Hi,
lockdep support for seqlock was discussed at:
http://marc.info/?l=linux-kernel&m=137875860600648&w=2
This is very draft patch to add lockdep support for seqlock.
Actually, it doesn't work yet. I throw it out to get comments and help.
TODO:
- fix seqlock usage in vdso code
I got below with this patch applied.
There should be something wrong with this patch.
Any idea?
Thanks.
[ 0.000000] =================================
[ 0.000000] [ INFO: inconsistent lock state ]
[ 0.000000] 3.11.0-08716-g26b0332-dirty #54 Not tainted
[ 0.000000] ---------------------------------
[ 0.000000] inconsistent {IN-HARDIRQ-W} -> {HARDIRQ-ON-R} usage.
[ 0.000000] swapper/0/0 [HC0[0]:SC1[1]:HE1:SE0] takes:
[ 0.000000] (timekeeper_seq){-+.-..}, at: [<c10730d9>] tick_check_oneshot_change+0x3a/0xcc
[ 0.000000] {IN-HARDIRQ-W} state was registered at:
[ 0.000000] [<c10795b1>] __lock_acquire+0x516/0x8e7
[ 0.000000] [<c1079a10>] lock_acquire+0x8e/0xad
[ 0.000000] [<c106d967>] do_timer+0x8c8/0x933
[ 0.000000] [<c107226b>] tick_periodic+0x58/0x99
[ 0.000000] [<c10722ca>] tick_handle_periodic+0x1e/0x6f
[ 0.000000] [<c1003962>] timer_interrupt+0x12/0x19
[ 0.000000] [<c1090f70>] handle_irq_event_percpu+0x5c/0x171
[ 0.000000] [<c10910b6>] handle_irq_event+0x31/0x48
[ 0.000000] [<c109325b>] handle_level_irq+0x8c/0xb8
[ 0.000000] irq event stamp: 352
[ 0.000000] hardirqs last enabled at (352): [<c103a488>] __do_softirq+0x76/0x1b0
[ 0.000000] hardirqs last disabled at (348): [<c132d2ca>] common_interrupt+0x2a/0x38
[ 0.000000] softirqs last enabled at (350): [<c103a694>] _local_bh_enable+0x12/0x14
[ 0.000000] softirqs last disabled at (351): [<c103a648>] irq_exit+0x49/0x83
[ 0.000000]
[ 0.000000] other info that might help us debug this:
[ 0.000000] Possible unsafe locking scenario:
[ 0.000000]
[ 0.000000] CPU0
[ 0.000000] ----
[ 0.000000] lock(timekeeper_seq);
[ 0.000000] <Interrupt>
[ 0.000000] lock(timekeeper_seq);
[ 0.000000]
[ 0.000000] *** DEADLOCK ***
[ 0.000000]
[ 0.000000] no locks held by swapper/0/0.
[ 0.000000]
[ 0.000000] stack backtrace:
[ 0.000000] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.11.0-08716-g26b0332-dirty #54
[ 0.000000] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
[ 0.000000] c1868358 c14e3da8 c1327f1d c14ed934 c14e3dc8 c1075e2f c145afe6 c145b32a
[ 0.000000] c145bc44 c14eddb8 00000003 00000000 c14e3de8 c1076107 00000003 c1076c27
[ 0.000000] c14ed934 c14ed934 c14eddb4 c1735fb4 c14e3e24 c10795fb c14e3e28 c132fb1c
[ 0.000000] Call Trace:
[ 0.000000] [<c1327f1d>] dump_stack+0x41/0x54
[ 0.000000] [<c1075e2f>] print_usage_bug+0x230/0x23c
[ 0.000000] [<c1076107>] mark_lock+0x2cc/0x4f0
[ 0.000000] [<c1076c27>] ? check_usage_forwards+0xb8/0xb8
[ 0.000000] [<c10795fb>] __lock_acquire+0x560/0x8e7
[ 0.000000] [<c1079a10>] lock_acquire+0x8e/0xad
[ 0.000000] [<c10730d9>] ? tick_check_oneshot_change+0x3a/0xcc
[ 0.000000] [<c106b94c>] timekeeping_valid_for_hres+0x21/0x61
[ 0.000000] [<c10730d9>] ? tick_check_oneshot_change+0x3a/0xcc
[ 0.000000] [<c10730d9>] tick_check_oneshot_change+0x3a/0xcc
[ 0.000000] [<c1051ef9>] hrtimer_run_pending+0x2a/0xe8
[ 0.000000] [<c103fbb6>] run_timer_softirq+0x1a/0x1b2
[ 0.000000] [<c103a488>] ? __do_softirq+0x76/0x1b0
[ 0.000000] [<c1076695>] ? trace_hardirqs_on_caller+0x121/0x16e
[ 0.000000] [<c103a4e3>] __do_softirq+0xd1/0x1b0
[ 0.000000] [<c103a648>] irq_exit+0x49/0x83
[ 0.000000] [<c1002e89>] do_IRQ+0x81/0x95
[ 0.000000] [<c132d2d1>] common_interrupt+0x31/0x38
[ 0.000000] [<c107007b>] ? timer_list_next+0x1d/0x4d
[ 0.000000] [<c132c176>] ? _raw_spin_unlock_irqrestore+0x3d/0x41
[ 0.000000] [<c1091faf>] __setup_irq+0x2bb/0x35c
[ 0.000000] [<c1092328>] setup_irq+0x50/0x68
[ 0.000000] [<c1547d82>] setup_default_timer_irq+0xf/0x11
[ 0.000000] [<c1547d9a>] hpet_time_init+0x16/0x18
[ 0.000000] [<c1547d6c>] x86_late_time_init+0x9/0x10
[ 0.000000] [<c15459fc>] start_kernel+0x28a/0x311
[ 0.000000] [<c15452a3>] i386_start_kernel+0x79/0x7d
---
include/linux/lockdep.h | 4 ++++
include/linux/seqlock.h | 32 +++++++++++++++++++++++++++++---
2 files changed, 33 insertions(+), 3 deletions(-)
diff --git a/include/linux/lockdep.h b/include/linux/lockdep.h
index cfc2f11..d190e7d 100644
--- a/include/linux/lockdep.h
+++ b/include/linux/lockdep.h
@@ -506,6 +506,10 @@ static inline void print_irqtrace_events(struct task_struct *curr)
#define rwsem_acquire_read(l, s, t, i) lock_acquire_shared(l, s, t, NULL, i)
# define rwsem_release(l, n, i) lock_release(l, n, i)
+#define seqlock_acquire(l, s, t, i) lock_acquire_exclusive(l, s, t, NULL, i)
+#define seqlock_acquire_read(l, s, t, i) lock_acquire_shared(l, s, t, NULL, i)
+#define seqlock_release(l, n, i) lock_release(l, n, i)
+
#define lock_map_acquire(l) lock_acquire_exclusive(l, 0, 0, NULL, _THIS_IP_)
#define lock_map_acquire_read(l) lock_acquire_shared_recursive(l, 0, 0, NULL, _THIS_IP_)
# define lock_map_release(l) lock_release(l, 1, _THIS_IP_)
diff --git a/include/linux/seqlock.h b/include/linux/seqlock.h
index 1829905..a1bc7c1 100644
--- a/include/linux/seqlock.h
+++ b/include/linux/seqlock.h
@@ -38,10 +38,19 @@
*/
typedef struct seqcount {
unsigned sequence;
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+ struct lockdep_map dep_map;
+#endif
} seqcount_t;
#define SEQCNT_ZERO { 0 }
-#define seqcount_init(x) do { *(x) = (seqcount_t) SEQCNT_ZERO; } while (0)
+#define seqcount_init(x) \
+do { \
+ static struct lock_class_key __key; \
+ \
+ lockdep_init_map(&(x)->dep_map, #x, &__key, 0); \
+ *(x) = (seqcount_t) SEQCNT_ZERO; \
+} while (0)
/**
* __read_seqcount_begin - begin a seq-read critical section (without barrier)
@@ -60,6 +69,8 @@ static inline unsigned __read_seqcount_begin(const seqcount_t *s)
{
unsigned ret;
+ seqlock_acquire_read((struct lockdep_map *)&s->dep_map, 0, 0, _RET_IP_);
+
repeat:
ret = ACCESS_ONCE(s->sequence);
if (unlikely(ret & 1)) {
@@ -101,7 +112,11 @@ static inline unsigned read_seqcount_begin(const seqcount_t *s)
*/
static inline unsigned raw_seqcount_begin(const seqcount_t *s)
{
- unsigned ret = ACCESS_ONCE(s->sequence);
+ unsigned ret;
+
+ seqlock_acquire_read((struct lockdep_map *)&s->dep_map, 0, 0, _RET_IP_);
+
+ ret = ACCESS_ONCE(s->sequence);
smp_rmb();
return ret & ~1;
}
@@ -122,6 +137,7 @@ static inline unsigned raw_seqcount_begin(const seqcount_t *s)
*/
static inline int __read_seqcount_retry(const seqcount_t *s, unsigned start)
{
+ seqlock_release((struct lockdep_map *)&s->dep_map, 1, _RET_IP_);
return unlikely(s->sequence != start);
}
@@ -148,12 +164,14 @@ static inline int read_seqcount_retry(const seqcount_t *s, unsigned start)
*/
static inline void write_seqcount_begin(seqcount_t *s)
{
+ seqlock_acquire((struct lockdep_map *)&s->dep_map, 0, 0, _RET_IP_);
s->sequence++;
smp_wmb();
}
static inline void write_seqcount_end(seqcount_t *s)
{
+ seqlock_release((struct lockdep_map *)&s->dep_map, 1, _RET_IP_);
smp_wmb();
s->sequence++;
}
@@ -167,6 +185,7 @@ static inline void write_seqcount_end(seqcount_t *s)
*/
static inline void write_seqcount_barrier(seqcount_t *s)
{
+ seqlock_acquire((struct lockdep_map *)&s->dep_map, 0, 0, _RET_IP_);
smp_wmb();
s->sequence+=2;
}
@@ -176,13 +195,20 @@ typedef struct {
spinlock_t lock;
} seqlock_t;
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+# define __SEQLOCK_DEP_MAP_INIT(lockname) , .dep_map = { .name = #lockname }
+#else
+# define __SEQLOCK_DEP_MAP_INIT(lockname)
+#endif
+
/*
* These macros triggered gcc-3.x compile-time problems. We think these are
* OK now. Be cautious.
*/
#define __SEQLOCK_UNLOCKED(lockname) \
{ \
- .seqcount = SEQCNT_ZERO, \
+ .seqcount = {.sequence = 0 \
+ __SEQLOCK_DEP_MAP_INIT(lockname)}, \
.lock = __SPIN_LOCK_UNLOCKED(lockname) \
}
On Wed, Sep 18, 2013 at 02:57:13PM +0800, Lin Ming wrote:
> Hi,
>
> lockdep support for seqlock was discussed at:
> http://marc.info/?l=linux-kernel&m=137875860600648&w=2
>
> This is very draft patch to add lockdep support for seqlock.
> Actually, it doesn't work yet. I throw it out to get comments and help.
>
> TODO:
> - fix seqlock usage in vdso code
>
> I got below with this patch applied.
> There should be something wrong with this patch.
> Any idea?
You're aware of John Stultz's patches?
lkml.kernel.org/r/[email protected]
On Wed, 2013-09-18 at 11:15 +0200, Peter Zijlstra wrote:
> On Wed, Sep 18, 2013 at 02:57:13PM +0800, Lin Ming wrote:
> > Hi,
> >
> > lockdep support for seqlock was discussed at:
> > http://marc.info/?l=linux-kernel&m=137875860600648&w=2
> >
> > This is very draft patch to add lockdep support for seqlock.
> > Actually, it doesn't work yet. I throw it out to get comments and help.
> >
> > TODO:
> > - fix seqlock usage in vdso code
> >
> > I got below with this patch applied.
> > There should be something wrong with this patch.
> > Any idea?
>
> You're aware of John Stultz's patches?
No.
Glad that John has wrote this.
Reading it ...
Thanks for the info.
>
> lkml.kernel.org/r/[email protected]