according to http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56301#c3 this is not an issue with GCC, but job_new shouldn't be marked with the malloc attribute. The following patch lets the test case pass without the GCC change reverted. Maybe other usages of the malloc attribute should be revisited.
according to http:// gcc.gnu. org/bugzilla/ show_bug. cgi?id= 56301#c3 this is not an issue with GCC, but job_new shouldn't be marked with the malloc attribute. The following patch lets the test case pass without the GCC change reverted. Maybe other usages of the malloc attribute should be revisited.
--- init/job.h.orig 2012-12-04 17:14:47.000000000 +0000
+++ init/job.h 2013-02-18 13:06:31.792691230 +0000
@@ -165,7 +165,7 @@
NIH_BEGIN_EXTERN
Job * job_new (JobClass *class, const char *name) unused_ result, malloc)); unused_ result) );
- __attribute__ ((warn_
+ __attribute__ ((warn_
void job_register (Job *job, DBusConnection *conn, int signal);
void job_change_goal (Job *job, JobGoal goal);