public inbox for gcc-cvs@sourceware.org help / color / mirror / Atom feed
From: Nathan Sidwell <nathan@gcc.gnu.org> To: gcc-cvs@gcc.gnu.org Subject: [gcc/devel/c++-modules] d: Move d_gimplify_expr and dependencies to d-gimplify.cc Date: Fri, 28 Aug 2020 16:04:37 +0000 (GMT) [thread overview] Message-ID: <20200828160437.0F0EE3948024@sourceware.org> (raw) https://gcc.gnu.org/g:e966361c8258d07ef78948d0110b27f2a5a4c69e commit e966361c8258d07ef78948d0110b27f2a5a4c69e Author: Iain Buclaw <ibuclaw@gdcproject.org> Date: Fri Aug 21 11:58:51 2020 +0200 d: Move d_gimplify_expr and dependencies to d-gimplify.cc gcc/d/ChangeLog: * Make-lang.in (D_OBJS): Add d-gimplify.o. * d-lang.cc (empty_modify_p): Move to d-gimplify.cc. (d_gimplify_expr): Likewise. * d-tree.h (d_gimplify_expr): Declare. * d-gimplify.cc: New file. Diff: --- gcc/d/Make-lang.in | 1 + gcc/d/d-gimplify.cc | 187 ++++++++++++++++++++++++++++++++++++++++++++++++++++ gcc/d/d-lang.cc | 160 -------------------------------------------- gcc/d/d-tree.h | 3 + 4 files changed, 191 insertions(+), 160 deletions(-) diff --git a/gcc/d/Make-lang.in b/gcc/d/Make-lang.in index 6c8c611a0f7..18d99abab35 100644 --- a/gcc/d/Make-lang.in +++ b/gcc/d/Make-lang.in @@ -141,6 +141,7 @@ D_OBJS = \ d/d-ctfloat.o \ d/d-diagnostic.o \ d/d-frontend.o \ + d/d-gimplify.o \ d/d-incpath.o \ d/d-lang.o \ d/d-longdouble.o \ diff --git a/gcc/d/d-gimplify.cc b/gcc/d/d-gimplify.cc new file mode 100644 index 00000000000..ed49e5f8dc4 --- /dev/null +++ b/gcc/d/d-gimplify.cc @@ -0,0 +1,187 @@ +/* D-specific tree lowering bits; see also gimple.c. + Copyright (C) 2020 Free Software Foundation, Inc. + +GCC is free software; you can redistribute it and/or modify +it under the terms of the GNU General Public License as published by +the Free Software Foundation; either version 3, or (at your option) +any later version. + +GCC is distributed in the hope that it will be useful, +but WITHOUT ANY WARRANTY; without even the implied warranty of +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +GNU General Public License for more details. + +You should have received a copy of the GNU General Public License +along with GCC; see the file COPYING3. If not see +<http://www.gnu.org/licenses/>. */ + +#include "config.h" +#include "system.h" +#include "coretypes.h" + +#include "dmd/globals.h" + +#include "tree.h" +#include "gimple-expr.h" +#include "gimplify.h" + +#include "d-tree.h" + + +/* Return TRUE if an operand OP of a given TYPE being copied has no data. + The middle-end does a similar check with zero sized types. */ + +static bool +empty_modify_p (tree type, tree op) +{ + tree_code code = TREE_CODE (op); + switch (code) + { + case COMPOUND_EXPR: + return empty_modify_p (type, TREE_OPERAND (op, 1)); + + case CONSTRUCTOR: + /* Non-empty construcors are valid. */ + if (CONSTRUCTOR_NELTS (op) != 0 || TREE_CLOBBER_P (op)) + return false; + break; + + case CALL_EXPR: + /* Leave nrvo alone because it isn't a copy. */ + if (CALL_EXPR_RETURN_SLOT_OPT (op)) + return false; + break; + + default: + /* If the operand doesn't have a simple form. */ + if (!is_gimple_lvalue (op) && !INDIRECT_REF_P (op)) + return false; + break; + } + + return empty_aggregate_p (type); +} + +/* Implements the lang_hooks.gimplify_expr routine for language D. + Do gimplification of D specific expression trees in EXPR_P. */ + +int +d_gimplify_expr (tree *expr_p, gimple_seq *pre_p, + gimple_seq *post_p ATTRIBUTE_UNUSED) +{ + tree_code code = TREE_CODE (*expr_p); + enum gimplify_status ret = GS_UNHANDLED; + tree op0, op1; + tree type; + + switch (code) + { + case INIT_EXPR: + case MODIFY_EXPR: + op0 = TREE_OPERAND (*expr_p, 0); + op1 = TREE_OPERAND (*expr_p, 1); + + if (!error_operand_p (op0) && !error_operand_p (op1) + && (AGGREGATE_TYPE_P (TREE_TYPE (op0)) + || AGGREGATE_TYPE_P (TREE_TYPE (op1))) + && !useless_type_conversion_p (TREE_TYPE (op1), TREE_TYPE (op0))) + { + /* If the back end isn't clever enough to know that the lhs and rhs + types are the same, add an explicit conversion. */ + TREE_OPERAND (*expr_p, 1) = build1 (VIEW_CONVERT_EXPR, + TREE_TYPE (op0), op1); + ret = GS_OK; + } + else if (empty_modify_p (TREE_TYPE (op0), op1)) + { + /* Remove any copies of empty aggregates. */ + gimplify_expr (&TREE_OPERAND (*expr_p, 0), pre_p, post_p, + is_gimple_lvalue, fb_lvalue); + + if (TREE_SIDE_EFFECTS (op1)) + gimplify_and_add (op1, pre_p); + + *expr_p = TREE_OPERAND (*expr_p, 0); + ret = GS_OK; + } + break; + + case ADDR_EXPR: + op0 = TREE_OPERAND (*expr_p, 0); + /* Constructors are not lvalues, so make them one. */ + if (TREE_CODE (op0) == CONSTRUCTOR) + { + TREE_OPERAND (*expr_p, 0) = force_target_expr (op0); + ret = GS_OK; + } + break; + + case CALL_EXPR: + if (CALL_EXPR_ARGS_ORDERED (*expr_p)) + { + /* Strictly evaluate all arguments from left to right. */ + int nargs = call_expr_nargs (*expr_p); + location_t loc = EXPR_LOC_OR_LOC (*expr_p, input_location); + + /* No need to enforce evaluation order if only one argument. */ + if (nargs < 2) + break; + + /* Or if all arguments are already free of side-effects. */ + bool has_side_effects = false; + for (int i = 0; i < nargs; i++) + { + if (TREE_SIDE_EFFECTS (CALL_EXPR_ARG (*expr_p, i))) + { + has_side_effects = true; + break; + } + } + + if (!has_side_effects) + break; + + /* Leave the last argument for gimplify_call_expr. */ + for (int i = 0; i < nargs - 1; i++) + { + tree new_arg = CALL_EXPR_ARG (*expr_p, i); + + /* If argument has a side-effect, gimplify_arg will handle it. */ + if (gimplify_arg (&new_arg, pre_p, loc) == GS_ERROR) + ret = GS_ERROR; + + /* Even if an argument itself doesn't have any side-effects, it + might be altered by another argument in the list. */ + if (new_arg == CALL_EXPR_ARG (*expr_p, i) + && !really_constant_p (new_arg)) + new_arg = get_formal_tmp_var (new_arg, pre_p); + + CALL_EXPR_ARG (*expr_p, i) = new_arg; + } + + if (ret != GS_ERROR) + ret = GS_OK; + } + break; + + case UNSIGNED_RSHIFT_EXPR: + /* Convert op0 to an unsigned type. */ + op0 = TREE_OPERAND (*expr_p, 0); + op1 = TREE_OPERAND (*expr_p, 1); + + type = d_unsigned_type (TREE_TYPE (op0)); + + *expr_p = convert (TREE_TYPE (*expr_p), + build2 (RSHIFT_EXPR, type, convert (type, op0), op1)); + ret = GS_OK; + break; + + case FLOAT_MOD_EXPR: + gcc_unreachable (); + + default: + break; + } + + return ret; +} diff --git a/gcc/d/d-lang.cc b/gcc/d/d-lang.cc index 6412c1c204c..6183389007a 100644 --- a/gcc/d/d-lang.cc +++ b/gcc/d/d-lang.cc @@ -48,8 +48,6 @@ along with GCC; see the file COPYING3. If not see #include "varasm.h" #include "output.h" #include "print-tree.h" -#include "gimple-expr.h" -#include "gimplify.h" #include "debug.h" #include "d-tree.h" @@ -843,164 +841,6 @@ d_post_options (const char ** fn) return false; } -/* Return TRUE if an operand OP of a given TYPE being copied has no data. - The middle-end does a similar check with zero sized types. */ - -static bool -empty_modify_p (tree type, tree op) -{ - tree_code code = TREE_CODE (op); - switch (code) - { - case COMPOUND_EXPR: - return empty_modify_p (type, TREE_OPERAND (op, 1)); - - case CONSTRUCTOR: - /* Non-empty construcors are valid. */ - if (CONSTRUCTOR_NELTS (op) != 0 || TREE_CLOBBER_P (op)) - return false; - break; - - case CALL_EXPR: - /* Leave nrvo alone because it isn't a copy. */ - if (CALL_EXPR_RETURN_SLOT_OPT (op)) - return false; - break; - - default: - /* If the operand doesn't have a simple form. */ - if (!is_gimple_lvalue (op) && !INDIRECT_REF_P (op)) - return false; - break; - } - - return empty_aggregate_p (type); -} - -/* Implements the lang_hooks.gimplify_expr routine for language D. - Do gimplification of D specific expression trees in EXPR_P. */ - -static int -d_gimplify_expr (tree *expr_p, gimple_seq *pre_p, - gimple_seq *post_p ATTRIBUTE_UNUSED) -{ - tree_code code = TREE_CODE (*expr_p); - enum gimplify_status ret = GS_UNHANDLED; - tree op0, op1; - tree type; - - switch (code) - { - case INIT_EXPR: - case MODIFY_EXPR: - op0 = TREE_OPERAND (*expr_p, 0); - op1 = TREE_OPERAND (*expr_p, 1); - - if (!error_operand_p (op0) && !error_operand_p (op1) - && (AGGREGATE_TYPE_P (TREE_TYPE (op0)) - || AGGREGATE_TYPE_P (TREE_TYPE (op1))) - && !useless_type_conversion_p (TREE_TYPE (op1), TREE_TYPE (op0))) - { - /* If the back end isn't clever enough to know that the lhs and rhs - types are the same, add an explicit conversion. */ - TREE_OPERAND (*expr_p, 1) = build1 (VIEW_CONVERT_EXPR, - TREE_TYPE (op0), op1); - ret = GS_OK; - } - else if (empty_modify_p (TREE_TYPE (op0), op1)) - { - /* Remove any copies of empty aggregates. */ - gimplify_expr (&TREE_OPERAND (*expr_p, 0), pre_p, post_p, - is_gimple_lvalue, fb_lvalue); - - if (TREE_SIDE_EFFECTS (op1)) - gimplify_and_add (op1, pre_p); - - *expr_p = TREE_OPERAND (*expr_p, 0); - ret = GS_OK; - } - break; - - case ADDR_EXPR: - op0 = TREE_OPERAND (*expr_p, 0); - /* Constructors are not lvalues, so make them one. */ - if (TREE_CODE (op0) == CONSTRUCTOR) - { - TREE_OPERAND (*expr_p, 0) = force_target_expr (op0); - ret = GS_OK; - } - break; - - case CALL_EXPR: - if (CALL_EXPR_ARGS_ORDERED (*expr_p)) - { - /* Strictly evaluate all arguments from left to right. */ - int nargs = call_expr_nargs (*expr_p); - location_t loc = EXPR_LOC_OR_LOC (*expr_p, input_location); - - /* No need to enforce evaluation order if only one argument. */ - if (nargs < 2) - break; - - /* Or if all arguments are already free of side-effects. */ - bool has_side_effects = false; - for (int i = 0; i < nargs; i++) - { - if (TREE_SIDE_EFFECTS (CALL_EXPR_ARG (*expr_p, i))) - { - has_side_effects = true; - break; - } - } - - if (!has_side_effects) - break; - - /* Leave the last argument for gimplify_call_expr. */ - for (int i = 0; i < nargs - 1; i++) - { - tree new_arg = CALL_EXPR_ARG (*expr_p, i); - - /* If argument has a side-effect, gimplify_arg will handle it. */ - if (gimplify_arg (&new_arg, pre_p, loc) == GS_ERROR) - ret = GS_ERROR; - - /* Even if an argument itself doesn't have any side-effects, it - might be altered by another argument in the list. */ - if (new_arg == CALL_EXPR_ARG (*expr_p, i) - && !really_constant_p (new_arg)) - new_arg = get_formal_tmp_var (new_arg, pre_p); - - CALL_EXPR_ARG (*expr_p, i) = new_arg; - } - - if (ret != GS_ERROR) - ret = GS_OK; - } - break; - - case UNSIGNED_RSHIFT_EXPR: - /* Convert op0 to an unsigned type. */ - op0 = TREE_OPERAND (*expr_p, 0); - op1 = TREE_OPERAND (*expr_p, 1); - - type = d_unsigned_type (TREE_TYPE (op0)); - - *expr_p = convert (TREE_TYPE (*expr_p), - build2 (RSHIFT_EXPR, type, convert (type, op0), op1)); - ret = GS_OK; - break; - - case FLOAT_MOD_EXPR: - gcc_unreachable (); - - default: - break; - } - - return ret; -} - /* Add the module M to the list of modules that may declare GCC builtins. These are scanned after first semantic and before codegen passes. See d_maybe_set_builtin() for the implementation. */ diff --git a/gcc/d/d-tree.h b/gcc/d/d-tree.h index 072de7e6543..b01a133fe62 100644 --- a/gcc/d/d-tree.h +++ b/gcc/d/d-tree.h @@ -593,6 +593,9 @@ extern tree convert_for_condition (tree, Type *); extern tree d_array_convert (Expression *); extern tree d_array_convert (Type *, Expression *); +/* In d-gimplify.cc. */ +extern int d_gimplify_expr (tree *, gimple_seq *, gimple_seq *); + /* In d-incpath.cc. */ extern void add_import_paths (const char *, const char *, bool);
reply other threads:[~2020-08-28 16:04 UTC|newest] Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20200828160437.0F0EE3948024@sourceware.org \ --to=nathan@gcc.gnu.org \ --cc=gcc-cvs@gcc.gnu.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).