Attached is a small patch for a cut'n'paste bug in linux 2.5.59
net/ipv4/proc.c:fold_field()
This causes some /proc/net counters that are incremented in BH to be counted
twice and the counters incremented in user context not counted.
--- proc.c Tue Jan 21 17:18:52 2003
+++ proc.c.new Fri Feb 7 09:39:34 2003
@@ -99,7 +99,7 @@
*((unsigned long *) (((void *) per_cpu_ptr(mib[0], i)) +
sizeof (unsigned long) * nr));
res +=
- *((unsigned long *) (((void *) per_cpu_ptr(mib[0], i)) +
+ *((unsigned long *) (((void *) per_cpu_ptr(mib[1], i)) +
sizeof (unsigned long) * nr));
}
return res;
The same routine(fold_field()) is duplicated in net/ipv6/proc.c, but it doesn't
have the bug.
Could we convert this to a static inline and move to a header file to avoid
the duplication and 'sctp' too can use it.
Thanks
Sridhar
|