xen-blkfront: fix kernel panic with negotiate_mq error path
authorManjunath Patil <manjunath.b.patil@oracle.com>
Tue, 30 Oct 2018 16:49:21 +0000 (09:49 -0700)
committerJuergen Gross <jgross@suse.com>
Tue, 6 Nov 2018 09:33:36 +0000 (10:33 +0100)
info->nr_rings isn't adjusted in case of ENOMEM error from
negotiate_mq(). This leads to kernel panic in error path.

Typical call stack involving panic -
 #8 page_fault at ffffffff8175936f
    [exception RIP: blkif_free_ring+33]
    RIP: ffffffffa0149491  RSP: ffff8804f7673c08  RFLAGS: 00010292
 ...
 #9 blkif_free at ffffffffa0149aaa [xen_blkfront]
 #10 talk_to_blkback at ffffffffa014c8cd [xen_blkfront]
 #11 blkback_changed at ffffffffa014ea8b [xen_blkfront]
 #12 xenbus_otherend_changed at ffffffff81424670
 #13 backend_changed at ffffffff81426dc3
 #14 xenwatch_thread at ffffffff81422f29
 #15 kthread at ffffffff810abe6a
 #16 ret_from_fork at ffffffff81754078

Cc: stable@vger.kernel.org
Fixes: 7ed8ce1c5fc7 ("xen-blkfront: move negotiate_mq to cover all cases of new VBDs")
Signed-off-by: Manjunath Patil <manjunath.b.patil@oracle.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Signed-off-by: Juergen Gross <jgross@suse.com>
drivers/block/xen-blkfront.c

index 9eea83ae01c6fdfd2b4696dba313a43a9b26bdd1..9ac3999a55b7a678cc3a91962ca268e82b535df1 100644 (file)
@@ -1919,6 +1919,7 @@ static int negotiate_mq(struct blkfront_info *info)
                              GFP_KERNEL);
        if (!info->rinfo) {
                xenbus_dev_fatal(info->xbdev, -ENOMEM, "allocating ring_info structure");
+               info->nr_rings = 0;
                return -ENOMEM;
        }