* [RFC PATCH] lockdep: fix lockdep_stats_show() by using lockdep_state magic
@ 2011-02-14 15:19 Yong Zhang
0 siblings, 0 replies; only message in thread
From: Yong Zhang @ 2011-02-14 15:19 UTC (permalink / raw)
To: linux-kernel; +Cc: Peter Zijlstra, Ingo Molnar
Hi Peter/Ingo,
Now lockdep_stats_show() doesn't show every thing we need, like RECLAIM_FS
status is missed. So I come up with a patch for it(but not for included).
1) In this patch, RECLAIM_FS-* locks is added:
RECLAIM_FS-safe locks: 4
RECLAIM_FS-unsafe locks: 134
RECLAIM_FS-read-safe locks: 0
RECLAIM_FS-read-unsafe locks: 74
2) But the difference is HARDIRQ, SOFRIRQ, RECLAIM_FS becomes to
upcase. Like:
HARDIRQ-safe locks: 63
HARDIRQ-unsafe locks: 590
SOFTIRQ-safe locks: 115
SOFTIRQ-unsafe locks: 506
RECLAIM_FS-safe locks: 4
RECLAIM_FS-unsafe locks: 134
Is this kind of change acceptable?
3) When looking through lockdep_stats_show(), I think there is more
thing to rewrite, like:
a) "all direct dependencies" calculation, actually I'm not very sure how
it is worked out(teach please).
b) factor is not used.
c) Is there any other message should be showed?
So could we make a decision on what lockdep_stats should be, then
I can make a patchset to reach the requirement?
BTW, the below is just to show some point, not for included.
Thanks,
Yong
---
Subject: [PATCH] lockdep: fix lockdep_stats_show() by using lockdep_state magic
Signed-off-by: Yong Zhang <yong.zhang0@gmail.com>
---
kernel/lockdep_proc.c | 79 ++++++++++++++++++++++++++-----------------------
1 files changed, 42 insertions(+), 37 deletions(-)
diff --git a/kernel/lockdep_proc.c b/kernel/lockdep_proc.c
index 1969d2f..c979679 100644
--- a/kernel/lockdep_proc.c
+++ b/kernel/lockdep_proc.c
@@ -220,12 +220,16 @@ static int lockdep_stats_show(struct seq_file *m, void *v)
struct lock_class *class;
unsigned long nr_unused = 0, nr_uncategorized = 0,
nr_irq_safe = 0, nr_irq_unsafe = 0,
- nr_softirq_safe = 0, nr_softirq_unsafe = 0,
- nr_hardirq_safe = 0, nr_hardirq_unsafe = 0,
nr_irq_read_safe = 0, nr_irq_read_unsafe = 0,
- nr_softirq_read_safe = 0, nr_softirq_read_unsafe = 0,
- nr_hardirq_read_safe = 0, nr_hardirq_read_unsafe = 0,
sum_forward_deps = 0, factor = 0;
+#define LOCKDEP_STATE(__STATE) \
+ unsigned long nr_##__STATE##_safe = 0; \
+ unsigned long nr_##__STATE##_unsafe = 0; \
+ unsigned long nr_##__STATE##_read_safe = 0; \
+ unsigned long nr_##__STATE##_read_unsafe = 0;
+
+#include "lockdep_states.h"
+#undef LOCKDEP_STATE
list_for_each_entry(class, &all_lock_classes, lock_entry) {
@@ -233,35 +237,33 @@ static int lockdep_stats_show(struct seq_file *m, void *v)
nr_unused++;
if (class->usage_mask == LOCKF_USED)
nr_uncategorized++;
+#define LOCKDEP_STATE(__STATE) \
+ if (class->usage_mask & LOCKF_USED_IN_##__STATE) \
+ nr_##__STATE##_safe++; \
+ if (class->usage_mask & LOCKF_ENABLED_##__STATE) \
+ nr_##__STATE##_unsafe++; \
+ if (class->usage_mask & LOCKF_USED_IN_##__STATE##_READ) \
+ nr_##__STATE##_read_safe++; \
+ if (class->usage_mask & LOCKF_ENABLED_##__STATE##_READ) \
+ nr_##__STATE##_read_unsafe++;
+
+#include "lockdep_states.h"
+#undef LOCKDEP_STATE
+
if (class->usage_mask & LOCKF_USED_IN_IRQ)
nr_irq_safe++;
if (class->usage_mask & LOCKF_ENABLED_IRQ)
nr_irq_unsafe++;
- if (class->usage_mask & LOCKF_USED_IN_SOFTIRQ)
- nr_softirq_safe++;
- if (class->usage_mask & LOCKF_ENABLED_SOFTIRQ)
- nr_softirq_unsafe++;
- if (class->usage_mask & LOCKF_USED_IN_HARDIRQ)
- nr_hardirq_safe++;
- if (class->usage_mask & LOCKF_ENABLED_HARDIRQ)
- nr_hardirq_unsafe++;
if (class->usage_mask & LOCKF_USED_IN_IRQ_READ)
nr_irq_read_safe++;
if (class->usage_mask & LOCKF_ENABLED_IRQ_READ)
nr_irq_read_unsafe++;
- if (class->usage_mask & LOCKF_USED_IN_SOFTIRQ_READ)
- nr_softirq_read_safe++;
- if (class->usage_mask & LOCKF_ENABLED_SOFTIRQ_READ)
- nr_softirq_read_unsafe++;
- if (class->usage_mask & LOCKF_USED_IN_HARDIRQ_READ)
- nr_hardirq_read_safe++;
- if (class->usage_mask & LOCKF_ENABLED_HARDIRQ_READ)
- nr_hardirq_read_unsafe++;
#ifdef CONFIG_PROVE_LOCKING
sum_forward_deps += lockdep_count_forward_deps(class);
#endif
}
+
#ifdef CONFIG_DEBUG_LOCKDEP
DEBUG_LOCKS_WARN_ON(debug_atomic_read(nr_unused_locks) != nr_unused);
#endif
@@ -280,7 +282,7 @@ static int lockdep_stats_show(struct seq_file *m, void *v)
*/
seq_printf(m, " all direct dependencies: %11lu\n",
nr_irq_unsafe * nr_irq_safe +
- nr_hardirq_unsafe * nr_hardirq_safe +
+ nr_HARDIRQ_unsafe * nr_HARDIRQ_safe +
nr_list_entries);
/*
@@ -312,27 +314,30 @@ static int lockdep_stats_show(struct seq_file *m, void *v)
(nr_softirq_chains + 1) *
(nr_process_chains + 1)
);
- seq_printf(m, " hardirq-safe locks: %11lu\n",
- nr_hardirq_safe);
- seq_printf(m, " hardirq-unsafe locks: %11lu\n",
- nr_hardirq_unsafe);
- seq_printf(m, " softirq-safe locks: %11lu\n",
- nr_softirq_safe);
- seq_printf(m, " softirq-unsafe locks: %11lu\n",
- nr_softirq_unsafe);
+
+#define LOCKDEP_STATE(__STATE) \
+ seq_printf(m, " %-31s%11lu\n", \
+ __stringify(__STATE)"-safe locks:", \
+ nr_##__STATE##_safe); \
+ seq_printf(m, " %-31s%11lu\n", \
+ __stringify(__STATE)"-unsafe locks:", \
+ nr_##__STATE##_unsafe);
+#include "lockdep_states.h"
+#undef LOCKDEP_STATE
seq_printf(m, " irq-safe locks: %11lu\n",
nr_irq_safe);
seq_printf(m, " irq-unsafe locks: %11lu\n",
nr_irq_unsafe);
- seq_printf(m, " hardirq-read-safe locks: %11lu\n",
- nr_hardirq_read_safe);
- seq_printf(m, " hardirq-read-unsafe locks: %11lu\n",
- nr_hardirq_read_unsafe);
- seq_printf(m, " softirq-read-safe locks: %11lu\n",
- nr_softirq_read_safe);
- seq_printf(m, " softirq-read-unsafe locks: %11lu\n",
- nr_softirq_read_unsafe);
+#define LOCKDEP_STATE(__STATE) \
+ seq_printf(m, " %-31s%11lu\n", \
+ __stringify(__STATE)"-read-safe locks:", \
+ nr_##__STATE##_read_safe); \
+ seq_printf(m, " %-31s%11lu\n", \
+ __stringify(__STATE)"-read-unsafe locks:", \
+ nr_##__STATE##_read_unsafe);
+#include "lockdep_states.h"
+#undef LOCKDEP_STATE
seq_printf(m, " irq-read-safe locks: %11lu\n",
nr_irq_read_safe);
seq_printf(m, " irq-read-unsafe locks: %11lu\n",
--
1.7.1
^ permalink raw reply related [flat|nested] only message in thread
only message in thread, other threads:[~2011-02-14 15:19 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-02-14 15:19 [RFC PATCH] lockdep: fix lockdep_stats_show() by using lockdep_state magic Yong Zhang
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.