KVM: s390: fix locking order problem in enable_sie

There are potential locking problem in enable_sie. We take the task_lock
and the mmap_sem. As exit_mm uses the same locks vice versa, this triggers
a lockdep warning.
The second problem is that dup_mm and mmput might sleep, so we must not
hold the task_lock at that moment.

The solution is to dup the mm unconditional and use the task_lock before and
afterwards to check  if we can use the new mm. dup_mm and mmput are called
outside the task_lock, but we run update_mm while holding the task_lock,
protection us against ptrace.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

1 files changed, 27 insertions, 17 deletions

diff --git a/arch/s390/mm/pgtable.c b/arch/s390/mm/pgtable.c
index 5c1aea97cd12..3d98ba82ea67 100644
--- a/arch/s390/mm/pgtable.c
+++ b/arch/s390/mm/pgtable.c
@@ -254,36 +254,46 @@ void disable_noexec(struct mm_struct *mm, struct task_struct *tsk)
 int s390_enable_sie(void)
 {
         struct task_struct *tsk = current;
-        struct mm_struct *mm;
-        int rc;
+        struct mm_struct *mm, *old_mm;
 
-        task_lock(tsk);
-
-        rc = 0;
+        /* Do we have pgstes? if yes, we are done */
         if (tsk->mm->context.pgstes)
-                goto unlock;
+                return 0;
 
-        rc = -EINVAL;
+        /* lets check if we are allowed to replace the mm */
+        task_lock(tsk);
         if (!tsk->mm || atomic_read(&tsk->mm->mm_users) > 1 ||
-            tsk->mm != tsk->active_mm || tsk->mm->ioctx_list)
-                goto unlock;
+            tsk->mm != tsk->active_mm || tsk->mm->ioctx_list) {
+                task_unlock(tsk);
+                return -EINVAL;
+        }
+        task_unlock(tsk);
 
-        tsk->mm->context.pgstes = 1;    /* dirty little tricks .. */
+        /* we copy the mm with pgstes enabled */
+        tsk->mm->context.pgstes = 1;
         mm = dup_mm(tsk);
         tsk->mm->context.pgstes = 0;
-
-        rc = -ENOMEM;
         if (!mm)
-                goto unlock;
-        mmput(tsk->mm);
+                return -ENOMEM;
+
+        /* Now lets check again if somebody attached ptrace etc */
+        task_lock(tsk);
+        if (!tsk->mm || atomic_read(&tsk->mm->mm_users) > 1 ||
+            tsk->mm != tsk->active_mm || tsk->mm->ioctx_list) {
+                mmput(mm);
+                task_unlock(tsk);
+                return -EINVAL;
+        }
+
+        /* ok, we are alone. No ptrace, no threads, etc. */
+        old_mm = tsk->mm;
         tsk->mm = tsk->active_mm = mm;
         preempt_disable();
         update_mm(mm, tsk);
         cpu_set(smp_processor_id(), mm->cpu_vm_mask);
         preempt_enable();
-        rc = 0;
-unlock:
         task_unlock(tsk);
-        return rc;
+        mmput(old_mm);
+        return 0;
 }
 EXPORT_SYMBOL_GPL(s390_enable_sie);