Skip to content

R
riscv-gcc-1
Commit 02bb8bce by Uros Bizjak
Options

re PR rtl-optimization/37544 (Conversion double -> unsigned long long ->… 

re PR rtl-optimization/37544 (Conversion double -> unsigned long long -> unsigned -> double gives wrong results)

	PR rtl-optimization/37544
	* regrename.c (maybe_mode_change): Exit early when copy_mode
	is narrower than orig_mode and narrower than new_mode.

testsuite/ChangeLog:

	PR rtl-optimization/37544
	* gcc.dg/pr37544.c: New test.

From-SVN: r140446
parent d9e74dfc
Showing with 122 additions and 50 deletions
gcc/ChangeLog
2008-09-18 Uros Bizjak <ubizjak@gmail.com>
PR rtl-optimization/37544
* regrename.c (maybe_mode_change): Exit early when copy_mode
is narrower than orig_mode and narrower than new_mode.
2008-09-18 Alexander Monakov <amonakov@ispras.ru>
PR middle-end/37499
......@@ -234,8 +240,8 @@
2008-09-13 Jan Hubicka <jh@suse.cz>
* i386.c (ix86_expand_movmem, ix86_expand_movstr): Avoid processing too
many bytes on misalligned blocks.
* i386.c (ix86_expand_movmem, ix86_expand_movstr): Avoid processing
too many bytes on misalligned blocks.
2008-09-13 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
......@@ -316,7 +322,8 @@
2008-09-12 Sebastian Pop <sebastian.pop@amd.com>
* tree-scalar-evolution.c (set_instantiated_value): Set instantiated_below.
* tree-scalar-evolution.c (set_instantiated_value): Set
instantiated_below.
2008-09-12 Vladimir Makarov <vmakarov@redhat.com>
......@@ -459,7 +466,8 @@
TRUTH_AND_EXPR for boolean (1-bit precision) expressions.
(do_jump) [BIT_IOR_EXPR]: Compile as TRUTH_OR_EXPR.
* tree-flow.h (simplify_stmt_using_ranges): Accept a GSI, return a bool.
* tree-flow.h (simplify_stmt_using_ranges): Accept a GSI, return
a bool.
* tree-ssa-propagate.c (substitute_and_fold): Pass a GSI to
VRP's simplify_stmt_using_ranges. Do simplify_stmt_using_ranges
before finalizing the changes.
......@@ -480,8 +488,8 @@
2008-09-11 Jan Hubicka <jh@suse.cz>
* cgraph.c (cgraph_release_function_body): Plug memory leak on
ipa_transforms_to_apply and CFG; ggc_free gimple df and struct function
so we are sure we do not dangle it forever.
ipa_transforms_to_apply and CFG; ggc_free gimple df and struct
function so we are sure we do not dangle it forever.
* function.c (outer_function_chain, find_function_data): Remove.
(function_context_stack): New.
(push_function_context, pop_function_context): Update.
......@@ -538,7 +546,8 @@
register_overhead, free_overhead, vec_heap_free): New functions.
(vec_gc_o_reserve_1): ggc_free when resizing to 0.
(vec_heap_o_reserve_1): free when resizing to 0; add statistics.
(cmp_statistic, add_statistics, dump_vec_loc_statistics): New functions.
(cmp_statistic, add_statistics, dump_vec_loc_statistics):
New functions.
2008-09-11 Ira Rosen <irar@il.ibm.com>
......@@ -692,8 +701,8 @@
calls.
PR target/37438
* config/i386/i386.md (zero_extendqihi2_movzbl): Enable when optimizing
for size, not speed.
* config/i386/i386.md (zero_extendqihi2_movzbl): Enable when
optimizing for size, not speed.
PR rtl-optimization/37408
* function.c (assign_parm_find_stack_rtl): Set correct MEM_SIZE
......@@ -702,18 +711,20 @@
2008-09-09 Sebastian Pop <sebastian.pop@amd.com>
PR tree-optimization/37375
* tree-scalar-evolution.c (scev_info_str): Add field instantiated_below.
* tree-scalar-evolution.c (scev_info_str): Add field
instantiated_below.
(new_scev_info_str, eq_scev_info, find_var_scev_info,
set_scalar_evolution, get_scalar_evolution, get_instantiated_value,
set_instantiated_value): Pass instantiated_below.
(analyze_scalar_evolution_1, analyze_scalar_evolution): Update calls to
above functions.
(analyze_scalar_evolution_1, analyze_scalar_evolution): Update calls
to above functions.
(instantiate_scev_1, instantiate_scev): Pass a basic block above which
the definitions are not instantiated.
* tree-scalar-evolution.h (instantiate_scev): Update declaration.
(block_before_loop): New.
* tree-data-ref.c (dr_analyze_indices): Update uses of instantiate_scev.
* tree-data-ref.c (dr_analyze_indices): Update uses of
instantiate_scev.
* graphite.c (block_before_scop): New.
(loop_affine_expr, stmt_simple_for_scop_p, harmful_stmt_in_bb): Pass a
basic block, not a loop for determining the parameters.
......@@ -1671,8 +1682,8 @@
openmp applications.
* pa-hpux11.h (LIB_SPEC): Likewise.
* ira.c (setup_eliminable_regset): Only declare i if ELIMINABLE_REGS is
defined.
* ira.c (setup_eliminable_regset): Only declare i if ELIMINABLE_REGS
is defined.
2008-09-01 Jan Hubicka <jh@suse.cz>
......@@ -1782,8 +1793,7 @@
* sel-sched.h, sel-sched-dump.h, sel-sched-ir.h, sel-sched.c,
sel-sched-dump.c, sel-sched-ir.c: New files.
* Makefile.in (OBJS-common): Add selective scheduling object
files.
* Makefile.in (OBJS-common): Add selective scheduling object files.
(sel-sched.o, sel-sched-dump.o, sel-sched-ir.o): New entries.
(SEL_SCHED_IR_H, SEL_SCHED_DUMP_H): New entries.
(sched-vis.o): Add dependency on $(INSN_ATTR_H).
......@@ -1905,8 +1915,7 @@
(setup_sched_dump): New function.
(haifa_init_only_bb): New static function.
(haifa_speculate_insn): New static function.
(try_ready): Use haifa_* instead of speculate_insn and
change_pattern.
(try_ready): Use haifa_* instead of speculate_insn and change_pattern.
(extend_ready, extend_all): Remove.
(sched_extend_ready_list, sched_finish_ready_list): New functions.
(create_check_block_twin, add_to_speculative_block): Use
......@@ -1936,8 +1945,7 @@
(current_sched_info_var): Remove.
(move_block_after_check): Use common_sched_info.
(haifa_luid_for_non_insn): New static function.
(init_before_recovery): Use haifa_init_only_bb instead of
add_block.
(init_before_recovery): Use haifa_init_only_bb instead of add_block.
(increase_insn_priority): New.
* modulo-sched.c: (issue_rate): Remove static declaration.
(sms_sched_info): Change type to haifa_sched_info.
......@@ -2351,7 +2359,8 @@
(noce_try_cmove_arith): Likewise.
(noce_try_sign_mask): LIkewise.
* expr.c (compress_float_constant): Update rtx_cost calls.
* tree-ssa-address.c (most_expensive_mult_to_index): Add speed argument.
* tree-ssa-address.c (most_expensive_mult_to_index): Add speed
argument.
(addr_to_parts): Likewise.
(create_mem_ref): Likewise.
* dse.c (find_shift_sequence): Add speed argument.
......@@ -2370,7 +2379,8 @@
(expand_mult_highpart): Use profile.
(expand_smod_pow2): Use profile.
(expand_divmod): Use profile.
* simplify-rtx.c (simplify_binary_operation_1): Update call of rtx_cost.
* simplify-rtx.c (simplify_binary_operation_1): Update call of
rtx_cost.
* loop-invariant.c (create_new_invariant): Use profile.
(gain_for_invariant): Add speed parameter.
(best_gain_for_invariant): Likewise.
......@@ -2638,7 +2648,8 @@
* config/sparc/sparc.h (BRANCH_COST): Update.
* config/m32r/m32r.h (BRANCH_COST): Update.
* config/i386/i386.h (BRANCH_COST): Update.
* config/i386/i386.c (ix86_expand_int_movcc): Update use of BRANCH_COST.
* config/i386/i386.c (ix86_expand_int_movcc): Update use of
BRANCH_COST.
* config/sh/sh.h (BRANCH_COST): Update.
* config/pdp11/pdp11.h (BRANCH_COST): Update.
* config/avr/avr.h (BRANCH_COST): Update.
......@@ -2688,7 +2699,8 @@
* tree-ssa-pre.c (do_pre): Use optimize_function_for_speed_p.
* tree-predcom.c (tree_predictive_commoning): Use
optimize_loop_for_speed_p.
* varasm.c (assemble_start_function): Use optimize_function_for_speed_p.
* varasm.c (assemble_start_function): Use
optimize_function_for_speed_p.
* bb-reorder.c (rest_of_handle_reorder_blocks): Likewise.
* predict.c (optimize_loop_for_speed_p): Fix walk.
......@@ -2909,16 +2921,17 @@
2008-08-29 Jan Hubicka <jh@suse.cz>
* tree.c (build_function_type_skip_args, build_function_decl_skip_args):
New functions.
* tree.h (build_function_type_skip_args, build_function_decl_skip_args):
Declare.
* tree.c (build_function_type_skip_args,
build_function_decl_skip_args): New functions.
* tree.h (build_function_type_skip_args,
build_function_decl_skip_args): Declare.
* gimple.c (giple_copy_call_skip_args): New function.
(giple_copy_call_skip_args): Declare.
* cgraph.h (cgraph_function_versioning): Add skip_args arugmnet
* ipa-cp.c (ipcp_node_not_modifiable_p): Rename to ...
(ipcp_node_modifiable_p): ... this one; use tree_versionable_function_p.
(ipcp_node_modifiable_p): ... this one; use
tree_versionable_function_p.
(ipcp_create_replace_map): Improve debug output.
(ipcp_need_redirect_p): Return false when not clonning.
(ipcp_update_callgraph): Skip args.
......@@ -2927,13 +2940,15 @@
* cgraphunit.c (cgraph_function_versioning): Add skip_args argument.
(save_inline_function_body): Update call of tree_function_versioning.
* ipa-prop.c (ipa_edge_removal_hook): Do not ICE on unanalyzed nodes.
* tree-inline.c (copy_arguments_for_versioning): Add skip_args argument.
* tree-inline.c (copy_arguments_for_versioning): Add skip_args
argument.
(tree_function_versioning): Likewise.
* tree-inline.h (tree_function_versioning): Update prototype.
2008-08-29 Jan Hubicka <jh@suse.cz>
* loop-unswitch.c (unswitch_single_loop): Use optimize_loop_for_speed_p.
* loop-unswitch.c (unswitch_single_loop): Use
optimize_loop_for_speed_p.
* tree-ssa-threadupdate.c (mark_threaded_blocks):
Use optimize_function_for_size_p.
* tracer.c (ignore_bb_p): Use optimize_bb_for_size_p.
......@@ -2947,7 +2962,8 @@
* builtins.c (fold_builtin_cabs): Use optimize_function_for_speed_p.
(fold_builtin_strcpy, fold_builtin_fputs): Use
optimize_function_for_size_p.
* fold-const.c (tree_swap_operands_p): Use optimize_function_for_size_p.
* fold-const.c (tree_swap_operands_p): Use
optimize_function_for_size_p.
* recog.c (relax_delay_slots): Likewise.
* tree-ssa-math-opts.c (replace_reciprocal):
Use optimize_bb_for_speed_p.
......@@ -2963,11 +2979,12 @@
work it out from profile.
* tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely): Use
optimize_loop_for_speed_p.
* predict.c (optimize_bb_for_size_p, optimize_bb_for_speed_p): Constify
argument.
* predict.c (optimize_bb_for_size_p, optimize_bb_for_speed_p):
Constify argument.
(optimize_loop_nest_for_size_p, optimize_loop_nest_for_speed_p): New.
* tree-parloops.c (parallelize_loops): Use optimize_loop_for_size_p.
* tree-eh.c (decide_copy_try_finally): Use optimize_function_for_size_p.
* tree-eh.c (decide_copy_try_finally): Use
optimize_function_for_size_p.
* local-alloc.c (block_alloc): Pass BB pointer.
(find_free_reg): Add BB pointer, use optimize_bb_for_size_p.
* gcse.c (gcse_main): Use optimize_function_for_size_p.
......@@ -4365,8 +4382,8 @@
(update_call_notes_after_inlining): Expect ADDR_EXPR in operand.
* ipa-prop.h (jump_func_type): Remove IPA_CONST_REF.
(jump_func_type): Remove IPA_CONST_VALUE_REF.
* tree-inline.c (tree_function_versioning): Add variables referenced by
replacing trees.
* tree-inline.c (tree_function_versioning): Add variables referenced
by replacing trees.
2008-08-23 Jan Hubicka <jh@suse.cz>
......@@ -4766,9 +4783,9 @@
(vect_get_and_check_slp_defs): Detect patterns.
(vect_build_slp_tree): Add an argument, don't fail in case of multiple
types.
(vect_analyze_slp_instance): Don't fail in case of multiple types. Call
vect_build_slp_tree with correct arguments. Calculate unrolling factor
according to the smallest type in the loop.
(vect_analyze_slp_instance): Don't fail in case of multiple types.
Call vect_build_slp_tree with correct arguments. Calculate unrolling
factor according to the smallest type in the loop.
(vect_detect_hybrid_slp_stmts): Include statements from patterns.
* tree-vect-patterns.c (vect_recog_widen_mult_pattern): Call
supportable_widening_operation with correct arguments.
......@@ -5581,7 +5598,8 @@
2008-08-09 Richard Sandiford <rdsandiford@googlemail.com>
* config/mips/mips.c (mips_unspec_address_offset): Move earlier in file.
* config/mips/mips.c (mips_unspec_address_offset): Move earlier
in file.
(mips_unspec_address, mips_unspec_offset_high): Likewise.
(mips_ok_for_lazy_binding_p, mips_load_call_address): Likewise.
(mips16_cfun_returns_in_fpr_p): Likewise.
......@@ -6150,7 +6168,8 @@
* configure.ac (HAVE_GAS_CFI_DIRECTIVE): New.
* config.in, configure: Rebuild.
* dwarf2asm.c (dw2_asm_output_data_raw): New.
(dw2_asm_output_data_uleb128_raw, dw2_asm_output_data_sleb128_raw): New.
(dw2_asm_output_data_uleb128_raw, dw2_asm_output_data_sleb128_raw):
New.
(dw2_force_const_mem): Externalize.
* dwarf2asm.h: Update.
* dwarf2out.c (dwarf2out_cfi_label): If flag_dwarf2_cfi_asm, don't
......@@ -6185,8 +6204,10 @@
(MFC_MIN_DMA_LIST_ELEMENTS): New define.
(MFC_MAX_DMA_LIST_ELEMENTS): Likewise.
(MFC_MIN_DMA_LIST_SIZE): Redefine in terms of MFC_MIN_DMA_LIST_ELEMENTS.
(MFC_MAX_DMA_LIST_SIZE): Redefine in terms of MFC_MAX_DMA_LIST_ELEMENTS.
(MFC_MIN_DMA_LIST_SIZE): Redefine in terms of
MFC_MIN_DMA_LIST_ELEMENTS.
(MFC_MAX_DMA_LIST_SIZE): Redefine in terms of
MFC_MAX_DMA_LIST_ELEMENTS.
(MFC_START_ENABLE): Remove PPU-only define.
(MFC_PUTS_CMD, MFC_PUTFS_CMD, MFC_PUTBS_CMD): Likewise.
......@@ -6861,7 +6882,8 @@
* configure.ac: Use the m4_do macro to concatenate the warnings into
one string in ACX_PROG_CC_WARNING_OPTS,
ACX_PROG_CC_WARNING_ALMOST_PEDANTIC and ACX_PROG_CC_WARNINGS_ARE_ERRORS.
ACX_PROG_CC_WARNING_ALMOST_PEDANTIC and
ACX_PROG_CC_WARNINGS_ARE_ERRORS.
* configure: Regenerate.
2008-07-28 Richard Guenther <rguenther@suse.de>
......@@ -7363,7 +7385,8 @@
(ipa_analyze_params_uses): New function.
(ipa_free_node_params_substructures): Also free the param_calls linked
list.
(ipa_node_duplication_hook): Also duplicate the param_calls linked list.
(ipa_node_duplication_hook): Also duplicate the param_calls
linked list.
(ipa_print_node_param_flags): New function.
(ipa_print_all_params_modified): Renamed to ipa_print_all_param_flags.
(ipa_print_all_param_flags): Calls ipa_print_node_param_flags.
......@@ -7912,15 +7935,16 @@
not optimizing so unused user vars remains visible.
* common.opt (flag_tree_ter): Always enable by default.
* tree-ssa-ter.c: Include flags.h
(is_replaceable_p): Check that locations match; when aliasing is missing
be conservative about loads.
(is_replaceable_p): Check that locations match; when aliasing
is missing be conservative about loads.
* tree-optimize.c (gate_init_datastructures): Remove.
(pass_init_datastructures): New.
* passes.c: Reorder passes so we always go into SSA.
2008-07-19 Jan Hubicka <jh@suse.cz>
* doc/extend.texi (flatten attribute): Remove note about unit-at-a-time
* doc/extend.texi (flatten attribute): Remove note about
unit-at-a-time.
* doc/invoke.texi (--combine): Likewise.
(-finline-functions-called-once): Update levels when enabled.
(-funit-at-a-time): Document new behaviour.
......
gcc/regrename.c
......@@ -1314,6 +1314,10 @@ maybe_mode_change (enum machine_mode orig_mode, enum machine_mode copy_mode,
enum machine_mode new_mode, unsigned int regno,
unsigned int copy_regno ATTRIBUTE_UNUSED)
{
if (GET_MODE_SIZE (copy_mode) < GET_MODE_SIZE (orig_mode)
&& GET_MODE_SIZE (copy_mode) < GET_MODE_SIZE (new_mode))
return NULL_RTX;
if (orig_mode == new_mode)
return gen_rtx_raw_REG (new_mode, regno);
else if (mode_change_ok (orig_mode, new_mode, regno))
......
gcc/testsuite/ChangeLog
2008-09-18 Uros Bizjak <ubizjak@gmail.com>
PR rtl-optimization/37544
* gcc.dg/pr37544.c: New test.
2008-09-17 Janis Johnson <janis187@us.ibm.com>
PR testsuite/25241
......
gcc/testsuite/gcc.dg/pr37544.c 0 → 100644
/* { dg-do run } */
/* { dg-options "-O2" } */
/* { dg-options "-O2 -msse2 -mtune=core2 -mfpmath=387" { target { i?86-*-* x86_64-*-* } } } */
#ifdef __i386__
#include "cpuid.h"
#endif
extern void abort (void);
int main(void)
{
double arr[1000];
double a, b;
int i;
#ifdef __i386__
unsigned int eax, ebx, ecx, edx;
if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
return 0;
/* Run SSE2 test only if host has SSE2 support. */
if (!(edx & bit_SSE2))
return 0;
#endif
for (i = 0; i < 1000; i++)
arr[i] = 4294967296.0 + (double)i;
a = arr[0];
b = (unsigned int)((unsigned long long int)a % 4294967296ULL);
if (b >= 4294967296.0)
abort ();
return 0;
}
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment