[PATCH] e1000: prevent statistics from getting garbled during reset
authorLinas Vepstas <linas@austin.ibm.com>
Fri, 9 Jun 2006 05:19:44 +0000 (22:19 -0700)
committerJeff Garzik <jeff@garzik.org>
Sun, 11 Jun 2006 14:35:15 +0000 (10:35 -0400)
If a PCI bus error/fault triggers a PCI bus reset, attempts to get the
ethernet packet count statistics from the hardware will fail, returning
garbage data upstream.  This patch skips statistics data collection if the
PCI device is not on the bus.

This patch presumes that an earlier patch,
[PATCH] PCI Error Recovery: e1000 network device driver
has already been applied.

Signed-off-by: Linas Vepstas <linas@austin.ibm.com>
Cc: John Ronciak <john.ronciak@intel.com>
Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>
Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
drivers/net/e1000/e1000_main.c

index 56c7492e3e91f191acf2c042661a03b3f9ec9ff6..a373ccb308d8d3281298bee38c1a5dcbfcc8ca7f 100644 (file)
@@ -3045,14 +3045,20 @@ void
 e1000_update_stats(struct e1000_adapter *adapter)
 {
        struct e1000_hw *hw = &adapter->hw;
+       struct pci_dev *pdev = adapter->pdev;
        unsigned long flags;
        uint16_t phy_tmp;
 
 #define PHY_IDLE_ERROR_COUNT_MASK 0x00FF
 
-       /* Prevent stats update while adapter is being reset */
+       /*
+        * Prevent stats update while adapter is being reset, or if the pci
+        * connection is down.
+        */
        if (adapter->link_speed == 0)
                return;
+       if (pdev->error_state && pdev->error_state != pci_channel_io_normal)
+               return;
 
        spin_lock_irqsave(&adapter->stats_lock, flags);