hold task->mempolicy while numa_maps scans.

  /proc/<pid>/numa_maps scans vma and show mempolicy under
  mmap_sem. It sometimes accesses task->mempolicy which can
  be freed without mmap_sem and numa_maps can show some
  garbage while scanning.

This patch tries to take reference count of task->mempolicy at reading
numa_maps before calling get_vma_policy(). By this, task->mempolicy
will not be freed until numa_maps reaches its end.

V2->v3
  -  updated comments to be more verbose.
  -  removed task_lock() in numa_maps code.
V1->V2
  -  access task->mempolicy only once and remember it.  Becase kernel/exit.c
     can overwrite it.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

2 files changed, 51 insertions, 3 deletions

diff --git a/fs/proc/internal.h b/fs/proc/internal.h
index cceaab07ad5..43973b084ab 100644
--- a/fs/proc/internal.h
+++ b/fs/proc/internal.h
@@ -12,6 +12,7 @@
 #include <linux/sched.h>
 #include <linux/proc_fs.h>
 struct  ctl_table_header;
+struct  mempolicy;
 
 extern struct proc_dir_entry proc_root;
 #ifdef CONFIG_PROC_SYSCTL
@@ -74,6 +75,9 @@ struct proc_maps_private {
 #ifdef CONFIG_MMU
 	struct vm_area_struct *tail_vma;
 #endif
+#ifdef CONFIG_NUMA
+	struct mempolicy *task_mempolicy;
+#endif
 };
 
 void proc_init_inodecache(void);
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 14df8806ff2..90c63f9392a 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -90,10 +90,55 @@ static void pad_len_spaces(struct seq_file *m, int len)
 	seq_printf(m, "%*c", len, ' ');
 }
 
+#ifdef CONFIG_NUMA
+/*
+ * These functions are for numa_maps but called in generic **maps seq_file
+ * ->start(), ->stop() ops.
+ *
+ * numa_maps scans all vmas under mmap_sem and checks their mempolicy.
+ * Each mempolicy object is controlled by reference counting. The problem here
+ * is how to avoid accessing dead mempolicy object.
+ *
+ * Because we're holding mmap_sem while reading seq_file, it's safe to access
+ * each vma's mempolicy, no vma objects will never drop refs to mempolicy.
+ *
+ * A task's mempolicy (task->mempolicy) has different behavior. task->mempolicy
+ * is set and replaced under mmap_sem but unrefed and cleared under task_lock().
+ * So, without task_lock(), we cannot trust get_vma_policy() because we cannot
+ * gurantee the task never exits under us. But taking task_lock() around
+ * get_vma_plicy() causes lock order problem.
+ *
+ * To access task->mempolicy without lock, we hold a reference count of an
+ * object pointed by task->mempolicy and remember it. This will guarantee
+ * that task->mempolicy points to an alive object or NULL in numa_maps accesses.
+ */
+static void hold_task_mempolicy(struct proc_maps_private *priv)
+{
+	struct task_struct *task = priv->task;
+
+	task_lock(task);
+	priv->task_mempolicy = task->mempolicy;
+	mpol_get(priv->task_mempolicy);
+	task_unlock(task);
+}
+static void release_task_mempolicy(struct proc_maps_private *priv)
+{
+	mpol_put(priv->task_mempolicy);
+}
+#else
+static void hold_task_mempolicy(struct proc_maps_private *priv)
+{
+}
+static void release_task_mempolicy(struct proc_maps_private *priv)
+{
+}
+#endif
+
 static void vma_stop(struct proc_maps_private *priv, struct vm_area_struct *vma)
 {
 	if (vma && vma != priv->tail_vma) {
 		struct mm_struct *mm = vma->vm_mm;
+		release_task_mempolicy(priv);
 		up_read(&mm->mmap_sem);
 		mmput(mm);
 	}
@@ -132,7 +177,7 @@ static void *m_start(struct seq_file *m, loff_t *pos)
 
 	tail_vma = get_gate_vma(priv->task->mm);
 	priv->tail_vma = tail_vma;
-
+	hold_task_mempolicy(priv);
 	/* Start with last addr hint */
 	vma = find_vma(mm, last_addr);
 	if (last_addr && vma) {
@@ -159,6 +204,7 @@ out:
 	if (vma)
 		return vma;
 
+	release_task_mempolicy(priv);
 	/* End of vmas has been reached */
 	m->version = (tail_vma != NULL)? 0: -1UL;
 	up_read(&mm->mmap_sem);
@@ -1178,11 +1224,9 @@ static int show_numa_map(struct seq_file *m, void *v, int is_pid)
 	walk.private = md;
 	walk.mm = mm;
 
-	task_lock(task);
 	pol = get_vma_policy(task, vma, vma->vm_start);
 	mpol_to_str(buffer, sizeof(buffer), pol, 0);
 	mpol_cond_put(pol);
-	task_unlock(task);
 
 	seq_printf(m, "%08lx %s", vma->vm_start, buffer);