blktap: fix cleanup after unclean application exit #2

When an application using blktap devices doesn't close the mmap-s of
/dev/xen/blktapN and the frontend driver never connects, we cannot
defer the mmput() on the stored mm until blktap_release() or the exit
path of the worker thread, as the former will never be called without
the mm's reference count dropping to zero, and the worker thread
would never get started.

Also remove a left-over from c/s 1013.

Signed-off-by: Jan Beulich <jbeulich@novell.com>

diff --git a/drivers/xen/blktap/blktap.c b/drivers/xen/blktap/blktap.c
index ffa9802215a19d936111b4783589b52f910d64e3..390885348129b72f4a9a93b7a3bee387b8b0608d 100644 (file)
--- a/drivers/xen/blktap/blktap.c
+++ b/drivers/xen/blktap/blktap.c
@@ -553,6 +553,7 @@ void signal_tapdisk(int idx)
 {
        tap_blkif_t *info;
        struct task_struct *ptask;
+       struct mm_struct *mm;
 
        /*
         * if the userland tools set things up wrong, this could be negative;
@@ -572,7 +573,9 @@ void signal_tapdisk(int idx)
        }
        info->blkif = NULL;
 
-       return;
+       mm = xchg(&info->mm, NULL);
+       if (mm)
+               mmput(mm);
 }
 
 static int blktap_open(struct inode *inode, struct file *filp)
@@ -652,7 +655,6 @@ static int blktap_release(struct inode *inode, struct file *filp)
        mm = xchg(&info->mm, NULL);
        if (mm)
                mmput(mm);
-       info->mm = NULL;
        kfree(info->foreign_map.map);
        info->foreign_map.map = NULL;