Thread (9 messages) read the whole thread 9 messages, 3 authors, 2024-12-07

Re: [PATCH RFC v4 1/3] page_pool: fix timing for checking and disabling napi_local

From: Jakub Kicinski <kuba@kernel.org>
Date: 2024-12-03 02:49:56
Also in: lkml

On Wed, 20 Nov 2024 18:34:53 +0800 Yunsheng Lin wrote:
page_pool page may be freed from skb_defer_free_flush() in
softirq context without binding to any specific napi, it
may cause use-after-free problem due to the below time window,
as below, CPU1 may still access napi->list_owner after CPU0
free the napi memory:

            CPU 0                           CPU1
      page_pool_destroy()          skb_defer_free_flush()
             .                               .
             .                napi = READ_ONCE(pool->p.napi);
             .                               .
page_pool_disable_direct_recycling()         .
   driver free napi memory                   .
             .                               .
             .       napi && READ_ONCE(napi->list_owner) == cpuid
             .                               .

Use rcu mechanism to avoid the above problem.

Note, the above was found during code reviewing on how to fix
the problem in [1].

1. https://lore.kernel.org/lkml/8067f204-1380-4d37-8ffd-007fc6f26738@kernel.org/T/ (local)
Please split it from the rest of the series, it's related but not
really logically connected (dma mappings and NAPI recycling are 
different features of page pool).
quoted hunk ↗ jump to hunk
@@ -1126,6 +1133,12 @@ void page_pool_destroy(struct page_pool *pool)
 	if (!page_pool_release(pool))
 		return;
 
+	/* Paired with rcu lock in page_pool_napi_local() to enable clearing
+	 * of pool->p.napi in page_pool_disable_direct_recycling() is seen
+	 * before returning to driver to free the napi instance.
+	 */
+	synchronize_rcu();
I don't think this is in the right place.
Why not inside page_pool_disable_direct_recycling() ?

Hopefully this doesn't cause long stalls during reconfig, normally
NAPIs and page pools are freed together, and NAPI removal implies
synchronize_rcu(). That's why it's not really a problem for majority
of drivers. You should perhaps make a note of this in the commit
message.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help