Probably an obvious patch but...
The testcase fails because it looks like recent glibc headers (2.27 at least)
no longer contain a declaration for __strdup, which results in warning messages
being generated and failure for excess errors. Fixed by calling the standard
name.
Verified the testcase now passes, ok for trunk?
-Pat
testsuite/ChangeLog:
2018-07-26 Pat Haugen <pthau...@us.ibm.com>
PR target/86612
* gcc.target/powerpc/pr58673-2.c: Call strdup.
Index: testsuite/gcc.target/powerpc/pr58673-2.c
===================================================================
--- testsuite/gcc.target/powerpc/pr58673-2.c (revision 262974)
+++ testsuite/gcc.target/powerpc/pr58673-2.c (working copy)
@@ -140,7 +140,7 @@ pr_ff (t_coupl_rec * tcr, real time, t_i
malloc
(__len);
__retval;}
- )): __strdup (eoNames[i])));
+ )): strdup (eoNames[i])));
raleg[j++] =
(__extension__
(__builtin_constant_p (buf)
@@ -165,7 +165,7 @@ pr_ff (t_coupl_rec * tcr, real time, t_i
malloc
(__len);
__retval;}
- )): __strdup (buf)));
+ )): strdup (buf)));
}
}
if (tcr->nLJ)
