[BUGFIX] lockdep: output lock_class key instead of address for forward dependency output

[BUGFIX] lockdep: output lock_class key instead of address for forward dependency output

[Huang, Ying]

The key instead of address of lock_class should be output in
/proc/lockdep when forward dependency is output, because key is
output for lock_class itself as identifier too.

This patch is based on x86/auto-latest branch of git-x86 tree, and has
been tested on x86_64 platform.

Signed-off-by: Huang Ying <ying.huang@intel.com>

---
 kernel/lockdep_proc.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- a/kernel/lockdep_proc.c
+++ b/kernel/lockdep_proc.c
@@ -139,7 +139,7 @@ static int l_show(struct seq_file *m, vo
 
 	list_for_each_entry(entry, &class->locks_after, entry) {
 		if (entry->distance == 1) {
-			seq_printf(m, " -> [%p] ", entry->class);
+			seq_printf(m, " -> [%p] ", entry->class->key);
 			print_name(m, entry->class);
 			seq_puts(m, "\n");
 		}

Re: [BUGFIX] lockdep: output lock_class key instead of address for forward dependency output

[Ingo Molnar]

* Huang, Ying <ying.huang@intel.com> wrote:

> The key instead of address of lock_class should be output in
> /proc/lockdep when forward dependency is output, because key is
> output for lock_class itself as identifier too.
> 
> This patch is based on x86/auto-latest branch of git-x86 tree, and has
> been tested on x86_64 platform.

applied to tip/core/locking - thanks Ying.

i'm wondering, can entry->class never be NULL here?

	Ingo

Re: [BUGFIX] lockdep: output lock_class key instead of address for forward dependency output

[Huang, Ying]

On Fri, 2008-06-13 at 08:58 +0200, Ingo Molnar wrote:
> * Huang, Ying <ying.huang@intel.com> wrote:
> 
> > The key instead of address of lock_class should be output in
> > /proc/lockdep when forward dependency is output, because key is
> > output for lock_class itself as identifier too.
> > 
> > This patch is based on x86/auto-latest branch of git-x86 tree, and has
> > been tested on x86_64 platform.
> 
> applied to tip/core/locking - thanks Ying.
> 
> i'm wondering, can entry->class never be NULL here?

As far as I know, it can never be NULL. Because entry is added into
class->locks_after in following call chain:

lock_acquire -> __lock_acquire -> validate_chain -> check_prevs_add ->
check_prev_add -> add_lock_to_list

in __lock_acquire the lock_class is look up or registered from
lockdep_map and put in held_lock, which is used in following function in
call chain.

Best Regards,
Huang Ying