* [PATCH net v3] net: pch_gbe: handle TX skb allocation failure
@ 2026-06-15 12:50 Ruoyu Wang
2026-06-17 8:24 ` Simon Horman
0 siblings, 1 reply; 2+ messages in thread
From: Ruoyu Wang @ 2026-06-15 12:50 UTC (permalink / raw)
To: Andrew Lunn, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni
Cc: Simon Horman, Masayuki Ohtake, netdev, linux-kernel, Ruoyu Wang
pch_gbe_alloc_tx_buffers() allocates an skb for each TX descriptor and
then passes the returned pointer to skb_reserve(). If netdev_alloc_skb()
fails, skb_reserve() dereferences NULL.
Make pch_gbe_alloc_tx_buffers() return an error when an skb allocation
fails. On failure, let pch_gbe_alloc_tx_buffers() clean the partially
allocated TX ring before returning the error. While bringing the device
up, release the RX buffer pool through a shared cleanup helper before
unwinding the IRQ setup.
Fixes: 77555ee72282 ("net: Add Gigabit Ethernet driver of Topcliff PCH")
Signed-off-by: Ruoyu Wang <ruoyuw560@gmail.com>
---
Changes in v3:
- Move the partial TX ring cleanup into pch_gbe_alloc_tx_buffers(), as
suggested by Simon Horman.
Changes in v2:
- Add the kernel-doc return value description for
pch_gbe_alloc_tx_buffers().
.../ethernet/oki-semi/pch_gbe/pch_gbe_main.c | 38 ++++++++++++++-----
1 file changed, 29 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c b/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c
index e5a6f59..98091fb 100644
--- a/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c
+++ b/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c
@@ -1411,13 +1411,25 @@ pch_gbe_alloc_rx_buffers_pool(struct pch_gbe_adapter *adapter,
return 0;
}
+static void pch_gbe_free_rx_buffers_pool(struct pch_gbe_adapter *adapter,
+ struct pch_gbe_rx_ring *rx_ring)
+{
+ dma_free_coherent(&adapter->pdev->dev, rx_ring->rx_buff_pool_size,
+ rx_ring->rx_buff_pool, rx_ring->rx_buff_pool_logic);
+ rx_ring->rx_buff_pool_logic = 0;
+ rx_ring->rx_buff_pool_size = 0;
+ rx_ring->rx_buff_pool = NULL;
+}
+
/**
* pch_gbe_alloc_tx_buffers - Allocate transmit buffers
* @adapter: Board private structure
* @tx_ring: Tx descriptor ring
+ *
+ * Return: 0 on success, -ENOMEM if a TX skb allocation fails.
*/
-static void pch_gbe_alloc_tx_buffers(struct pch_gbe_adapter *adapter,
- struct pch_gbe_tx_ring *tx_ring)
+static int pch_gbe_alloc_tx_buffers(struct pch_gbe_adapter *adapter,
+ struct pch_gbe_tx_ring *tx_ring)
{
struct pch_gbe_buffer *buffer_info;
struct sk_buff *skb;
@@ -1431,12 +1443,17 @@ static void pch_gbe_alloc_tx_buffers(struct pch_gbe_adapter *adapter,
for (i = 0; i < tx_ring->count; i++) {
buffer_info = &tx_ring->buffer_info[i];
skb = netdev_alloc_skb(adapter->netdev, bufsz);
+ if (!skb) {
+ pch_gbe_clean_tx_ring(adapter, tx_ring);
+ return -ENOMEM;
+ }
skb_reserve(skb, PCH_GBE_DMA_ALIGN);
buffer_info->skb = skb;
tx_desc = PCH_GBE_TX_DESC(*tx_ring, i);
tx_desc->gbec_status = (DSC_INIT16);
}
- return;
+
+ return 0;
}
/**
@@ -1878,7 +1895,12 @@ int pch_gbe_up(struct pch_gbe_adapter *adapter)
"Error: can't bring device up - alloc rx buffers pool failed\n");
goto freeirq;
}
- pch_gbe_alloc_tx_buffers(adapter, tx_ring);
+ err = pch_gbe_alloc_tx_buffers(adapter, tx_ring);
+ if (err) {
+ netdev_err(netdev,
+ "Error: can't bring device up - alloc tx buffers failed\n");
+ goto freebuf;
+ }
pch_gbe_alloc_rx_buffers(adapter, rx_ring, rx_ring->count);
adapter->tx_queue_len = netdev->tx_queue_len;
pch_gbe_enable_dma_rx(&adapter->hw);
@@ -1892,6 +1914,8 @@ int pch_gbe_up(struct pch_gbe_adapter *adapter)
return 0;
+freebuf:
+ pch_gbe_free_rx_buffers_pool(adapter, rx_ring);
freeirq:
pch_gbe_free_irq(adapter);
out:
@@ -1927,11 +1951,7 @@ void pch_gbe_down(struct pch_gbe_adapter *adapter)
pch_gbe_clean_tx_ring(adapter, adapter->tx_ring);
pch_gbe_clean_rx_ring(adapter, adapter->rx_ring);
- dma_free_coherent(&adapter->pdev->dev, rx_ring->rx_buff_pool_size,
- rx_ring->rx_buff_pool, rx_ring->rx_buff_pool_logic);
- rx_ring->rx_buff_pool_logic = 0;
- rx_ring->rx_buff_pool_size = 0;
- rx_ring->rx_buff_pool = NULL;
+ pch_gbe_free_rx_buffers_pool(adapter, rx_ring);
}
/**
--
2.51.0
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [PATCH net v3] net: pch_gbe: handle TX skb allocation failure
2026-06-15 12:50 [PATCH net v3] net: pch_gbe: handle TX skb allocation failure Ruoyu Wang
@ 2026-06-17 8:24 ` Simon Horman
0 siblings, 0 replies; 2+ messages in thread
From: Simon Horman @ 2026-06-17 8:24 UTC (permalink / raw)
To: Ruoyu Wang
Cc: Andrew Lunn, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Masayuki Ohtake, netdev, linux-kernel
On Mon, Jun 15, 2026 at 08:50:42PM +0800, Ruoyu Wang wrote:
> pch_gbe_alloc_tx_buffers() allocates an skb for each TX descriptor and
> then passes the returned pointer to skb_reserve(). If netdev_alloc_skb()
> fails, skb_reserve() dereferences NULL.
>
> Make pch_gbe_alloc_tx_buffers() return an error when an skb allocation
> fails. On failure, let pch_gbe_alloc_tx_buffers() clean the partially
> allocated TX ring before returning the error. While bringing the device
> up, release the RX buffer pool through a shared cleanup helper before
> unwinding the IRQ setup.
>
> Fixes: 77555ee72282 ("net: Add Gigabit Ethernet driver of Topcliff PCH")
> Signed-off-by: Ruoyu Wang <ruoyuw560@gmail.com>
> ---
> Changes in v3:
> - Move the partial TX ring cleanup into pch_gbe_alloc_tx_buffers(), as
> suggested by Simon Horman.
>
> Changes in v2:
> - Add the kernel-doc return value description for
> pch_gbe_alloc_tx_buffers().
Thanks for the updates.
Reviewed-by: Simon Horman <horms@kernel.org>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2026-06-17 8:24 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-06-15 12:50 [PATCH net v3] net: pch_gbe: handle TX skb allocation failure Ruoyu Wang
2026-06-17 8:24 ` Simon Horman
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox