Re: [PATCH] xen-blkback: fix two memleaks

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Fri, Dec 11, 2015 at 12:24:01PM -0500, Konrad Rzeszutek Wilk wrote:
> > diff --git a/drivers/block/xen-blkback/xenbus.c b/drivers/block/xen-blkback/xenbus.c
> > index 44396b8..dabdb18 100644
> > --- a/drivers/block/xen-blkback/xenbus.c
> > +++ b/drivers/block/xen-blkback/xenbus.c
> > @@ -246,6 +246,9 @@ static int xen_blkif_disconnect(struct xen_blkif *blkif)
> >  	struct pending_req *req, *n;
> >  	unsigned int j, r;
> >  
> > +	if (!blkif->rings)
> > +		goto out;
> > +
> 
> I dropped this and instead added: blkif->nr_rings = 0.
> 
> Which will mean we can still run through this function many
> times but that should not be a problem.

And I realized that there is still a latent bug. We take
the refcount on the rings whenever we go to connect_ring.

But we don't put it in disconnect. So a guest could change
our state to go in (say we have 4 rings). Initially our 
refcount is 1.

XenbusStateConnected (we refcount to five).
XenbusStateClosed (we don't do a put on the refcount, so we are still
 at 5).
XenbusStateConnected (we refcount to ten now).
XenbusStateClosed (refcount still at ten).
Guest dies, we go to xen_blkbk_remove, which clears the rings but
only refcounts down to five.

My first instict was to move the loop from xen_blkbk_remove:
               for (i = 0; i < be->blkif->nr_rings; i++)                       
                        xen_blkif_put(be->blkif);                

in the xen_blkif_disconnect().

That would take care of that, and also of the inadvert bug I added
with doign blkif->nr_rings = 0 in here - that is the loop above
would never run (oops).


But the problem we have is that the xen_blkif_put calls the 
xen_blkif_deferred_free (when the refcount is zero) which calls
xen_blkif_free which calls xen_blkif_disconnect.

How about this patch:
>From a95337a898fa452eb417e429ef270283393fd994 Mon Sep 17 00:00:00 2001
From: Bob Liu <bob.liu@oracle.com>
Date: Thu, 10 Dec 2015 09:16:48 +0800
Subject: [PATCH] xen/blkback: Fix two memory leaks.

This patch fixs two memleaks:
  backtrace:
    [<ffffffff817ba5e8>] kmemleak_alloc+0x28/0x50
    [<ffffffff81205e3b>] kmem_cache_alloc+0xbb/0x1d0
    [<ffffffff81534028>] xen_blkbk_probe+0x58/0x230
    [<ffffffff8146adb6>] xenbus_dev_probe+0x76/0x130
    [<ffffffff81511716>] driver_probe_device+0x166/0x2c0
    [<ffffffff815119bc>] __device_attach_driver+0xac/0xb0
    [<ffffffff8150fa57>] bus_for_each_drv+0x67/0x90
    [<ffffffff81511ab7>] __device_attach+0xc7/0x120
    [<ffffffff81511b23>] device_initial_probe+0x13/0x20
    [<ffffffff8151059a>] bus_probe_device+0x9a/0xb0
    [<ffffffff8150f0a1>] device_add+0x3b1/0x5c0
    [<ffffffff8150f47e>] device_register+0x1e/0x30
    [<ffffffff8146a9e8>] xenbus_probe_node+0x158/0x170
    [<ffffffff8146abaf>] xenbus_dev_changed+0x1af/0x1c0
    [<ffffffff8146b1bb>] backend_changed+0x1b/0x20
    [<ffffffff81468ca6>] xenwatch_thread+0xb6/0x160
unreferenced object 0xffff880007ba8ef8 (size 224):

  backtrace:
    [<ffffffff817ba5e8>] kmemleak_alloc+0x28/0x50
    [<ffffffff81205c73>] __kmalloc+0xd3/0x1e0
    [<ffffffff81534d87>] frontend_changed+0x2c7/0x580
    [<ffffffff8146af12>] xenbus_otherend_changed+0xa2/0xb0
    [<ffffffff8146b2c0>] frontend_changed+0x10/0x20
    [<ffffffff81468ca6>] xenwatch_thread+0xb6/0x160
    [<ffffffff810d3e97>] kthread+0xd7/0xf0
    [<ffffffff817c4a9f>] ret_from_fork+0x3f/0x70
    [<ffffffffffffffff>] 0xffffffffffffffff
unreferenced object 0xffff8800048dcd38 (size 224):

The first leak is caused by not put() the be->blkif reference
which we had gotten in xen_blkif_alloc(), while the second is
us not freeing blkif->rings in the right place.

Signed-off-by: Bob Liu <bob.liu@oracle.com>
Reported-and-Tested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
 drivers/block/xen-blkback/xenbus.c | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/drivers/block/xen-blkback/xenbus.c b/drivers/block/xen-blkback/xenbus.c
index 44396b8..ab793ac 100644
--- a/drivers/block/xen-blkback/xenbus.c
+++ b/drivers/block/xen-blkback/xenbus.c
@@ -297,8 +297,16 @@ static int xen_blkif_disconnect(struct xen_blkif *blkif)
 		BUG_ON(ring->free_pages_num != 0);
 		BUG_ON(ring->persistent_gnt_c != 0);
 		WARN_ON(i != (XEN_BLKIF_REQS_PER_PAGE * blkif->nr_ring_pages));
+		xen_blkif_put(be->blkif);
 	}
 	blkif->nr_ring_pages = 0;
+	/*
+	 * blkif->rings was allocated in connect_ring, so we should free it in
+	 * here.
+	 */
+	kfree(blkif->rings);
+	blkif->rings = NULL;
+	blkif->nr_rings = 0;
 
 	return 0;
 }
@@ -310,7 +318,6 @@ static void xen_blkif_free(struct xen_blkif *blkif)
 	xen_vbd_free(&blkif->vbd);
 
 	/* Make sure everything is drained before shutting down */
-	kfree(blkif->rings);
 	kmem_cache_free(xen_blkif_cachep, blkif);
 }
 
@@ -499,12 +506,11 @@ static int xen_blkbk_remove(struct xenbus_device *dev)
 
 	dev_set_drvdata(&dev->dev, NULL);
 
-	if (be->blkif) {
+	if (be->blkif)
 		xen_blkif_disconnect(be->blkif);
-		for (i = 0; i < be->blkif->nr_rings; i++)
-			xen_blkif_put(be->blkif);
-	}
 
+	/* Put the reference we set in xen_blkif_alloc(). */
+	xen_blkif_put(be->blkif);
 	kfree(be->mode);
 	kfree(be);
 	return 0;
-- 
2.1.0