On Sat, Dec 30, 2023 at 11:58:29AM +0000, Greg Kroah-Hartman wrote:
6.6-stable review patch. If anyone has any objections, please let me know.
No objections from me.
For reference I do have an implementation of this functionality to v6.1 if/when it should be added. It is different as the bnxt_en driver did not use the page pool to manage DMA mapping until v6.6.
The minimally disruptive patch to prevent this memory leak is below:
From dc82f8b57e2692ec987628b53e6446ab9f4fa615 Mon Sep 17 00:00:00 2001 From: Andy Gospodarek andrew.gospodarek@broadcom.com Date: Thu, 7 Dec 2023 16:23:21 -0500 Subject: [PATCH] bnxt_en: unmap frag buffers before returning page to pool
If pages are not unmapped before calling page_pool_recycle_direct they will not be freed back to the pool. This will lead to a memory leak and messages like the following in dmesg:
[ 8229.436920] page_pool_release_retry() stalled pool shutdown 340 inflight 5437 sec
Fixes: a7559bc8c17c ("bnxt: support transmit and free of aggregation buffers") Signed-off-by: Andy Gospodarek andrew.gospodarek@broadcom.com --- drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c index aa56db138d6b..5b4548fad870 100644 --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c @@ -80,6 +80,7 @@ struct bnxt_sw_tx_bd *bnxt_xmit_bd(struct bnxt *bp, return NULL;
dma_unmap_addr_set(frag_tx_buf, mapping, frag_mapping); + dma_unmap_addr_set(frag_tx_buf, len, frag_len);
flags = frag_len << TX_BD_LEN_SHIFT; txbd->tx_bd_len_flags_type = cpu_to_le32(flags); @@ -154,8 +155,14 @@ void bnxt_tx_int_xdp(struct bnxt *bp, struct bnxt_napi *bnapi, int nr_pkts)
frags = tx_buf->nr_frags; for (j = 0; j < frags; j++) { + struct pci_dev *pdev = bp->pdev; + tx_cons = NEXT_TX(tx_cons); tx_buf = &txr->tx_buf_ring[tx_cons]; + dma_unmap_single(&pdev->dev, + dma_unmap_addr(tx_buf, mapping), + dma_unmap_len(tx_buf, len), + DMA_TO_DEVICE); page_pool_recycle_direct(rxr->page_pool, tx_buf->page); } }
From: Andy Gospodarek andrew.gospodarek@broadcom.com
[ Upstream commit 23c93c3b6275a59f2a685f4a693944b53c31df4e ]
Remove double-mapping of DMA buffers as it can prevent page pool entries from being freed. Mapping is managed by page pool infrastructure and was previously managed by the driver in __bnxt_alloc_rx_page before allowing the page pool infrastructure to manage it.
Fixes: 578fcfd26e2a ("bnxt_en: Let the page pool manage the DMA mapping") Reviewed-by: Somnath Kotur somnath.kotur@broadcom.com Signed-off-by: Andy Gospodarek andrew.gospodarek@broadcom.com Signed-off-by: Michael Chan michael.chan@broadcom.com Reviewed-by: David Wei dw@davidwei.uk Link: https://lore.kernel.org/r/20231214213138.98095-1-michael.chan@broadcom.com Signed-off-by: Jakub Kicinski kuba@kernel.org Signed-off-by: Sasha Levin sashal@kernel.org
drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c | 11 ++--------- 1 file changed, 2 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c index 96f5ca778c67d..8cb9a99154aad 100644 --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c @@ -59,7 +59,6 @@ struct bnxt_sw_tx_bd *bnxt_xmit_bd(struct bnxt *bp, for (i = 0; i < num_frags ; i++) { skb_frag_t *frag = &sinfo->frags[i]; struct bnxt_sw_tx_bd *frag_tx_buf;
struct pci_dev *pdev = bp->pdev;@@ -73,16 +72,10 @@ struct bnxt_sw_tx_bd *bnxt_xmit_bd(struct bnxt *bp, txbd = &txr->tx_desc_ring[TX_RING(prod)][TX_IDX(prod)]; frag_len = skb_frag_size(frag);
frag_mapping = skb_frag_dma_map(&pdev->dev, frag, 0,frag_len, DMA_TO_DEVICE);if (unlikely(dma_mapping_error(&pdev->dev, frag_mapping)))return NULL;dma_unmap_addr_set(frag_tx_buf, mapping, frag_mapping);- flags = frag_len << TX_BD_LEN_SHIFT; txbd->tx_bd_len_flags_type = cpu_to_le32(flags);
frag_mapping = page_pool_get_dma_addr(skb_frag_page(frag)) +skb_frag_off(frag);len = frag_len; -- 2.43.0