2021-06-28 15:39:52

by Xiongwei Song

[permalink] [raw]
Subject: [PATCH v2] locking/lockdep: Fix meaningless usages output of lock classes

From: Xiongwei Song <[email protected]>

When enabling CONFIG_LOCK_STAT, then CONFIG_LOCKDEP is forcedly enabled.
We can get output from /proc/lockdep, which currently includes usages of
lock classes. But the usages are meaningless, see the output below:

/ # cat /proc/lockdep
all lock classes:
ffffffff9af63350 ....: cgroup_mutex

ffffffff9af54eb8 ....: (console_sem).lock

ffffffff9af54e60 ....: console_lock

ffffffff9ae74c38 ....: console_owner_lock

ffffffff9ae74c80 ....: console_owner

ffffffff9ae66e60 ....: cpu_hotplug_lock

Only one usage context for each lock, this is because each usage is only
changed in mark_lock() that is in CONFIG_PROVE_LOCKING defined section,
however in the test situation, it's not.

The fix is to move the usages reading and seq_print from
CONFIG_PROVE_LOCKING undefined setcion to its defined section. Also,
locks_after list of lock_class is empty when CONFIG_PROVE_LOCKING
undefined, so do the same thing as what have done for usages of lock
classes.

Signed-off-by: Xiongwei Song <[email protected]>
---
kernel/locking/lockdep_proc.c | 24 +++++++++++++-----------
1 file changed, 13 insertions(+), 11 deletions(-)

diff --git a/kernel/locking/lockdep_proc.c b/kernel/locking/lockdep_proc.c
index 806978314496..a1ec2652d492 100644
--- a/kernel/locking/lockdep_proc.c
+++ b/kernel/locking/lockdep_proc.c
@@ -70,23 +70,25 @@ static int l_show(struct seq_file *m, void *v)
#ifdef CONFIG_DEBUG_LOCKDEP
seq_printf(m, " OPS:%8ld", debug_class_ops_read(class));
#endif
-#ifdef CONFIG_PROVE_LOCKING
- seq_printf(m, " FD:%5ld", lockdep_count_forward_deps(class));
- seq_printf(m, " BD:%5ld", lockdep_count_backward_deps(class));
-#endif
+ if (IS_ENABLED(CONFIG_PROVE_LOCKING)) {
+ seq_printf(m, " FD:%5ld", lockdep_count_forward_deps(class));
+ seq_printf(m, " BD:%5ld", lockdep_count_backward_deps(class));

- get_usage_chars(class, usage);
- seq_printf(m, " %s", usage);
+ get_usage_chars(class, usage);
+ seq_printf(m, " %s", usage);
+ }

seq_printf(m, ": ");
print_name(m, class);
seq_puts(m, "\n");

- list_for_each_entry(entry, &class->locks_after, entry) {
- if (entry->distance == 1) {
- seq_printf(m, " -> [%p] ", entry->class->key);
- print_name(m, entry->class);
- seq_puts(m, "\n");
+ if (IS_ENABLED(CONFIG_PROVE_LOCKING)) {
+ list_for_each_entry(entry, &class->locks_after, entry) {
+ if (entry->distance == 1) {
+ seq_printf(m, " -> [%p] ", entry->class->key);
+ print_name(m, entry->class);
+ seq_puts(m, "\n");
+ }
}
}
seq_puts(m, "\n");
--
2.30.2


2021-06-29 00:32:29

by Waiman Long

[permalink] [raw]
Subject: Re: [PATCH v2] locking/lockdep: Fix meaningless usages output of lock classes

On 6/28/21 11:17 AM, Xiongwei Song wrote:
> From: Xiongwei Song <[email protected]>
>
> When enabling CONFIG_LOCK_STAT, then CONFIG_LOCKDEP is forcedly enabled.
> We can get output from /proc/lockdep, which currently includes usages of
> lock classes. But the usages are meaningless, see the output below:
>
> / # cat /proc/lockdep
> all lock classes:
> ffffffff9af63350 ....: cgroup_mutex
>
> ffffffff9af54eb8 ....: (console_sem).lock
>
> ffffffff9af54e60 ....: console_lock
>
> ffffffff9ae74c38 ....: console_owner_lock
>
> ffffffff9ae74c80 ....: console_owner
>
> ffffffff9ae66e60 ....: cpu_hotplug_lock
>
> Only one usage context for each lock, this is because each usage is only
> changed in mark_lock() that is in CONFIG_PROVE_LOCKING defined section,
> however in the test situation, it's not.
>
> The fix is to move the usages reading and seq_print from
> CONFIG_PROVE_LOCKING undefined setcion to its defined section. Also,
> locks_after list of lock_class is empty when CONFIG_PROVE_LOCKING
> undefined, so do the same thing as what have done for usages of lock
> classes.
With this patch, CONFIG_LOCKDEP without CONFIG_PROVE_LOCKING will make
/proc/lockdep displays just the list of lock classes with their
associated lock keys. I think it is worth explicitly saying that in the
commit log.
> Signed-off-by: Xiongwei Song <[email protected]>
> ---
> kernel/locking/lockdep_proc.c | 24 +++++++++++++-----------
> 1 file changed, 13 insertions(+), 11 deletions(-)
>
> diff --git a/kernel/locking/lockdep_proc.c b/kernel/locking/lockdep_proc.c
> index 806978314496..a1ec2652d492 100644
> --- a/kernel/locking/lockdep_proc.c
> +++ b/kernel/locking/lockdep_proc.c
> @@ -70,23 +70,25 @@ static int l_show(struct seq_file *m, void *v)
> #ifdef CONFIG_DEBUG_LOCKDEP
> seq_printf(m, " OPS:%8ld", debug_class_ops_read(class));
> #endif
> -#ifdef CONFIG_PROVE_LOCKING
> - seq_printf(m, " FD:%5ld", lockdep_count_forward_deps(class));
> - seq_printf(m, " BD:%5ld", lockdep_count_backward_deps(class));
> -#endif
> + if (IS_ENABLED(CONFIG_PROVE_LOCKING)) {
> + seq_printf(m, " FD:%5ld", lockdep_count_forward_deps(class));
> + seq_printf(m, " BD:%5ld", lockdep_count_backward_deps(class));
>
> - get_usage_chars(class, usage);
> - seq_printf(m, " %s", usage);
> + get_usage_chars(class, usage);
> + seq_printf(m, " %s", usage);
> + }
>
> seq_printf(m, ": ");
> print_name(m, class);
> seq_puts(m, "\n");
>
> - list_for_each_entry(entry, &class->locks_after, entry) {
> - if (entry->distance == 1) {
> - seq_printf(m, " -> [%p] ", entry->class->key);
> - print_name(m, entry->class);
> - seq_puts(m, "\n");
> + if (IS_ENABLED(CONFIG_PROVE_LOCKING)) {
> + list_for_each_entry(entry, &class->locks_after, entry) {
> + if (entry->distance == 1) {
> + seq_printf(m, " -> [%p] ", entry->class->key);
> + print_name(m, entry->class);
> + seq_puts(m, "\n");
> + }
> }
> }
> seq_puts(m, "\n");

Maybe you can remove the blank lines in this case by moving the last
seq_puts() inside the if loop. The blank lines are not really needed
without the associated locks_after information.

Cheers,
Longman

2021-06-29 02:54:49

by Xiongwei Song

[permalink] [raw]
Subject: Re: [PATCH v2] locking/lockdep: Fix meaningless usages output of lock classes

On Tue, Jun 29, 2021 at 5:17 AM Waiman Long <[email protected]> wrote:
>
> On 6/28/21 11:17 AM, Xiongwei Song wrote:
> > From: Xiongwei Song <[email protected]>
> >
> > When enabling CONFIG_LOCK_STAT, then CONFIG_LOCKDEP is forcedly enabled.
> > We can get output from /proc/lockdep, which currently includes usages of
> > lock classes. But the usages are meaningless, see the output below:
> >
> > / # cat /proc/lockdep
> > all lock classes:
> > ffffffff9af63350 ....: cgroup_mutex
> >
> > ffffffff9af54eb8 ....: (console_sem).lock
> >
> > ffffffff9af54e60 ....: console_lock
> >
> > ffffffff9ae74c38 ....: console_owner_lock
> >
> > ffffffff9ae74c80 ....: console_owner
> >
> > ffffffff9ae66e60 ....: cpu_hotplug_lock
> >
> > Only one usage context for each lock, this is because each usage is only
> > changed in mark_lock() that is in CONFIG_PROVE_LOCKING defined section,
> > however in the test situation, it's not.
> >
> > The fix is to move the usages reading and seq_print from
> > CONFIG_PROVE_LOCKING undefined setcion to its defined section. Also,
> > locks_after list of lock_class is empty when CONFIG_PROVE_LOCKING
> > undefined, so do the same thing as what have done for usages of lock
> > classes.
> With this patch, CONFIG_LOCKDEP without CONFIG_PROVE_LOCKING will make
> /proc/lockdep displays just the list of lock classes with their
> associated lock keys. I think it is worth explicitly saying that in the
> commit log.

Make sense. Will update.

> > Signed-off-by: Xiongwei Song <[email protected]>
> > ---
> > kernel/locking/lockdep_proc.c | 24 +++++++++++++-----------
> > 1 file changed, 13 insertions(+), 11 deletions(-)
> >
> > diff --git a/kernel/locking/lockdep_proc.c b/kernel/locking/lockdep_proc.c
> > index 806978314496..a1ec2652d492 100644
> > --- a/kernel/locking/lockdep_proc.c
> > +++ b/kernel/locking/lockdep_proc.c
> > @@ -70,23 +70,25 @@ static int l_show(struct seq_file *m, void *v)
> > #ifdef CONFIG_DEBUG_LOCKDEP
> > seq_printf(m, " OPS:%8ld", debug_class_ops_read(class));
> > #endif
> > -#ifdef CONFIG_PROVE_LOCKING
> > - seq_printf(m, " FD:%5ld", lockdep_count_forward_deps(class));
> > - seq_printf(m, " BD:%5ld", lockdep_count_backward_deps(class));
> > -#endif
> > + if (IS_ENABLED(CONFIG_PROVE_LOCKING)) {
> > + seq_printf(m, " FD:%5ld", lockdep_count_forward_deps(class));
> > + seq_printf(m, " BD:%5ld", lockdep_count_backward_deps(class));
> >
> > - get_usage_chars(class, usage);
> > - seq_printf(m, " %s", usage);
> > + get_usage_chars(class, usage);
> > + seq_printf(m, " %s", usage);
> > + }
> >
> > seq_printf(m, ": ");
> > print_name(m, class);
> > seq_puts(m, "\n");
> >
> > - list_for_each_entry(entry, &class->locks_after, entry) {
> > - if (entry->distance == 1) {
> > - seq_printf(m, " -> [%p] ", entry->class->key);
> > - print_name(m, entry->class);
> > - seq_puts(m, "\n");
> > + if (IS_ENABLED(CONFIG_PROVE_LOCKING)) {
> > + list_for_each_entry(entry, &class->locks_after, entry) {
> > + if (entry->distance == 1) {
> > + seq_printf(m, " -> [%p] ", entry->class->key);
> > + print_name(m, entry->class);
> > + seq_puts(m, "\n");
> > + }
> > }
> > }
> > seq_puts(m, "\n");
>
> Maybe you can remove the blank lines in this case by moving the last
> seq_puts() inside the if loop. The blank lines are not really needed
> without the associated locks_after information.

Yeah, I agree. Thank you.

Regards,
Xiongwei

> Cheers,
> Longman
>