fix SMP lockup in eepro100 with ethtool on unused interface
PATCH 1/1 fix SMP lockup in eepro100 with ethtool on unused interface
When support for the GSET and SSET ethtool ioctls was added to eepro100.c in 2.4.20, the tx lock was overloaded to serialize their use. Unfortunately, this lock is only initialized in dev->open(), causing ethtool to deadlock the machine when used on an unconfigured eepro100 interface.
The fix is to initialize the spinlock at probe time. Jeff, the patch below is against 2.4.21-pre5, but I'm assuming you'll integrate it into 2.5 and push it to Linus as well as Marcelo. It applies to both trees.
Jason
drivers/net/eepro100.c +1 -1
@@ -842,6 +842,7 @@ sp->lstats = (struct speedo_stats *)(sp->tx_ring +TX_RING_SIZE ); sp->lstats_dma =TX_RING_ELEM_DMA (sp,TX_RING_SIZE );init_timer (&sp->timer);/* used in ioctl() */ +spin_lock_init (&sp->lock); sp->mii_if.full_duplex = option >=0 && (option &0x10 ) ?1 :0 ;if (card_idx >=0 ) {
@@ -993,7 +994,6 @@ sp->dirty_tx =0 ; sp->last_cmd =0 ; sp->tx_full =0 ; -spin_lock_init (&sp->lock); sp->in_interrupt =0 ;/* .. we can safely take handler calls during init. */
PATCH 1/1 fix SMP deadlock (uninitialized spinlock)
When support for the ETHTOOL_GSET and ETHTOOL_SSET ethtool methods was added to eepro100.c in 2.4.20, the tx lock was overloaded to serialize their use. Unfortunately, this lock is only initialized in dev->open(), causing ethtool to deadlock the machine when used on a never-opened eepro100 interface.
The fix is to initialize the spinlock at probe time. The patch below is against 2.4.21-pre5, but also applies to 2.5.
please apply,
Jason
drivers/net/eepro100.c +1 -1
@@ -842,6 +842,7 @@ sp->lstats = (struct speedo_stats *)(sp->tx_ring +TX_RING_SIZE ); sp->lstats_dma =TX_RING_ELEM_DMA (sp,TX_RING_SIZE );init_timer (&sp->timer);/* used in ioctl() */ +spin_lock_init (&sp->lock); sp->mii_if.full_duplex = option >=0 && (option &0x10 ) ?1 :0 ;if (card_idx >=0 ) {
@@ -993,7 +994,6 @@ sp->dirty_tx =0 ; sp->last_cmd =0 ; sp->tx_full =0 ; -spin_lock_init (&sp->lock); sp->in_interrupt =0 ;/* .. we can safely take handler calls during init. */