linux/drivers/net/igb
Jeff Kirsher 2d0bb1c1f4 e100/e1000*/igb*/ixgb*: Add missing read memory barrier
Based on patches from Sonny Rao and Milton Miller...

Combined the patches to fix up clean_tx_irq and clean_rx_irq.

The PowerPC architecture does not require loads to independent bytes
to be ordered without adding an explicit barrier.

In ixgbe_clean_rx_irq we load the status bit then load the packet data.
With packet split disabled if these loads go out of order we get a
stale packet, but we will notice the bad sequence numbers and drop it.

The problem occurs with packet split enabled where the TCP/IP header
and data are in different descriptors. If the reads go out of order
we may have data that doesn't match the TCP/IP header. Since we use
hardware checksumming this bad data is never verified and it makes it
all the way to the application.

This bug was found during stress testing and adding this barrier has
been shown to fix it.  The bug can manifest as a data integrity issue
(bad payload data) or as a BUG in skb_pull().

This was a nasty bug to hunt down, if people agree with the fix I think
it's a candidate for stable.

Previously Submitted to e1000-devel only for ixgbe

http://marc.info/?l=e1000-devel&m=126593062701537&w=3

We've now seen this problem hit with other device drivers (e1000e mostly)
So I'm resubmitting with fixes for other Intel Device Drivers with
similar issues.

CC: Milton Miller <miltonm@bga.com>
CC: Anton Blanchard <anton@samba.org>
CC: Sonny Rao <sonnyrao@us.ibm.com>
CC: stable <stable@kernel.org>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-08-08 20:02:45 -07:00
..
e1000_82575.c igb: Program MDICNFG register prior to PHY init 2010-08-03 16:40:53 -07:00
e1000_82575.h igb: add per-packet timestamping 2010-03-30 23:42:27 -07:00
e1000_defines.h igb: restore EEPROM values of MDICNFG on reset with 82580 2010-07-26 18:42:53 -07:00
e1000_hw.h igb: update hw_debug macro to make use of netdev_dbg call 2010-03-30 23:02:26 -07:00
e1000_mac.c igb: add support for reporting 5GT/s during probe on PCIe Gen2 2010-04-27 12:53:28 -07:00
e1000_mac.h igb: remove adaptive IFS from driver 2010-02-17 13:21:34 -08:00
e1000_mbx.c igb: fix a few items where weren't correctly setup for mbx timeout 2009-10-06 14:59:21 -07:00
e1000_mbx.h igb: rework use of VMOLR in regards to PF and VFs 2009-10-28 03:25:40 -07:00
e1000_nvm.c igb: remove microwire support from igb 2009-10-06 14:59:22 -07:00
e1000_nvm.h
e1000_phy.c igb: Power down link when interface is down 2010-02-17 13:21:33 -08:00
e1000_phy.h igb: Power down link when interface is down 2010-02-17 13:21:33 -08:00
e1000_regs.h igb: support for VF configuration tools 2010-02-12 16:56:09 -08:00
igb_ethtool.c igb: correct link test not being run when link is down 2010-07-01 22:45:57 -07:00
igb_main.c e100/e1000*/igb*/ixgb*: Add missing read memory barrier 2010-08-08 20:02:45 -07:00
igb.h igb: reduce cache misses on tx cleanup 2010-05-05 21:30:12 -07:00
Makefile igb: Add support for enabling VFs to PF driver. 2009-02-20 00:22:54 -08:00