This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

PR middle-end/36902 Array bound warning with dead code after optimization

Updated to a recent revision, bootstrapped and regression tested on
x86_64-unknown-linux-gnu with
--enable-languages=all,obj-c++ --enable-decimal-float.

2009-04-09  Manuel López-Ibáñez  <>

	PR middle-end/36902
	* tree-vrp.c (check_array_ref): Pass a location_t instead of a
	pointer. Use warning_at instead of warning.
	(search_for_addr_array): Likewise.
	(check_array_bounds): Likewise.
	(check_all_array_refs): Check that the incoming edge is not in the
	list of edges to be removed.
	(check_all_array_refs): Avoid the temporal pointer.
	(vrp_visit_cond_stmt): Fix typo.
	(simplify_switch_using_ranges): Handle the case where the switch
	index is an integer constant.
	* gcc.dg/pr36902.c: New.

OK for trunk?


2008/10/26 Manuel López-Ibáñez <>:
> Bootstrapped and regression tested on x86_64-unknown-linux-gnu with
> --enable-languages=all,obj-c++ --enable-decimal-float.
> This could be improved further by:
> * Marking the edges to be removed in some way, so we don't have to
> search in the list to_remove_edges.
> * Considering BBs that have more than one incoming edge.
> * Augmenting get_value_range to handle integer constants (or constants
> in general).
> * Augmenting find_case_label_range to handle more efficiently the case
> where i == j.
> Nonetheless, the current patch is enough to fix the PR. I could
> include the above proposals in the comments.

Attachment: fix-pr36902.diff
Description: Text document

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]