Hi!
While it would be perhaps nice to pass explicit location_t in the target
option handling code, there are hundreds of error/warning/sorry calls
in lots of backends, and lots of those routines are used not just
for the process_options time (i.e. command line options), but also for
pragma GCC target and target option handling, so at least for the time being
I think it is easier to just use UNKNOWN_LOCATION for the command line
option diagnostics.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2016-09-05 Jakub Jelinek <ja...@redhat.com>
PR middle-end/77475
* toplev.c (process_options): Temporarily set input_location
to UNKNOWN_LOCATION around targetm.target_option.override () call.
--- gcc/toplev.c.jj 2016-09-03 11:18:55.000000000 +0200
+++ gcc/toplev.c 2016-09-05 15:05:19.819995470 +0200
@@ -1220,7 +1220,10 @@ process_options (void)
no_backend = lang_hooks.post_options (&main_input_filename);
/* Some machines may reject certain combinations of options. */
+ location_t saved_location = input_location;
+ input_location = UNKNOWN_LOCATION;
targetm.target_option.override ();
+ input_location = saved_location;
if (flag_diagnostics_generate_patch)
global_dc->edit_context_ptr = new edit_context ();
Jakub
