Re: [PATCH net] page_pool: fix inconsistency for page_pool_ring_[un]lock()

From: Jakub Kicinski
Date: Fri May 26 2023 - 15:35:00 EST

On Mon, 22 May 2023 11:17:14 +0800 Yunsheng Lin wrote:
> page_pool_ring_[un]lock() use in_softirq() to decide which
> spin lock variant to use, and when they are called in the
> context with in_softirq() being false, spin_lock_bh() is
> called in page_pool_ring_lock() while spin_unlock() is
> called in page_pool_ring_unlock(), because spin_lock_bh()
> has disabled the softirq in page_pool_ring_lock(), which
> causes inconsistency for spin lock pair calling.
> This patch fixes it by returning in_softirq state from
> page_pool_producer_lock(), and use it to decide which
> spin lock variant to use in page_pool_producer_unlock().
> As pool->ring has both producer and consumer lock, so
> rename it to page_pool_producer_[un]lock() to reflect
> the actual usage. Also move them to page_pool.c as they
> are only used there, and remove the 'inline' as the
> compiler may have better idea to do inlining or not.
> Fixes: 7886244736a4 ("net: page_pool: Add bulk support for ptr_ring")
> Signed-off-by: Yunsheng Lin <linyunsheng@xxxxxxxxxx>

I just realized now while doing backports that the Fixes tag is
incorrect here. The correct Fixes tag is:

Fixes: 542bcea4be86 ("net: page_pool: use in_softirq() instead")

Before that we used in_serving_softirq() which was perfectly fine.
This explains the major mystery of how such a serious bug would survive
for 10+ releases... it didn't, it wasn't there :) It only came in 6.3.
We can't change the tag now but at least the universe makes sense again.