bnxt: make sure we return pages to the pool

[ Upstream commit 97f5e03 ]

Before the commit under Fixes the page would have been released
from the pool before the napi_alloc_skb() call, so normal page
freeing was fine (released page == no longer in the pool).

After the change we just mark the page for recycling so it's still
in the pool if the skb alloc fails, we need to recycle.

Same commit added the same bug in the new bnxt_rx_multi_page_skb().

Fixes: 1dc4c55 ("bnxt: adding bnxt_xdp_build_skb to build skb from multibuffer xdp_buff")
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Link: https://lore.kernel.org/r/20230111042547.987749-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drivers/net/ethernet/broadcom/bnxt/bnxt.c

@@ -990,7 +990,7 @@ static struct sk_buff *bnxt_rx_multi_page_skb(struct bnxt *bp,
 			     DMA_ATTR_WEAK_ORDERING);
 	skb = build_skb(page_address(page), PAGE_SIZE);
 	if (!skb) {
-		__free_page(page);
+		page_pool_recycle_direct(rxr->page_pool, page);
 		return NULL;
 	}
 	skb_mark_for_recycle(skb);
@@ -1028,7 +1028,7 @@ static struct sk_buff *bnxt_rx_page_skb(struct bnxt *bp,
 
 	skb = napi_alloc_skb(&rxr->bnapi->napi, payload);
 	if (!skb) {
-		__free_page(page);
+		page_pool_recycle_direct(rxr->page_pool, page);
 		return NULL;
 	}