In the last few kernels (at least as far back as 2.6.10), exit_mm()
looks like this:
static void exit_mm(struct task_struct * tsk)
{
struct mm_struct *mm = tsk->mm;
mm_release(tsk, mm);
if (!mm)
return;
while mm_release() looks like this:
void mm_release(struct task_struct *tsk, struct mm_struct *mm)
{
<snip>
if (tsk->clear_child_tid && atomic_read(&mm->mm_users) > 1) {
If it's valid to call exit_mm() with tsk->mm being NULL, then it seems
like we'll get problems when we dereference it unconditionally in
mm_release().
If mm_release() is valid, then the check for !mm in exit_mm() is
unnecessary but there should probably be a comment.
Can someone tell me which one it is?
Thanks,
Chris