On Thu, Nov 08, 2018 at 04:58:28PM -0800, Manjunath Patil wrote:

If a hot-attaching device fails inside domU[ex:negotiate_mq() returns with ENOMEM] we clear the blkfront_info struct in talk_to_blkback()

When we try to detach the device, blkfront_remove() tries to access blkfront_info leading to kernel panic.

Typical call stack involving panic - #10 page_fault at ffffffff816f25df [exception RIP: blkif_free+46] #11 blkfront_remove at ffffffffa004de47 [xen_blkfront] #12 xenbus_dev_remove at ffffffff813faa4c #13 __device_release_driver at ffffffff814769aa #14 device_release_driver at ffffffff81476a63 #15 bus_remove_device at ffffffff814762eb #16 device_del at ffffffff81472684 #17 device_unregister at ffffffff814727e2 #18 xenbus_dev_changed at ffffffff813fa89f #19 frontend_changed at ffffffff813fca19 #20 xenwatch_thread at ffffffff813f88f9 #21 kthread at ffffffff810a486b #22 ret_from_fork at ffffffff816ed2a8

Cc: stable@vger.kernel.org Signed-off-by: Manjunath Patil manjunath.b.patil@oracle.com

---

drivers/block/xen-blkfront.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c index dc8fe25..144cda8 100644 --- a/drivers/block/xen-blkfront.c +++ b/drivers/block/xen-blkfront.c @@ -2496,6 +2496,9 @@ static int blkfront_remove(struct xenbus_device *xbdev) dev_dbg(&xbdev->dev, "%s removed", xbdev->nodename);

• if(!info)

• return 0;
  

Always run scripts/checkpatch.pl on your patches so you do not get a grumpy maintainer telling you to run scripts/checkpatch.pl on your patches.

thanks,

greg k-h