From nobody Fri Nov 14 16:50:29 2025 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=gmail.com ARC-Seal: i=1; a=rsa-sha256; t=1761876975; cv=none; d=zohomail.com; s=zohoarc; b=Yk9N+IDtFMAszUGpG73bk+YMiT3jFGb8Sj+ADxMnFAcc5ZuXq6xYrAiIVAmrhAwU2ekLQ1gg1hEp7L/qC+RL5KHUWkSiOf67JeWRWUM1L5/1kBwXrzLlQfw28Dk6n/Mw2TSIs1ZwY9xHvyxBcM0/9CClVi0S2o4ke4nL53qGrTw= ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=zohomail.com; s=zohoarc; t=1761876975; h=Content-Type: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=kHBh4sfMHS3WxddSSw4wXbBZtVAKEG+HZPV0N4S5SJU=; b=crl0PLvyAXv8plJKMWJpuF22uP3NwSerzdl3YYWl+OtqjY4aG60Ih5iN6ZmjhsxQg/3g1HhynSwSRPFFOwSn+xqBPt82/hmcDazunKsn1x7t4fjgLcNBqVlQZAP3MOUqqWYNHKiQInUU6k6Xl9ykYZcv0Pwy53J8m3WydodItl8= 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 1761876975807301.84777376441616; Thu, 30 Oct 2025 19:16:15 -0700 (PDT) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1vEeew-0005FE-13; Thu, 30 Oct 2025 22:14:34 -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 1vEeeu-0005EF-LK for qemu-devel@nongnu.org; Thu, 30 Oct 2025 22:14:32 -0400 Received: from mail-il1-x132.google.com ([2607:f8b0:4864:20::132]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_128_GCM_SHA256:128) (Exim 4.90_1) (envelope-from ) id 1vEeek-0005vN-71 for qemu-devel@nongnu.org; Thu, 30 Oct 2025 22:14:32 -0400 Received: by mail-il1-x132.google.com with SMTP id e9e14a558f8ab-430abca3354so15829195ab.2 for ; Thu, 30 Oct 2025 19:14:20 -0700 (PDT) Received: from taylor-ubuntu.. (c-67-172-136-152.hsd1.co.comcast.net. [67.172.136.152]) by smtp.gmail.com with ESMTPSA id e9e14a558f8ab-43310274e21sm1631055ab.9.2025.10.30.19.14.14 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 30 Oct 2025 19:14:16 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1761876858; x=1762481658; 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=kHBh4sfMHS3WxddSSw4wXbBZtVAKEG+HZPV0N4S5SJU=; b=ac0zpk+rHxUCvORhMNycoC2DotsoeKUL3s5hMC34M6rEtNg7vtmXPYr9GP/piXiK+x 7R0fERQI74Qqw9myXJXXBOjfqpM4my0zzOjsit62F+qFV2sAwZijdox3CCqc2nX5T38A Rhr22q/xggWAKbQNdceFBIVFGhBa6zuC+lvY5qN9C6J1as8e2eG2yZRvjZGgg5S3qu7G usDmkX0I7T+Gw82Rr3j758UyofGWT0lloimSrXDZBcTsrej3S+xq/JHb8d5+RXYg2kCj DSM5B0YZsDSbkj0onf5KLE+s/j//VAi4ePSNDdzP+ZfexJfG4pT03BOQmNSPBtfhbLRd 9KKA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1761876858; x=1762481658; 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=kHBh4sfMHS3WxddSSw4wXbBZtVAKEG+HZPV0N4S5SJU=; b=HtWtSgSjAlpCv9CjLyNlsusSX6EZlxk/ePOSRpoQcfivYCrbq+Zy7WAE9R2ixHe4Tu aeUJkJCeCoa4Jr5/rF3owMPUSrXcSoshkehVd3xwI4Fh6QVTyE565ocKLQS07gehnzJs jtEffHkpnq3yf258jxyemAavEfn/cvoLbGAvF+IUcPw1eQ+SIiiyd9YvZHycHLituluB sFmBsHeGsQlyduX+3qMiPXQQxWb0DmbToPTZQDWbS5k63fepMnkE3MjjYkIwClwpfKcw 9mIIhW8eImar4WbxX08S5MeI+iBNRBYhX3rA9bpL+Ygyw1jTj+SDG3tsIGz/DbLNTHuX URbQ== X-Gm-Message-State: AOJu0Yyj+HfBFagrWAOaNz4U0FJKJP03Cuyx+P13ZACgM+KX5vG+bdm5 bLtsdTYHrbw6Ayb2ymvG6hVg81gm1jL8sRbU/hC4Y42GWILr6NcoSmsqJTt4c/gI9i8= X-Gm-Gg: ASbGnctZPBuVP6pwr6QEObeSmAxM/BR0THObRckKdSJ0JdgW0BhNCJBK5IfoqxmClm6 4preII/XxraBHxcc2OfZ8qUPnsP9FXgo+mNXM85WZ1QcisKAvy+UkK0cGSbIPJeVJzcsloKplNO Qpfx6DsuLWD1QHZetbqORLttnc32K/kTugWBeDX/HfKatGayQgu69/KM7G+VFEglHJmlHjGXAiB hsyERrHQ2ClsozloSGPw7MBaTFifXlCNElgyUQbo8l2umuIAipt6EwBzvKFRO4EsnciUFpodz1A vjnGBfYdsr9hNTf48iAbkjqpsvTAgQ4HAmf48ru5cwOmTYsO6xSM71STsCt8qXcSZRCye/Vai8Z 8gQq5H9noqj7pYiPDM7BVvJeu1GLfTNCAvFN1+MElaSHbDxa94br2YiBUpglrocofBgPZUUeTg7 n4311NiNjRzVqVRgkllM9njoP8zGFgO84OO7Thl8FvUjn3RvepqGbdxq5yk6QYc2YcQA== X-Google-Smtp-Source: AGHT+IE//aCgl7j94hWzj/ZdRmq1GEPtphVAfqFnu4umviUlRNtl6d9lFjgzrlTY8ppXpiqzX2DEag== X-Received: by 2002:a05:6e02:1d9d:b0:42d:86f3:a2d1 with SMTP id e9e14a558f8ab-4330d1c9dc2mr30966255ab.25.1761876857600; Thu, 30 Oct 2025 19:14:17 -0700 (PDT) From: Taylor Simpson To: qemu-devel@nongnu.org Cc: brian.cain@oss.qualcomm.com, matheus.bernardino@oss.qualcomm.com, sid.manning@oss.qualcomm.com, marco.liebel@oss.qualcomm.com, richard.henderson@linaro.org, philmd@linaro.org, ale@rev.ng, anjo@rev.ng, ltaylorsimpson@gmail.com Subject: [PATCH 1/2] Hexagon (target/hexagon) Add pkt_need_commit argument to arch_fpop_end Date: Thu, 30 Oct 2025 20:14:10 -0600 Message-ID: <20251031021411.111365-2-ltaylorsimpson@gmail.com> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20251031021411.111365-1-ltaylorsimpson@gmail.com> References: <20251031021411.111365-1-ltaylorsimpson@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" 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::132; envelope-from=ltaylorsimpson@gmail.com; helo=mail-il1-x132.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, FREEMAIL_FROM=0.001, 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 @gmail.com) X-ZM-MESSAGEID: 1761876977870158500 The arch_fpop_end function converts the softfloat fp_status flags to bits set in Hexagon USR. It is hard-coded that the packet must need a commit. We add an argument to the function and update all the helper call sites. In a subsequent commit, we will change the code that forces a commit when the packet contains an floating point instruction. Signed-off-by: Taylor Simpson --- target/hexagon/arch.h | 2 +- target/hexagon/gen_tcg.h | 114 ++++++------ target/hexagon/helper.h | 114 ++++++------ target/hexagon/arch.c | 3 +- target/hexagon/op_helper.c | 287 +++++++++++++++++------------ target/hexagon/gen_helper_funcs.py | 8 +- target/hexagon/gen_tcg_funcs.py | 4 + 7 files changed, 294 insertions(+), 238 deletions(-) diff --git a/target/hexagon/arch.h b/target/hexagon/arch.h index 70918065d3..c295a86600 100644 --- a/target/hexagon/arch.h +++ b/target/hexagon/arch.h @@ -28,7 +28,7 @@ uint64_t interleave(uint32_t odd, uint32_t even); uint64_t deinterleave(uint64_t src); int32_t conv_round(int32_t a, int n); void arch_fpop_start(CPUHexagonState *env); -void arch_fpop_end(CPUHexagonState *env); +void arch_fpop_end(CPUHexagonState *env, bool pkt_need_commit); int arch_sf_recip_common(float32 *Rs, float32 *Rt, float32 *Rd, int *adjust, float_status *fp_status); int arch_sf_invsqrt_common(float32 *Rs, float32 *Rd, int *adjust, diff --git a/target/hexagon/gen_tcg.h b/target/hexagon/gen_tcg.h index 8a3b801287..f17eac2f1a 100644 --- a/target/hexagon/gen_tcg.h +++ b/target/hexagon/gen_tcg.h @@ -613,7 +613,7 @@ #define fGEN_TCG_F2_sfrecipa(SHORTCODE) \ do { \ TCGv_i64 tmp =3D tcg_temp_new_i64(); \ - gen_helper_sfrecipa(tmp, tcg_env, RsV, RtV); \ + gen_helper_sfrecipa(tmp, tcg_env, RsV, RtV, pkt_need_commit); \ tcg_gen_extrh_i64_i32(RdV, tmp); \ tcg_gen_extrl_i64_i32(PeV, tmp); \ } while (0) @@ -628,7 +628,7 @@ #define fGEN_TCG_F2_sfinvsqrta(SHORTCODE) \ do { \ TCGv_i64 tmp =3D tcg_temp_new_i64(); \ - gen_helper_sfinvsqrta(tmp, tcg_env, RsV); \ + gen_helper_sfinvsqrta(tmp, tcg_env, RsV, pkt_need_commit); \ tcg_gen_extrh_i64_i32(RdV, tmp); \ tcg_gen_extrl_i64_i32(PeV, tmp); \ } while (0) @@ -1204,122 +1204,122 @@ =20 /* Floating point */ #define fGEN_TCG_F2_conv_sf2df(SHORTCODE) \ - gen_helper_conv_sf2df(RddV, tcg_env, RsV) + gen_helper_conv_sf2df(RddV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_df2sf(SHORTCODE) \ - gen_helper_conv_df2sf(RdV, tcg_env, RssV) + gen_helper_conv_df2sf(RdV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_uw2sf(SHORTCODE) \ - gen_helper_conv_uw2sf(RdV, tcg_env, RsV) + gen_helper_conv_uw2sf(RdV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_uw2df(SHORTCODE) \ - gen_helper_conv_uw2df(RddV, tcg_env, RsV) + gen_helper_conv_uw2df(RddV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_w2sf(SHORTCODE) \ - gen_helper_conv_w2sf(RdV, tcg_env, RsV) + gen_helper_conv_w2sf(RdV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_w2df(SHORTCODE) \ - gen_helper_conv_w2df(RddV, tcg_env, RsV) + gen_helper_conv_w2df(RddV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_ud2sf(SHORTCODE) \ - gen_helper_conv_ud2sf(RdV, tcg_env, RssV) + gen_helper_conv_ud2sf(RdV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_ud2df(SHORTCODE) \ - gen_helper_conv_ud2df(RddV, tcg_env, RssV) + gen_helper_conv_ud2df(RddV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_d2sf(SHORTCODE) \ - gen_helper_conv_d2sf(RdV, tcg_env, RssV) + gen_helper_conv_d2sf(RdV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_d2df(SHORTCODE) \ - gen_helper_conv_d2df(RddV, tcg_env, RssV) + gen_helper_conv_d2df(RddV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_sf2uw(SHORTCODE) \ - gen_helper_conv_sf2uw(RdV, tcg_env, RsV) + gen_helper_conv_sf2uw(RdV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_sf2w(SHORTCODE) \ - gen_helper_conv_sf2w(RdV, tcg_env, RsV) + gen_helper_conv_sf2w(RdV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_sf2ud(SHORTCODE) \ - gen_helper_conv_sf2ud(RddV, tcg_env, RsV) + gen_helper_conv_sf2ud(RddV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_sf2d(SHORTCODE) \ - gen_helper_conv_sf2d(RddV, tcg_env, RsV) + gen_helper_conv_sf2d(RddV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_df2uw(SHORTCODE) \ - gen_helper_conv_df2uw(RdV, tcg_env, RssV) + gen_helper_conv_df2uw(RdV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_df2w(SHORTCODE) \ - gen_helper_conv_df2w(RdV, tcg_env, RssV) + gen_helper_conv_df2w(RdV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_df2ud(SHORTCODE) \ - gen_helper_conv_df2ud(RddV, tcg_env, RssV) + gen_helper_conv_df2ud(RddV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_df2d(SHORTCODE) \ - gen_helper_conv_df2d(RddV, tcg_env, RssV) + gen_helper_conv_df2d(RddV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_sf2uw_chop(SHORTCODE) \ - gen_helper_conv_sf2uw_chop(RdV, tcg_env, RsV) + gen_helper_conv_sf2uw_chop(RdV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_sf2w_chop(SHORTCODE) \ - gen_helper_conv_sf2w_chop(RdV, tcg_env, RsV) + gen_helper_conv_sf2w_chop(RdV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_sf2ud_chop(SHORTCODE) \ - gen_helper_conv_sf2ud_chop(RddV, tcg_env, RsV) + gen_helper_conv_sf2ud_chop(RddV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_sf2d_chop(SHORTCODE) \ - gen_helper_conv_sf2d_chop(RddV, tcg_env, RsV) + gen_helper_conv_sf2d_chop(RddV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_conv_df2uw_chop(SHORTCODE) \ - gen_helper_conv_df2uw_chop(RdV, tcg_env, RssV) + gen_helper_conv_df2uw_chop(RdV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_df2w_chop(SHORTCODE) \ - gen_helper_conv_df2w_chop(RdV, tcg_env, RssV) + gen_helper_conv_df2w_chop(RdV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_df2ud_chop(SHORTCODE) \ - gen_helper_conv_df2ud_chop(RddV, tcg_env, RssV) + gen_helper_conv_df2ud_chop(RddV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_conv_df2d_chop(SHORTCODE) \ - gen_helper_conv_df2d_chop(RddV, tcg_env, RssV) + gen_helper_conv_df2d_chop(RddV, tcg_env, RssV, pkt_need_commit) #define fGEN_TCG_F2_sfadd(SHORTCODE) \ - gen_helper_sfadd(RdV, tcg_env, RsV, RtV) + gen_helper_sfadd(RdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sfsub(SHORTCODE) \ - gen_helper_sfsub(RdV, tcg_env, RsV, RtV) + gen_helper_sfsub(RdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sfcmpeq(SHORTCODE) \ - gen_helper_sfcmpeq(PdV, tcg_env, RsV, RtV) + gen_helper_sfcmpeq(PdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sfcmpgt(SHORTCODE) \ - gen_helper_sfcmpgt(PdV, tcg_env, RsV, RtV) + gen_helper_sfcmpgt(PdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sfcmpge(SHORTCODE) \ - gen_helper_sfcmpge(PdV, tcg_env, RsV, RtV) + gen_helper_sfcmpge(PdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sfcmpuo(SHORTCODE) \ - gen_helper_sfcmpuo(PdV, tcg_env, RsV, RtV) + gen_helper_sfcmpuo(PdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sfmax(SHORTCODE) \ - gen_helper_sfmax(RdV, tcg_env, RsV, RtV) + gen_helper_sfmax(RdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sfmin(SHORTCODE) \ - gen_helper_sfmin(RdV, tcg_env, RsV, RtV) + gen_helper_sfmin(RdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sfclass(SHORTCODE) \ do { \ TCGv imm =3D tcg_constant_tl(uiV); \ - gen_helper_sfclass(PdV, tcg_env, RsV, imm); \ + gen_helper_sfclass(PdV, tcg_env, RsV, imm, pkt_need_commit); \ } while (0) #define fGEN_TCG_F2_sffixupn(SHORTCODE) \ - gen_helper_sffixupn(RdV, tcg_env, RsV, RtV) + gen_helper_sffixupn(RdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sffixupd(SHORTCODE) \ - gen_helper_sffixupd(RdV, tcg_env, RsV, RtV) + gen_helper_sffixupd(RdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sffixupr(SHORTCODE) \ - gen_helper_sffixupr(RdV, tcg_env, RsV) + gen_helper_sffixupr(RdV, tcg_env, RsV, pkt_need_commit) #define fGEN_TCG_F2_dfadd(SHORTCODE) \ - gen_helper_dfadd(RddV, tcg_env, RssV, RttV) + gen_helper_dfadd(RddV, tcg_env, RssV, RttV, pkt_need_commit) #define fGEN_TCG_F2_dfsub(SHORTCODE) \ - gen_helper_dfsub(RddV, tcg_env, RssV, RttV) + gen_helper_dfsub(RddV, tcg_env, RssV, RttV, pkt_need_commit) #define fGEN_TCG_F2_dfmax(SHORTCODE) \ - gen_helper_dfmax(RddV, tcg_env, RssV, RttV) + gen_helper_dfmax(RddV, tcg_env, RssV, RttV, pkt_need_commit) #define fGEN_TCG_F2_dfmin(SHORTCODE) \ - gen_helper_dfmin(RddV, tcg_env, RssV, RttV) + gen_helper_dfmin(RddV, tcg_env, RssV, RttV, pkt_need_commit) #define fGEN_TCG_F2_dfcmpeq(SHORTCODE) \ - gen_helper_dfcmpeq(PdV, tcg_env, RssV, RttV) + gen_helper_dfcmpeq(PdV, tcg_env, RssV, RttV, pkt_need_commit) #define fGEN_TCG_F2_dfcmpgt(SHORTCODE) \ - gen_helper_dfcmpgt(PdV, tcg_env, RssV, RttV) + gen_helper_dfcmpgt(PdV, tcg_env, RssV, RttV, pkt_need_commit) #define fGEN_TCG_F2_dfcmpge(SHORTCODE) \ - gen_helper_dfcmpge(PdV, tcg_env, RssV, RttV) + gen_helper_dfcmpge(PdV, tcg_env, RssV, RttV, pkt_need_commit) #define fGEN_TCG_F2_dfcmpuo(SHORTCODE) \ - gen_helper_dfcmpuo(PdV, tcg_env, RssV, RttV) + gen_helper_dfcmpuo(PdV, tcg_env, RssV, RttV, pkt_need_commit) #define fGEN_TCG_F2_dfclass(SHORTCODE) \ do { \ TCGv imm =3D tcg_constant_tl(uiV); \ - gen_helper_dfclass(PdV, tcg_env, RssV, imm); \ + gen_helper_dfclass(PdV, tcg_env, RssV, imm, pkt_need_commit); \ } while (0) #define fGEN_TCG_F2_sfmpy(SHORTCODE) \ - gen_helper_sfmpy(RdV, tcg_env, RsV, RtV) + gen_helper_sfmpy(RdV, tcg_env, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sffma(SHORTCODE) \ - gen_helper_sffma(RxV, tcg_env, RxV, RsV, RtV) + gen_helper_sffma(RxV, tcg_env, RxV, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sffma_sc(SHORTCODE) \ - gen_helper_sffma_sc(RxV, tcg_env, RxV, RsV, RtV, PuV) + gen_helper_sffma_sc(RxV, tcg_env, RxV, RsV, RtV, PuV, pkt_need_commit) #define fGEN_TCG_F2_sffms(SHORTCODE) \ - gen_helper_sffms(RxV, tcg_env, RxV, RsV, RtV) + gen_helper_sffms(RxV, tcg_env, RxV, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sffma_lib(SHORTCODE) \ - gen_helper_sffma_lib(RxV, tcg_env, RxV, RsV, RtV) + gen_helper_sffma_lib(RxV, tcg_env, RxV, RsV, RtV, pkt_need_commit) #define fGEN_TCG_F2_sffms_lib(SHORTCODE) \ - gen_helper_sffms_lib(RxV, tcg_env, RxV, RsV, RtV) + gen_helper_sffms_lib(RxV, tcg_env, RxV, RsV, RtV, pkt_need_commit) =20 #define fGEN_TCG_F2_dfmpyfix(SHORTCODE) \ - gen_helper_dfmpyfix(RddV, tcg_env, RssV, RttV) + gen_helper_dfmpyfix(RddV, tcg_env, RssV, RttV, pkt_need_commit) #define fGEN_TCG_F2_dfmpyhh(SHORTCODE) \ - gen_helper_dfmpyhh(RxxV, tcg_env, RxxV, RssV, RttV) + gen_helper_dfmpyhh(RxxV, tcg_env, RxxV, RssV, RttV, pkt_need_commit) =20 /* Nothing to do for these in qemu, need to suppress compiler warnings */ #define fGEN_TCG_Y4_l2fetch(SHORTCODE) \ diff --git a/target/hexagon/helper.h b/target/hexagon/helper.h index f8baa599c8..64456822bc 100644 --- a/target/hexagon/helper.h +++ b/target/hexagon/helper.h @@ -24,72 +24,72 @@ DEF_HELPER_3(gather_store, void, env, i32, int) DEF_HELPER_1(commit_hvx_stores, void, env) DEF_HELPER_FLAGS_4(fcircadd, TCG_CALL_NO_RWG_SE, s32, s32, s32, s32, s32) DEF_HELPER_FLAGS_1(fbrev, TCG_CALL_NO_RWG_SE, i32, i32) -DEF_HELPER_3(sfrecipa, i64, env, f32, f32) -DEF_HELPER_2(sfinvsqrta, i64, env, f32) +DEF_HELPER_4(sfrecipa, i64, env, f32, f32, i32) +DEF_HELPER_3(sfinvsqrta, i64, env, f32, i32) DEF_HELPER_5(vacsh_val, s64, env, s64, s64, s64, i32) DEF_HELPER_FLAGS_4(vacsh_pred, TCG_CALL_NO_RWG_SE, s32, env, s64, s64, s64) DEF_HELPER_FLAGS_2(cabacdecbin_val, TCG_CALL_NO_RWG_SE, s64, s64, s64) DEF_HELPER_FLAGS_2(cabacdecbin_pred, TCG_CALL_NO_RWG_SE, s32, s64, s64) =20 /* Floating point */ -DEF_HELPER_2(conv_sf2df, f64, env, f32) -DEF_HELPER_2(conv_df2sf, f32, env, f64) -DEF_HELPER_2(conv_uw2sf, f32, env, s32) -DEF_HELPER_2(conv_uw2df, f64, env, s32) -DEF_HELPER_2(conv_w2sf, f32, env, s32) -DEF_HELPER_2(conv_w2df, f64, env, s32) -DEF_HELPER_2(conv_ud2sf, f32, env, s64) -DEF_HELPER_2(conv_ud2df, f64, env, s64) -DEF_HELPER_2(conv_d2sf, f32, env, s64) -DEF_HELPER_2(conv_d2df, f64, env, s64) -DEF_HELPER_2(conv_sf2uw, i32, env, f32) -DEF_HELPER_2(conv_sf2w, s32, env, f32) -DEF_HELPER_2(conv_sf2ud, i64, env, f32) -DEF_HELPER_2(conv_sf2d, s64, env, f32) -DEF_HELPER_2(conv_df2uw, i32, env, f64) -DEF_HELPER_2(conv_df2w, s32, env, f64) -DEF_HELPER_2(conv_df2ud, i64, env, f64) -DEF_HELPER_2(conv_df2d, s64, env, f64) -DEF_HELPER_2(conv_sf2uw_chop, i32, env, f32) -DEF_HELPER_2(conv_sf2w_chop, s32, env, f32) -DEF_HELPER_2(conv_sf2ud_chop, i64, env, f32) -DEF_HELPER_2(conv_sf2d_chop, s64, env, f32) -DEF_HELPER_2(conv_df2uw_chop, i32, env, f64) -DEF_HELPER_2(conv_df2w_chop, s32, env, f64) -DEF_HELPER_2(conv_df2ud_chop, i64, env, f64) -DEF_HELPER_2(conv_df2d_chop, s64, env, f64) -DEF_HELPER_3(sfadd, f32, env, f32, f32) -DEF_HELPER_3(sfsub, f32, env, f32, f32) -DEF_HELPER_3(sfcmpeq, s32, env, f32, f32) -DEF_HELPER_3(sfcmpgt, s32, env, f32, f32) -DEF_HELPER_3(sfcmpge, s32, env, f32, f32) -DEF_HELPER_3(sfcmpuo, s32, env, f32, f32) -DEF_HELPER_3(sfmax, f32, env, f32, f32) -DEF_HELPER_3(sfmin, f32, env, f32, f32) -DEF_HELPER_3(sfclass, s32, env, f32, s32) -DEF_HELPER_3(sffixupn, f32, env, f32, f32) -DEF_HELPER_3(sffixupd, f32, env, f32, f32) -DEF_HELPER_2(sffixupr, f32, env, f32) +DEF_HELPER_3(conv_sf2df, f64, env, f32, i32) +DEF_HELPER_3(conv_df2sf, f32, env, f64, i32) +DEF_HELPER_3(conv_uw2sf, f32, env, s32, i32) +DEF_HELPER_3(conv_uw2df, f64, env, s32, i32) +DEF_HELPER_3(conv_w2sf, f32, env, s32, i32) +DEF_HELPER_3(conv_w2df, f64, env, s32, i32) +DEF_HELPER_3(conv_ud2sf, f32, env, s64, i32) +DEF_HELPER_3(conv_ud2df, f64, env, s64, i32) +DEF_HELPER_3(conv_d2sf, f32, env, s64, i32) +DEF_HELPER_3(conv_d2df, f64, env, s64, i32) +DEF_HELPER_3(conv_sf2uw, i32, env, f32, i32) +DEF_HELPER_3(conv_sf2w, s32, env, f32, i32) +DEF_HELPER_3(conv_sf2ud, i64, env, f32, i32) +DEF_HELPER_3(conv_sf2d, s64, env, f32, i32) +DEF_HELPER_3(conv_df2uw, i32, env, f64, i32) +DEF_HELPER_3(conv_df2w, s32, env, f64, i32) +DEF_HELPER_3(conv_df2ud, i64, env, f64, i32) +DEF_HELPER_3(conv_df2d, s64, env, f64, i32) +DEF_HELPER_3(conv_sf2uw_chop, i32, env, f32, i32) +DEF_HELPER_3(conv_sf2w_chop, s32, env, f32, i32) +DEF_HELPER_3(conv_sf2ud_chop, i64, env, f32, i32) +DEF_HELPER_3(conv_sf2d_chop, s64, env, f32, i32) +DEF_HELPER_3(conv_df2uw_chop, i32, env, f64, i32) +DEF_HELPER_3(conv_df2w_chop, s32, env, f64, i32) +DEF_HELPER_3(conv_df2ud_chop, i64, env, f64, i32) +DEF_HELPER_3(conv_df2d_chop, s64, env, f64, i32) +DEF_HELPER_4(sfadd, f32, env, f32, f32, i32) +DEF_HELPER_4(sfsub, f32, env, f32, f32, i32) +DEF_HELPER_4(sfcmpeq, s32, env, f32, f32, i32) +DEF_HELPER_4(sfcmpgt, s32, env, f32, f32, i32) +DEF_HELPER_4(sfcmpge, s32, env, f32, f32, i32) +DEF_HELPER_4(sfcmpuo, s32, env, f32, f32, i32) +DEF_HELPER_4(sfmax, f32, env, f32, f32, i32) +DEF_HELPER_4(sfmin, f32, env, f32, f32, i32) +DEF_HELPER_4(sfclass, s32, env, f32, s32, i32) +DEF_HELPER_4(sffixupn, f32, env, f32, f32, i32) +DEF_HELPER_4(sffixupd, f32, env, f32, f32, i32) +DEF_HELPER_3(sffixupr, f32, env, f32, i32) =20 -DEF_HELPER_3(dfadd, f64, env, f64, f64) -DEF_HELPER_3(dfsub, f64, env, f64, f64) -DEF_HELPER_3(dfmax, f64, env, f64, f64) -DEF_HELPER_3(dfmin, f64, env, f64, f64) -DEF_HELPER_3(dfcmpeq, s32, env, f64, f64) -DEF_HELPER_3(dfcmpgt, s32, env, f64, f64) -DEF_HELPER_3(dfcmpge, s32, env, f64, f64) -DEF_HELPER_3(dfcmpuo, s32, env, f64, f64) -DEF_HELPER_3(dfclass, s32, env, f64, s32) +DEF_HELPER_4(dfadd, f64, env, f64, f64, i32) +DEF_HELPER_4(dfsub, f64, env, f64, f64, i32) +DEF_HELPER_4(dfmax, f64, env, f64, f64, i32) +DEF_HELPER_4(dfmin, f64, env, f64, f64, i32) +DEF_HELPER_4(dfcmpeq, s32, env, f64, f64, i32) +DEF_HELPER_4(dfcmpgt, s32, env, f64, f64, i32) +DEF_HELPER_4(dfcmpge, s32, env, f64, f64, i32) +DEF_HELPER_4(dfcmpuo, s32, env, f64, f64, i32) +DEF_HELPER_4(dfclass, s32, env, f64, s32, i32) =20 -DEF_HELPER_3(sfmpy, f32, env, f32, f32) -DEF_HELPER_4(sffma, f32, env, f32, f32, f32) -DEF_HELPER_5(sffma_sc, f32, env, f32, f32, f32, f32) -DEF_HELPER_4(sffms, f32, env, f32, f32, f32) -DEF_HELPER_4(sffma_lib, f32, env, f32, f32, f32) -DEF_HELPER_4(sffms_lib, f32, env, f32, f32, f32) +DEF_HELPER_4(sfmpy, f32, env, f32, f32, i32) +DEF_HELPER_5(sffma, f32, env, f32, f32, f32, i32) +DEF_HELPER_6(sffma_sc, f32, env, f32, f32, f32, f32, i32) +DEF_HELPER_5(sffms, f32, env, f32, f32, f32, i32) +DEF_HELPER_5(sffma_lib, f32, env, f32, f32, f32, i32) +DEF_HELPER_5(sffms_lib, f32, env, f32, f32, f32, i32) =20 -DEF_HELPER_3(dfmpyfix, f64, env, f64, f64) -DEF_HELPER_4(dfmpyhh, f64, env, f64, f64, f64) +DEF_HELPER_4(dfmpyfix, f64, env, f64, f64, i32) +DEF_HELPER_5(dfmpyhh, f64, env, f64, f64, f64, i32) =20 /* Histogram instructions */ DEF_HELPER_1(vhist, void, env) diff --git a/target/hexagon/arch.c b/target/hexagon/arch.c index d053d68487..e17e714a6a 100644 --- a/target/hexagon/arch.c +++ b/target/hexagon/arch.c @@ -222,9 +222,8 @@ void arch_fpop_start(CPUHexagonState *env) } \ } while (0) =20 -void arch_fpop_end(CPUHexagonState *env) +void arch_fpop_end(CPUHexagonState *env, bool pkt_need_commit) { - const bool pkt_need_commit =3D true; int flags =3D get_float_exception_flags(&env->fp_status); if (flags !=3D 0) { SOFTFLOAT_TEST_FLAG(float_flag_inexact, FPINPF, FPINPE); diff --git a/target/hexagon/op_helper.c b/target/hexagon/op_helper.c index e2e80ca7ef..554e7dd447 100644 --- a/target/hexagon/op_helper.c +++ b/target/hexagon/op_helper.c @@ -200,7 +200,8 @@ static float32 build_float32(uint8_t sign, uint32_t exp= , uint32_t mant) * Since helpers can only return a single value, we pack the two results * into a 64-bit value. */ -uint64_t HELPER(sfrecipa)(CPUHexagonState *env, float32 RsV, float32 RtV) +uint64_t HELPER(sfrecipa)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { int32_t PeV =3D 0; float32 RdV; @@ -217,11 +218,12 @@ uint64_t HELPER(sfrecipa)(CPUHexagonState *env, float= 32 RsV, float32 RtV) exp =3D SF_BIAS - (float32_getexp(RtV) - SF_BIAS) - 1; RdV =3D build_float32(extract32(RtV, 31, 1), exp, mant); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return ((uint64_t)RdV << 32) | PeV; } =20 -uint64_t HELPER(sfinvsqrta)(CPUHexagonState *env, float32 RsV) +uint64_t HELPER(sfinvsqrta)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { int PeV =3D 0; float32 RdV; @@ -238,7 +240,7 @@ uint64_t HELPER(sfinvsqrta)(CPUHexagonState *env, float= 32 RsV) exp =3D SF_BIAS - ((float32_getexp(RsV) - SF_BIAS) >> 1) - 1; RdV =3D build_float32(extract32(RsV, 31, 1), exp, mant); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return ((uint64_t)RdV << 32) | PeV; } =20 @@ -476,97 +478,108 @@ static void check_noshuf(CPUHexagonState *env, bool = pkt_has_scalar_store_s1, #endif =20 /* Floating point */ -float64 HELPER(conv_sf2df)(CPUHexagonState *env, float32 RsV) +float64 HELPER(conv_sf2df)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { float64 out_f64; arch_fpop_start(env); out_f64 =3D float32_to_float64(RsV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return out_f64; } =20 -float32 HELPER(conv_df2sf)(CPUHexagonState *env, float64 RssV) +float32 HELPER(conv_df2sf)(CPUHexagonState *env, float64 RssV, + uint32_t pkt_need_commit) { float32 out_f32; arch_fpop_start(env); out_f32 =3D float64_to_float32(RssV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return out_f32; } =20 -float32 HELPER(conv_uw2sf)(CPUHexagonState *env, int32_t RsV) +float32 HELPER(conv_uw2sf)(CPUHexagonState *env, int32_t RsV, + uint32_t pkt_need_commit) { float32 RdV; arch_fpop_start(env); RdV =3D uint32_to_float32(RsV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -float64 HELPER(conv_uw2df)(CPUHexagonState *env, int32_t RsV) +float64 HELPER(conv_uw2df)(CPUHexagonState *env, int32_t RsV, + uint32_t pkt_need_commit) { float64 RddV; arch_fpop_start(env); RddV =3D uint32_to_float64(RsV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -float32 HELPER(conv_w2sf)(CPUHexagonState *env, int32_t RsV) +float32 HELPER(conv_w2sf)(CPUHexagonState *env, int32_t RsV, + uint32_t pkt_need_commit) { float32 RdV; arch_fpop_start(env); RdV =3D int32_to_float32(RsV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -float64 HELPER(conv_w2df)(CPUHexagonState *env, int32_t RsV) +float64 HELPER(conv_w2df)(CPUHexagonState *env, int32_t RsV, + uint32_t pkt_need_commit) { float64 RddV; arch_fpop_start(env); RddV =3D int32_to_float64(RsV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -float32 HELPER(conv_ud2sf)(CPUHexagonState *env, int64_t RssV) +float32 HELPER(conv_ud2sf)(CPUHexagonState *env, int64_t RssV, + uint32_t pkt_need_commit) { float32 RdV; arch_fpop_start(env); RdV =3D uint64_to_float32(RssV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -float64 HELPER(conv_ud2df)(CPUHexagonState *env, int64_t RssV) +float64 HELPER(conv_ud2df)(CPUHexagonState *env, int64_t RssV, + uint32_t pkt_need_commit) { float64 RddV; arch_fpop_start(env); RddV =3D uint64_to_float64(RssV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -float32 HELPER(conv_d2sf)(CPUHexagonState *env, int64_t RssV) +float32 HELPER(conv_d2sf)(CPUHexagonState *env, int64_t RssV, + uint32_t pkt_need_commit) { float32 RdV; arch_fpop_start(env); RdV =3D int64_to_float32(RssV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -float64 HELPER(conv_d2df)(CPUHexagonState *env, int64_t RssV) +float64 HELPER(conv_d2df)(CPUHexagonState *env, int64_t RssV, + uint32_t pkt_need_commit) { float64 RddV; arch_fpop_start(env); RddV =3D int64_to_float64(RssV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -uint32_t HELPER(conv_sf2uw)(CPUHexagonState *env, float32 RsV) +uint32_t HELPER(conv_sf2uw)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { uint32_t RdV; arch_fpop_start(env); @@ -577,11 +590,12 @@ uint32_t HELPER(conv_sf2uw)(CPUHexagonState *env, flo= at32 RsV) } else { RdV =3D float32_to_uint32(RsV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -int32_t HELPER(conv_sf2w)(CPUHexagonState *env, float32 RsV) +int32_t HELPER(conv_sf2w)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { int32_t RdV; arch_fpop_start(env); @@ -592,11 +606,12 @@ int32_t HELPER(conv_sf2w)(CPUHexagonState *env, float= 32 RsV) } else { RdV =3D float32_to_int32(RsV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -uint64_t HELPER(conv_sf2ud)(CPUHexagonState *env, float32 RsV) +uint64_t HELPER(conv_sf2ud)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { uint64_t RddV; arch_fpop_start(env); @@ -607,11 +622,12 @@ uint64_t HELPER(conv_sf2ud)(CPUHexagonState *env, flo= at32 RsV) } else { RddV =3D float32_to_uint64(RsV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -int64_t HELPER(conv_sf2d)(CPUHexagonState *env, float32 RsV) +int64_t HELPER(conv_sf2d)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { int64_t RddV; arch_fpop_start(env); @@ -622,11 +638,12 @@ int64_t HELPER(conv_sf2d)(CPUHexagonState *env, float= 32 RsV) } else { RddV =3D float32_to_int64(RsV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -uint32_t HELPER(conv_df2uw)(CPUHexagonState *env, float64 RssV) +uint32_t HELPER(conv_df2uw)(CPUHexagonState *env, float64 RssV, + uint32_t pkt_need_commit) { uint32_t RdV; arch_fpop_start(env); @@ -637,11 +654,12 @@ uint32_t HELPER(conv_df2uw)(CPUHexagonState *env, flo= at64 RssV) } else { RdV =3D float64_to_uint32(RssV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -int32_t HELPER(conv_df2w)(CPUHexagonState *env, float64 RssV) +int32_t HELPER(conv_df2w)(CPUHexagonState *env, float64 RssV, + uint32_t pkt_need_commit) { int32_t RdV; arch_fpop_start(env); @@ -652,11 +670,12 @@ int32_t HELPER(conv_df2w)(CPUHexagonState *env, float= 64 RssV) } else { RdV =3D float64_to_int32(RssV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -uint64_t HELPER(conv_df2ud)(CPUHexagonState *env, float64 RssV) +uint64_t HELPER(conv_df2ud)(CPUHexagonState *env, float64 RssV, + uint32_t pkt_need_commit) { uint64_t RddV; arch_fpop_start(env); @@ -667,11 +686,12 @@ uint64_t HELPER(conv_df2ud)(CPUHexagonState *env, flo= at64 RssV) } else { RddV =3D float64_to_uint64(RssV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -int64_t HELPER(conv_df2d)(CPUHexagonState *env, float64 RssV) +int64_t HELPER(conv_df2d)(CPUHexagonState *env, float64 RssV, + uint32_t pkt_need_commit) { int64_t RddV; arch_fpop_start(env); @@ -682,11 +702,12 @@ int64_t HELPER(conv_df2d)(CPUHexagonState *env, float= 64 RssV) } else { RddV =3D float64_to_int64(RssV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -uint32_t HELPER(conv_sf2uw_chop)(CPUHexagonState *env, float32 RsV) +uint32_t HELPER(conv_sf2uw_chop)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { uint32_t RdV; arch_fpop_start(env); @@ -697,11 +718,12 @@ uint32_t HELPER(conv_sf2uw_chop)(CPUHexagonState *env= , float32 RsV) } else { RdV =3D float32_to_uint32_round_to_zero(RsV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -int32_t HELPER(conv_sf2w_chop)(CPUHexagonState *env, float32 RsV) +int32_t HELPER(conv_sf2w_chop)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { int32_t RdV; arch_fpop_start(env); @@ -712,11 +734,12 @@ int32_t HELPER(conv_sf2w_chop)(CPUHexagonState *env, = float32 RsV) } else { RdV =3D float32_to_int32_round_to_zero(RsV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -uint64_t HELPER(conv_sf2ud_chop)(CPUHexagonState *env, float32 RsV) +uint64_t HELPER(conv_sf2ud_chop)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { uint64_t RddV; arch_fpop_start(env); @@ -727,11 +750,12 @@ uint64_t HELPER(conv_sf2ud_chop)(CPUHexagonState *env= , float32 RsV) } else { RddV =3D float32_to_uint64_round_to_zero(RsV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -int64_t HELPER(conv_sf2d_chop)(CPUHexagonState *env, float32 RsV) +int64_t HELPER(conv_sf2d_chop)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { int64_t RddV; arch_fpop_start(env); @@ -742,11 +766,12 @@ int64_t HELPER(conv_sf2d_chop)(CPUHexagonState *env, = float32 RsV) } else { RddV =3D float32_to_int64_round_to_zero(RsV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -uint32_t HELPER(conv_df2uw_chop)(CPUHexagonState *env, float64 RssV) +uint32_t HELPER(conv_df2uw_chop)(CPUHexagonState *env, float64 RssV, + uint32_t pkt_need_commit) { uint32_t RdV; arch_fpop_start(env); @@ -757,11 +782,12 @@ uint32_t HELPER(conv_df2uw_chop)(CPUHexagonState *env= , float64 RssV) } else { RdV =3D float64_to_uint32_round_to_zero(RssV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -int32_t HELPER(conv_df2w_chop)(CPUHexagonState *env, float64 RssV) +int32_t HELPER(conv_df2w_chop)(CPUHexagonState *env, float64 RssV, + uint32_t pkt_need_commit) { int32_t RdV; arch_fpop_start(env); @@ -772,11 +798,12 @@ int32_t HELPER(conv_df2w_chop)(CPUHexagonState *env, = float64 RssV) } else { RdV =3D float64_to_int32_round_to_zero(RssV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -uint64_t HELPER(conv_df2ud_chop)(CPUHexagonState *env, float64 RssV) +uint64_t HELPER(conv_df2ud_chop)(CPUHexagonState *env, float64 RssV, + uint32_t pkt_need_commit) { uint64_t RddV; arch_fpop_start(env); @@ -787,11 +814,12 @@ uint64_t HELPER(conv_df2ud_chop)(CPUHexagonState *env= , float64 RssV) } else { RddV =3D float64_to_uint64_round_to_zero(RssV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -int64_t HELPER(conv_df2d_chop)(CPUHexagonState *env, float64 RssV) +int64_t HELPER(conv_df2d_chop)(CPUHexagonState *env, float64 RssV, + uint32_t pkt_need_commit) { int64_t RddV; arch_fpop_start(env); @@ -802,49 +830,54 @@ int64_t HELPER(conv_df2d_chop)(CPUHexagonState *env, = float64 RssV) } else { RddV =3D float64_to_int64_round_to_zero(RssV, &env->fp_status); } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -float32 HELPER(sfadd)(CPUHexagonState *env, float32 RsV, float32 RtV) +float32 HELPER(sfadd)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { float32 RdV; arch_fpop_start(env); RdV =3D float32_add(RsV, RtV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -float32 HELPER(sfsub)(CPUHexagonState *env, float32 RsV, float32 RtV) +float32 HELPER(sfsub)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { float32 RdV; arch_fpop_start(env); RdV =3D float32_sub(RsV, RtV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -int32_t HELPER(sfcmpeq)(CPUHexagonState *env, float32 RsV, float32 RtV) +int32_t HELPER(sfcmpeq)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { int32_t PdV; arch_fpop_start(env); PdV =3D f8BITSOF(float32_eq_quiet(RsV, RtV, &env->fp_status)); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -int32_t HELPER(sfcmpgt)(CPUHexagonState *env, float32 RsV, float32 RtV) +int32_t HELPER(sfcmpgt)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { int cmp; int32_t PdV; arch_fpop_start(env); cmp =3D float32_compare_quiet(RsV, RtV, &env->fp_status); PdV =3D f8BITSOF(cmp =3D=3D float_relation_greater); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -int32_t HELPER(sfcmpge)(CPUHexagonState *env, float32 RsV, float32 RtV) +int32_t HELPER(sfcmpge)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { int cmp; int32_t PdV; @@ -852,38 +885,42 @@ int32_t HELPER(sfcmpge)(CPUHexagonState *env, float32= RsV, float32 RtV) cmp =3D float32_compare_quiet(RsV, RtV, &env->fp_status); PdV =3D f8BITSOF(cmp =3D=3D float_relation_greater || cmp =3D=3D float_relation_equal); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -int32_t HELPER(sfcmpuo)(CPUHexagonState *env, float32 RsV, float32 RtV) +int32_t HELPER(sfcmpuo)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { int32_t PdV; arch_fpop_start(env); PdV =3D f8BITSOF(float32_unordered_quiet(RsV, RtV, &env->fp_status)); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -float32 HELPER(sfmax)(CPUHexagonState *env, float32 RsV, float32 RtV) +float32 HELPER(sfmax)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { float32 RdV; arch_fpop_start(env); RdV =3D float32_maximum_number(RsV, RtV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -float32 HELPER(sfmin)(CPUHexagonState *env, float32 RsV, float32 RtV) +float32 HELPER(sfmin)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { float32 RdV; arch_fpop_start(env); RdV =3D float32_minimum_number(RsV, RtV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -int32_t HELPER(sfclass)(CPUHexagonState *env, float32 RsV, int32_t uiV) +int32_t HELPER(sfclass)(CPUHexagonState *env, float32 RsV, int32_t uiV, + uint32_t pkt_need_commit) { int32_t PdV =3D 0; arch_fpop_start(env); @@ -903,100 +940,110 @@ int32_t HELPER(sfclass)(CPUHexagonState *env, float= 32 RsV, int32_t uiV) PdV =3D 0xff; } set_float_exception_flags(0, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -float32 HELPER(sffixupn)(CPUHexagonState *env, float32 RsV, float32 RtV) +float32 HELPER(sffixupn)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { float32 RdV =3D 0; int adjust; arch_fpop_start(env); arch_sf_recip_common(&RsV, &RtV, &RdV, &adjust, &env->fp_status); RdV =3D RsV; - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -float32 HELPER(sffixupd)(CPUHexagonState *env, float32 RsV, float32 RtV) +float32 HELPER(sffixupd)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { float32 RdV =3D 0; int adjust; arch_fpop_start(env); arch_sf_recip_common(&RsV, &RtV, &RdV, &adjust, &env->fp_status); RdV =3D RtV; - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -float32 HELPER(sffixupr)(CPUHexagonState *env, float32 RsV) +float32 HELPER(sffixupr)(CPUHexagonState *env, float32 RsV, + uint32_t pkt_need_commit) { float32 RdV =3D 0; int adjust; arch_fpop_start(env); arch_sf_invsqrt_common(&RsV, &RdV, &adjust, &env->fp_status); RdV =3D RsV; - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 -float64 HELPER(dfadd)(CPUHexagonState *env, float64 RssV, float64 RttV) +float64 HELPER(dfadd)(CPUHexagonState *env, float64 RssV, float64 RttV, + uint32_t pkt_need_commit) { float64 RddV; arch_fpop_start(env); RddV =3D float64_add(RssV, RttV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -float64 HELPER(dfsub)(CPUHexagonState *env, float64 RssV, float64 RttV) +float64 HELPER(dfsub)(CPUHexagonState *env, float64 RssV, float64 RttV, + uint32_t pkt_need_commit) { float64 RddV; arch_fpop_start(env); RddV =3D float64_sub(RssV, RttV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -float64 HELPER(dfmax)(CPUHexagonState *env, float64 RssV, float64 RttV) +float64 HELPER(dfmax)(CPUHexagonState *env, float64 RssV, float64 RttV, + uint32_t pkt_need_commit) { float64 RddV; arch_fpop_start(env); RddV =3D float64_maximum_number(RssV, RttV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -float64 HELPER(dfmin)(CPUHexagonState *env, float64 RssV, float64 RttV) +float64 HELPER(dfmin)(CPUHexagonState *env, float64 RssV, float64 RttV, + uint32_t pkt_need_commit) { float64 RddV; arch_fpop_start(env); RddV =3D float64_minimum_number(RssV, RttV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 -int32_t HELPER(dfcmpeq)(CPUHexagonState *env, float64 RssV, float64 RttV) +int32_t HELPER(dfcmpeq)(CPUHexagonState *env, float64 RssV, float64 RttV, + uint32_t pkt_need_commit) { int32_t PdV; arch_fpop_start(env); PdV =3D f8BITSOF(float64_eq_quiet(RssV, RttV, &env->fp_status)); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -int32_t HELPER(dfcmpgt)(CPUHexagonState *env, float64 RssV, float64 RttV) +int32_t HELPER(dfcmpgt)(CPUHexagonState *env, float64 RssV, float64 RttV, + uint32_t pkt_need_commit) { int cmp; int32_t PdV; arch_fpop_start(env); cmp =3D float64_compare_quiet(RssV, RttV, &env->fp_status); PdV =3D f8BITSOF(cmp =3D=3D float_relation_greater); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -int32_t HELPER(dfcmpge)(CPUHexagonState *env, float64 RssV, float64 RttV) +int32_t HELPER(dfcmpge)(CPUHexagonState *env, float64 RssV, float64 RttV, + uint32_t pkt_need_commit) { int cmp; int32_t PdV; @@ -1004,20 +1051,22 @@ int32_t HELPER(dfcmpge)(CPUHexagonState *env, float= 64 RssV, float64 RttV) cmp =3D float64_compare_quiet(RssV, RttV, &env->fp_status); PdV =3D f8BITSOF(cmp =3D=3D float_relation_greater || cmp =3D=3D float_relation_equal); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -int32_t HELPER(dfcmpuo)(CPUHexagonState *env, float64 RssV, float64 RttV) +int32_t HELPER(dfcmpuo)(CPUHexagonState *env, float64 RssV, float64 RttV, + uint32_t pkt_need_commit) { int32_t PdV; arch_fpop_start(env); PdV =3D f8BITSOF(float64_unordered_quiet(RssV, RttV, &env->fp_status)); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -int32_t HELPER(dfclass)(CPUHexagonState *env, float64 RssV, int32_t uiV) +int32_t HELPER(dfclass)(CPUHexagonState *env, float64 RssV, int32_t uiV, + uint32_t pkt_need_commit) { int32_t PdV =3D 0; arch_fpop_start(env); @@ -1037,51 +1086,55 @@ int32_t HELPER(dfclass)(CPUHexagonState *env, float= 64 RssV, int32_t uiV) PdV =3D 0xff; } set_float_exception_flags(0, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return PdV; } =20 -float32 HELPER(sfmpy)(CPUHexagonState *env, float32 RsV, float32 RtV) +float32 HELPER(sfmpy)(CPUHexagonState *env, float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { float32 RdV; arch_fpop_start(env); RdV =3D float32_mul(RsV, RtV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RdV; } =20 float32 HELPER(sffma)(CPUHexagonState *env, float32 RxV, - float32 RsV, float32 RtV) + float32 RsV, float32 RtV, + uint32_t pkt_need_commit) { arch_fpop_start(env); RxV =3D float32_muladd(RsV, RtV, RxV, 0, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RxV; } =20 float32 HELPER(sffma_sc)(CPUHexagonState *env, float32 RxV, - float32 RsV, float32 RtV, float32 PuV) + float32 RsV, float32 RtV, float32 PuV, + uint32_t pkt_need_commit) { arch_fpop_start(env); RxV =3D float32_muladd_scalbn(RsV, RtV, RxV, fSXTN(8, 64, PuV), float_muladd_suppress_add_product_zero, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RxV; } =20 float32 HELPER(sffms)(CPUHexagonState *env, float32 RxV, - float32 RsV, float32 RtV) + float32 RsV, float32 RtV, uint32_t pkt_need_commit) { arch_fpop_start(env); RxV =3D float32_muladd(RsV, RtV, RxV, float_muladd_negate_product, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RxV; } =20 static float32 do_sffma_lib(CPUHexagonState *env, float32 RxV, - float32 RsV, float32 RtV, int negate) + float32 RsV, float32 RtV, int negate, + uint32_t pkt_need_commit) { int flags; =20 @@ -1103,23 +1156,25 @@ static float32 do_sffma_lib(CPUHexagonState *env, f= loat32 RxV, } } =20 - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RxV; } =20 float32 HELPER(sffma_lib)(CPUHexagonState *env, float32 RxV, - float32 RsV, float32 RtV) + float32 RsV, float32 RtV, uint32_t pkt_need_comm= it) { - return do_sffma_lib(env, RxV, RsV, RtV, 0); + return do_sffma_lib(env, RxV, RsV, RtV, 0, pkt_need_commit); } =20 float32 HELPER(sffms_lib)(CPUHexagonState *env, float32 RxV, - float32 RsV, float32 RtV) + float32 RsV, float32 RtV, uint32_t pkt_need_comm= it) { - return do_sffma_lib(env, RxV, RsV, RtV, float_muladd_negate_product); + return do_sffma_lib(env, RxV, RsV, RtV, float_muladd_negate_product, + pkt_need_commit); } =20 -float64 HELPER(dfmpyfix)(CPUHexagonState *env, float64 RssV, float64 RttV) +float64 HELPER(dfmpyfix)(CPUHexagonState *env, float64 RssV, float64 RttV, + uint32_t pkt_need_commit) { int64_t RddV; arch_fpop_start(env); @@ -1136,16 +1191,16 @@ float64 HELPER(dfmpyfix)(CPUHexagonState *env, floa= t64 RssV, float64 RttV) } else { RddV =3D RssV; } - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RddV; } =20 float64 HELPER(dfmpyhh)(CPUHexagonState *env, float64 RxxV, - float64 RssV, float64 RttV) + float64 RssV, float64 RttV, uint32_t pkt_need_comm= it) { arch_fpop_start(env); RxxV =3D internal_mpyhh(RssV, RttV, RxxV, &env->fp_status); - arch_fpop_end(env); + arch_fpop_end(env, pkt_need_commit); return RxxV; } =20 diff --git a/target/hexagon/gen_helper_funcs.py b/target/hexagon/gen_helper= _funcs.py index a9c0e27a80..6206104424 100755 --- a/target/hexagon/gen_helper_funcs.py +++ b/target/hexagon/gen_helper_funcs.py @@ -84,11 +84,6 @@ def gen_helper_function(f, tag, tagregs, tagimms): {hex_common.semdict[tag]} """)) =20 - if "A_FPOP" in hex_common.attribdict[tag]: - f.write(hex_common.code_fmt(f"""\ - arch_fpop_end(env); - """)) - ## Return the scalar result for regtype, regid in regs: reg =3D hex_common.get_register(tag, regtype, regid) @@ -116,6 +111,9 @@ def main(): ## Skip the guest instructions if "A_GUEST" in hex_common.attribdict[tag]: continue + ## Skip the floating point instructions + if "A_FPOP" in hex_common.attribdict[tag]: + continue ## Skip the diag instructions if tag =3D=3D "Y6_diag": continue diff --git a/target/hexagon/gen_tcg_funcs.py b/target/hexagon/gen_tcg_funcs= .py index c2ba91ddc0..91b6a1e9d0 100755 --- a/target/hexagon/gen_tcg_funcs.py +++ b/target/hexagon/gen_tcg_funcs.py @@ -77,6 +77,10 @@ def gen_tcg_func(f, tag, regs, imms): f.write(f" emit_{tag}({arguments});\n") =20 elif hex_common.skip_qemu_helper(tag): + if "A_FPOP" in hex_common.attribdict[tag]: + f.write(" TCGv pkt_need_commit =3D ") + f.write("tcg_constant_tl(ctx->need_commit);\n") + f.write(f" fGEN_TCG_{tag}({hex_common.semdict[tag]});\n") else: ## Generate the call to the helper --=20 2.43.0 From nobody Fri Nov 14 16:50:29 2025 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=gmail.com ARC-Seal: i=1; a=rsa-sha256; t=1761876970; cv=none; d=zohomail.com; s=zohoarc; b=n2CD/N18ykITiRlaXz5UucUhrgIQpwgUOuPeyT0uQAg95QOhX+F718D8NOvi3yiLjccUeAOoTj2iYfBLssHbO6oqearziou67lMpXW0+BCYLEpBvTJAa0EmMuY5wdcPHWScdYwi65j4lrfukHS72jy7O2j1FlsHgB6UuQqolXYk= ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=zohomail.com; s=zohoarc; t=1761876970; h=Content-Type: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=GDDzVpIEYFtan7kEC7miAi1QvrklbQFsgIlEd92Gf0I=; b=hqHPbnwy+UWGRRRKlRWviKaCG4+zCS26McwtndcipUkhBwlc//Ed1amU8rSt/t0F+LSn62rqSBYrGtTuMVLUZTsjSvuCwnwmHkFZfiekRwL93W6e0nL2Z69SBO2S+HYfT37Xx69k43FfBcjyezOp4y1wPhsxhJSgzH/T9OtJQuk= 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 1761876970037479.2425522798143; Thu, 30 Oct 2025 19:16:10 -0700 (PDT) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1vEeev-0005El-EW; Thu, 30 Oct 2025 22:14:33 -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 1vEeet-0005Dq-HP for qemu-devel@nongnu.org; Thu, 30 Oct 2025 22:14:31 -0400 Received: from mail-il1-x129.google.com ([2607:f8b0:4864:20::129]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_128_GCM_SHA256:128) (Exim 4.90_1) (envelope-from ) id 1vEeen-0005vd-Uu for qemu-devel@nongnu.org; Thu, 30 Oct 2025 22:14:29 -0400 Received: by mail-il1-x129.google.com with SMTP id e9e14a558f8ab-430c17e29d9so7091535ab.1 for ; Thu, 30 Oct 2025 19:14:22 -0700 (PDT) Received: from taylor-ubuntu.. (c-67-172-136-152.hsd1.co.comcast.net. [67.172.136.152]) by smtp.gmail.com with ESMTPSA id e9e14a558f8ab-43310274e21sm1631055ab.9.2025.10.30.19.14.17 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 30 Oct 2025 19:14:18 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1761876859; x=1762481659; 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=GDDzVpIEYFtan7kEC7miAi1QvrklbQFsgIlEd92Gf0I=; b=HnLC6u9z0ySf19D1FjwNAqDGAdwvWRHoCNJ822wQ623N4bJjd3qu9hG2FLodQn25Nh 8pFQbnVGWTfOjCz5DOZmrOWq10I4WWKfcYKDfZ/PcqNEbmlkesXq1Fb0MSMS7MGdA0d0 GsxLWo8fJvWLZvfUxe1pmbdxCRDucQ3SZLC7yblwbMORHUgxeCuGhWIBTzNjRmqlFDKl RBryVdwrMbFzXODCsPlDVkUrpNXxhgEH+9Fd2Xp+ISNYXYnJjRSSzaEtHDALVTDU3iPO oEKP6knZtJQpL8rr9IGBM2fib+rvStNjcDdLs7tC9W8FVV+rOl0sceyAKOEQu7RKrs5f zKtQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1761876859; x=1762481659; 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=GDDzVpIEYFtan7kEC7miAi1QvrklbQFsgIlEd92Gf0I=; b=vKUijO2KnWD31Xu8BMUWT0XoPUHxd9QeIdf7A/pSmKombf0U3URxrKX9fZqcIqKtya kBd4RaQkB7SCRRAJOKsSi7mRTkDGHGtMQ0YdwNmJmOjm0SO0vt/mpnUU//xJDOKKf5Pg 7Cl0MwtBGT2O0atnMiIsW+Jb+Hi+xjY74XNysG5zlYLUjDPHHx4KCwYRt1ya+Py9X1Dy dJG322bcuasrK6KoglQN5DQqC3+aanNcIte/+XM0G2ungg6irVgpCJfrNPmYCUlQXL2q sGFcQrKqGJcJZbLim1qcZLQMawKvPk4jsR0GB01geRdCSEBcopd39NIOpd7KjfPtxrHr E8RA== X-Gm-Message-State: AOJu0Yz1WoMeWG685CePXj+NorSUKNp/49phdLMIISqpn4cvCbL7Babm +GKw/NmU86S7pWEGCk6vlp5ypvI+bXT4oYmJFYlzegnzfkGShZwrOMoRhso92QHRaaw= X-Gm-Gg: ASbGncvWZ4YniV9vSZ6fBSsd4ziEg0ZEJesrIaGLnTsa4HibpaenA+xpD0xYLdolq1d k0oQ+AQzQ5pcki8J23D6jv79yKsf5ZozoGIv7obl44AckBqTpoyZbfRlVpWkXW/Zcz46w97hVwu y/OqBuL6hgHoJ6RxWy/7+jhOgdCC7jIwI0IKrOWYxgQkwqdipjV8mN4ngSdNC/YE5pYh7VGvkUo c+LsTun5xHMzqUXJQc/e90/5YyAvlCtBzWroHCChSxGrwc/Zfa0JRplY+x6WpTOXEXgEYTwuw41 HNypkHORrfFeZ9eOhDErqp4orY51FaQ5nqJbgmTy/P/ENJLfu/6bl3WH9IymAbJnbdnzyprOExF Jop9r7VEcYCPCTCJ0etnNuDrTCBj7ZwFm5Lx66K2xCyIGKOzeGIM08vJeqrn5tnIgypixP2Vh4T iOOueubsEdButZSlLe4LmbvWKBAeRq0BGdlUI9kVniuE3MSJB4f9fuwYZEA2hPPgB5TQ== X-Google-Smtp-Source: AGHT+IExN6uya5YAjveaT/1KOgGZbnkr9ZTmMk8mb1KdmLmmRM85vCiUDvlQ3tgfEDnxfV9liOwKmw== X-Received: by 2002:a05:6e02:97:b0:430:d21d:7b24 with SMTP id e9e14a558f8ab-4330d235d0emr34303135ab.32.1761876858873; Thu, 30 Oct 2025 19:14:18 -0700 (PDT) From: Taylor Simpson To: qemu-devel@nongnu.org Cc: brian.cain@oss.qualcomm.com, matheus.bernardino@oss.qualcomm.com, sid.manning@oss.qualcomm.com, marco.liebel@oss.qualcomm.com, richard.henderson@linaro.org, philmd@linaro.org, ale@rev.ng, anjo@rev.ng, ltaylorsimpson@gmail.com Subject: [PATCH 2/2] Hexagon (target/hexagon) Implicit writes to USR don't force packet commit Date: Thu, 30 Oct 2025 20:14:11 -0600 Message-ID: <20251031021411.111365-3-ltaylorsimpson@gmail.com> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20251031021411.111365-1-ltaylorsimpson@gmail.com> References: <20251031021411.111365-1-ltaylorsimpson@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" 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::129; envelope-from=ltaylorsimpson@gmail.com; helo=mail-il1-x129.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, FREEMAIL_FROM=0.001, 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 @gmail.com) X-ZM-MESSAGEID: 1761876974493158500 Implicit writes to USR are only to specific fields of USR to indicate side effects (e.g., saturation overflow, floating point status). In these cases, we don't force a packet commit. This will allow more packets to be short-circuited (avoid writing the results to temporaries). Signed-off-by: Taylor Simpson --- target/hexagon/translate.c | 14 +------------- 1 file changed, 1 insertion(+), 13 deletions(-) diff --git a/target/hexagon/translate.c b/target/hexagon/translate.c index 8fce219c0d..3da01c2b5e 100644 --- a/target/hexagon/translate.c +++ b/target/hexagon/translate.c @@ -272,12 +272,7 @@ static void mark_implicit_reg_write(DisasContext *ctx,= int attrib, int rnum) { uint16_t opcode =3D ctx->insn->opcode; if (GET_ATTRIB(opcode, attrib)) { - /* - * USR is used to set overflow and FP exceptions, - * so treat it as conditional - */ - bool is_predicated =3D GET_ATTRIB(opcode, A_CONDEXEC) || - rnum =3D=3D HEX_REG_USR; + bool is_predicated =3D GET_ATTRIB(opcode, A_CONDEXEC); =20 /* LC0/LC1 is conditionally written by endloop instructions */ if ((rnum =3D=3D HEX_REG_LC0 || rnum =3D=3D HEX_REG_LC1) && @@ -300,8 +295,6 @@ static void mark_implicit_reg_writes(DisasContext *ctx) mark_implicit_reg_write(ctx, A_IMPLICIT_WRITES_SA0, HEX_REG_SA0); mark_implicit_reg_write(ctx, A_IMPLICIT_WRITES_LC1, HEX_REG_LC1); mark_implicit_reg_write(ctx, A_IMPLICIT_WRITES_SA1, HEX_REG_SA1); - mark_implicit_reg_write(ctx, A_IMPLICIT_WRITES_USR, HEX_REG_USR); - mark_implicit_reg_write(ctx, A_FPOP, HEX_REG_USR); } =20 static void mark_implicit_pred_write(DisasContext *ctx, int attrib, int pn= um) @@ -351,11 +344,6 @@ static bool need_commit(DisasContext *ctx) } } =20 - /* Floating point instructions are hard-coded to use new_value */ - if (check_for_attrib(pkt, A_FPOP)) { - return true; - } - if (ctx->read_after_write || ctx->has_hvx_overlap) { return true; } --=20 2.43.0