The patch titled
lockdep: re-annotate scheduler runqueues
has been added to the -mm tree. Its filename is
lockdep-re-annotate-scheduler-runqueues.patch
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/SubmitChecklist when testing your code ***
See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
out what to do about this
The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/
------------------------------------------------------
Subject: lockdep: re-annotate scheduler runqueues
From: Peter Zijlstra <a.p.zijlstra@xxxxxxxxx>
Instead of using a per-rq lock class, use the regular nesting operations.
However, take extra care with double_lock_balance() as it can release the
already held rq->lock (and therefore change its nesting class).
So what can happen is:
spin_lock(rq->lock); // this rq subclass 0
double_lock_balance(rq, other_rq);
// release rq
// acquire other_rq->lock subclass 0
// acquire rq->lock subclass 1
spin_unlock(other_rq->lock);
leaving you with rq->lock in subclass 1
So a subsequent double_lock_balance() call can try to nest a subclass 1
lock while already holding a subclass 1 lock.
Fix this by introducing double_unlock_balance() which releases the other
rq's lock, but also re-sets the subclass for this rq's lock to 0.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@xxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
kernel/sched.c | 21 +++++++++++++--------
kernel/sched_rt.c | 8 +++++---
2 files changed, 18 insertions(+), 11 deletions(-)
diff -puN kernel/sched.c~lockdep-re-annotate-scheduler-runqueues kernel/sched.c
--- a/kernel/sched.c~lockdep-re-annotate-scheduler-runqueues
+++ a/kernel/sched.c
@@ -600,7 +600,6 @@ struct rq {
/* BKL stats */
unsigned int bkl_count;
#endif
- struct lock_class_key rq_lock_key;
};
static DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues);
@@ -2756,10 +2755,10 @@ static void double_rq_lock(struct rq *rq
} else {
if (rq1 < rq2) {
spin_lock(&rq1->lock);
- spin_lock(&rq2->lock);
+ spin_lock_nested(&rq2->lock, SINGLE_DEPTH_NESTING);
} else {
spin_lock(&rq2->lock);
- spin_lock(&rq1->lock);
+ spin_lock_nested(&rq1->lock, SINGLE_DEPTH_NESTING);
}
}
update_rq_clock(rq1);
@@ -2802,14 +2801,21 @@ static int double_lock_balance(struct rq
if (busiest < this_rq) {
spin_unlock(&this_rq->lock);
spin_lock(&busiest->lock);
- spin_lock(&this_rq->lock);
+ spin_lock_nested(&this_rq->lock, SINGLE_DEPTH_NESTING);
ret = 1;
} else
- spin_lock(&busiest->lock);
+ spin_lock_nested(&busiest->lock, SINGLE_DEPTH_NESTING);
}
return ret;
}
+static void double_unlock_balance(struct rq *this_rq, struct rq *busiest)
+ __releases(busiest->lock)
+{
+ spin_unlock(&busiest->lock);
+ lock_set_subclass(&this_rq->lock.dep_map, 0, _RET_IP_);
+}
+
/*
* If dest_cpu is allowed for this process, migrate the task to it.
* This is accomplished by forcing the cpu_allowed mask to only
@@ -3634,7 +3640,7 @@ redo:
ld_moved = move_tasks(this_rq, this_cpu, busiest,
imbalance, sd, CPU_NEWLY_IDLE,
&all_pinned);
- spin_unlock(&busiest->lock);
+ double_unlock_balance(this_rq, busiest);
if (unlikely(all_pinned)) {
cpu_clear(cpu_of(busiest), *cpus);
@@ -3749,7 +3755,7 @@ static void active_load_balance(struct r
else
schedstat_inc(sd, alb_failed);
}
- spin_unlock(&target_rq->lock);
+ double_unlock_balance(busiest_rq, target_rq);
}
#ifdef CONFIG_NO_HZ
@@ -7987,7 +7993,6 @@ void __init sched_init(void)
rq = cpu_rq(i);
spin_lock_init(&rq->lock);
- lockdep_set_class(&rq->lock, &rq->rq_lock_key);
rq->nr_running = 0;
init_cfs_rq(&rq->cfs, rq);
init_rt_rq(&rq->rt, rq);
diff -puN kernel/sched_rt.c~lockdep-re-annotate-scheduler-runqueues kernel/sched_rt.c
--- a/kernel/sched_rt.c~lockdep-re-annotate-scheduler-runqueues
+++ a/kernel/sched_rt.c
@@ -861,6 +861,8 @@ static void put_prev_task_rt(struct rq *
#define RT_MAX_TRIES 3
static int double_lock_balance(struct rq *this_rq, struct rq *busiest);
+static void double_unlock_balance(struct rq *this_rq, struct rq *busiest);
+
static void deactivate_task(struct rq *rq, struct task_struct *p, int sleep);
static int pick_rt_task(struct rq *rq, struct task_struct *p, int cpu)
@@ -1022,7 +1024,7 @@ static struct rq *find_lock_lowest_rq(st
break;
/* try again */
- spin_unlock(&lowest_rq->lock);
+ double_unlock_balance(rq, lowest_rq);
lowest_rq = NULL;
}
@@ -1091,7 +1093,7 @@ static int push_rt_task(struct rq *rq)
resched_task(lowest_rq->curr);
- spin_unlock(&lowest_rq->lock);
+ double_unlock_balance(rq, lowest_rq);
ret = 1;
out:
@@ -1197,7 +1199,7 @@ static int pull_rt_task(struct rq *this_
}
skip:
- spin_unlock(&src_rq->lock);
+ double_unlock_balance(this_rq, src_rq);
}
return ret;
_
Patches currently in -mm which might be from a.p.zijlstra@xxxxxxxxx are
linux-next.patch
sched-type-fix.patch
lockdep-fix-combinatorial-explosion-in-lock-subgraph-traversal.patch
lockdep-lock_set_subclass-reset-a-held-locks-subclass.patch
lockdep-re-annotate-scheduler-runqueues.patch
lockdep-shrink-held_lock-structure.patch
lockdep-map_acquire.patch
lockdep-lock-protection-locks.patch
lockdep-spin_lock_nest_lock.patch
--
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
