Data Structures
struct	edge_info

Functions
void	mcf_smooth_cfg (void)
gcov_type	sum_edge_counts (vec< edge, va_gc > *edges)
void	init_node_map (bool)
void	del_node_map (void)
void	get_working_sets (void)
gcov_type	get_hot_bb_threshold (void)
void	set_hot_bb_threshold (gcov_type)

Function Documentation

void del_node_map ( void )

Delete the CGRAPH_NODE_MAP.

References pointer_map_destroy().

Referenced by ipa_profile(), and tree_profiling().

gcov_type get_hot_bb_threshold ( void )

In predict.c.

Determine the threshold for hot BB counts.

References find_working_set(), min_count, and gcov_working_set_info::min_counter.

Referenced by ipa_profile(), maybe_hot_count_p(), and output_profile_summary().

void get_working_sets ( void )

Fill the working set information into the profile_info structure.

References compute_working_sets(), dump_file, HOST_WIDEST_INT, HOST_WIDEST_INT_PRINT_DEC, gcov_working_set_info::min_counter, and gcov_working_set_info::num_counters.

Referenced by get_exec_counts(), and input_symtab().

void init_node_map ( bool )

Referenced by ipa_profile(), and tree_profiling().

void mcf_smooth_cfg ( void )

Smoothes the initial assigned basic block and edge counts using
   a minimum cost flow algorithm.

Main routine. Smoothes the initial assigned basic block and edge counts using
   a minimum cost flow algorithm, to ensure that the flow consistency rule is
   obeyed: sum of outgoing edges = sum of incoming edges for each basic
   block.

References adjust_cfg_counts(), create_fixup_graph(), delete_fixup_graph(), find_minimum_cost_flow(), and memset().

Referenced by compute_branch_probabilities().

void set_hot_bb_threshold ( gcov_type )

gcov_type sum_edge_counts ( vec< edge, va_gc > * edges )

Data Structures

Functions

Function Documentation