Good evening -

I sent a similar diff in a little while ago; it were, however,
a rather ugly thing...

There's a subtraction of a known zero value in uipc_mbuf.c:

The only way copyhdr evaluates to true is if the offset is 0;
in the code as it is, it's saved as off0, and off is the manipulated value.
There is then a subtraction of off0 in the first iteration of the loop if
copyhdr is non-zero; this ends up as a subtraction of zero, and can be
removed.

- Bert

Index: src/sys/kern/uipc_mbuf.c
================================================== =================
RCS file: /cvs/src/sys/kern/uipc_mbuf.c,v
retrieving revision 1.76
diff -u -r1.76 uipc_mbuf.c
--- src/sys/kern/uipc_mbuf.c 14 Jul 2006 01:58:58 -0000 1.76
+++ src/sys/kern/uipc_mbuf.c 4 Oct 2006 23:36:19 -0000
@@ -262,10 +262,9 @@
}

struct mbuf *
-m_copym0(struct mbuf *m, int off0, int len, int wait, int deep)
+m_copym0(struct mbuf *m, int off, int len, int wait, int deep)
{
struct mbuf *n, **np;
- int off = off0;
struct mbuf *top;
int copyhdr = 0;

@@ -295,9 +294,7 @@
goto nospace;
if (copyhdr) {
M_DUP_PKTHDR(n, m);
- if (len == M_COPYALL)
- n->m_pkthdr.len -= off0;
- else
+ if (len != M_COPYALL)
n->m_pkthdr.len = len;
copyhdr = 0;
}