[PATCH 2/2] w35und: plug memory leak in wbsoft_tx()

[PATCH 2/2] w35und: plug memory leak in wbsoft_tx()

[Pekka J Enberg]

From: Pekka Enberg <penberg@cs.helsinki.fi>

There's no reason to duplicate the skb in wbsoft_tx() and leak GFP_ATOMIC
memory as the contents are copied to ->TxBuffer in MdxTx() anyway before
MLMESendFrame() returns.

Cc: Pavel Machek <pavel@suse.cz>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
---
 drivers/staging/winbond/linux/wbusb.c |    7 ++-----
 1 files changed, 2 insertions(+), 5 deletions(-)

diff --git a/drivers/staging/winbond/linux/wbusb.c b/drivers/staging/winbond/linux/wbusb.c
index 128b70a..8957634 100644
--- a/drivers/staging/winbond/linux/wbusb.c
+++ b/drivers/staging/winbond/linux/wbusb.c
@@ -114,11 +114,8 @@ static void wbsoft_configure_filter(struct ieee80211_hw *dev,
 
 static int wbsoft_tx(struct ieee80211_hw *dev, struct sk_buff *skb)
 {
-	char *buffer = kmalloc(skb->len, GFP_ATOMIC);
-	printk("Sending frame %d bytes\n", skb->len);
-	memcpy(buffer, skb->data, skb->len);
-	if (1 == MLMESendFrame(my_adapter, buffer, skb->len, FRAME_TYPE_802_11_MANAGEMENT))
-		printk("frame sent ok (%d bytes)?\n", skb->len);
+	MLMESendFrame(my_adapter, skb->data, skb->len, FRAME_TYPE_802_11_MANAGEMENT);
+
 	return NETDEV_TX_OK;
 }
 
-- 
1.5.3.7

Re: [PATCH 2/2] w35und: plug memory leak in wbsoft_tx()

[Pavel Machek]

On Tue 2008-10-28 00:14:38, Pekka J Enberg wrote:
> From: Pekka Enberg <penberg@cs.helsinki.fi>
> 
> There's no reason to duplicate the skb in wbsoft_tx() and leak GFP_ATOMIC
> memory as the contents are copied to ->TxBuffer in MdxTx() anyway before
> MLMESendFrame() returns.

Acked-by: Pavel Machek <pavel@suse.cz>

...but I guess I should test the this after it hits the mainline...
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: [PATCH 2/2] w35und: plug memory leak in wbsoft_tx()

[Pekka Enberg]

On Tue 2008-10-28 00:14:38, Pekka J Enberg wrote:
>> From: Pekka Enberg <penberg@cs.helsinki.fi>
>>
>> There's no reason to duplicate the skb in wbsoft_tx() and leak GFP_ATOMIC
>> memory as the contents are copied to ->TxBuffer in MdxTx() anyway before
>> MLMESendFrame() returns.

On Tue, Oct 28, 2008 at 9:32 PM, Pavel Machek <pavel@suse.cz> wrote:
> Acked-by: Pavel Machek <pavel@suse.cz>
>
> ...but I guess I should test the this after it hits the mainline...

For sure. I'm bit surprised no one noticed this before as it seems to
be leaking memory for every transmitted frame...

Re: [PATCH 2/2] w35und: plug memory leak in wbsoft_tx()

[Pavel Machek]

On Tue 2008-10-28 23:08:50, Pekka Enberg wrote:
> On Tue 2008-10-28 00:14:38, Pekka J Enberg wrote:
> >> From: Pekka Enberg <penberg@cs.helsinki.fi>
> >>
> >> There's no reason to duplicate the skb in wbsoft_tx() and leak GFP_ATOMIC
> >> memory as the contents are copied to ->TxBuffer in MdxTx() anyway before
> >> MLMESendFrame() returns.
> 
> On Tue, Oct 28, 2008 at 9:32 PM, Pavel Machek <pavel@suse.cz> wrote:
> > Acked-by: Pavel Machek <pavel@suse.cz>
> >
> > ...but I guess I should test the this after it hits the mainline...
> 
> For sure. I'm bit surprised no one noticed this before as it seems to
> be leaking memory for every transmitted frame...

Well, I'm not quite surprised.  I noticed the driver stops working
after a while, but at that point it was a small miracle that it worked
at all...

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html