Re: irda weirdness

[Martin Diehl]

On Sun, 5 Oct 2003, Mikko Korhonen wrote:

> I'm getting
> Debug: sleeping function called from invalid context at 
> include/asm/uaccess.h:498

copy_from_user called from context where sleeping is not allowed. No idea 
where this is comming from - it's not in your calltrace.

> and Badness in local_bh_enable at kernel/softirq.c:119

local_bh_enable called with irqs_disabled.

Your calltrace suggests it's called from sirdev_hard_xmit, i.e. the final 
spin_unlock_bh - definitedly a bad idea because netdev->hard_start_xmit
is already called with bh disabled. However, I'm not sure is this is the 
whole story because I fail to see who might call us with interrupts off.

Could you try the patch below?

Martin

------------------------

--- linux-2.6.0-test6-bk5/drivers/net/irda/sir_dev.c	Sat Oct  4 11:45:44 2003
+++ v2.6.0-test6-bk5-md/drivers/net/irda/sir_dev.c	Sun Oct  5 17:20:13 2003
@@ -306,8 +306,8 @@ static int sirdev_hard_xmit(struct sk_bu
 		IRDA_DEBUG(3, "%s(), write not completed\n", __FUNCTION__);
 	}
 
-	/* serialize with write completion */
-	spin_lock_bh(&dev->tx_lock);
+	/* serialize with write completion - bh's already disabled */
+	spin_lock(&dev->tx_lock);
 
         /* Copy skb to tx_buff while wrapping, stuffing and making CRC */
 	dev->tx_buff.len = async_wrap_skb(skb, dev->tx_buff.data, dev->tx_buff.truesize); 
@@ -337,7 +337,7 @@ static int sirdev_hard_xmit(struct sk_bu
 		dev->stats.tx_dropped++;		      
 		netif_wake_queue(ndev);
 	}
-	spin_unlock_bh(&dev->tx_lock);
+	spin_unlock(&dev->tx_lock);
 
 	return 0;
 }

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/