Re: [patch] gcc.dg/compare1.c: Add -fno-short-enums.

[Gabriel Dos Reis <gdr at integrable-solutions dot net>]

Kazu Hirata <kazu@codesourcery.com> writes:

| Hi,
| 
| Attached is a patch to add -fno-short-enums.
| 
| Consider
| 
| int i(enum mm2 x)
| {
|   return x == (tf?DI2:-1); /* { dg-warning "signed and unsigned" "case 4" } */
| }
| 
| The rhs of == is of type int because of -1.  Now, 'x' is of type mm2
| and promoted to int.  Since int covers the entire range of mm2,
| c-typeck.c:build_binary_op decides not to issue a "comparison between
| signed and unsigned" warning.  The following "if" statement is
| responsible for this decision:
| 
| 		 Do not warn if the comparison is being done in a signed type,
| 		 since the signed type will only be chosen if it can represent
| 		 all the values of the unsigned type.  */
| 	      if (!TYPE_UNSIGNED (result_type))
| 		/* OK */;
| 
| The patch solves the problem as suggested by Paul Brook.  (Thanks
| Paul!)  Specifically, if we add -fno-short-enums, we can force targets
| like arm*-*-eabi* and arm*-*-symbianelf* as well as other targets to
| use enum of the same size as int.  This way, even if somebody adds a
| port with short enum, we won't have to touch this testcase.
| 
| Tested on x86_64-pc-linux-gnu and arm-none-eabi.  OK to apply?

I do not understand.  Why don't you add a new separate testcase for
the kind of thing you would like to test?  If this specific testcase
is expected to fail on arm*-*-eabi* and arm*-*-symbianelf* then you
should mark it as XFAIL for those targets.

-- Gaby