In blkif_free_ring(), the driver tears down the ring's persistent grants,
shadow request arrays, and shared ring structure (xenbus_teardown_ring),
and only calls unbind_from_irqhandler() at the very end.

While blkif_free_ring() is freeing persistent grants and clearing the
shadow array, the event channel interrupt (blkif_interrupt) is still
registered and active.  If an interrupt arrives from the backend during
this teardown window, blkif_interrupt() reads rinfo->ring.sring and,
via blkif_completion(), accesses rinfo->shadow[id].grants_used and
rinfo->shadow[id].sg.  blkif_free_ring() tears these structures down
without holding rinfo->ring_lock, and the handler only checks
info->connected at entry, so this is a real race resulting in a
use-after-free or NULL pointer dereference.

Fix this by moving unbind_from_irqhandler() to the beginning of
blkif_free_ring().  Calling unbind_from_irqhandler() first frees the
IRQ and synchronizes with any in-flight interrupt handlers on other CPUs
before ring memory and shadow request structures are deallocated,
matching the teardown order in drivers/net/xen-netfront.c.

Fixes: 11659569f720 ("xen/blkfront: split per device io_lock")
Cc: stable@vger.kernel.org
Signed-off-by: Yuchao Zhang <ndaugoing@gmail.com>
---
v2:
 - Update Fixes tag to 11659569f720 ("xen/blkfront: split per device io_lock")
   per Roger Pau Monné.
 - Add comment in blkif_free_ring() noting that interrupt teardown must precede
   freeing queue-related data.

 drivers/block/xen-blkfront.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
index 8dad7bf5f664..86f5dd3aced1 100644
--- a/drivers/block/xen-blkfront.c
+++ b/drivers/block/xen-blkfront.c
@@ -1210,6 +1210,14 @@ static void blkif_free_ring(struct blkfront_ring_info *rinfo)
 	struct blkfront_info *info = rinfo->dev_info;
 	int i, j, segs;
 
+	/*
+	 * Interrupt teardown must be done ahead of freeing queue-related
+	 * data, otherwise the interrupt handler can race with the cleanup.
+	 */
+	if (rinfo->irq)
+		unbind_from_irqhandler(rinfo->irq, rinfo);
+	rinfo->evtchn = rinfo->irq = 0;
+
 	/*
 	 * Remove indirect pages, this only happens when using indirect
 	 * descriptors but not persistent grants
@@ -1292,10 +1300,6 @@ static void blkif_free_ring(struct blkfront_ring_info *rinfo)
 	/* Free resources associated with old device channel. */
 	xenbus_teardown_ring((void **)&rinfo->ring.sring, info->nr_ring_pages,
 			     rinfo->ring_ref);
-
-	if (rinfo->irq)
-		unbind_from_irqhandler(rinfo->irq, rinfo);
-	rinfo->evtchn = rinfo->irq = 0;
 }
 
 static void blkif_free(struct blkfront_info *info, int suspend)
-- 
2.53.0

On Tue, Sep 22, 2026 at 03:21:55PM +0800, Yuchao Zhang wrote:
> In blkif_free_ring(), the driver tears down the ring's persistent grants,
> shadow request arrays, and shared ring structure (xenbus_teardown_ring),
> and only calls unbind_from_irqhandler() at the very end.
> 
> While blkif_free_ring() is freeing persistent grants and clearing the
> shadow array, the event channel interrupt (blkif_interrupt) is still
> registered and active.  If an interrupt arrives from the backend during
> this teardown window, blkif_interrupt() reads rinfo->ring.sring and,
> via blkif_completion(), accesses rinfo->shadow[id].grants_used and
> rinfo->shadow[id].sg.  blkif_free_ring() tears these structures down
> without holding rinfo->ring_lock, and the handler only checks
> info->connected at entry, so this is a real race resulting in a
> use-after-free or NULL pointer dereference.
> 
> Fix this by moving unbind_from_irqhandler() to the beginning of
> blkif_free_ring().  Calling unbind_from_irqhandler() first frees the
> IRQ and synchronizes with any in-flight interrupt handlers on other CPUs
> before ring memory and shadow request structures are deallocated,
> matching the teardown order in drivers/net/xen-netfront.c.
> 
> Fixes: 11659569f720 ("xen/blkfront: split per device io_lock")
> Cc: stable@vger.kernel.org
> Signed-off-by: Yuchao Zhang <ndaugoing@gmail.com>

Acked-by: Roger Pau Monné <roger@xenproject.org>

Thanks, Roger.