From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 7531 invoked by alias); 28 Feb 2018 17:51:39 -0000 Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org Received: (qmail 7512 invoked by uid 89); 28 Feb 2018 17:51:39 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-11.0 required=5.0 tests=AWL,BAYES_00,GIT_PATCH_2,GIT_PATCH_3,KAM_ASCII_DIVIDERS,RCVD_IN_DNSWL_NONE,SPF_PASS autolearn=ham version=3.3.2 spammy= X-HELO: mail-wm0-f54.google.com Received: from mail-wm0-f54.google.com (HELO mail-wm0-f54.google.com) (74.125.82.54) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Wed, 28 Feb 2018 17:51:37 +0000 Received: by mail-wm0-f54.google.com with SMTP id q83so6672277wme.5 for ; Wed, 28 Feb 2018 09:51:37 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:mail-followup-to:subject:date:message-id :user-agent:mime-version; bh=UNuqEml/na8bXMhrsLbc7TNDGjo26/ysTd6XHmLGqek=; b=bt5mAEg7r87YyVgDSGd8veOYl2MyiMLckBqWbOlUG7wiY1+/10dLjUzfimvmQCO+FJ uOb8ltK+KDOcfTL+1AOeqefd/haliy+yCznyZx7UzbsNT3xqawUQVzxjxHjaFScRoarD AuSU8THl1nU5Z6dMH22NeZuEvN3FzrkQ5c8XotKyOnfaaZ/+dlgL7hoEKmfHD5WcW4N6 TcZeNWcRbF4XyfyoBAUA8GZ8glDINRjHJGV9S+IKeCxUOkA2wXIgKBQ5JVPzveouQmgy dKDDQ6IpcawAXdj74ZmA5dcCmVhcZFlRo4DpWbT/vRpnKn+7BXHVkk12SEmVfLVG1I0Q Waew== X-Gm-Message-State: APf1xPBO4zdEnYPmOQAnCt9fO09Cu6vopuIQAgrfk7DbmEE0ZLMmGNaJ TOjTGPQ2GGvzgm6fyXTp3FD8AVSrXtI= X-Google-Smtp-Source: AG47ELumqb60/Bt8QcxcRqPVZDzE1udvQSNF+/MhZhZpDxZop5V7gLZ54aMe4VxCgZ7VAbIcpqQp9A== X-Received: by 10.28.147.12 with SMTP id v12mr14607098wmd.139.1519840295140; Wed, 28 Feb 2018 09:51:35 -0800 (PST) Received: from localhost (94.197.120.165.threembb.co.uk. [94.197.120.165]) by smtp.gmail.com with ESMTPSA id e18sm4675587wmc.21.2018.02.28.09.51.32 for (version=TLS1_2 cipher=ECDHE-RSA-CHACHA20-POLY1305 bits=256/256); Wed, 28 Feb 2018 09:51:33 -0800 (PST) From: Richard Sandiford To: gcc-patches@gcc.gnu.org Mail-Followup-To: gcc-patches@gcc.gnu.org, richard.sandiford@linaro.org Subject: Tighten use of HARD_FRAME_POINTER_REGNUM in alias.c (PR 84538) Date: Wed, 28 Feb 2018 17:51:00 -0000 Message-ID: <87efl580kb.fsf@linaro.org> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/25.3 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain X-SW-Source: 2018-02/txt/msg01608.txt.bz2 RTL code needs to be consistent about whether it uses the stack pointer, the frame pointer or the argument pointer to access a given part of the frame. alias.c used this to divide accesses into three independent areas. The problem in the PR is that we did this for HARD_FRAME_POINTER_REGNUM even when the register wasn't being used as a frame pointer. We can't do that because the frame pointer is then just any old allocatable register and could certainly point to info accessed through the argument pointer or stack pointer. In this case the bug triggered wrong code, but diffing the before and after assembly output for one target per CPU shows that the bug also introduced an unnecessary scheduling dependence between B and C in things like: A: (set hfp (symbol_ref "...")) B: ...(mem hfp)... C: (clobber (mem sp)) where the hard frame pointer points to something that is obviously distinct from the frame. So it looks like this patch is both a correctness fix and a very minor scheduling improvement. Tested on aarch64-linux-gnu, and by diffing output as above. OK to install? Richard 2018-02-28 Richard Sandiford gcc/ PR rtl-optimization/84538 * alias.c (init_alias_target): Add commentary. (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM a unique base value if the frame pointer is not eliminated to the stack pointer. gcc/testsuite/ PR rtl-optimization/84538 * gcc.dg/torture/pr84538.c: New test. Index: gcc/alias.c =================================================================== --- gcc/alias.c 2018-01-14 08:42:44.497155977 +0000 +++ gcc/alias.c 2018-02-28 17:47:36.486754143 +0000 @@ -3191,12 +3191,21 @@ init_alias_target (void) && targetm.hard_regno_mode_ok (i, Pmode)) static_reg_base_value[i] = arg_base_value; + /* RTL code is required to be consistent about whether it uses the + stack pointer, the frame pointer or the argument pointer to + access a given area of the frame. We can therefore use the + base address to distinguish between the different areas. */ static_reg_base_value[STACK_POINTER_REGNUM] = unique_base_value (UNIQUE_BASE_VALUE_SP); static_reg_base_value[ARG_POINTER_REGNUM] = unique_base_value (UNIQUE_BASE_VALUE_ARGP); static_reg_base_value[FRAME_POINTER_REGNUM] = unique_base_value (UNIQUE_BASE_VALUE_FP); + + /* The above rules extend post-reload, with eliminations applying + consistently to each of the three pointers. Cope with cases in + which the frame pointer is eliminated to the hard frame pointer + rather than the stack pointer. */ if (!HARD_FRAME_POINTER_IS_FRAME_POINTER) static_reg_base_value[HARD_FRAME_POINTER_REGNUM] = unique_base_value (UNIQUE_BASE_VALUE_HFP); @@ -3329,7 +3338,14 @@ init_alias_analysis (void) /* Initialize the alias information for this pass. */ for (i = 0; i < FIRST_PSEUDO_REGISTER; i++) - if (static_reg_base_value[i]) + if (static_reg_base_value[i] + /* Don't treat the hard frame pointer as special if we + eliminated the frame pointer to the stack pointer instead. */ + && !(i == HARD_FRAME_POINTER_REGNUM + && reload_completed + && !frame_pointer_needed + && targetm.can_eliminate (FRAME_POINTER_REGNUM, + STACK_POINTER_REGNUM))) { new_reg_base_value[i] = static_reg_base_value[i]; bitmap_set_bit (reg_seen, i); Index: gcc/testsuite/gcc.dg/torture/pr84538.c =================================================================== --- /dev/null 2018-02-26 10:26:41.624847060 +0000 +++ gcc/testsuite/gcc.dg/torture/pr84538.c 2018-02-28 17:47:36.491754008 +0000 @@ -0,0 +1,16 @@ +/* { dg-do run } */ +/* { dg-additional-options "-fno-omit-frame-pointer -w" } */ + +#define SIZE 8 + +main() +{ + int a[SIZE] = {1}; + int i; + + for (i = 1; i < SIZE; i++) + if (a[i] != 0) + abort(); + + exit (0); +}