Running with kmemleak shows that this driver leaks every skb allocated in
routine htc_connect_service() whether the service connection succeeds or
fails.
Signed-off-by: Larry Finger <[email protected]>
Cc: Stable <[email protected]>
Cc: "Luis R. Rodriguez" <[email protected]>
Cc: Vasanthakumar Thiagarajan <[email protected]>
Cc: Senthil Balasubramanian <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
---
drivers/net/wireless/ath/ath9k/htc_hst.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/net/wireless/ath/ath9k/htc_hst.c b/drivers/net/wireless/ath/ath9k/htc_hst.c
index 4a9570d..a304748 100644
--- a/drivers/net/wireless/ath/ath9k/htc_hst.c
+++ b/drivers/net/wireless/ath/ath9k/htc_hst.c
@@ -278,10 +278,12 @@ int htc_connect_service(struct htc_target *target,
if (!time_left) {
dev_err(target->dev, "Service connection timeout for: %d\n",
service_connreq->service_id);
- return -ETIMEDOUT;
+ ret = -ETIMEDOUT;
+ goto err;
}
*conn_rsp_epid = target->conn_rsp_epid;
+ kfree_skb(skb);
return 0;
err:
kfree_skb(skb);
--
1.7.10.4
Larry Finger wrote:
> My only counter argument is that none of the other paths that get to
> ath9k_htc_txcompletion_cb() leak the skb. It only happens for the path that goes
> through htc_connect_service().
Sure, but the TX completion handler would be invoked for every skb that is passed
to the USB layer, including the ones that are allocated in htc_hst.c. When this
skb goes down to hif_usb.c, a URB is allocated and I am not sure how freeing
the skb before the USB layer invokes the completion handler for the URB is correct.
I'll come up with a patch and see if kmemleak still complains.
Sujith
On 01/01/2013 10:30 PM, Sujith Manoharan wrote:
> Larry Finger wrote:
>> diff --git a/drivers/net/wireless/ath/ath9k/htc_hst.c b/drivers/net/wireless/ath/ath9k/htc_hst.c
>> index 4a9570d..a304748 100644
>> --- a/drivers/net/wireless/ath/ath9k/htc_hst.c
>> +++ b/drivers/net/wireless/ath/ath9k/htc_hst.c
>> @@ -278,10 +278,12 @@ int htc_connect_service(struct htc_target *target,
>> if (!time_left) {
>> dev_err(target->dev, "Service connection timeout for: %d\n",
>> service_connreq->service_id);
>> - return -ETIMEDOUT;
>> + ret = -ETIMEDOUT;
>> + goto err;
>> }
>>
>> *conn_rsp_epid = target->conn_rsp_epid;
>> + kfree_skb(skb);
>> return 0;
>> err:
>> kfree_skb(skb);
> ath9k_htc_txcompletion_cb()
> The allocated skb should be freed in the TX completion hander,
> ath9k_htc_txcompletion_cb() - doing it in htc_connect_service() is wrong, I think.
My only counter argument is that none of the other paths that get to
ath9k_htc_txcompletion_cb() leak the skb. It only happens for the path that goes
through htc_connect_service().
Larry
Larry Finger wrote:
> diff --git a/drivers/net/wireless/ath/ath9k/htc_hst.c b/drivers/net/wireless/ath/ath9k/htc_hst.c
> index 4a9570d..a304748 100644
> --- a/drivers/net/wireless/ath/ath9k/htc_hst.c
> +++ b/drivers/net/wireless/ath/ath9k/htc_hst.c
> @@ -278,10 +278,12 @@ int htc_connect_service(struct htc_target *target,
> if (!time_left) {
> dev_err(target->dev, "Service connection timeout for: %d\n",
> service_connreq->service_id);
> - return -ETIMEDOUT;
> + ret = -ETIMEDOUT;
> + goto err;
> }
>
> *conn_rsp_epid = target->conn_rsp_epid;
> + kfree_skb(skb);
> return 0;
> err:
> kfree_skb(skb);
The allocated skb should be freed in the TX completion hander,
ath9k_htc_txcompletion_cb() - doing it in htc_connect_service() is wrong, I think.
Sujith
Larry Finger wrote:
> > I'll come up with a patch and see if kmemleak still complains.
>
> Did I miss your posting on this issue?
Nope, I haven't got to this yet, sorry. I'll try to send a patch by today.
Sujith
On 01/01/2013 11:26 PM, Sujith Manoharan wrote:
> Larry Finger wrote:
>> My only counter argument is that none of the other paths that get to
>> ath9k_htc_txcompletion_cb() leak the skb. It only happens for the path that goes
>> through htc_connect_service().
>
> Sure, but the TX completion handler would be invoked for every skb that is passed
> to the USB layer, including the ones that are allocated in htc_hst.c. When this
> skb goes down to hif_usb.c, a URB is allocated and I am not sure how freeing
> the skb before the USB layer invokes the completion handler for the URB is correct.
>
> I'll come up with a patch and see if kmemleak still complains.
Did I miss your posting on this issue?
Larry