From patchwork Fri Feb 11 12:35:17 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Richard Biener X-Patchwork-Id: 51046 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 0694E385843F for ; Fri, 11 Feb 2022 12:35:49 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 0694E385843F DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org; s=default; t=1644582949; bh=NewHQQjG+4V99STfj51eY2bKR2aHVg0VcP788GzGVT8=; h=Date:To:Subject:List-Id:List-Unsubscribe:List-Archive:List-Post: List-Help:List-Subscribe:From:Reply-To:Cc:From; b=dspx/5g+eC7xWnug0ZyXR8aK867TQnmIzQNbXSyi8klwodbservtAv/0RIZHQ/sBQ O21BlRYysvbgbjL/aEofGEXjG3KVFPr1FtxKiv08Um+Zl89Dy4PA2DaPeivWoLkNHy hSjlseVNvxMMQGkxziKRsLBZvj62N06N62hmmw9k= X-Original-To: gcc-patches@gcc.gnu.org Delivered-To: gcc-patches@gcc.gnu.org Received: from smtp-out1.suse.de (smtp-out1.suse.de [195.135.220.28]) by sourceware.org (Postfix) with ESMTPS id 4F4633858D1E for ; Fri, 11 Feb 2022 12:35:19 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 4F4633858D1E Received: from imap2.suse-dmz.suse.de (imap2.suse-dmz.suse.de [192.168.254.74]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (P-521) server-digest SHA512) (No client certificate requested) by smtp-out1.suse.de (Postfix) with ESMTPS id 133992113A; Fri, 11 Feb 2022 12:35:18 +0000 (UTC) Received: from imap2.suse-dmz.suse.de (imap2.suse-dmz.suse.de [192.168.254.74]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (P-521) server-digest SHA512) (No client certificate requested) by imap2.suse-dmz.suse.de (Postfix) with ESMTPS id ED01E13C6D; Fri, 11 Feb 2022 12:35:17 +0000 (UTC) Received: from dovecot-director2.suse.de ([192.168.254.65]) by imap2.suse-dmz.suse.de with ESMTPSA id SnjHOAVYBmJxcgAAMHmgww (envelope-from ); Fri, 11 Feb 2022 12:35:17 +0000 Date: Fri, 11 Feb 2022 13:35:17 +0100 (CET) To: gcc-patches@gcc.gnu.org Subject: [PATCH] [gimplefe] Add vector_mask attribute to get access to vector bools MIME-Version: 1.0 Message-Id: <20220211123517.ED01E13C6D@imap2.suse-dmz.suse.de> X-Spam-Status: No, score=-12.1 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, RCVD_IN_DNSWL_LOW, SPF_HELO_NONE, SPF_PASS, TXREP, T_SCC_BODY_TEXT_LINE 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 Biener via Gcc-patches From: Richard Biener Reply-To: Richard Biener Cc: richard.sandiford@arm.com Errors-To: gcc-patches-bounces+patchwork=sourceware.org@gcc.gnu.org Sender: "Gcc-patches" The following adds __attribute__((vector_mask)) to get access to the corresponding mask type for a vector type. The implementation simply uses truth_type_for so creating a mask type that's not what the target would choose as canonical, say a AVX2 style one when AVX512VL is enabled, is not possible. It might be possible to provide access to that with an optional argument specifying the precision of the bool element. The syntax is as simple as typedef vector_type mask_type __attribute__((vector_mask)); In theory this allows to create unit testcases for vector lowering and ISEL. Bootstrapped and tested on x86_64-unknown-linux-gnu, I've also checked that it works with SVE and -msve-vector-bits=256 but I'm unsure how to build the vector type for a variable size case. Without -msve-vector-bits=256 the testcase is decomposed by vector lowering (I suppose there's no good way to "decompose" a fixed width vector to a variable length one without creating a loop around each stmt...). OK? Thanks, Richard. 2022-02-11 Richard Biener gcc/c-family/ * c-attribs.cc (c_common_attribute_table): Add entry for vector_mask. (handle_vector_mask_attribute): New. gcc/c/ * gimple-fe.cc (c_parser_gimple_statement): Properly parse VEC_COND_EXPRs. gcc/testsuite/ * gcc.dg/gimplefe-48.c: New testcase. --- gcc/c-family/c-attribs.cc | 36 ++++++++++++++++++++++++++++++ gcc/c/gimple-parser.cc | 12 ++++++---- gcc/testsuite/gcc.dg/gimplefe-48.c | 19 ++++++++++++++++ 3 files changed, 63 insertions(+), 4 deletions(-) create mode 100644 gcc/testsuite/gcc.dg/gimplefe-48.c diff --git a/gcc/c-family/c-attribs.cc b/gcc/c-family/c-attribs.cc index 4fb5dbd1409..3849dba90b2 100644 --- a/gcc/c-family/c-attribs.cc +++ b/gcc/c-family/c-attribs.cc @@ -129,6 +129,8 @@ static tree handle_unavailable_attribute (tree *, tree, tree, int, bool *); static tree handle_vector_size_attribute (tree *, tree, tree, int, bool *) ATTRIBUTE_NONNULL(3); +static tree handle_vector_mask_attribute (tree *, tree, tree, int, + bool *) ATTRIBUTE_NONNULL(3); static tree handle_nonnull_attribute (tree *, tree, tree, int, bool *); static tree handle_nonstring_attribute (tree *, tree, tree, int, bool *); static tree handle_nothrow_attribute (tree *, tree, tree, int, bool *); @@ -417,6 +419,8 @@ const struct attribute_spec c_common_attribute_table[] = handle_unavailable_attribute, NULL }, { "vector_size", 1, 1, false, true, false, true, handle_vector_size_attribute, NULL }, + { "vector_mask", 0, 0, false, true, false, true, + handle_vector_mask_attribute, NULL }, { "visibility", 1, 1, false, false, false, false, handle_visibility_attribute, NULL }, { "tls_model", 1, 1, true, false, false, false, @@ -4419,6 +4423,38 @@ handle_vector_size_attribute (tree *node, tree name, tree args, return NULL_TREE; } +/* Handle a "vector_mask" attribute; arguments as in + struct attribute_spec.handler. */ + +static tree +handle_vector_mask_attribute (tree *node, tree name, tree, + int ARG_UNUSED (flags), + bool *no_add_attrs) +{ + *no_add_attrs = true; + if (!flag_gimple) + { + warning (OPT_Wattributes, "%qE attribute ignored", name); + return NULL_TREE; + } + + /* Determine the "base" type to apply the attribute to. */ + tree type = type_for_vector_size (*node); + if (!VECTOR_TYPE_P (type) || VECTOR_BOOLEAN_TYPE_P (type)) + { + warning (OPT_Wattributes, "%qE attribute only supported on " + "non-mask vector types", name); + return NULL_TREE; + } + + tree new_type = truth_type_for (type); + + /* Build back pointers if needed. */ + *node = lang_hooks.types.reconstruct_complex_type (*node, new_type); + + return NULL_TREE; +} + /* Handle the "nonnull" attribute. */ static tree diff --git a/gcc/c/gimple-parser.cc b/gcc/c/gimple-parser.cc index 51ddd86f23a..31075237c98 100644 --- a/gcc/c/gimple-parser.cc +++ b/gcc/c/gimple-parser.cc @@ -860,9 +860,10 @@ c_parser_gimple_statement (gimple_parser &parser, gimple_seq *seq) if (lhs.value != error_mark_node && rhs.value != error_mark_node) { - /* If we parsed a comparison and the next token is a '?' then - parse a conditional expression. */ - if (COMPARISON_CLASS_P (rhs.value) + /* If we parsed a comparison or an identifier and the next token + is a '?' then parse a conditional expression. */ + if ((COMPARISON_CLASS_P (rhs.value) + || SSA_VAR_P (rhs.value)) && c_parser_next_token_is (parser, CPP_QUERY)) { struct c_expr trueval, falseval; @@ -874,7 +875,10 @@ c_parser_gimple_statement (gimple_parser &parser, gimple_seq *seq) if (trueval.value == error_mark_node || falseval.value == error_mark_node) return; - rhs.value = build3_loc (loc, COND_EXPR, TREE_TYPE (trueval.value), + rhs.value = build3_loc (loc, + VECTOR_TYPE_P (TREE_TYPE (rhs.value)) + ? VEC_COND_EXPR : COND_EXPR, + TREE_TYPE (trueval.value), rhs.value, trueval.value, falseval.value); } if (get_gimple_rhs_class (TREE_CODE (rhs.value)) == GIMPLE_INVALID_RHS) diff --git a/gcc/testsuite/gcc.dg/gimplefe-48.c b/gcc/testsuite/gcc.dg/gimplefe-48.c new file mode 100644 index 00000000000..13d534a40c1 --- /dev/null +++ b/gcc/testsuite/gcc.dg/gimplefe-48.c @@ -0,0 +1,19 @@ +/* { dg-do compile } */ +/* { dg-options "-O -fgimple" } */ + +typedef int v8si __attribute__((vector_size(8*sizeof(int)))); +typedef v8si v8sib __attribute__((vector_mask)); + +v8si res; + +void __GIMPLE (ssa) foo (v8si v1, v8si v2, v8si v3, v8si v4) +{ + v8sib tem; + v8si resr; + +__BB(2): + tem_3 = v1_1(D) <= v2_2(D); + resr_4 = tem_3 ? v3_5(D) : v4_6(D); + res = resr_4; + return; +}