Merge branch 'fix-late-dma-unmap-crash-for-page-pool'

Toke Høiland-Jørgensen says:

====================
Fix late DMA unmap crash for page pool

This series fixes the late dma_unmap crash for page pool first reported
by Yonglong Liu in [0]. It is an alternative approach to the one
submitted by Yunsheng Lin, most recently in [1]. The first commit just
wraps some tests in a helper function, in preparation of the main change
in patch 2. See the commit message of patch 2 for the details.

[0] https://lore.kernel.org/8067f204-1380-4d37-8ffd-007fc6f26738@kernel.org
[1] https://lore.kernel.org/20250307092356.638242-1-linyunsheng@huawei.com

v8: https://lore.kernel.org/20250407-page-pool-track-dma-v8-0-da9500d4ba21@redhat.com
v7: https://lore.kernel.org/20250404-page-pool-track-dma-v7-0-ad34f069bc18@redhat.com
v6: https://lore.kernel.org/20250401-page-pool-track-dma-v6-0-8b83474870d4@redhat.com
v5: https://lore.kernel.org/20250328-page-pool-track-dma-v5-0-55002af683ad@redhat.com
v4: https://lore.kernel.org/20250327-page-pool-track-dma-v4-0-b380dc6706d0@redhat.com
v3: https://lore.kernel.org/20250326-page-pool-track-dma-v3-0-8e464016e0ac@redhat.com
v2: https://lore.kernel.org/20250325-page-pool-track-dma-v2-0-113ebc1946f3@redhat.com
v1: https://lore.kernel.org/20250314-page-pool-track-dma-v1-0-c212e57a74c2@redhat.com
====================

Link: https://patch.msgid.link/20250409-page-pool-track-dma-v9-0-6a9ef2e0cba8@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

4 files changed, 104 insertions, 30 deletions

diff --git a/net/core/netmem_priv.h b/net/core/netmem_priv.h
index 7eadb8393e00..cd95394399b4 100644
--- a/net/core/netmem_priv.h
+++ b/net/core/netmem_priv.h
@@ -5,7 +5,7 @@
 
 static inline unsigned long netmem_get_pp_magic(netmem_ref netmem)
 {
-	return __netmem_clear_lsb(netmem)->pp_magic;
+	return __netmem_clear_lsb(netmem)->pp_magic & ~PP_DMA_INDEX_MASK;
 }
 
 static inline void netmem_or_pp_magic(netmem_ref netmem, unsigned long pp_magic)
@@ -15,9 +15,16 @@ static inline void netmem_or_pp_magic(netmem_ref netmem, unsigned long pp_magic)
 
 static inline void netmem_clear_pp_magic(netmem_ref netmem)
 {
+	WARN_ON_ONCE(__netmem_clear_lsb(netmem)->pp_magic & PP_DMA_INDEX_MASK);
+
 	__netmem_clear_lsb(netmem)->pp_magic = 0;
 }
 
+static inline bool netmem_is_pp(netmem_ref netmem)
+{
+	return (netmem_get_pp_magic(netmem) & PP_MAGIC_MASK) == PP_SIGNATURE;
+}
+
 static inline void netmem_set_pp(netmem_ref netmem, struct page_pool *pool)
 {
 	__netmem_clear_lsb(netmem)->pp = pool;
@@ -28,4 +35,28 @@ static inline void netmem_set_dma_addr(netmem_ref netmem,
 {
 	__netmem_clear_lsb(netmem)->dma_addr = dma_addr;
 }
+
+static inline unsigned long netmem_get_dma_index(netmem_ref netmem)
+{
+	unsigned long magic;
+
+	if (WARN_ON_ONCE(netmem_is_net_iov(netmem)))
+		return 0;
+
+	magic = __netmem_clear_lsb(netmem)->pp_magic;
+
+	return (magic & PP_DMA_INDEX_MASK) >> PP_DMA_INDEX_SHIFT;
+}
+
+static inline void netmem_set_dma_index(netmem_ref netmem,
+					unsigned long id)
+{
+	unsigned long magic;
+
+	if (WARN_ON_ONCE(netmem_is_net_iov(netmem)))
+		return;
+
+	magic = netmem_get_pp_magic(netmem) | (id << PP_DMA_INDEX_SHIFT);
+	__netmem_clear_lsb(netmem)->pp_magic = magic;
+}
 #endif
diff --git a/net/core/page_pool.c b/net/core/page_pool.c
index 7745ad924ae2..2b7684865941 100644
--- a/net/core/page_pool.c
+++ b/net/core/page_pool.c
@@ -276,8 +276,7 @@ static int page_pool_init(struct page_pool *pool,
 	/* Driver calling page_pool_create() also call page_pool_destroy() */
 	refcount_set(&pool->user_cnt, 1);
 
-	if (pool->dma_map)
-		get_device(pool->p.dev);
+	xa_init_flags(&pool->dma_mapped, XA_FLAGS_ALLOC1);
 
 	if (pool->slow.flags & PP_FLAG_ALLOW_UNREADABLE_NETMEM) {
 		netdev_assert_locked(pool->slow.netdev);
@@ -320,9 +319,7 @@ free_ptr_ring:
 static void page_pool_uninit(struct page_pool *pool)
 {
 	ptr_ring_cleanup(&pool->ring, NULL);
-
-	if (pool->dma_map)
-		put_device(pool->p.dev);
+	xa_destroy(&pool->dma_mapped);
 
 #ifdef CONFIG_PAGE_POOL_STATS
 	if (!pool->system)
@@ -463,13 +460,21 @@ page_pool_dma_sync_for_device(const struct page_pool *pool,
 			      netmem_ref netmem,
 			      u32 dma_sync_size)
 {
-	if (pool->dma_sync && dma_dev_need_sync(pool->p.dev))
-		__page_pool_dma_sync_for_device(pool, netmem, dma_sync_size);
+	if (pool->dma_sync && dma_dev_need_sync(pool->p.dev)) {
+		rcu_read_lock();
+		/* re-check under rcu_read_lock() to sync with page_pool_scrub() */
+		if (pool->dma_sync)
+			__page_pool_dma_sync_for_device(pool, netmem,
+							dma_sync_size);
+		rcu_read_unlock();
+	}
 }
 
-static bool page_pool_dma_map(struct page_pool *pool, netmem_ref netmem)
+static bool page_pool_dma_map(struct page_pool *pool, netmem_ref netmem, gfp_t gfp)
 {
 	dma_addr_t dma;
+	int err;
+	u32 id;
 
 	/* Setup DMA mapping: use 'struct page' area for storing DMA-addr
 	 * since dma_addr_t can be either 32 or 64 bits and does not always fit
@@ -483,15 +488,30 @@ static bool page_pool_dma_map(struct page_pool *pool, netmem_ref netmem)
 	if (dma_mapping_error(pool->p.dev, dma))
 		return false;
 
-	if (page_pool_set_dma_addr_netmem(netmem, dma))
+	if (page_pool_set_dma_addr_netmem(netmem, dma)) {
+		WARN_ONCE(1, "unexpected DMA address, please report to netdev@");
 		goto unmap_failed;
+	}
+
+	if (in_softirq())
+		err = xa_alloc(&pool->dma_mapped, &id, netmem_to_page(netmem),
+			       PP_DMA_INDEX_LIMIT, gfp);
+	else
+		err = xa_alloc_bh(&pool->dma_mapped, &id, netmem_to_page(netmem),
+				  PP_DMA_INDEX_LIMIT, gfp);
+	if (err) {
+		WARN_ONCE(err != -ENOMEM, "couldn't track DMA mapping, please report to netdev@");
+		goto unset_failed;
+	}
 
+	netmem_set_dma_index(netmem, id);
 	page_pool_dma_sync_for_device(pool, netmem, pool->p.max_len);
 
 	return true;
 
+unset_failed:
+	page_pool_set_dma_addr_netmem(netmem, 0);
 unmap_failed:
-	WARN_ONCE(1, "unexpected DMA address, please report to netdev@");
 	dma_unmap_page_attrs(pool->p.dev, dma,
 			     PAGE_SIZE << pool->p.order, pool->p.dma_dir,
 			     DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_WEAK_ORDERING);
@@ -508,7 +528,7 @@ static struct page *__page_pool_alloc_page_order(struct page_pool *pool,
 	if (unlikely(!page))
 		return NULL;
 
-	if (pool->dma_map && unlikely(!page_pool_dma_map(pool, page_to_netmem(page)))) {
+	if (pool->dma_map && unlikely(!page_pool_dma_map(pool, page_to_netmem(page), gfp))) {
 		put_page(page);
 		return NULL;
 	}
@@ -554,7 +574,7 @@ static noinline netmem_ref __page_pool_alloc_pages_slow(struct page_pool *pool,
 	 */
 	for (i = 0; i < nr_pages; i++) {
 		netmem = pool->alloc.cache[i];
-		if (dma_map && unlikely(!page_pool_dma_map(pool, netmem))) {
+		if (dma_map && unlikely(!page_pool_dma_map(pool, netmem, gfp))) {
 			put_page(netmem_to_page(netmem));
 			continue;
 		}
@@ -656,6 +676,8 @@ void page_pool_clear_pp_info(netmem_ref netmem)
 static __always_inline void __page_pool_release_page_dma(struct page_pool *pool,
 							 netmem_ref netmem)
 {
+	struct page *old, *page = netmem_to_page(netmem);
+	unsigned long id;
 	dma_addr_t dma;
 
 	if (!pool->dma_map)
@@ -664,6 +686,17 @@ static __always_inline void __page_pool_release_page_dma(struct page_pool *pool,
 		 */
 		return;
 
+	id = netmem_get_dma_index(netmem);
+	if (!id)
+		return;
+
+	if (in_softirq())
+		old = xa_cmpxchg(&pool->dma_mapped, id, page, NULL, 0);
+	else
+		old = xa_cmpxchg_bh(&pool->dma_mapped, id, page, NULL, 0);
+	if (old != page)
+		return;
+
 	dma = page_pool_get_dma_addr_netmem(netmem);
 
 	/* When page is unmapped, it cannot be returned to our pool */
@@ -671,6 +704,7 @@ static __always_inline void __page_pool_release_page_dma(struct page_pool *pool,
 			     PAGE_SIZE << pool->p.order, pool->p.dma_dir,
 			     DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_WEAK_ORDERING);
 	page_pool_set_dma_addr_netmem(netmem, 0);
+	netmem_set_dma_index(netmem, 0);
 }
 
 /* Disconnects a page (from a page_pool).  API users can have a need
@@ -1080,8 +1114,29 @@ static void page_pool_empty_alloc_cache_once(struct page_pool *pool)
 
 static void page_pool_scrub(struct page_pool *pool)
 {
+	unsigned long id;
+	void *ptr;
+
 	page_pool_empty_alloc_cache_once(pool);
-	pool->destroy_cnt++;
+	if (!pool->destroy_cnt++ && pool->dma_map) {
+		if (pool->dma_sync) {
+			/* Disable page_pool_dma_sync_for_device() */
+			pool->dma_sync = false;
+
+			/* Make sure all concurrent returns that may see the old
+			 * value of dma_sync (and thus perform a sync) have
+			 * finished before doing the unmapping below. Skip the
+			 * wait if the device doesn't actually need syncing, or
+			 * if there are no outstanding mapped pages.
+			 */
+			if (dma_dev_need_sync(pool->p.dev) &&
+			    !xa_empty(&pool->dma_mapped))
+				synchronize_net();
+		}
+
+		xa_for_each(&pool->dma_mapped, id, ptr)
+			__page_pool_release_page_dma(pool, page_to_netmem(ptr));
+	}
 
 	/* No more consumers should exist, but producers could still
 	 * be in-flight.
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 6cbf77bc61fc..74a2d886a35b 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -893,11 +893,6 @@ static void skb_clone_fraglist(struct sk_buff *skb)
 		skb_get(list);
 }
 
-static bool is_pp_netmem(netmem_ref netmem)
-{
-	return (netmem_get_pp_magic(netmem) & ~0x3UL) == PP_SIGNATURE;
-}
-
 int skb_pp_cow_data(struct page_pool *pool, struct sk_buff **pskb,
 		    unsigned int headroom)
 {
@@ -995,14 +990,7 @@ bool napi_pp_put_page(netmem_ref netmem)
 {
 	netmem = netmem_compound_head(netmem);
 
-	/* page->pp_magic is OR'ed with PP_SIGNATURE after the allocation
-	 * in order to preserve any existing bits, such as bit 0 for the
-	 * head page of compound page and bit 1 for pfmemalloc page, so
-	 * mask those bits for freeing side when doing below checking,
-	 * and page_is_pfmemalloc() is checked in __page_pool_put_page()
-	 * to avoid recycling the pfmemalloc page.
-	 */
-	if (unlikely(!is_pp_netmem(netmem)))
+	if (unlikely(!netmem_is_pp(netmem)))
 		return false;
 
 	page_pool_put_full_netmem(netmem_get_pp(netmem), netmem, false);
@@ -1042,7 +1030,7 @@ static int skb_pp_frag_ref(struct sk_buff *skb)
 
 	for (i = 0; i < shinfo->nr_frags; i++) {
 		head_netmem = netmem_compound_head(shinfo->frags[i].netmem);
-		if (likely(is_pp_netmem(head_netmem)))
+		if (likely(netmem_is_pp(head_netmem)))
 			page_pool_ref_netmem(head_netmem);
 		else
 			page_ref_inc(netmem_to_page(head_netmem));
diff --git a/net/core/xdp.c b/net/core/xdp.c
index 3cd0db9c9d2d..a014df88620c 100644
--- a/net/core/xdp.c
+++ b/net/core/xdp.c
@@ -438,8 +438,8 @@ void __xdp_return(netmem_ref netmem, enum xdp_mem_type mem_type,
 		netmem = netmem_compound_head(netmem);
 		if (napi_direct && xdp_return_frame_no_direct())
 			napi_direct = false;
-		/* No need to check ((page->pp_magic & ~0x3UL) == PP_SIGNATURE)
-		 * as mem->type knows this a page_pool page
+		/* No need to check netmem_is_pp() as mem->type knows this a
+		 * page_pool page
 		 */
 		page_pool_put_full_netmem(netmem_get_pp(netmem), netmem,
 					  napi_direct);