net/mlx5e: RX, Fix page_pool page fragment tracking for XDP

Currently mlx5e releases pages directly to the page_pool for XDP_TX and
does page fragment counting for XDP_REDIRECT. RX pages from the
page_pool are leaking on XDP_REDIRECT because the xdp core will release
only one fragment out of MLX5E_PAGECNT_BIAS_MAX and subsequently the page
is marked as "skip release" which avoids the driver release.

A fix would be to take an extra fragment for XDP_REDIRECT and not set the
"skip release" bit so that the release on the driver side can handle the
remaining bias fragments. But this would be a shortsighted solution.
Instead, this patch converges the two XDP paths (XDP_TX and XDP_REDIRECT) to
always do fragment tracking. The "skip release" bit is no longer
necessary for XDP.

Fixes: 6f574284 ("net/mlx5e: RX, Enable skb page recycling through the page_pool")
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>

drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c

@@ -662,8 +662,7 @@ static void mlx5e_free_xdpsq_desc(struct mlx5e_xdpsq *sq,
 				/* No need to check ((page->pp_magic & ~0x3UL) == PP_SIGNATURE)
 				 * as we know this is a page_pool page.
 				 */
-				page_pool_put_defragged_page(page->pp,
-							     page, -1, true);
+				page_pool_recycle_direct(page->pp, page);
 			} while (++n < num);
 
 			break;

drivers/net/ethernet/mellanox/mlx5/core/en_rx.c

@@ -1751,11 +1751,11 @@ mlx5e_skb_from_cqe_nonlinear(struct mlx5e_rq *rq, struct mlx5e_wqe_frag_info *wi
 
 	prog = rcu_dereference(rq->xdp_prog);
 	if (prog && mlx5e_xdp_handle(rq, prog, &mxbuf)) {
-		if (test_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)) {
+		if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)) {
 			struct mlx5e_wqe_frag_info *pwi;
 
 			for (pwi = head_wi; pwi < wi; pwi++)
-				pwi->flags |= BIT(MLX5E_WQE_FRAG_SKIP_RELEASE);
+				pwi->frag_page->frags++;
 		}
 		return NULL; /* page/packet was consumed by XDP */
 	}
@@ -1825,12 +1825,8 @@ static void mlx5e_handle_rx_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
 			      rq, wi, cqe, cqe_bcnt);
 	if (!skb) {
 		/* probably for XDP */
-		if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)) {
-			/* do not return page to cache,
-			 * it will be returned on XDP_TX completion.
-			 */
-			wi->flags |= BIT(MLX5E_WQE_FRAG_SKIP_RELEASE);
-		}
+		if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags))
+			wi->frag_page->frags++;
 		goto wq_cyc_pop;
 	}
 
@@ -1876,12 +1872,8 @@ static void mlx5e_handle_rx_cqe_rep(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
 			      rq, wi, cqe, cqe_bcnt);
 	if (!skb) {
 		/* probably for XDP */
-		if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)) {
-			/* do not return page to cache,
-			 * it will be returned on XDP_TX completion.
-			 */
-			wi->flags |= BIT(MLX5E_WQE_FRAG_SKIP_RELEASE);
-		}
+		if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags))
+			wi->frag_page->frags++;
 		goto wq_cyc_pop;
 	}
 
@@ -2060,12 +2052,12 @@ mlx5e_skb_from_cqe_mpwrq_nonlinear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *w
 	if (prog) {
 		if (mlx5e_xdp_handle(rq, prog, &mxbuf)) {
 			if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)) {
-				int i;
+				struct mlx5e_frag_page *pfp;
+
+				for (pfp = head_page; pfp < frag_page; pfp++)
+					pfp->frags++;
 
-				for (i = 0; i < sinfo->nr_frags; i++)
-					/* non-atomic */
-					__set_bit(page_idx + i, wi->skip_release_bitmap);
-				return NULL;
+				wi->linear_page.frags++;
 			}
 			mlx5e_page_release_fragmented(rq, &wi->linear_page);
 			return NULL; /* page/packet was consumed by XDP */
@@ -2163,7 +2155,7 @@ mlx5e_skb_from_cqe_mpwrq_linear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi,
 				 cqe_bcnt, &mxbuf);
 		if (mlx5e_xdp_handle(rq, prog, &mxbuf)) {
 			if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags))
-				__set_bit(page_idx, wi->skip_release_bitmap); /* non-atomic */
+				frag_page->frags++;
 			return NULL; /* page/packet was consumed by XDP */
 		}