From patchwork Mon May 6 02:42:14 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Brian Cain X-Patchwork-Id: 13654833 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 242EBC04FFE for ; Mon, 6 May 2024 02:43:33 +0000 (UTC) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1s3oJd-0000Wv-P6; Sun, 05 May 2024 22:42:57 -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 1s3oJa-0000UV-7d for qemu-devel@nongnu.org; Sun, 05 May 2024 22:42:55 -0400 Received: from mx0b-0031df01.pphosted.com ([205.220.180.131]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1s3oJY-0007WT-4N for qemu-devel@nongnu.org; Sun, 05 May 2024 22:42:53 -0400 Received: from pps.filterd (m0279868.ppops.net [127.0.0.1]) by mx0a-0031df01.pphosted.com (8.18.1.2/8.18.1.2) with ESMTP id 4462H2Jk012804; Mon, 6 May 2024 02:42:49 GMT DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=quicinc.com; h= from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-type:content-transfer-encoding; s= qcppdkim1; bh=anKUYCzf7FN5QOzr+ywcEC0NOPXtLnLvx/j+P7sM2+M=; b=Fu 92q4cjOl9oHx65DqmNGn3eF1bYrWLxCJhi2hgOMVFjO3Ru/0SRTFO33zW217QK4t rCeRj806X7YKiK0vSFvp6WgJSMefIbRt9hX6k8GnEdbPuroDGpel2gQ2zYMj8blz k++ibJ2Skve9El4epOIA0OKcQqJ+wQb98fVQ5wtN7xKbBUe8wk4rk2XeWFA+lmqF yoQ5gSH83DKTxd54ryXqIYC8o0IFUdScoK2kmbJxr5mzMNqTBufiHAcHYbFWof/R razfqO88lhm0YWMDCalJfdzK583cGjjMnjnvjXuJj8jn35nCzM8urlP63aikw6rB aTqlt/Z5aEOAYRtqkOGA== Received: from nalasppmta02.qualcomm.com (Global_NAT1.qualcomm.com [129.46.96.20]) by mx0a-0031df01.pphosted.com (PPS) with ESMTPS id 3xwcbpjg44-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Mon, 06 May 2024 02:42:49 +0000 (GMT) Received: from nalasex01a.na.qualcomm.com (nalasex01a.na.qualcomm.com [10.47.209.196]) by NALASPPMTA02.qualcomm.com (8.17.1.5/8.17.1.5) with ESMTPS id 4462gmP0007514 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Mon, 6 May 2024 02:42:48 GMT Received: from hu-bcain-lv.qualcomm.com (10.49.16.6) by nalasex01a.na.qualcomm.com (10.47.209.196) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.2.1544.9; Sun, 5 May 2024 19:42:47 -0700 From: Brian Cain To: CC: , , , , , , , , , , Subject: [PULL 02/15] Hexagon (target/hexagon) Enable more short-circuit packets (scalar core) Date: Sun, 5 May 2024 19:42:14 -0700 Message-ID: <20240506024227.3345791-3-bcain@quicinc.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20240506024227.3345791-1-bcain@quicinc.com> References: <20240506024227.3345791-1-bcain@quicinc.com> MIME-Version: 1.0 X-Originating-IP: [10.49.16.6] X-ClientProxiedBy: nalasex01c.na.qualcomm.com (10.47.97.35) To nalasex01a.na.qualcomm.com (10.47.209.196) X-QCInternal: smtphost X-Proofpoint-Virus-Version: vendor=nai engine=6200 definitions=5800 signatures=585085 X-Proofpoint-ORIG-GUID: Tlua68FX_Q9JxN4xOewItOldiEWhshcY X-Proofpoint-GUID: Tlua68FX_Q9JxN4xOewItOldiEWhshcY X-Proofpoint-Virus-Version: vendor=baseguard engine=ICAP:2.0.293,Aquarius:18.0.1011,Hydra:6.0.650,FMLib:17.11.176.26 definitions=2024-05-05_17,2024-05-03_02,2023-05-22_02 X-Proofpoint-Spam-Details: rule=outbound_notspam policy=outbound score=0 lowpriorityscore=0 bulkscore=0 malwarescore=0 adultscore=0 impostorscore=0 mlxscore=0 suspectscore=0 phishscore=0 priorityscore=1501 clxscore=1015 spamscore=0 mlxlogscore=644 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.19.0-2404010003 definitions=main-2405060012 Received-SPF: pass client-ip=205.220.180.131; envelope-from=bcain@quicinc.com; helo=mx0b-0031df01.pphosted.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, 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+qemu-devel=archiver.kernel.org@nongnu.org Sender: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org From: Taylor Simpson Look for read-after-write instead of overlap of reads and writes Here is an example with overalp but no read-after-write: 0x000200fc: 0x38103876 { R0 = add(R0,R1); R6 = add(R6,R7) } BEFORE: ---- 00000000000200fc mov_i32 loc2,$0x0 mov_i32 loc2,r0 add_i32 loc3,loc2,r1 mov_i32 loc2,loc3 mov_i32 loc4,$0x0 mov_i32 loc4,r6 add_i32 loc5,loc4,r7 mov_i32 loc4,loc5 mov_i32 r0,loc2 mov_i32 r6,loc4 AFTER: ---- 00000000000200fc add_i32 loc2,r0,r1 mov_i32 r0,loc2 add_i32 loc3,r6,r7 mov_i32 r6,loc3 We can also short-circuit packets with .new values by reading from the real destination instead of the temporary. 0x00020100: 0x78005ff3 { R19 = #0xff 0x00020104: 0x2002e204 if (cmp.eq(N19.new,R2)) jump:t PC+8 } BEFORE: ---- 0000000000020100 mov_i32 pc,$0x20108 mov_i32 loc8,$0x0 mov_i32 loc8,$0xff setcond_i32 loc10,loc8,r2,eq mov_i32 loc6,loc10 mov_i32 r19,loc8 add_i32 pkt_cnt,pkt_cnt,$0x2 add_i32 insn_cnt,insn_cnt,$0x4 brcond_i32 loc6,$0x0,eq,$L1 goto_tb $0x0 mov_i32 pc,$0x20108 exit_tb $0x7fbb54000040 set_label $L1 goto_tb $0x1 exit_tb $0x7fbb54000041 set_label $L0 exit_tb $0x7fbb54000043 AFTER: ---- 0000000000020100 mov_i32 pc,$0x20108 mov_i32 r19,$0xff setcond_i32 loc7,r19,r2,eq mov_i32 loc4,loc7 add_i32 pkt_cnt,pkt_cnt,$0x2 add_i32 insn_cnt,insn_cnt,$0x4 brcond_i32 loc4,$0x0,eq,$L1 goto_tb $0x0 mov_i32 pc,$0x20108 exit_tb $0x7f9764000040 set_label $L1 goto_tb $0x1 exit_tb $0x7f9764000041 set_label $L0 exit_tb $0x7f9764000043 Signed-off-by: Taylor Simpson Reviewed-by: Brian Cain Message-Id: <20240201103340.119081-3-ltaylorsimpson@gmail.com> Signed-off-by: Brian Cain --- target/hexagon/translate.c | 21 ++++----------------- target/hexagon/translate.h | 13 +++++++------ 2 files changed, 11 insertions(+), 23 deletions(-) diff --git a/target/hexagon/translate.c b/target/hexagon/translate.c index 47a870f42d..8d42ebd91c 100644 --- a/target/hexagon/translate.c +++ b/target/hexagon/translate.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2019-2023 Qualcomm Innovation Center, Inc. All Rights Reserved. + * Copyright(c) 2019-2024 Qualcomm Innovation Center, Inc. All Rights Reserved. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -396,20 +396,8 @@ static bool need_commit(DisasContext *ctx) } } - /* Check for overlap between register reads and writes */ - for (int i = 0; i < ctx->reg_log_idx; i++) { - int rnum = ctx->reg_log[i]; - if (test_bit(rnum, ctx->regs_read)) { - return true; - } - } - - /* Check for overlap between predicate reads and writes */ - for (int i = 0; i < ctx->preg_log_idx; i++) { - int pnum = ctx->preg_log[i]; - if (test_bit(pnum, ctx->pregs_read)) { - return true; - } + if (ctx->read_after_write) { + return true; } /* Check for overlap between HVX reads and writes */ @@ -468,6 +456,7 @@ static void analyze_packet(DisasContext *ctx) { Packet *pkt = ctx->pkt; ctx->has_hvx_helper = false; + ctx->read_after_write = false; for (int i = 0; i < pkt->num_insns; i++) { Insn *insn = &pkt->insn[i]; ctx->insn = insn; @@ -492,11 +481,9 @@ static void gen_start_packet(DisasContext *ctx) ctx->next_PC = next_PC; ctx->reg_log_idx = 0; bitmap_zero(ctx->regs_written, TOTAL_PER_THREAD_REGS); - bitmap_zero(ctx->regs_read, TOTAL_PER_THREAD_REGS); bitmap_zero(ctx->predicated_regs, TOTAL_PER_THREAD_REGS); ctx->preg_log_idx = 0; bitmap_zero(ctx->pregs_written, NUM_PREGS); - bitmap_zero(ctx->pregs_read, NUM_PREGS); ctx->future_vregs_idx = 0; ctx->tmp_vregs_idx = 0; ctx->vreg_log_idx = 0; diff --git a/target/hexagon/translate.h b/target/hexagon/translate.h index f06d71fc53..d5e7f49ad8 100644 --- a/target/hexagon/translate.h +++ b/target/hexagon/translate.h @@ -38,12 +38,10 @@ typedef struct DisasContext { int reg_log[REG_WRITES_MAX]; int reg_log_idx; DECLARE_BITMAP(regs_written, TOTAL_PER_THREAD_REGS); - DECLARE_BITMAP(regs_read, TOTAL_PER_THREAD_REGS); DECLARE_BITMAP(predicated_regs, TOTAL_PER_THREAD_REGS); int preg_log[PRED_WRITES_MAX]; int preg_log_idx; DECLARE_BITMAP(pregs_written, NUM_PREGS); - DECLARE_BITMAP(pregs_read, NUM_PREGS); uint8_t store_width[STORES_MAX]; bool s1_store_processed; int future_vregs_idx; @@ -68,6 +66,7 @@ typedef struct DisasContext { bool is_tight_loop; bool short_circuit; bool has_hvx_helper; + bool read_after_write; TCGv new_value[TOTAL_PER_THREAD_REGS]; TCGv new_pred_value[NUM_PREGS]; TCGv pred_written; @@ -88,13 +87,14 @@ static inline void ctx_log_pred_write(DisasContext *ctx, int pnum) static inline void ctx_log_pred_read(DisasContext *ctx, int pnum) { - set_bit(pnum, ctx->pregs_read); + if (test_bit(pnum, ctx->pregs_written)) { + ctx->read_after_write = true; + } } static inline void ctx_log_pred_read_new(DisasContext *ctx, int pnum) { g_assert(test_bit(pnum, ctx->pregs_written)); - set_bit(pnum, ctx->pregs_read); } static inline void ctx_log_reg_write(DisasContext *ctx, int rnum, @@ -125,13 +125,14 @@ static inline void ctx_log_reg_write_pair(DisasContext *ctx, int rnum, static inline void ctx_log_reg_read(DisasContext *ctx, int rnum) { - set_bit(rnum, ctx->regs_read); + if (test_bit(rnum, ctx->regs_written)) { + ctx->read_after_write = true; + } } static inline void ctx_log_reg_read_new(DisasContext *ctx, int rnum) { g_assert(test_bit(rnum, ctx->regs_written)); - set_bit(rnum, ctx->regs_read); } static inline void ctx_log_reg_read_pair(DisasContext *ctx, int rnum)