[PATCH] Detect early free of a live mm

[Mark Rutland]

KASAN splats indicate that in some cases we free a live mm, then
continue to access it, with potentially disastrous results. This is
likely due to a mismatched mmdrop() somewhere in the kernel, but so far
the culprit remains elusive.

Let's have __mmdrop() verify that the mm isn't live for the current
task, similar to the existing check for init_mm. This way, we can catch
this class of issue earlier, and without requiring KASAN.

Signed-off-by: Mark Rutland <mark.rutland@xxxxxxx>
Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
Cc: Michal Hocko <mhocko@xxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxx>
Cc: Rik van Riel <riel@xxxxxxxxxx>
Cc: Will Deacon <will.deacon@xxxxxxx>
---
 kernel/fork.c | 2 ++
 1 file changed, 2 insertions(+)

Hi,

For context, we're seeing an intermittent use-after-free of an mm on
arm64 [1], where it looks like an mm has been freed earlier than
expected. So far KASAN has only caught legitimate mmdrop() uses, where
mm_count is presumably already bogus.

Mark.

[1] https://lkml.kernel.org/r/20180214120254.qq4w4s42ecxio7lu@xxxxxxxxxxxxxxxxxxxxxxxxx

diff --git a/kernel/fork.c b/kernel/fork.c
index e5d9d405ae4e..6922d93551b8 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -595,6 +595,8 @@ static void check_mm(struct mm_struct *mm)
 void __mmdrop(struct mm_struct *mm)
 {
 	BUG_ON(mm == &init_mm);
+	BUG_ON(mm == current->mm);
+	BUG_ON(mm == current->active_mm);
 	mm_free_pgd(mm);
 	destroy_context(mm);
 	hmm_mm_destroy(mm);
-- 
2.11.0