From patchwork Tue Nov 30 17:34:57 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Richard Sandiford X-Patchwork-Id: 48302 Return-Path: X-Original-To: patchwork@sourceware.org Delivered-To: patchwork@sourceware.org Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id CBE15385AC3A for ; Tue, 30 Nov 2021 17:35:28 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org CBE15385AC3A DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org; s=default; t=1638293728; bh=5WI74hMSIR2h0wL4VHBlhYKuM5kTeEab7Gh5aQ0ZNKY=; h=To:Subject:Date:List-Id:List-Unsubscribe:List-Archive:List-Post: List-Help:List-Subscribe:From:Reply-To:From; b=VTrTSjZ8JNgrIkOaXSJ3GnfJT22wpnabF0Al0xOVBLLKggen07LeS9GuZdK/pi2j5 8lJTohYVfZSzeiQB3rMjlRWJk/yR/99eQZoG7yxshOkLFjYV/tFs/jKJiWp9PrK/NQ A/Hyei6XppbriAeimS9I6pw9nqo2+IJHvry9SmcQ= 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 9253D3858D28 for ; Tue, 30 Nov 2021 17:34:59 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 9253D3858D28 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 E9904106F for ; Tue, 30 Nov 2021 09:34:58 -0800 (PST) Received: from localhost (unknown [10.32.98.88]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id 90A3B3F694 for ; Tue, 30 Nov 2021 09:34:58 -0800 (PST) To: gcc-patches@gcc.gnu.org Mail-Followup-To: gcc-patches@gcc.gnu.org, richard.sandiford@arm.com Subject: [committed] vect: Fix ncopies calculation for emulated gather/scatter [PR103494] Date: Tue, 30 Nov 2021 17:34:57 +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=-12.4 required=5.0 tests=BAYES_00, GIT_PATCH_0, KAM_DMARC_STATUS, SPF_HELO_NONE, SPF_PASS, TXREP autolearn=ham autolearn_force=no version=3.4.4 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) 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+patchwork=sourceware.org@gcc.gnu.org Sender: "Gcc-patches" I was too eager about removing ncopies calculations in g:10833849b55. When emulating gather/scatter, the offset ncopies can be different from the data ncopies. This patch restores the original calculation. Tested on aarch64-linux-gnu and x86_64-linux-gnu. Pushed as obvious, since it's essentially reverting part of my earlier patch (except for obvious adjustments to keep slp_node). Richard gcc/ PR tree-optimization/103494 * tree-vect-stmts.c (vect_get_gather_scatter_ops): Remove ncopies argument and calculate ncopies from gs_info->offset_vectype where necessary. (vectorizable_store, vectorizable_load): Update accordingly. gcc/testsuite/ PR tree-optimization/103494 * gcc.dg/vect/pr103494.c: New test. * g++.dg/vect/pr103494.cc: Likewise. --- gcc/testsuite/g++.dg/vect/pr103494.cc | 26 ++++++++++++++++++++++++++ gcc/testsuite/gcc.dg/vect/pr103494.c | 14 ++++++++++++++ gcc/tree-vect-stmts.c | 21 ++++++++++++--------- 3 files changed, 52 insertions(+), 9 deletions(-) create mode 100644 gcc/testsuite/g++.dg/vect/pr103494.cc create mode 100644 gcc/testsuite/gcc.dg/vect/pr103494.c diff --git a/gcc/testsuite/g++.dg/vect/pr103494.cc b/gcc/testsuite/g++.dg/vect/pr103494.cc new file mode 100644 index 00000000000..c0b078105c2 --- /dev/null +++ b/gcc/testsuite/g++.dg/vect/pr103494.cc @@ -0,0 +1,26 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-O3" } */ + +void glFinish(); +struct _Vector_base { + struct { + unsigned _M_start; + } _M_impl; +}; +class vector : _Vector_base { +public: + vector(long) {} + unsigned *data() { return &_M_impl._M_start; } +}; +void *PutBitsIndexedImpl_color_table; +int PutBitsIndexedImpl_dstRectHeight; +char *PutBitsIndexedImpl_src_ptr; +void PutBitsIndexedImpl() { + vector unpacked_buf(PutBitsIndexedImpl_dstRectHeight); + unsigned *dst_ptr = unpacked_buf.data(); + for (int x; x; x++) { + char i = *PutBitsIndexedImpl_src_ptr++; + dst_ptr[x] = static_cast(PutBitsIndexedImpl_color_table)[i]; + } + glFinish(); +} diff --git a/gcc/testsuite/gcc.dg/vect/pr103494.c b/gcc/testsuite/gcc.dg/vect/pr103494.c new file mode 100644 index 00000000000..b544bf2379c --- /dev/null +++ b/gcc/testsuite/gcc.dg/vect/pr103494.c @@ -0,0 +1,14 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-O3" } */ + +typedef int T1; +typedef signed char T2; + +T1 +f (T1 *d, T2 *x, int n) +{ + unsigned char res = 0; + for (int i = 0; i < n; ++i) + res += d[x[i]]; + return res; +} diff --git a/gcc/tree-vect-stmts.c b/gcc/tree-vect-stmts.c index 8642acbc0b4..9726450ab2d 100644 --- a/gcc/tree-vect-stmts.c +++ b/gcc/tree-vect-stmts.c @@ -2962,8 +2962,7 @@ vect_build_gather_load_calls (vec_info *vinfo, stmt_vec_info stmt_info, static void vect_get_gather_scatter_ops (loop_vec_info loop_vinfo, class loop *loop, stmt_vec_info stmt_info, - slp_tree slp_node, unsigned int ncopies, - gather_scatter_info *gs_info, + slp_tree slp_node, gather_scatter_info *gs_info, tree *dataref_ptr, vec *vec_offset) { gimple_seq stmts = NULL; @@ -2978,9 +2977,13 @@ vect_get_gather_scatter_ops (loop_vec_info loop_vinfo, if (slp_node) vect_get_slp_defs (SLP_TREE_CHILDREN (slp_node)[0], vec_offset); else - vect_get_vec_defs_for_operand (loop_vinfo, stmt_info, ncopies, - gs_info->offset, vec_offset, - gs_info->offset_vectype); + { + unsigned ncopies + = vect_get_num_copies (loop_vinfo, gs_info->offset_vectype); + vect_get_vec_defs_for_operand (loop_vinfo, stmt_info, ncopies, + gs_info->offset, vec_offset, + gs_info->offset_vectype); + } } /* Prepare to implement a grouped or strided load or store using @@ -8149,8 +8152,8 @@ vectorizable_store (vec_info *vinfo, else if (STMT_VINFO_GATHER_SCATTER_P (stmt_info)) { vect_get_gather_scatter_ops (loop_vinfo, loop, stmt_info, - slp_node, ncopies, &gs_info, - &dataref_ptr, &vec_offsets); + slp_node, &gs_info, &dataref_ptr, + &vec_offsets); vec_offset = vec_offsets[0]; } else @@ -9454,8 +9457,8 @@ vectorizable_load (vec_info *vinfo, else if (STMT_VINFO_GATHER_SCATTER_P (stmt_info)) { vect_get_gather_scatter_ops (loop_vinfo, loop, stmt_info, - slp_node, ncopies, &gs_info, - &dataref_ptr, &vec_offsets); + slp_node, &gs_info, &dataref_ptr, + &vec_offsets); } else dataref_ptr