On Wed, 23 Dec 2009 23:01:02 -0800
Andrew Morton [off-list ref] wrote:

quoted

Dec 23 17:32:58 mail kernel: DRHD: handling fault status reg 2
Dec 23 17:32:58 mail kernel: DMAR:[DMA Read] Request device [06:00.0] 
fault addr fff8423fe000
Dec 23 17:32:58 mail kernel: DMAR:[fault reason 06] PTE Read access is 
not set
Dec 23 17:32:58 mail kernel: sky2 0000:06:00.0: error interrupt 
status=0x80000000
Dec 23 17:32:58 mail kernel: sky2 0000:06:00.0: PCI hardware error (0x2010)
Dec 23 17:33:06 mail kernel: ------------[ cut here ]------------
Dec 23 17:33:06 mail kernel: WARNING: at net/sched/sch_generic.c:255 
dev_watchdog+0xf3/0x161()
Dec 23 17:33:06 mail kernel: Hardware name: System Product Name
Dec 23 17:33:06 mail kernel: NETDEV WATCHDOG: eth0 (sky2): transmit 

Just to reiterate, these are consistent with a DMA error on transmit.
Since is 4 seconds after the DHCPACK, they probably are not related, except
that when DHCP renew may have caused more traffic.

Are you using something like network console or kgdb, or anything that could
be using some unusual network I/O?