Re: pthread-mutex debugging (PR#44387)

[Frank Wille <frank%phoenix.owl.de@localhost>]

On Sat, 22 Sep 2012 21:20:13 +0200
Frank Wille <frank%phoenix.owl.de@localhost> wrote:

> I could now isolate the problem on sys/kern/kern_mutex.c. It was
> sufficient to compile kern_mutex.c with -DFULL to make the problem
> disappear. The rest of the kernel was compiled without DIAGNOSTIC.

As we are unable to solve this critical bug I would suggest to do
something like the following temporary workaround in kern_mutex.c
for 6.0:

--- kern_mutex.c.orig   2012-10-13 20:22:43.000000000 +0200
+++ kern_mutex.c        2012-10-13 20:24:14.000000000 +0200
@@ -66,7 +66,8 @@
  * more than an splraiseipl() and splx() pair.
  */
 
-#if defined(DIAGNOSTIC) || defined(MULTIPROCESSOR) || defined(LOCKDEBUG)
+#if defined(DIAGNOSTIC) || defined(MULTIPROCESSOR) || defined(LOCKDEBUG) \
+    || defined(__powerpc__)
 #define        FULL
 #endif

This makes sure that PPC always uses real spin mutexes, and the problems
are gone.


I would also like to know if PR #44387 affects ibm4xx and booke as well,
or is it an oea-only problem? Unfortunately I have only oea-machines
here (and a not-yet-supported 440).

-- 
Frank Wille