e1000: save skb counts in TX to avoid cache misses
authorDean Nelson <dnelson@redhat.com>
Thu, 25 Aug 2011 14:39:24 +0000 (14:39 +0000)
committerDavid S. Miller <davem@davemloft.net>
Fri, 26 Aug 2011 16:55:18 +0000 (12:55 -0400)
commit31c15a2f24ebdab14333d9bf5df49757842ae2ec
treeef89d4b94c4f41f6d18c923490441189ff95b52d
parent363437f40a23bacdead80bb80d08d8193a20cfce
e1000: save skb counts in TX to avoid cache misses

Virtual Machines with emulated e1000 network adapter running on Parallels'
server were seeing kernel panics due to the e1000 driver dereferencing an
unexpected NULL pointer retrieved from buffer_info->skb.

The problem has been addressed for the e1000e driver, but not for the e1000.
Since the two drivers share similar code in the affected area, a port of the
following e1000e driver commit solves the issue for the e1000 driver:

commit 9ed318d546a29d7a591dbe648fd1a2efe3be1180
Author: Tom Herbert <therbert@google.com>
Date:   Wed May 5 14:02:27 2010 +0000

    e1000e: save skb counts in TX to avoid cache misses

    In e1000_tx_map, precompute number of segements and bytecounts which
    are derived from fields in skb; these are stored in buffer_info.  When
    cleaning tx in e1000_clean_tx_irq use the values in the associated
    buffer_info for statistics counting, this eliminates cache misses
    on skb fields.

Signed-off-by: Dean Nelson <dnelson@redhat.com>
Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/intel/e1000/e1000.h
drivers/net/ethernet/intel/e1000/e1000_main.c