return fold_and(ctx, op); + } + + /* + * If the portion of the value outside len that remains after + * shifting is zero, we can elide the mask and just shift. + */ + need_mask =3D t2->z_mask & ~len_mask; + need_mask =3D (need_mask << ofs) & type_mask; + if (!need_mask) { + op->opc =3D INDEX_op_shl; + op->args[1] =3D arg2; + op->args[2] =3D arg_new_constant(ctx, ofs); + goto done; + } + + /* Lower invalid deposit into zero as AND + SHL. */ + if (!valid) { + op2 =3D opt_insert_before(ctx, op, INDEX_op_and, 3); + op2->args[0] =3D ret; + op2->args[1] =3D arg2; + op2->args[2] =3D arg_new_constant(ctx, len_mask); + fold_and(ctx, op2); + + op->opc =3D INDEX_op_shl; + op->args[1] =3D ret; + op->args[2] =3D arg_new_constant(ctx, ofs); + goto done; + } + } + + /* After special cases, lower invalid deposit. */ + if (!valid) { + TCGArg tmp; + + if (tcg_op_supported(INDEX_op_extract2, ctx->type, 0)) { + if (ofs =3D=3D 0 && tcg_op_supported(INDEX_op_rotl, ctx->type,= 0)) { + /* + * ret =3D arg2:arg1 >> len + * ret =3D rotl(ret, len) + */ + op2 =3D opt_insert_before(ctx, op, INDEX_op_extract2, 4); + op2->args[0] =3D ret; + op2->args[1] =3D arg1; + op2->args[2] =3D arg2; + op2->args[3] =3D len; + + op->opc =3D INDEX_op_rotl; + op->args[1] =3D ret; + op->args[2] =3D arg_new_constant(ctx, len); + goto done; + } + if (ofs + len =3D=3D width) { + /* + * tmp =3D arg1 << len + * ret =3D arg2:tmp >> len + */ + tmp =3D ret =3D=3D arg2 ? arg_new_temp(ctx) : ret; + + op2 =3D opt_insert_before(ctx, op, INDEX_op_shl, 4); + op2->args[0] =3D tmp; + op2->args[1] =3D arg1; + op2->args[2] =3D arg_new_constant(ctx, len); + + op->opc =3D INDEX_op_extract2; + op->args[0] =3D ret; + op->args[1] =3D tmp; + op->args[2] =3D arg2; + op->args[3] =3D len; + goto done; + } + } + + /* + * tmp =3D arg2 & mask + * ret =3D arg1 & ~(mask << ofs) + * tmp =3D tmp << ofs + * ret =3D ret | tmp + */ + tmp =3D arg_new_temp(ctx); + + op2 =3D opt_insert_before(ctx, op, INDEX_op_and, 3); + op2->args[0] =3D tmp; + op2->args[1] =3D arg2; + op2->args[2] =3D arg_new_constant(ctx, len_mask); + fold_and(ctx, op2); + + op2 =3D opt_insert_before(ctx, op, INDEX_op_shl, 3); + op2->args[0] =3D tmp; + op2->args[1] =3D tmp; + op2->args[2] =3D arg_new_constant(ctx, ofs); + + op2 =3D opt_insert_before(ctx, op, INDEX_op_and, 3); + op2->args[0] =3D ret; + op2->args[1] =3D arg1; + op2->args[2] =3D arg_new_constant(ctx, ~(len_mask << ofs)); + fold_and(ctx, op2); + + op->opc =3D INDEX_op_or; + op->args[1] =3D ret; + op->args[2] =3D tmp; + } + + done: return fold_masks_zos(ctx, op, z_mask, o_mask, s_mask); } =20 diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c index b95b07efb5..96f72ba381 100644 --- a/tcg/tcg-op.c +++ b/tcg/tcg-op.c @@ -876,9 +876,6 @@ void tcg_gen_rotri_i32(TCGv_i32 ret, TCGv_i32 arg1, int= 32_t arg2) void tcg_gen_deposit_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2, unsigned int ofs, unsigned int len) { - uint32_t mask; - TCGv_i32 t1; - tcg_debug_assert(ofs < 32); tcg_debug_assert(len > 0); tcg_debug_assert(len <=3D 32); @@ -886,39 +883,9 @@ void tcg_gen_deposit_i32(TCGv_i32 ret, TCGv_i32 arg1, = TCGv_i32 arg2, =20 if (len =3D=3D 32) { tcg_gen_mov_i32(ret, arg2); - return; - } - if (TCG_TARGET_deposit_valid(TCG_TYPE_I32, ofs, len)) { - tcg_gen_op5ii_i32(INDEX_op_deposit, ret, arg1, arg2, ofs, len); - return; - } - - t1 =3D tcg_temp_ebb_new_i32(); - - if (tcg_op_supported(INDEX_op_extract2, TCG_TYPE_I32, 0)) { - if (ofs + len =3D=3D 32) { - tcg_gen_shli_i32(t1, arg1, len); - tcg_gen_extract2_i32(ret, t1, arg2, len); - goto done; - } - if (ofs =3D=3D 0) { - tcg_gen_extract2_i32(ret, arg1, arg2, len); - tcg_gen_rotli_i32(ret, ret, len); - goto done; - } - } - - mask =3D (1u << len) - 1; - if (ofs + len < 32) { - tcg_gen_andi_i32(t1, arg2, mask); - tcg_gen_shli_i32(t1, t1, ofs); } else { - tcg_gen_shli_i32(t1, arg2, ofs); + tcg_gen_op5ii_i32(INDEX_op_deposit, ret, arg1, arg2, ofs, len); } - tcg_gen_andi_i32(ret, arg1, ~(mask << ofs)); - tcg_gen_or_i32(ret, ret, t1); - done: - tcg_temp_free_i32(t1); } =20 void tcg_gen_deposit_z_i32(TCGv_i32 ret, TCGv_i32 arg, @@ -932,13 +899,10 @@ void tcg_gen_deposit_z_i32(TCGv_i32 ret, TCGv_i32 arg, if (ofs + len =3D=3D 32) { tcg_gen_shli_i32(ret, arg, ofs); } else if (ofs =3D=3D 0) { - tcg_gen_andi_i32(ret, arg, (1u << len) - 1); - } else if (TCG_TARGET_deposit_valid(TCG_TYPE_I32, ofs, len)) { + tcg_gen_extract_i32(ret, arg, 0, len); + } else { TCGv_i32 zero =3D tcg_constant_i32(0); tcg_gen_op5ii_i32(INDEX_op_deposit, ret, zero, arg, ofs, len); - } else { - tcg_gen_andi_i32(ret, arg, (1u << len) - 1); - tcg_gen_shli_i32(ret, ret, ofs); } } =20 @@ -2133,9 +2097,6 @@ void tcg_gen_rotri_i64(TCGv_i64 ret, TCGv_i64 arg1, i= nt64_t arg2) void tcg_gen_deposit_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2, unsigned int ofs, unsigned int len) { - uint64_t mask; - TCGv_i64 t1; - tcg_debug_assert(ofs < 64); tcg_debug_assert(len > 0); tcg_debug_assert(len <=3D 64); @@ -2143,40 +2104,9 @@ void tcg_gen_deposit_i64(TCGv_i64 ret, TCGv_i64 arg1= , TCGv_i64 arg2, =20 if (len =3D=3D 64) { tcg_gen_mov_i64(ret, arg2); - return; - } - - if (TCG_TARGET_deposit_valid(TCG_TYPE_I64, ofs, len)) { - tcg_gen_op5ii_i64(INDEX_op_deposit, ret, arg1, arg2, ofs, len); - return; - } - - t1 =3D tcg_temp_ebb_new_i64(); - - if (tcg_op_supported(INDEX_op_extract2, TCG_TYPE_I64, 0)) { - if (ofs + len =3D=3D 64) { - tcg_gen_shli_i64(t1, arg1, len); - tcg_gen_extract2_i64(ret, t1, arg2, len); - goto done; - } - if (ofs =3D=3D 0) { - tcg_gen_extract2_i64(ret, arg1, arg2, len); - tcg_gen_rotli_i64(ret, ret, len); - goto done; - } - } - - mask =3D (1ull << len) - 1; - if (ofs + len < 64) { - tcg_gen_andi_i64(t1, arg2, mask); - tcg_gen_shli_i64(t1, t1, ofs); } else { - tcg_gen_shli_i64(t1, arg2, ofs); + tcg_gen_op5ii_i64(INDEX_op_deposit, ret, arg1, arg2, ofs, len); } - tcg_gen_andi_i64(ret, arg1, ~(mask << ofs)); - tcg_gen_or_i64(ret, ret, t1); - done: - tcg_temp_free_i64(t1); } =20 void tcg_gen_deposit_z_i64(TCGv_i64 ret, TCGv_i64 arg, @@ -2191,12 +2121,9 @@ void tcg_gen_deposit_z_i64(TCGv_i64 ret, TCGv_i64 ar= g, tcg_gen_shli_i64(ret, arg, ofs); } else if (ofs =3D=3D 0) { tcg_gen_andi_i64(ret, arg, (1ull << len) - 1); - } else if (TCG_TARGET_deposit_valid(TCG_TYPE_I64, ofs, len)) { + } else { TCGv_i64 zero =3D tcg_constant_i64(0); tcg_gen_op5ii_i64(INDEX_op_deposit, ret, zero, arg, ofs, len); - } else { - tcg_gen_andi_i64(ret, arg, (1ull << len) - 1); - tcg_gen_shli_i64(ret, ret, ofs); } } =20 --=20 2.43.0