Re: [PATCH 2.6.12-rc3-mm3] connector: add a fork connector

[Nick Piggin <nickpiggin@yahoo.com.au>]

Guillaume Thouvenin wrote:
> On Mon, 2005-05-09 at 11:31 +0200, Alexander Nyberg wrote:

>>>Index: linux-2.6.12-rc3-mm3/kernel/fork.c
>>>===================================================================
>>>--- linux-2.6.12-rc3-mm3.orig/kernel/fork.c	2005-05-09 07:45:56.000000000 +0200
>>>+++ linux-2.6.12-rc3-mm3/kernel/fork.c	2005-05-09 08:03:15.000000000 +0200
>>>@@ -41,6 +41,7 @@
>>> #include <linux/profile.h>
>>> #include <linux/rmap.h>
>>> #include <linux/acct.h>
>>>+#include <linux/cn_fork.h>
>>> 
>>> #include <asm/pgtable.h>
>>> #include <asm/pgalloc.h>
>>>@@ -63,6 +64,14 @@ DEFINE_PER_CPU(unsigned long, process_co
>>> 
>>> EXPORT_SYMBOL(tasklist_lock);
>>> 
>>>+#ifdef CONFIG_FORK_CONNECTOR
>>>+/* 
>>>+ * fork_counts is used by the fork_connector() inline routine as 
>>>+ * the sequence number of the netlink message.
>>>+ */
>>>+static DEFINE_PER_CPU(unsigned long, fork_counts); 
>>>+#endif /* CONFIG_FORK_CONNECTOR */
>>>+
>>
>>The above should go into cn_fork.c
> 
> 
> I don't see why.
> 

Because you get these ugly ifdefs and things just spilling into
conceptually the wrong place. Why should anyone apart from the
fork connecter care how `fork_counts` is stored? And why would
any generic code care that it is defined when CONFIG_FORK_CONNECTOR
is set? Alexander is right, unless I missed something that requires
reading the code :)

-- 
SUSE Labs, Novell Inc.