Re: [PATCH] uml: fix net_kern workqueue abuse

[Peter Zijlstra <a.p.zijlstra@chello.nl>]

Garh, it seems I forgot to finish the subject.

On Fri, 2006-12-08 at 11:20 +0100, Peter Zijlstra wrote:
> fixup the work on stack and exit scope trouble by placing the work_struct in
> the uml_net_private data.
> 
> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
> ---
>  arch/um/drivers/net_kern.c |   15 +++++++++------
>  arch/um/include/net_kern.h |    2 ++
>  2 files changed, 11 insertions(+), 6 deletions(-)
> 
> Index: linux-2.6-git/arch/um/drivers/net_kern.c
> ===================================================================
> --- linux-2.6-git.orig/arch/um/drivers/net_kern.c	2006-12-08 10:32:56.000000000 +0100
> +++ linux-2.6-git/arch/um/drivers/net_kern.c	2006-12-08 11:14:28.000000000 +0100
> @@ -72,9 +72,11 @@ static int uml_net_rx(struct net_device 
>  	return pkt_len;
>  }
>  
> -static void uml_dev_close(void* dev)
> +static void uml_dev_close(struct work_struct *work)
>  {
> -	dev_close( (struct net_device *) dev);
> +	struct uml_net_private *lp =
> +		container_of(work, struct uml_net_private, work);
> +	dev_close(lp->dev);
>  }
>  
>  irqreturn_t uml_net_interrupt(int irq, void *dev_id)
> @@ -89,7 +91,6 @@ irqreturn_t uml_net_interrupt(int irq, v
>  	spin_lock(&lp->lock);
>  	while((err = uml_net_rx(dev)) > 0) ;
>  	if(err < 0) {
> -		DECLARE_WORK(close_work, uml_dev_close, dev);
>  		printk(KERN_ERR 
>  		       "Device '%s' read returned %d, shutting it down\n", 
>  		       dev->name, err);
> @@ -97,9 +98,10 @@ irqreturn_t uml_net_interrupt(int irq, v
>  		 * again lp->lock.
>  		 * And dev_close() can be safely called multiple times on the
>  		 * same device, since it tests for (dev->flags & IFF_UP). So
> -		 * there's no harm in delaying the device shutdown. */
> -		schedule_work(&close_work);
> -#error this is not permitted - close_work will go out of scope
> +		 * there's no harm in delaying the device shutdown.
> +		 * Furthermore, the workqueue will not re-enqueue an already
> +		 * enqueued work item. */
> +		schedule_work(&lp->work);
>  		goto out;
>  	}
>  	reactivate_fd(lp->fd, UM_ETH_IRQ);
> @@ -366,6 +368,7 @@ static int eth_configure(int n, void *in
>  	/* This points to the transport private data. It's still clear, but we
>  	 * must memset it to 0 *now*. Let's help the drivers. */
>  	memset(lp, 0, size);
> +	INIT_WORK(&lp->work, uml_dev_close);
>  
>  	/* sysfs register */
>  	if (!driver_registered) {
> Index: linux-2.6-git/arch/um/include/net_kern.h
> ===================================================================
> --- linux-2.6-git.orig/arch/um/include/net_kern.h	2006-12-08 10:55:25.000000000 +0100
> +++ linux-2.6-git/arch/um/include/net_kern.h	2006-12-08 11:00:26.000000000 +0100
> @@ -11,6 +11,7 @@
>  #include <linux/skbuff.h>
>  #include <linux/socket.h>
>  #include <linux/list.h>
> +#include <linux/workqueue.h>
>  
>  struct uml_net {
>  	struct list_head list;
> @@ -26,6 +27,7 @@ struct uml_net_private {
>  	struct net_device *dev;
>  	struct timer_list tl;
>  	struct net_device_stats stats;
> +	struct work_struct work;
>  	int fd;
>  	unsigned char mac[ETH_ALEN];
>  	unsigned short (*protocol)(struct sk_buff *);
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/