From patchwork Mon Feb 13 10:41:30 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Richard Sandiford X-Patchwork-Id: 1741581 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@legolas.ozlabs.org Authentication-Results: legolas.ozlabs.org; spf=pass (sender SPF authorized) smtp.mailfrom=gcc.gnu.org (client-ip=2620:52:3:1:0:246e:9693:128c; helo=sourceware.org; envelope-from=gcc-patches-bounces+incoming=patchwork.ozlabs.org@gcc.gnu.org; receiver=) Authentication-Results: legolas.ozlabs.org; dkim=pass (1024-bit key; unprotected) header.d=gcc.gnu.org header.i=@gcc.gnu.org header.a=rsa-sha256 header.s=default header.b=HIeQaauS; dkim-atps=neutral Received: from sourceware.org (server2.sourceware.org [IPv6:2620:52:3:1:0:246e:9693:128c]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (P-384) server-digest SHA384) (No client certificate requested) by legolas.ozlabs.org (Postfix) with ESMTPS id 4PFgrS2mymz23hX for ; Mon, 13 Feb 2023 21:41:58 +1100 (AEDT) Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id F1FD13844053 for ; Mon, 13 Feb 2023 10:41:52 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org F1FD13844053 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org; s=default; t=1676284913; bh=zOQfc+TYU/T2kLTV7ldkpEVQTc7pHD7Z6z9T910qPfE=; h=To:Subject:References:Date:In-Reply-To:List-Id:List-Unsubscribe: List-Archive:List-Post:List-Help:List-Subscribe:From:Reply-To: From; b=HIeQaauSN4B9mbptAdQHSeISpd2BI+xVvKIRIBoyyiZRWwQeL5qQ8vVsv8PVYI0ec I/aJSGYELsblnCPXDUramaJKl/Z4xEnoZka+4p3gcsMJN5D+Ek/RypleMo395khmDt hqRiybLdDlqSpY5gFcit1fdrGu7wBSD01tKrShI4= X-Original-To: gcc-patches@gcc.gnu.org Delivered-To: gcc-patches@gcc.gnu.org Received: from foss.arm.com (foss.arm.com [217.140.110.172]) by sourceware.org (Postfix) with ESMTP id B11EB385482D for ; Mon, 13 Feb 2023 10:41:32 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org B11EB385482D Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.121.207.14]) by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id CCE034B3 for ; Mon, 13 Feb 2023 02:42:14 -0800 (PST) Received: from localhost (e121540-lin.manchester.arm.com [10.32.99.50]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id D5FBF3F703 for ; Mon, 13 Feb 2023 02:41:31 -0800 (PST) To: gcc-patches@gcc.gnu.org Mail-Followup-To: gcc-patches@gcc.gnu.org, richard.sandiford@arm.com Subject: [Ping] ifcvt: Fix regression in aarch64/fcsel_1.c References: Date: Mon, 13 Feb 2023 10:41:30 +0000 In-Reply-To: (Richard Sandiford's message of "Fri, 03 Feb 2023 09:15:35 +0000") Message-ID: User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/26.3 (gnu/linux) MIME-Version: 1.0 X-Spam-Status: No, score=-35.3 required=5.0 tests=BAYES_00, GIT_PATCH_0, KAM_DMARC_NONE, KAM_DMARC_STATUS, KAM_LAZY_DOMAIN_SECURITY, SPF_HELO_NONE, SPF_NONE, TXREP autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org X-BeenThere: gcc-patches@gcc.gnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Gcc-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-Patchwork-Original-From: Richard Sandiford via Gcc-patches From: Richard Sandiford Reply-To: Richard Sandiford Errors-To: gcc-patches-bounces+incoming=patchwork.ozlabs.org@gcc.gnu.org Sender: "Gcc-patches" Ping for the patch below ---- aarch64/fcsel_1.c contains: double f_2 (double a, double b, double c, double d) { if (a > b) return c; else return d; } which started failing in the GCC 12 timeframe. When it passed, the RTL had the form: [A] (set (reg ret) (reg c)) (set (pc) (if_then_else (gt ...) (label_ref ret) (pc))) edge to ret, fallthru to else else: (set (reg ret) (reg d)) fallthru to ret ret: ...exit... i.e. a branch around. Now the RTL has form: [B] (set (reg ret) (reg d)) (set (pc) (if_then_else (gt ...) (label_ref then) (pc))) edge to then, fallthru to ret ret: ...exit... then: (set (reg ret) (reg c)) edge to ret i.e. a branch out. Both are valid, of course, and there's no easy way to predict which we'll get. But ifcvt canonicalises its representation on: if (cond) goto fallthru else goto non-fallthru That is, it canoncalises on the branch-around case for half-diamonds. It therefore wants to invert the comparison in [B] to get: if (...) goto ret else goto then But that isn't possible for strict FP gt, so the optimisation fails. Canonicalising on the branch-around case seems like the wrong choice for half diamonds. The natural way of expressing a conditional branch is for the label_ref to be the "then" destination and pc to be the "else" destination. And the natural choice of condition seems to be the one under which extra stuff *is* done, rather than the one under which extra stuff *isn't* done. But that decision goes back at least 20 years and it doesn't seem like a good idea to change it in stage 4. This patch instead allows the internal structure to store the condition in inverted form. For simplicity it handles only conditional moves, which is the one case that is needed to fix the known regression. (There are probably unknown regressions too, but still.) Tested on aarch64-linux-gnu and x86_64-linux-gnu. OK to install? Richard gcc/ * ifcvt.h (noce_if_info::cond_inverted): New field. * ifcvt.cc (cond_move_convert_if_block): Swap the then and else values when cond_inverted is true. (noce_find_if_block): Allow the condition to be inverted when handling conditional moves. --- gcc/ifcvt.cc | 31 +++++++++++++++++++------------ gcc/ifcvt.h | 8 ++++++++ 2 files changed, 27 insertions(+), 12 deletions(-) diff --git a/gcc/ifcvt.cc b/gcc/ifcvt.cc index 008796838f7..63ef42b3c34 100644 --- a/gcc/ifcvt.cc +++ b/gcc/ifcvt.cc @@ -4253,6 +4253,9 @@ cond_move_convert_if_block (struct noce_if_info *if_infop, e = dest; } + if (if_infop->cond_inverted) + std::swap (t, e); + target = noce_emit_cmove (if_infop, dest, code, cond_arg0, cond_arg1, t, e); if (!target) @@ -4405,7 +4408,6 @@ noce_find_if_block (basic_block test_bb, edge then_edge, edge else_edge, basic_block then_bb, else_bb, join_bb; bool then_else_reversed = false; rtx_insn *jump; - rtx cond; rtx_insn *cond_earliest; struct noce_if_info if_info; bool speed_p = optimize_bb_for_speed_p (test_bb); @@ -4481,25 +4483,28 @@ noce_find_if_block (basic_block test_bb, edge then_edge, edge else_edge, if (! onlyjump_p (jump)) return FALSE; - /* If this is not a standard conditional jump, we can't parse it. */ - cond = noce_get_condition (jump, &cond_earliest, then_else_reversed); - if (!cond) - return FALSE; - - /* We must be comparing objects whose modes imply the size. */ - if (GET_MODE (XEXP (cond, 0)) == BLKmode) - return FALSE; - /* Initialize an IF_INFO struct to pass around. */ memset (&if_info, 0, sizeof if_info); if_info.test_bb = test_bb; if_info.then_bb = then_bb; if_info.else_bb = else_bb; if_info.join_bb = join_bb; - if_info.cond = cond; + if_info.cond = noce_get_condition (jump, &cond_earliest, + then_else_reversed);; rtx_insn *rev_cond_earliest; if_info.rev_cond = noce_get_condition (jump, &rev_cond_earliest, !then_else_reversed); + if (!if_info.cond && !if_info.rev_cond) + return FALSE; + if (!if_info.cond) + { + std::swap (if_info.cond, if_info.rev_cond); + std::swap (cond_earliest, rev_cond_earliest); + if_info.cond_inverted = true; + } + /* We must be comparing objects whose modes imply the size. */ + if (GET_MODE (XEXP (if_info.cond, 0)) == BLKmode) + return FALSE; gcc_assert (if_info.rev_cond == NULL_RTX || rev_cond_earliest == cond_earliest); if_info.cond_earliest = cond_earliest; @@ -4518,7 +4523,9 @@ noce_find_if_block (basic_block test_bb, edge then_edge, edge else_edge, /* Do the real work. */ - if (noce_process_if_block (&if_info)) + /* ??? noce_process_if_block has not yet been updated to handle + inverted conditions. */ + if (!if_info.cond_inverted && noce_process_if_block (&if_info)) return TRUE; if (HAVE_conditional_move diff --git a/gcc/ifcvt.h b/gcc/ifcvt.h index c6ea244aae1..be1385aabe4 100644 --- a/gcc/ifcvt.h +++ b/gcc/ifcvt.h @@ -86,6 +86,14 @@ struct noce_if_info form as well. */ bool then_else_reversed; + /* True if THEN_BB is conditional on !COND rather than COND. + This is used if: + + - JUMP branches to THEN_BB on COND + - JUMP falls through to JOIN_BB on !COND + - COND cannot be reversed. */ + bool cond_inverted; + /* True if the contents of then_bb and else_bb are a simple single set instruction. */ bool then_simple;