[PATCH] net: airoha: Clean up RX queues in airoha_dev_stop

[PATCH] net: airoha: Clean up RX queues in airoha_dev_stop

[Wayen Yan]

When the last port is stopped, airoha_dev_stop() clears TX queues
but neglects to clean up RX queues. This can lead to:
- RX ring buffer descriptors remaining valid after device close
- Potential DMA synchronization issues on device reopen
- Risk of use-after-free if pages are freed while DMA is still active

Add cleanup loop for RX queues to mirror the TX queue cleanup,
ensuring symmetric resource management.

Fixes: 20bf7d07c956 ("net: airoha: add QDMA support for Airoha EN7581 Ethernet")
Signed-off-by: Wayen Yan <win847@gmail.com>
---
 drivers/net/ethernet/airoha/airoha_eth.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index 31cdb11cd7..9ca5bbf64d 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -1771,6 +1771,13 @@ static int airoha_dev_stop(struct net_device *dev)
 
 			airoha_qdma_cleanup_tx_queue(&qdma->q_tx[i]);
 		}
+
+		for (i = 0; i < ARRAY_SIZE(qdma->q_rx); i++) {
+			if (!qdma->q_rx[i].ndesc)
+				continue;
+
+			airoha_qdma_cleanup_rx_queue(&qdma->q_rx[i]);
+		}
 	}
 
 	return 0;
-- 
2.51.0

Re: [PATCH] net: airoha: Clean up RX queues in airoha_dev_stop

[Simon Horman]

On Tue, Jun 16, 2026 at 06:50:48PM +0800, Wayen Yan wrote:
> When the last port is stopped, airoha_dev_stop() clears TX queues
> but neglects to clean up RX queues. This can lead to:
> - RX ring buffer descriptors remaining valid after device close
> - Potential DMA synchronization issues on device reopen
> - Risk of use-after-free if pages are freed while DMA is still active
> 
> Add cleanup loop for RX queues to mirror the TX queue cleanup,
> ensuring symmetric resource management.
> 
> Fixes: 20bf7d07c956 ("net: airoha: add QDMA support for Airoha EN7581 Ethernet")
> Signed-off-by: Wayen Yan <win847@gmail.com>

Hi Wayen Yan,

There is AI-generated review of this patch-set available on both
https://sashiko.dev and https://netdev-ai.bots.linux.dev/sashiko/

I asked AI to summarise these concerns, it came up with the
following. I would appreciate it if you could look over this feedback.

1. NAPI Synchronization:
   While the TX path is managed by the netdev layer during stop, the RX path
   relies on the NAPI subsystem. Since NAPI remains active during
   `airoha_dev_stop()`, the new cleanup loop could race with the poller
   (`airoha_qdma_rx_process`). It would be safer to call `napi_disable()`
   before draining the queues to ensure exclusive access to the descriptors.

2. RX Queue Refill:
   Unlike TX, the RX hardware requires descriptors to be pre-allocated and
   posted by the driver to receive data. Because this patch empties the rings,
   `airoha_dev_open()` needs a corresponding update to refill them (e.g.,
   via `airoha_qdma_fill_rx_queue()`). Without this, the interface will
   encounter an empty ring on restart, leading to an RX stall.

3. SKB Accumulation:
   The cleanup should also account for the `q->skb` pointer used for
   fragmented packets. If a partial packet is sitting in the queue when the
   interface is stopped, freeing it and resetting the pointer to NULL will
   prevent a memory leak and ensure the next session starts with a clean
   state.