[PATCH 5.17 1114/1126] xsk: Do not write NULL in SW ring at allocation failure

[Greg Kroah-Hartman]

From: Magnus Karlsson <magnus.karlsson@xxxxxxxxx>

commit a95a4d9b39b0324402569ed7395aae59b8fd2b11 upstream.

For the case when xp_alloc_batch() is used but the batched allocation
cannot be used, there is a slow path that uses the non-batched
xp_alloc(). When it fails to allocate an entry, it returns NULL. The
current code wrote this NULL into the entry of the provided results
array (pointer to the driver SW ring usually) and returned. This might
not be what the driver expects and to make things simpler, just write
successfully allocated xdp_buffs into the SW ring,. The driver might
have information in there that is still important after an allocation
failure.

Note that at this point in time, there are no drivers using
xp_alloc_batch() that could trigger this slow path. But one might get
added.

Fixes: 47e4075df300 ("xsk: Batched buffer allocation for the pool")
Signed-off-by: Magnus Karlsson <magnus.karlsson@xxxxxxxxx>
Signed-off-by: Alexei Starovoitov <ast@xxxxxxxxxx>
Link: https://lore.kernel.org/bpf/20220328142123.170157-2-maciej.fijalkowski@xxxxxxxxx
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
 net/xdp/xsk_buff_pool.c |    8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

--- a/net/xdp/xsk_buff_pool.c
+++ b/net/xdp/xsk_buff_pool.c
@@ -584,9 +584,13 @@ u32 xp_alloc_batch(struct xsk_buff_pool
 	u32 nb_entries1 = 0, nb_entries2;
 
 	if (unlikely(pool->dma_need_sync)) {
+		struct xdp_buff *buff;
+
 		/* Slow path */
-		*xdp = xp_alloc(pool);
-		return !!*xdp;
+		buff = xp_alloc(pool);
+		if (buff)
+			*xdp = buff;
+		return !!buff;
 	}
 
 	if (unlikely(pool->free_list_cnt)) {