Re: [BUG Report] KASAN: slab-use-after-free in page_pool_recycle_in_ring

[Jakub Kicinski <kuba@kernel.org>]

On Tue, 13 May 2025 13:06:38 -0700 Mina Almasry wrote:
> diff --git a/net/core/page_pool.c b/net/core/page_pool.c
> index 2b76848659418..8654608734773 100644
> --- a/net/core/page_pool.c
> +++ b/net/core/page_pool.c
> @@ -1146,10 +1146,17 @@ static void page_pool_scrub(struct page_pool *pool)
> 
>  static int page_pool_release(struct page_pool *pool)
>  {
> +       bool in_softirq;
>         int inflight;
> 
> +
> +       /* Acquire producer lock to make sure we don't race with another thread
> +        * returning a netmem to the ptr_ring.
> +        */
> +       in_softirq = page_pool_producer_lock(pool);
>         page_pool_scrub(pool);
>         inflight = page_pool_inflight(pool, true);
> +       page_pool_producer_unlock(pool, in_softirq);

Makes sense! A couple minor notes.

Consumer lock should be outside, but really we only need to make 
sure producer has "exited" right? So lock/unlock, no need to wrap
any code in it.

I'd add a helper to ptr_ring.h, a "producer barrier" which just
takes/releases the producer lock. We can't be in softirq context
here but doesn't matter, let's take the lock in "any" mode IOW
irqsave() ?

The barrier is only needed if we're proceeding to destruction.
If inflight returns != 0 we won't destroy the pool so no need
to touch producer lock.