The patch titled
tulip: fix shutdown DMA/irq race
has been removed from the -mm tree. Its filename was
tulip-fix-shutdown-dma-irq-race.patch
This patch was dropped because it is obsolete
------------------------------------------------------
Subject: tulip: fix shutdown DMA/irq race
From: Grant Grundler <grundler@xxxxxxxxxxxxxxxx>
IRQs are racing with tulip_down(). DMA can be restarted by the interrupt
handler _after_ we call tulip_stop_rxtx() and the DMA buffers are unmapped.
The result is an MCA (hard crash on ia64) because of an IO TLB miss. The
long-term fix is to make the interrupt handler shutdown aware.
Signed-off-by: Grant Grundler <grundler@xxxxxxxxxxxxxxxx>
Acked-by: Valerie Henson <val_henson@xxxxxxxxxxxxxxx>
Cc: Jeff Garzik <jeff@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
drivers/net/tulip/tulip_core.c | 31 +++++++++++++++++++++----------
1 file changed, 21 insertions(+), 10 deletions(-)
diff -puN drivers/net/tulip/tulip_core.c~tulip-fix-shutdown-dma-irq-race drivers/net/tulip/tulip_core.c
--- a/drivers/net/tulip/tulip_core.c~tulip-fix-shutdown-dma-irq-race
+++ a/drivers/net/tulip/tulip_core.c
@@ -734,23 +734,36 @@ static void tulip_down (struct net_devic
#endif
spin_lock_irqsave (&tp->lock, flags);
+ /*
+ FIXME: We should really add a shutdown-in-progress flag and
+ check it in the interrupt handler to see whether we should
+ reenable DMA or not. The preferred ordering here would be:
+
+ stop DMA engine
+ disable interrupts
+ remove DMA resources
+ free_irq()
+
+ The below works but is non-obvious and doesn't match the
+ ordering of bring-up. -VAL
+ */
+
/* Disable interrupts by clearing the interrupt mask. */
iowrite32 (0x00000000, ioaddr + CSR7);
+ ioread32 (ioaddr + CSR7); /* flush posted write */
- /* Stop the Tx and Rx processes. */
- tulip_stop_rxtx(tp);
+ spin_unlock_irqrestore (&tp->lock, flags);
- /* prepare receive buffers */
- tulip_refill_rx(dev);
+ free_irq (dev->irq, dev); /* no more races after this */
+ tulip_stop_rxtx(tp); /* Stop DMA */
- /* release any unconsumed transmit buffers */
- tulip_clean_tx_ring(tp);
+ /* Put driver back into the state we start with */
+ tulip_refill_rx(dev); /* prepare RX buffers */
+ tulip_clean_tx_ring(tp); /* clean up unsent TX buffers */
if (ioread32 (ioaddr + CSR6) != 0xffffffff)
tp->stats.rx_missed_errors += ioread32 (ioaddr + CSR8) & 0xffff;
- spin_unlock_irqrestore (&tp->lock, flags);
-
init_timer(&tp->timer);
tp->timer.data = (unsigned long)dev;
tp->timer.function = tulip_tbl[tp->chip_id].media_timer;
@@ -776,7 +789,6 @@ static int tulip_close (struct net_devic
printk (KERN_DEBUG "%s: Shutting down ethercard, status was %2.2x.\n",
dev->name, ioread32 (ioaddr + CSR5));
- free_irq (dev->irq, dev);
/* Free all the skbuffs in the Rx queue. */
for (i = 0; i < RX_RING_SIZE; i++) {
@@ -1747,7 +1759,6 @@ static int tulip_suspend (struct pci_dev
tulip_down(dev);
netif_device_detach(dev);
- free_irq(dev->irq, dev);
pci_save_state(pdev);
pci_disable_device(pdev);
_
Patches currently in -mm which might be from grundler@xxxxxxxxxxxxxxxx are
git-netdev-all.patch
tulip-fix-shutdown-dma-irq-race.patch
atomich-add-atomic64-cmpxchg-xchg-and-add_unless-to-parisc.patch
local_t-parisc-cleanup.patch
-
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
