netdev - [PATCH net-next] page_pool: unlink from napi during destroy

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-Id: <20230419182006.719923-1-kuba@kernel.org>
Date:   Wed, 19 Apr 2023 11:20:06 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     davem@...emloft.net
Cc:     netdev@...r.kernel.org, edumazet@...gle.com, pabeni@...hat.com,
        Jakub Kicinski <kuba@...nel.org>,
        Jesper Dangaard Brouer <jbrouer@...hat.com>, hawk@...nel.org,
        ilias.apalodimas@...aro.org, tariqt@...dia.com
Subject: [PATCH net-next] page_pool: unlink from napi during destroy

Jesper points out that we must prevent recycling into cache
after page_pool_destroy() is called, because page_pool_destroy()
is not synchronized with recycling (some pages may still be
outstanding when destroy() gets called).

I assumed this will not happen because NAPI can't be scheduled
if its page pool is being destroyed. But I missed the fact that
NAPI may get reused. For instance when user changes ring configuration
driver may allocate a new page pool, stop NAPI, swap, start NAPI,
and then destroy the old pool. The NAPI is running so old page
pool will think it can recycle to the cache, but the consumer
at that point is the destroy() path, not NAPI.

To avoid extra synchronization let the drivers do "unlinking"
during the "swap" stage while NAPI is indeed disabled.

Fixes: 8c48eea3adf3 ("page_pool: allow caching from safely localized NAPI")
Reported-by: Jesper Dangaard Brouer <jbrouer@...hat.com>
Link: https://lore.kernel.org/all/e8df2654-6a5b-3c92-489d-2fe5e444135f@redhat.com/
Signed-off-by: Jakub Kicinski <kuba@...nel.org>
---
bnxt does not do the live swap so no need to change it.
But let's export the API, anyway, I'm sure others will
need it. And knowing driver authors they will hack some
workaround into the driver rather than export the helper.

CC: hawk@...nel.org
CC: ilias.apalodimas@...aro.org
CC: tariqt@...dia.com
---
 include/net/page_pool.h |  5 +++++
 net/core/page_pool.c    | 18 +++++++++++++++++-
 2 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/include/net/page_pool.h b/include/net/page_pool.h
index 91b808dade82..c8ec2f34722b 100644
--- a/include/net/page_pool.h
+++ b/include/net/page_pool.h
@@ -247,6 +247,7 @@ struct page_pool *page_pool_create(const struct page_pool_params *params);
 struct xdp_mem_info;
 
 #ifdef CONFIG_PAGE_POOL
+void page_pool_unlink_napi(struct page_pool *pool);
 void page_pool_destroy(struct page_pool *pool);
 void page_pool_use_xdp_mem(struct page_pool *pool, void (*disconnect)(void *),
 			   struct xdp_mem_info *mem);
@@ -254,6 +255,10 @@ void page_pool_release_page(struct page_pool *pool, struct page *page);
 void page_pool_put_page_bulk(struct page_pool *pool, void **data,
 			     int count);
 #else
+static inline void page_pool_unlink_napi(struct page_pool *pool)
+{
+}
+
 static inline void page_pool_destroy(struct page_pool *pool)
 {
 }
diff --git a/net/core/page_pool.c b/net/core/page_pool.c
index 2f6bf422ed30..caa30a5a5abc 100644
--- a/net/core/page_pool.c
+++ b/net/core/page_pool.c
@@ -838,6 +838,21 @@ void page_pool_use_xdp_mem(struct page_pool *pool, void (*disconnect)(void *),
 	pool->xdp_mem_id = mem->id;
 }
 
+void page_pool_unlink_napi(struct page_pool *pool)
+{
+	if (!pool->p.napi)
+		return;
+
+	/* To avoid races with recycling and additional barriers make sure
+	 * pool and NAPI are unlinked when NAPI is disabled.
+	 */
+	WARN_ON(!test_bit(NAPI_STATE_SCHED, &pool->p.napi->state) ||
+		READ_ONCE(pool->p.napi->list_owner) != -1);
+
+	WRITE_ONCE(pool->p.napi, NULL);
+}
+EXPORT_SYMBOL(page_pool_unlink_napi);
+
 void page_pool_destroy(struct page_pool *pool)
 {
 	if (!pool)
@@ -846,6 +861,7 @@ void page_pool_destroy(struct page_pool *pool)
 	if (!page_pool_put(pool))
 		return;
 
+	page_pool_unlink_napi(pool);
 	page_pool_free_frag(pool);
 
 	if (!page_pool_release(pool))
@@ -899,7 +915,7 @@ bool page_pool_return_skb_page(struct page *page, bool napi_safe)
 	 * in the same context as the consumer would run, so there's
 	 * no possible race.
 	 */
-	napi = pp->p.napi;
+	napi = READ_ONCE(pp->p.napi);
 	allow_direct = napi_safe && napi &&
 		READ_ONCE(napi->list_owner) == smp_processor_id();
 
-- 
2.39.2