It doesn't make sense to me to add a "sanitizer" that controls how another 
sanitizer reports failures; that is not how the `-fsanitize=` flags are 
supposed to work at all.

I think you should follow the existing model for frontend sanitizers here: if 
`-fsanitize-undefined-trap-on-error` is specified, then trap, otherwise produce 
a diagnostic. If it really makes sense to have a per-sanitizer trap-on-error 
mode (as opposed to the existing per-sanitizer `-fsanitize-recover=` setting), 
we should enable that for all the frontend sanitizers, not just this one, but 
it seems (perhaps naively) to me like `-fsanitize-recover=` should be 
sufficient here already?


http://reviews.llvm.org/D10268

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/



_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to