Hi Marc,
On Thu, 2024-03-07 at 13:23 +0100, Marc Kleine-Budde wrote:
Sorry, resend. I fsck up the linux-can ML's address.
Thanks for that.
On 07.03.2024 12:04:42, Vitor Soares wrote:
From: Vitor Soares vitor.soares@toradex.com
When the mcp251xfd_start_xmit() function fails, the driver stops processing messages, and the interrupt routine does not return, running indefinitely even after killing the running application.
Error messages: [ 441.298819] mcp251xfd spi2.0 can0: ERROR in mcp251xfd_start_xmit: -16 [ 441.306498] mcp251xfd spi2.0 can0: Transmit Event FIFO buffer not empty. (seq=0x000017c7, tef_tail=0x000017cf, tef_head=0x000017d0, tx_head=0x000017d3). ... and repeat forever.
The issue can be triggered when multiple devices share the same SPI interface. And there is concurrent access to the bus.
The problem occurs because tx_ring->head increments even if mcp251xfd_start_xmit() fails. Consequently, the driver skips one TX package while still expecting a response in mcp251xfd_handle_tefif_one().
This patch resolves the issue by decreasing tx_ring->head if mcp251xfd_start_xmit() fails. With the fix, if we attempt to trigger the issue again, the driver prints an error and discard the message.
What about returning NETDEV_TX_BUSY, then the networking stack will retry.
Do you mean when err == -EBUSY?
Fixes: 55e5b97f003e ("can: mcp25xxfd: add driver for Microchip MCP25xxFD SPI CAN") Cc: stable@vger.kernel.org Signed-off-by: Vitor Soares vitor.soares@toradex.com
drivers/net/can/spi/mcp251xfd/mcp251xfd-tx.c | 27 ++++++++++------
1 file changed, 14 insertions(+), 13 deletions(-)
diff --git a/drivers/net/can/spi/mcp251xfd/mcp251xfd-tx.c b/drivers/net/can/spi/mcp251xfd/mcp251xfd-tx.c index 160528d3cc26..a8eb941c1b95 100644 --- a/drivers/net/can/spi/mcp251xfd/mcp251xfd-tx.c +++ b/drivers/net/can/spi/mcp251xfd/mcp251xfd-tx.c @@ -181,25 +181,26 @@ netdev_tx_t mcp251xfd_start_xmit(struct sk_buff *skb, tx_obj = mcp251xfd_get_tx_obj_next(tx_ring); mcp251xfd_tx_obj_from_skb(priv, tx_obj, skb, tx_ring-
head);
- /* Stop queue if we occupy the complete TX FIFO */ tx_head = mcp251xfd_get_tx_head(tx_ring); - tx_ring->head++; - if (mcp251xfd_get_tx_free(tx_ring) == 0) - netif_stop_queue(ndev);
frame_len = can_skb_get_frame_len(skb); - err = can_put_echo_skb(skb, ndev, tx_head, frame_len); - if (!err) - netdev_sent_queue(priv->ndev, frame_len); + can_put_echo_skb(skb, ndev, tx_head, frame_len);
+ tx_ring->head++; err = mcp251xfd_tx_obj_write(priv, tx_obj); - if (err) - goto out_err; + if (err) { + can_free_echo_skb(ndev, tx_head, NULL); - return NETDEV_TX_OK; + tx_ring->head--;
I'm testing returning here when err = -EBUSY, but can_put_echo_skb() should be invoked after mcp251xfd_tx_obj_write(). Otherwise, I get a Kernel NULL pointer dereference error.
I'm not sure, if we want an error message for -EBUSY. We could add proper ethtool statistics.
+ netdev_err(priv->ndev, "ERROR in %s: %d\n", __func__, err); + } else { + /* Stop queue if we occupy the complete TX FIFO */ + if (mcp251xfd_get_tx_free(tx_ring) == 0) + netif_stop_queue(ndev);
- out_err:
- netdev_err(priv->ndev, "ERROR in %s: %d\n", __func__, err); + netdev_sent_queue(priv->ndev, frame_len); + } return NETDEV_TX_OK; }
regards, Marc
Best regards, Vitor Soares