On Fri, 2012-12-21 at 17:57 -0800, Tejun Heo wrote:
> i2400m_net_wake_tx() sets ->wake_tx_skb with the given skb if
> ->wake_tx_ws is not pending; however, i2400m_wake_tx_work() could have
> just started execution and haven't fetched -><wake_tx_skb yet. The
> previous packet will be leaked.
>
> Update ->wake_tx_skb handling.
>
> * i2400m_net_wake_tx() now tests whether the previous ->wake_tx_skb
> has been consumed by ->wake_tx_ws instead of testing work_pending().
>
> * i2400m_net_wake_stop() is simplified similarly. It always puts
> ->wake_tx_skb if non-NULL.
>
> * Spurious ->wake_tx_skb dereference outside critical section dropped
> from i2400m_wake_tx_work().
>
> Only compile tested.
>
> Signed-off-by: Tejun Heo <tj at kernel.org>
Acked-by: Dan Williams <dcbw at redhat.com>
No regressions in a quick connection check to Clear and browsing a bunch
of pages.
Dan
> Cc: Inaky Perez-Gonzalez <inaky.perez-gonzalez at intel.com>
> Cc: linux-wimax at intel.com
> Cc: wimax at linuxwimax.org
> ---
> Please let me know how this patch should be routed. I can take it
> through the workqueue tree if necessary.
>
> Thanks.
>
> drivers/net/wimax/i2400m/netdev.c | 31 +++++++++++++++++--------------
> 1 file changed, 17 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/net/wimax/i2400m/netdev.c b/drivers/net/wimax/i2400m/netdev.c
> index 1d76ae8..530581c 100644
> --- a/drivers/net/wimax/i2400m/netdev.c
> +++ b/drivers/net/wimax/i2400m/netdev.c
> @@ -156,7 +156,7 @@ void i2400m_wake_tx_work(struct work_struct *ws)
> struct i2400m *i2400m = container_of(ws, struct i2400m, wake_tx_ws);
> struct net_device *net_dev = i2400m->wimax_dev.net_dev;
> struct device *dev = i2400m_dev(i2400m);
> - struct sk_buff *skb = i2400m->wake_tx_skb;
> + struct sk_buff *skb;
> unsigned long flags;
>
> spin_lock_irqsave(&i2400m->tx_lock, flags);
> @@ -236,23 +236,26 @@ void i2400m_tx_prep_header(struct sk_buff *skb)
> void i2400m_net_wake_stop(struct i2400m *i2400m)
> {
> struct device *dev = i2400m_dev(i2400m);
> + struct sk_buff *wake_tx_skb;
> + unsigned long flags;
>
> d_fnstart(3, dev, "(i2400m %p)\n", i2400m);
> - /* See i2400m_hard_start_xmit(), references are taken there
> - * and here we release them if the work was still
> - * pending. Note we can't differentiate work not pending vs
> - * never scheduled, so the NULL check does that. */
> - if (cancel_work_sync(&i2400m->wake_tx_ws) == 0
> - && i2400m->wake_tx_skb != NULL) {
> - unsigned long flags;
> - struct sk_buff *wake_tx_skb;
> - spin_lock_irqsave(&i2400m->tx_lock, flags);
> - wake_tx_skb = i2400m->wake_tx_skb; /* compat help */
> - i2400m->wake_tx_skb = NULL; /* compat help */
> - spin_unlock_irqrestore(&i2400m->tx_lock, flags);
> + /*
> + * See i2400m_hard_start_xmit(), references are taken there and
> + * here we release them if the packet was still pending.
> + */
> + cancel_work_sync(&i2400m->wake_tx_ws);
> +
> + spin_lock_irqsave(&i2400m->tx_lock, flags);
> + wake_tx_skb = i2400m->wake_tx_skb;
> + i2400m->wake_tx_skb = NULL;
> + spin_unlock_irqrestore(&i2400m->tx_lock, flags);
> +
> + if (wake_tx_skb) {
> i2400m_put(i2400m);
> kfree_skb(wake_tx_skb);
> }
> +
> d_fnend(3, dev, "(i2400m %p) = void\n", i2400m);
> }
>
> @@ -288,7 +291,7 @@ int i2400m_net_wake_tx(struct i2400m *i2400m, struct net_device *net_dev,
> * and if pending, release those resources. */
> result = 0;
> spin_lock_irqsave(&i2400m->tx_lock, flags);
> - if (!work_pending(&i2400m->wake_tx_ws)) {
> + if (!i2400m->wake_tx_skb) {
> netif_stop_queue(net_dev);
> i2400m_get(i2400m);
> i2400m->wake_tx_skb = skb_get(skb); /* transfer ref count */
