Skip to contentSkip to conversation

can: mcba_usb: fix possible double dev_kfree_skb in mcba_usb_start_xmit

PATCH In review Hangyu Hua 2 participants 4 messages Mar 11, 2022 – Mar 23, 2022

Subsystems (4)
lore
Share & tools
lore

Hangyu Hua

author
#0 lore raw

PATCH 1/1 can: mcba_usb: fix possible double dev_kfree_skb in mcba_usb_start_xmit

There is no need to call dev_kfree_skb when usb_submit_urb fails beacause can_put_echo_skb deletes original skb and can_free_echo_skb deletes the cloned skb.

Fixes 51f3baad7de9 ("can: mcba_usb: Add support for Microchip CAN BUS Analyzer")
---
 drivers/net/can/usb/mcba_usb.c | 1 -
 1 file changed, 1 deletion(-)
drivers/net/can/usb/mcba_usb.c +0 -1
index 77bddff86252..7c198eb5bc9c 100644
@@ -364,7 +364,6 @@ static netdev_tx_t mcba_usb_start_xmit(struct sk_buff *skb,
 xmit_failed:
 	can_free_echo_skb(priv->netdev, ctx->ndx, NULL);
 	mcba_usb_free_ctx(ctx);
-	dev_kfree_skb(skb);
 	stats->tx_dropped++;
 
 	return NETDEV_TX_OK;
Signature · 1 line
2.25.1
author
#1 lore raw

Gentle ping.

Quoted 22 lines · Hangyu Hua

There is no need to call dev_kfree_skb when usb_submit_urb fails beacause can_put_echo_skb deletes original skb and can_free_echo_skb deletes the cloned skb.

Fixes 51f3baad7de9 ("can: mcba_usb: Add support for Microchip CAN BUS Analyzer")
---
  drivers/net/can/usb/mcba_usb.c | 1 -
  1 file changed, 1 deletion(-)
drivers/net/can/usb/mcba_usb.c +0 -1
index 77bddff86252..7c198eb5bc9c 100644
@@ -364,7 +364,6 @@ static netdev_tx_t mcba_usb_start_xmit(struct sk_buff *skb,
  xmit_failed:
  	can_free_echo_skb(priv->netdev, ctx->ndx, NULL);
  	mcba_usb_free_ctx(ctx);
-	dev_kfree_skb(skb);
  	stats->tx_dropped++;
 
  	return NETDEV_TX_OK;
reviewer
#2 lore raw
In reply to Hangyu Hua

Hi Hangyu,

Quoted 4 lines · Hangyu Hua

There is no need to call dev_kfree_skb when usb_submit_urb fails beacause can_put_echo_skb deletes original skb and can_free_echo_skb deletes the cloned skb.

So, it's more like, "we don't need to call dev_kfree_skb() after can_put_echo_skb()
because can_put_echo_skb() consumes the given skb.". It seems it doesn't depend on the condition of usb_submit_urb(). Plus, we don't see the "cloned skb" at the
call site.

Would you mind adding a comment on can_put_echo_skb(), in a separate patch, saying the fact that it consumes the skb?

ems_usb.c, gs_usb.c and possibly some others seem to call dev_kfree_skb() as well.
Are they affected?

Best,

Signature · 1 line
           yashi
author
#3 lore raw

Hi yashi,

You are right. There are a series of the same problems about can_put_echo_skb.

This issue was discovered while I was discussing a incorrect patch with Marc. You can check this in

https://lore.kernel.org/all/[email protected]/

So i submitted a new patch. This was the first place where the problem occurs. You can check this in

[1] https://lore.kernel.org/all/[email protected]/

After a week, I realized it could be a series of problems. So i submitted the following patches

[2] https://lore.kernel.org/all/[email protected]/
[3] https://lore.kernel.org/all/[email protected]/
[4] https://lore.kernel.org/all/[email protected]/

I think this is all affected. None of the four have been merged upstream. Do i need to remake all these four patches?

Thanks,
Hangyu

Quoted 23 lines · Yasushi SHOJI

Hi Hangyu,

Quoted 4 lines · Hangyu Hua

There is no need to call dev_kfree_skb when usb_submit_urb fails beacause can_put_echo_skb deletes original skb and can_free_echo_skb deletes the cloned skb.

So, it's more like, "we don't need to call dev_kfree_skb() after can_put_echo_skb()
because can_put_echo_skb() consumes the given skb.". It seems it doesn't depend on the condition of usb_submit_urb(). Plus, we don't see the "cloned skb" at the
call site.

Would you mind adding a comment on can_put_echo_skb(), in a separate patch, saying the fact that it consumes the skb?

ems_usb.c, gs_usb.c and possibly some others seem to call dev_kfree_skb() as well.
Are they affected?

Best,