Hi,
I'm using smc9194.c driver with my Laptop port replicator for a while.
I managed to fix Warning: kfree_sdb on hard IRQ that the driver sent to me.
The patch works great for me and I think very few persons use this driver.
For information, it has been market as OBSOLETE for a while and is only
recently back (partlty to my request) to 2.4.testxx kernel.
PLEASE ADD THIS TO THE NEXT test release.
*** smc9194-old.c Wed Dec 13 09:39:18 2000
--- smc9194.c Wed Dec 13 12:44:38 2000
***************
*** 50,55 ****
--- 50,56 ----
. 04/14/00 Heiko Pruessing (SMA Regelsysteme) Fixed bug in chip memory
. allocation
. 08/20/00 Arnaldo Melo fix kfree(skb) in
smc_hardware_send_packet
+ . 12/13/00 Christian Jullien fix Warning: kfree_skb on hard IRQ
-------------------------------------------------------------------------
---*/
static const char *version =
***************
*** 99,104 ****
--- 100,107 ----
0x300, 0x320, 0x340, 0x360, 0x380, 0x3A0, 0x3C0, 0x3E0, 0
};
+ #define SMC_dev_kfree_skb(ptr)
(in_irq()?dev_kfree_skb_irq(ptr):dev_kfree_skb(ptr))
+
/*
. Wait time for memory to be free. This probably shouldn't be
. tuned that much, as waiting for this means nothing else happens
***************
*** 525,531 ****
printk(CARDNAME": Far too big packet error. \n");
/* freeing the packet is a good thing here... but should
. any packets of this size get down here? */
! dev_kfree_skb (skb);
lp->saved_skb = NULL;
/* this IS an error, but, i don't want the skb saved */
netif_wake_queue(dev);
--- 528,534 ----
printk(CARDNAME": Far too big packet error. \n");
/* freeing the packet is a good thing here... but should
. any packets of this size get down here? */
! SMC_dev_kfree_skb (skb);
lp->saved_skb = NULL;
/* this IS an error, but, i don't want the skb saved */
netif_wake_queue(dev);
***************
*** 615,621 ****
if ( packet_no & 0x80 ) {
/* or isn't there? BAD CHIP! */
printk(KERN_DEBUG CARDNAME": Memory allocation failed. \n");
! dev_kfree_skb(skb);
lp->saved_skb = NULL;
netif_wake_queue(dev);
return;
--- 618,624 ----
if ( packet_no & 0x80 ) {
/* or isn't there? BAD CHIP! */
printk(KERN_DEBUG CARDNAME": Memory allocation failed. \n");
! SMC_dev_kfree_skb(skb);
lp->saved_skb = NULL;
netif_wake_queue(dev);
return;
***************
*** 678,684 ****
PRINTK2((CARDNAME": Sent packet of length %d \n",length));
lp->saved_skb = NULL;
! dev_kfree_skb (skb);
dev->trans_start = jiffies;
--- 681,687 ----
PRINTK2((CARDNAME": Sent packet of length %d \n",length));
lp->saved_skb = NULL;
! SMC_dev_kfree_skb (skb);
dev->trans_start = jiffies;
Regards
Christian Jullien
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
