Re: [PATCH] qla_target: Check refcount in find_sess_by_*

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



And below is a patch that allows speeding up the fast path a bit.
The code in tcm_qla2xxx_put_session() would turn into this:
	local_irq_save(flags);
	if (kref_put_and_lock(&se_sess->sess_kref, &ha->hardware_lock,
				target_release_session))
		spin_unlock(&ha->hardware_lock);
	local_irq_restore(flags);

I don't propose we do this yet, correct code is more important than
fast code at this point.  But we should be able to reclaim most of the
performance impact - if it is actually measureable.

Jörn

--
The grand essentials of happiness are: something to do, something to
love, and something to hope for.
-- Allan K. Chalmers

[PATCH] kref: add kref_put_and_lock()

Similar to kref_put(), but will only take a lock (and a performance hit)
if the refcount drops to zero.

Signed-off-by: Joern Engel <joern@xxxxxxxxx>
---
 include/linux/kref.h |    3 +++
 lib/kref.c           |   23 +++++++++++++++++++++++
 2 files changed, 26 insertions(+)

diff --git a/include/linux/kref.h b/include/linux/kref.h
index d4a62ab..9581493 100644
--- a/include/linux/kref.h
+++ b/include/linux/kref.h
@@ -16,6 +16,7 @@
 #define _KREF_H_
 
 #include <linux/types.h>
+#include <linux/spinlock.h>
 
 struct kref {
 	atomic_t refcount;
@@ -24,6 +25,8 @@ struct kref {
 void kref_init(struct kref *kref);
 void kref_get(struct kref *kref);
 int kref_put(struct kref *kref, void (*release) (struct kref *kref));
+int kref_put_and_lock(struct kref *kref, spinlock_t *lock,
+		void (*release)(struct kref *kref));
 int kref_sub(struct kref *kref, unsigned int count,
 	     void (*release) (struct kref *kref));
 
diff --git a/lib/kref.c b/lib/kref.c
index 3efb882..07a82f3 100644
--- a/lib/kref.c
+++ b/lib/kref.c
@@ -62,6 +62,29 @@ int kref_put(struct kref *kref, void (*release)(struct kref *kref))
 	return 0;
 }
 
+/**
+ * kref_put_and_lock - decrement refcount for object with locking
+ * @kref:	object
+ * @lock:	the lock to acquire when dropping the last refcount
+ * @release:	pointer to the function that will clean up the object when the
+ *		last reference to the object is released.
+ *
+ * Same as kref_put(), except that it will acquire the lock before dropping the
+ * last refcount.  If the last refcount is dropped, the lock will be held on
+ * return and the return value will be 1.
+ */
+int kref_put_and_lock(struct kref *kref, spinlock_t *lock,
+		void (*release)(struct kref *kref))
+{
+	WARN_ON(release == NULL);
+	WARN_ON(release == (void (*)(struct kref *))kfree);
+
+	if (atomic_dec_and_lock(&kref->refcount, lock)) {
+		release(kref);
+		return 1;
+	}
+	return 0;
+}
 
 /**
  * kref_sub - subtract a number of refcounts for object.
-- 
1.7.10

--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Index of Archives]     [SCSI Target Devel]     [Linux SCSI Target Infrastructure]     [Kernel Newbies]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Linux IIO]     [Samba]     [Device Mapper]
  Powered by Linux