On 02/02/2015 08:29 PM, gregkh@xxxxxxxxxxxxxxxxxxx wrote:
The patch below does not apply to the 3.14-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable@xxxxxxxxxxxxxxx>.
thanks,
greg k-h
Ilya, I hope you can handle this. I'm on vacation this week...
-Alex
------------------ original commit in Linus's tree ------------------
From ae43e9d05eb4bd324155292f889fbd001c4faea8 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@xxxxxxxxxx>
Date: Mon, 19 Jan 2015 18:13:43 +0300
Subject: [PATCH] rbd: fix rbd_dev_parent_get() when parent_overlap == 0
The comment for rbd_dev_parent_get() said
* We must get the reference before checking for the overlap to
* coordinate properly with zeroing the parent overlap in
* rbd_dev_v2_parent_info() when an image gets flattened. We
* drop it again if there is no overlap.
but the "drop it again if there is no overlap" part was missing from
the implementation. This lead to absurd parent_ref values for images
with parent_overlap == 0, as parent_ref was incremented for each
img_request and virtually never decremented.
Fix this by leveraging the fact that refresh path calls
rbd_dev_v2_parent_info() under header_rwsem and use it for read in
rbd_dev_parent_get(), instead of messing around with atomics. Get rid
of barriers in rbd_dev_v2_parent_info() while at it - I don't see what
they'd pair with now and I suspect we are in a pretty miserable
situation as far as proper locking goes regardless.
Cc: stable@xxxxxxxxxxxxxxx # 3.11+
Signed-off-by: Ilya Dryomov <idryomov@xxxxxxxxxx>
Reviewed-by: Josh Durgin <jdurgin@xxxxxxxxxx>
Reviewed-by: Alex Elder <elder@xxxxxxxxxx>
diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
index 3ec85dfce124..d511ecb35144 100644
--- a/drivers/block/rbd.c
+++ b/drivers/block/rbd.c
@@ -2098,32 +2098,26 @@ static void rbd_dev_parent_put(struct rbd_device *rbd_dev)
* If an image has a non-zero parent overlap, get a reference to its
* parent.
*
- * We must get the reference before checking for the overlap to
- * coordinate properly with zeroing the parent overlap in
- * rbd_dev_v2_parent_info() when an image gets flattened. We
- * drop it again if there is no overlap.
- *
* Returns true if the rbd device has a parent with a non-zero
* overlap and a reference for it was successfully taken, or
* false otherwise.
*/
static bool rbd_dev_parent_get(struct rbd_device *rbd_dev)
{
- int counter;
+ int counter = 0;
if (!rbd_dev->parent_spec)
return false;
- counter = atomic_inc_return_safe(&rbd_dev->parent_ref);
- if (counter > 0 && rbd_dev->parent_overlap)
- return true;
-
- /* Image was flattened, but parent is not yet torn down */
+ down_read(&rbd_dev->header_rwsem);
+ if (rbd_dev->parent_overlap)
+ counter = atomic_inc_return_safe(&rbd_dev->parent_ref);
+ up_read(&rbd_dev->header_rwsem);
if (counter < 0)
rbd_warn(rbd_dev, "parent reference overflow");
- return false;
+ return counter > 0;
}
/*
@@ -4239,7 +4233,6 @@ static int rbd_dev_v2_parent_info(struct rbd_device *rbd_dev)
*/
if (rbd_dev->parent_overlap) {
rbd_dev->parent_overlap = 0;
- smp_mb();
rbd_dev_parent_put(rbd_dev);
pr_info("%s: clone image has been flattened\n",
rbd_dev->disk->disk_name);
@@ -4285,7 +4278,6 @@ static int rbd_dev_v2_parent_info(struct rbd_device *rbd_dev)
* treat it specially.
*/
rbd_dev->parent_overlap = overlap;
- smp_mb();
if (!overlap) {
/* A null parent_spec indicates it's the initial probe */
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html