On Wed, 30 Jan 2019 09:42:06 +0900 Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx> wrote:
> > >
> > > If we want to allow vfree() to sleep, at least we need to test with
> > > kvmalloc() == vmalloc() (i.e. force kvmalloc()/kvfree() users to use
> > > vmalloc()/vfree() path). For now, reverting the
> > > "Context: Either preemptible task context or not-NMI interrupt." change
> > > will be needed for stable kernels.
> >
> > So, the comment for vfree "May sleep if called *not* from interrupt
> > context." is wrong?
>
> Commit bf22e37a641327e3 ("mm: add vfree_atomic()") says
>
> We are going to use sleeping lock for freeing vmap. However some
> vfree() users want to free memory from atomic (but not from interrupt)
> context. For this we add vfree_atomic() - deferred variation of vfree()
> which can be used in any atomic context (except NMIs).
>
> and commit 52414d3302577bb6 ("kvfree(): fix misleading comment") made
>
> - * Context: Any context except NMI.
> + * Context: Either preemptible task context or not-NMI interrupt.
>
> change. But I think that we converted kmalloc() to kvmalloc() without checking
> context of kvfree() callers. Therefore, I think that kvfree() needs to use
> vfree_atomic() rather than just saying "vfree() might sleep if called not in
> interrupt context."...
Whereabouts in the vfree() path can the kernel sleep?
