{"id":2226714,"url":"http://patchwork.ozlabs.org/api/patches/2226714/?format=json","web_url":"http://patchwork.ozlabs.org/project/gcc/patch/bmm.hhuom29jc4.gcc.gcc-TEST.pinskia.61.1.1@forge-stage.sourceware.org/","project":{"id":17,"url":"http://patchwork.ozlabs.org/api/projects/17/?format=json","name":"GNU Compiler Collection","link_name":"gcc","list_id":"gcc-patches.gcc.gnu.org","list_email":"gcc-patches@gcc.gnu.org","web_url":null,"scm_url":null,"webscm_url":null,"list_archive_url":"","list_archive_url_format":"","commit_url_format":""},"msgid":"","list_archive_url":null,"date":"2026-04-22T18:30:26","name":"[v1,1/6] forwprop: Change optimize_agr_copyprop into forward walk instead of backwards","commit_ref":null,"pull_url":null,"state":"new","archived":false,"hash":"a0a510b93f71d02a9ee150633e0dc330edb9c1dd","submitter":{"id":93219,"url":"http://patchwork.ozlabs.org/api/people/93219/?format=json","name":"Andrew Pinski via Sourceware Forge","email":"forge-bot+pinskia@forge-stage.sourceware.org"},"delegate":null,"mbox":"http://patchwork.ozlabs.org/project/gcc/patch/bmm.hhuom29jc4.gcc.gcc-TEST.pinskia.61.1.1@forge-stage.sourceware.org/mbox/","series":[{"id":501080,"url":"http://patchwork.ozlabs.org/api/series/501080/?format=json","web_url":"http://patchwork.ozlabs.org/project/gcc/list/?series=501080","date":"2026-04-22T18:30:28","name":"copy-propv3","version":1,"mbox":"http://patchwork.ozlabs.org/series/501080/mbox/"}],"comments":"http://patchwork.ozlabs.org/api/patches/2226714/comments/","check":"pending","checks":"http://patchwork.ozlabs.org/api/patches/2226714/checks/","tags":{},"related":[],"headers":{"Return-Path":"","X-Original-To":["incoming@patchwork.ozlabs.org","gcc-patches@gcc.gnu.org"],"Delivered-To":["patchwork-incoming@legolas.ozlabs.org","gcc-patches@gcc.gnu.org"],"Authentication-Results":["legolas.ozlabs.org;\n spf=pass (sender SPF authorized) smtp.mailfrom=gcc.gnu.org\n (client-ip=2620:52:6:3111::32; helo=vm01.sourceware.org;\n envelope-from=gcc-patches-bounces~incoming=patchwork.ozlabs.org@gcc.gnu.org;\n receiver=patchwork.ozlabs.org)","sourceware.org; dmarc=none (p=none dis=none)\n header.from=forge-stage.sourceware.org","sourceware.org;\n spf=pass smtp.mailfrom=forge-stage.sourceware.org","server2.sourceware.org;\n arc=none smtp.remote-ip=38.145.34.39"],"Received":["from vm01.sourceware.org (vm01.sourceware.org\n [IPv6:2620:52:6:3111::32])\n\t(using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits)\n\t key-exchange x25519 server-signature ECDSA (secp384r1) server-digest SHA384)\n\t(No client certificate requested)\n\tby legolas.ozlabs.org (Postfix) with ESMTPS id 4g17XH6fMyz1yD5\n\tfor ; Thu, 23 Apr 2026 04:48:15 +1000 (AEST)","from vm01.sourceware.org (localhost [127.0.0.1])\n\tby sourceware.org (Postfix) with ESMTP id EFA6E48EACF8\n\tfor ; Wed, 22 Apr 2026 18:48:13 +0000 (GMT)","from forge-stage.sourceware.org (vm08.sourceware.org [38.145.34.39])\n by sourceware.org (Postfix) with ESMTPS id 17B4D4BC8957\n for ; Wed, 22 Apr 2026 18:32:05 +0000 (GMT)","from forge-stage.sourceware.org (localhost [IPv6:::1])\n (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits)\n key-exchange x25519 server-signature ECDSA (prime256v1) server-digest SHA256)\n (No client certificate requested)\n by forge-stage.sourceware.org (Postfix) with ESMTPS id DEF32434CD\n for ; Wed, 22 Apr 2026 18:32:04 +0000 (UTC)"],"DKIM-Filter":["OpenDKIM Filter v2.11.0 sourceware.org EFA6E48EACF8","OpenDKIM Filter v2.11.0 sourceware.org 17B4D4BC8957"],"DMARC-Filter":"OpenDMARC Filter v1.4.2 sourceware.org 17B4D4BC8957","ARC-Filter":"OpenARC Filter v1.0.0 sourceware.org 17B4D4BC8957","ARC-Seal":"i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1776882725; cv=none;\n b=Hu0nHyadWCsJstnksHWDxkpbiDDQbJ2SC+ZrUvkvcqzw9rn3OfQQIvnPrrDEahMM1KzQBM78YQBI9iIMr3ksuEFVHoJwpfYKKOoile1P14Y3SBhkoc2qelfTv75H2Ii0APv5/BQZs8X9nDAYCXMFmNkViL3Lz6qmCwPdfSOfw+o=","ARC-Message-Signature":"i=1; a=rsa-sha256; d=sourceware.org; s=key;\n t=1776882725; c=relaxed/simple;\n bh=GJ8FIum5MmCtNL8n1KrSJonRtFiISJeDsfrHWfT8u6Q=;\n h=From:Date:Subject:To:Message-ID;\n b=BaHX1R16u/MrEULGXpuEoSOxNMjPljOBRfQEP4Kug2d2seEU2bA0oILz2PcSShT9AWQxvgAnGNKzoIwxf5+mF58s+6MHNuIhzeYcjUgAmXT9uOrIDCGfjKZLgpETxqxUUcjnPP/nh6loNQGzrL6hvmjdSHgfXz7a87M825ugZhg=","ARC-Authentication-Results":"i=1; server2.sourceware.org","From":"Andrew Pinski via Sourceware Forge\n ","Date":"Wed, 22 Apr 2026 18:30:26 +0000","Subject":"[PATCH v1 1/6] forwprop: Change optimize_agr_copyprop into forward\n walk instead of backwards","To":"gcc-patches mailing list ","Message-ID":"\n ","X-Mailer":"batrachomyomachia","X-Pull-Request-Organization":"gcc","X-Pull-Request-Repository":"gcc-TEST","X-Pull-Request":"https://forge.sourceware.org/gcc/gcc-TEST/pulls/61","References":"\n ","In-Reply-To":"\n ","X-Patch-URL":"\n https://forge.sourceware.org/pinskia/gcc-TEST/commit/936642cff643f387029762217aef68bcd3bf88b1","X-BeenThere":"gcc-patches@gcc.gnu.org","X-Mailman-Version":"2.1.30","Precedence":"list","List-Id":"Gcc-patches mailing list ","List-Unsubscribe":",\n ","List-Archive":"","List-Post":"","List-Help":"","List-Subscribe":",\n ","Reply-To":"gcc-patches mailing list ,\n pinskia@gcc.gnu.org","Errors-To":"gcc-patches-bounces~incoming=patchwork.ozlabs.org@gcc.gnu.org"},"content":"From: Andrew Pinski \n\nWhile thinking about how to implement the rest of the copy prop and makes sure not\nto introduce some compile time problems, optimize_agr_copyprop should be changed\ninto a forwproping rather than looking backwards.\n\nBootstrapped and tested on x86_64-linux-gnu.\n\ngcc/ChangeLog:\n\n\t* tree-ssa-forwprop.cc (optimize_agr_copyprop): Change into a\n\tforward looking (looking at vdef's uses) instead of a back\n\tlooking (vuse's def).\n\nSigned-off-by: Andrew Pinski \n---\n gcc/tree-ssa-forwprop.cc | 121 +++++++++++++++++++--------------------\n 1 file changed, 60 insertions(+), 61 deletions(-)","diff":"diff --git a/gcc/tree-ssa-forwprop.cc b/gcc/tree-ssa-forwprop.cc\nindex 43b1c9d696fd..2dc77ccba1d7 100644\n--- a/gcc/tree-ssa-forwprop.cc\n+++ b/gcc/tree-ssa-forwprop.cc\n@@ -1344,12 +1344,12 @@ optimize_memcpy_to_memset (gimple_stmt_iterator *gsip, tree dest, tree src, tree\n return true;\n }\n /* Optimizes\n- a = c;\n- b = a;\n+ DEST = SRC;\n+ DEST2 = DEST; # DEST2 = SRC2;\n into\n- a = c;\n- b = c;\n- GSIP is the second statement and SRC is the common\n+ DEST = SRC;\n+ DEST2 = SRC;\n+ GSIP is the first statement and SRC is the common\n between the statements.\n */\n static bool\n@@ -1365,65 +1365,64 @@ optimize_agr_copyprop (gimple_stmt_iterator *gsip)\n if (operand_equal_p (dest, src, 0))\n return false;\n \n- tree vuse = gimple_vuse (stmt);\n- /* If the vuse is the default definition, then there is no store beforehand. */\n- if (SSA_NAME_IS_DEFAULT_DEF (vuse))\n- return false;\n- gimple *defstmt = SSA_NAME_DEF_STMT (vuse);\n- if (!gimple_assign_load_p (defstmt)\n- || !gimple_store_p (defstmt))\n- return false;\n- if (gimple_has_volatile_ops (defstmt))\n- return false;\n-\n- tree dest2 = gimple_assign_lhs (defstmt);\n- tree src2 = gimple_assign_rhs1 (defstmt);\n-\n- /* If the original store is `src2 = src2;` skip over it. */\n- if (operand_equal_p (src2, dest2, 0))\n- return false;\n- if (!operand_equal_p (src, dest2, 0))\n- return false;\n-\n-\n- /* For 2 memory refences and using a temporary to do the copy,\n- don't remove the temporary as the 2 memory references might overlap.\n- Note t does not need to be decl as it could be field.\n- See PR 22237 for full details.\n- E.g.\n- t = *a;\n- *b = t;\n- Cannot be convert into\n- t = *a;\n- *b = *a;\n- Though the following is allowed to be done:\n- t = *a;\n- *a = t;\n- And convert it into:\n- t = *a;\n- *a = *a;\n- */\n- if (!operand_equal_p (src2, dest, 0)\n- && !DECL_P (dest) && !DECL_P (src2))\n- return false;\n-\n- if (dump_file && (dump_flags & TDF_DETAILS))\n+ tree vdef = gimple_vdef (stmt);\n+ imm_use_iterator iter;\n+ gimple *use_stmt;\n+ bool changed = false;\n+ FOR_EACH_IMM_USE_STMT (use_stmt, iter, vdef)\n {\n- fprintf (dump_file, \"Simplified\\n \");\n- print_gimple_stmt (dump_file, stmt, 0, dump_flags);\n- fprintf (dump_file, \"after previous\\n \");\n- print_gimple_stmt (dump_file, defstmt, 0, dump_flags);\n- }\n- gimple_assign_set_rhs_from_tree (gsip, unshare_expr (src2));\n- update_stmt (stmt);\n+ if (!gimple_assign_load_p (use_stmt)\n+\t || !gimple_store_p (use_stmt))\n+\tcontinue;\n+ if (gimple_has_volatile_ops (use_stmt))\n+\tcontinue;\n+ tree dest2 = gimple_assign_lhs (use_stmt);\n+ tree src2 = gimple_assign_rhs1 (use_stmt);\n+ /* If the new store is `src2 = src2;` skip over it. */\n+ if (operand_equal_p (src2, dest2, 0))\n+\tcontinue;\n+ if (!operand_equal_p (dest, src2, 0))\n+\tcontinue;\n+ /* For 2 memory refences and using a temporary to do the copy,\n+\t don't remove the temporary as the 2 memory references might overlap.\n+\t Note t does not need to be decl as it could be field.\n+\t See PR 22237 for full details.\n+\t E.g.\n+\t t = *a; #DEST = SRC;\n+\t *b = t; #DEST2 = SRC2;\n+\t Cannot be convert into\n+\t t = *a;\n+\t *b = *a;\n+\t Though the following is allowed to be done:\n+\t t = *a;\n+\t *a = t;\n+\t And convert it into:\n+\t t = *a;\n+\t *a = *a;\n+ */\n+ if (!operand_equal_p (dest2, src, 0)\n+\t && !DECL_P (dest2) && !DECL_P (src))\n+\tcontinue;\n+ if (dump_file && (dump_flags & TDF_DETAILS))\n+\t{\n+\t fprintf (dump_file, \"Simplified\\n \");\n+\t print_gimple_stmt (dump_file, use_stmt, 0, dump_flags);\n+\t fprintf (dump_file, \"after previous\\n \");\n+\t print_gimple_stmt (dump_file, stmt, 0, dump_flags);\n+\t}\n+ gimple_stmt_iterator gsi = gsi_for_stmt (use_stmt);\n+ gimple_assign_set_rhs_from_tree (&gsi, unshare_expr (src));\n+ update_stmt (use_stmt);\n \n- if (dump_file && (dump_flags & TDF_DETAILS))\n- {\n- fprintf (dump_file, \"into\\n \");\n- print_gimple_stmt (dump_file, stmt, 0, dump_flags);\n+ if (dump_file && (dump_flags & TDF_DETAILS))\n+\t{\n+\t fprintf (dump_file, \"into\\n \");\n+\t print_gimple_stmt (dump_file, use_stmt, 0, dump_flags);\n+\t}\n+ statistics_counter_event (cfun, \"copy prop for aggregate\", 1);\n+ changed = true;\n }\n- statistics_counter_event (cfun, \"copy prop for aggregate\", 1);\n- return true;\n+ return changed;\n }\n \n /* *GSI_P is a GIMPLE_CALL to a builtin function.\n","prefixes":["v1","1/6"]}