Ian Lance Taylor - Re: PATCH RFC: -Wstrict-overflow, take 2 (original) (raw)

This is the mail archive of the gcc-patches@gcc.gnu.orgmailing list for the GCC project.

• From: Ian Lance Taylor
• To: Jonathan Lennox
• Cc: gcc-patches at gcc dot gnu dot org
• Date: 16 Feb 2007 13:26:57 -0800
• Subject: Re: PATCH RFC: -Wstrict-overflow, take 2
• References: <m3hctwgi1d.fsf@localhost.localdomain> <17878.7709.414766.597566@metro-north.cs.columbia.edu>

Jonathan Lennox lennox@cs.columbia.edu writes:

Ian Lance Taylor wrote:

Here is a new version of my -Wstrict-overflow patch. This incorporates all the comments made on the first version.

At -Wstrict-overflow=3 we get 31 warnings:

../../trunk/gcc/expmed.c:3485: warning: assuming signed overflow does not occur when changing X +- C1 cmp C2 to X cmp C1 +- C2

How hard would it be to change this diagnostic to print the actual values of "X", "+-", "C1", "cmp", and "C2"?

For expmed.c:3485, for example, the warning would be

../../trunk/gcc/expmed.c:3485: warning: assuming signed overflow does not occur when changing "size - 1 < 32" to "size < 32 + 1"

I think this would be much more accessible to people not versed in the intricacies of compilers, and also make it much easier (as in this case) to see at a glance if signed overflow could actually be a problem.

Well, the X can, in general, be anything. As far as I know we don't have a good way to dump general expressions in warning contexts. But maybe I just don't know about it.

The C1 and C2 are integer constants, and would be easy to dump. It's line 8296 of fold-const.c. Go for it.

Ian

• Follow-Ups:
  • Re: PATCH RFC: -Wstrict-overflow, take 2
    * From: Joseph S. Myers
  • Re: PATCH RFC: -Wstrict-overflow, take 2
    * From: Gabriel Dos Reis
  • Re: PATCH RFC: -Wstrict-overflow, take 2
    * From: Jonathan Lennox
• References:
  • PATCH RFC: -Wstrict-overflow, take 2
    * From: Ian Lance Taylor
  • Re: PATCH RFC: -Wstrict-overflow, take 2
    * From: Jonathan Lennox