| Message ID | 4CC9ECFF.8020208@linux.vnet.ibm.com |
|---|---|
| State | New |
| Headers | show |
On Thu, 28 Oct 2010, Pat Haugen wrote: > 2010-10-29 Pat Haugen <pthaugen@us.ibm.com> > * gcc.c-torture/execute/20101011-1.c: Fix #ifdef. > > > Index: gcc.c-torture/execute/20101011-1.c > =================================================================== > --- gcc.c-torture/execute/20101011-1.c (revision 166039) > +++ gcc.c-torture/execute/20101011-1.c (working copy) > @@ -35,7 +35,7 @@ int k; > int > main () > { > -#ifdef DO_TEST > +#if DO_TEST That will run afoul -Wundef. Not sure what alternative is best, something like #ifdef DO_TEST # if DO_TEST # endif #endif perhaps, or #if defined(DO_TEST) && DO_TEST ? Or someone just says there are so many of these everywhere that we shouldn't worry about -Wundef, of course... Gerald
On 10/31/2010 2:43 PM, Gerald Pfeifer wrote: >> Index: gcc.c-torture/execute/20101011-1.c >> > =================================================================== >> > --- gcc.c-torture/execute/20101011-1.c (revision 166039) >> > +++ gcc.c-torture/execute/20101011-1.c (working copy) >> > @@ -35,7 +35,7 @@ int k; >> > int >> > main () >> > { >> > -#ifdef DO_TEST >> > +#if DO_TEST > That will run afoul -Wundef. Not sure what alternative is best, > something like > > #ifdef DO_TEST > # if DO_TEST > # endif > #endif > > perhaps, or > > #if defined(DO_TEST)&& DO_TEST > DO_TEST is always defined in the testcase, which is what led to the original problem. -Pat
Index: gcc.c-torture/execute/20101011-1.c =================================================================== --- gcc.c-torture/execute/20101011-1.c (revision 166039) +++ gcc.c-torture/execute/20101011-1.c (working copy) @@ -35,7 +35,7 @@ int k; int main () { -#ifdef DO_TEST +#if DO_TEST signal (SIGFPE, sigfpe); k = i / j; abort ();
Committed the following as obvious after noticing failure on PowerPC. testsuite/ChangeLog 2010-10-29 Pat Haugen <pthaugen@us.ibm.com> * gcc.c-torture/execute/20101011-1.c: Fix #ifdef.