The following data structure is the context of top_down_abc: The read_marked_list marks if one bound of one array's dimension is tested or not, for the READ regions. More...

Collaboration diagram for top_down_abc_context_t:

Data Fields
abc_checked	read_marked_list

abc_checked	write_marked_list

hash_table	read_saved_list

hash_table	write_saved_list

hash_table	statement_check_list

persistant_statement_to_control	map

stack	uns

Detailed Description

The following data structure is the context of top_down_abc: The read_marked_list marks if one bound of one array's dimension is tested or not, for the READ regions.

            TOP-DOWN ARRAY BOUND CHECK VERSION

This phase is based on the array region analyses Different strategies :

When a bounds violation is detected, we generate a STOP statement and we don't go down anymore => difficult for debugging all the program's error => finish as soon as possible
Generate comments for errors and warnings => convenient for debugging => can not detect bounds violation for dynamic tests with generated code
Test and STOP statement every where => the {0==1} every where , not very intelligent ?

The first strategy is implemented for this moment

Hypotheses : there is no write effect on the array bound expression.

There was a test for write effect on bound here but I put it away (in effect_on_array_bound.c) because it takes time to calculate the effect but in fact this case is rare The write_marked_list marks if one bound of one array's dimension is tested or not, for the WRITE regions. The saved list keeps the status before go down in the tree. The hash table statement_check_list associates each statement with its list of checks for each bound of array's dimension

Definition at line 90 of file array_bound_check_top_down.c.