[RFC PATCH 3/4] powerpc 2.6.16-rt17: to support CONFIG_MCOUNT

Sorry about the late response and thanks a lot for your comments.

On Fri, 11 Aug 2006 07:12:35 -0700, Daniel Walker [off-list ref] said:

quoted

On Fri, 2006-08-11 at 11:02 +0900, Tsutomu OWA wrote:

quoted

+
+#if defined(CONFIG_PREEMPT_TRACE) || defined(CONFIG_LATENCY_TRACE)
+        print_traces(current);
+#endif /* CONFIG_PREEMPT_TRACE || CONFIG_LATENCY_TRACE */

You shouldn't need the ifdef's here.

  Right.  I'll remove them.

quoted

+#ifdef CONFIG_DEBUG_MUTEXES
+	show_held_locks(current);
+#endif /* CONFIG_DEBUG_MUTEXES */
+
}
EXPORT_SYMBOL(dump_stack);

diff -rup -x CVS 2.6.16-rt17/arch/powerpc/kernel/prom.c rt-powerpc/arch/powerpc/kernel/prom.c

Your on an ancient version of the -rt patch. I see that your using CVS
which explains that. If you want to keep better pace you should consider
not using CVS. 

  Hmm... I'll think about that.

quoted

diff -rup -x CVS 2.6.16-rt17/lib/string.c rt-powerpc/lib/string.c

--- 2.6.16-rt17/lib/string.c	2006-03-20 14:53:29.000000000 +0900
+++ rt-powerpc/lib/string.c	2006-06-16 13:08:07.000000000 +0900

@@ -67,7 +67,7 @@ EXPORT_SYMBOL(strnicmp);

* @src: Where to copy the string from
*/
#undef strcpy
-char *strcpy(char *dest, const char *src)
+char * notrace strcpy(char *dest, const char *src)
{
char *tmp = dest;

Why are these notrace ?

  Because these are called from prom_init before the relocation.
I was a bit reluctant to calcurate the address of mcount_enabled,
every time _mcount is called and was not sure if it's ok to call
__trace() at this early stage.

thanks.
-- owa
Software Engineering Center, TOSHIBA.