Hi Scott,
thanks for your reply.
I've already looked at version 5.2.20 and I've ported the WOL fixes into
version 4.4.12. There are two fixes related to WOL. The first simply
resets the hardware to make sure that the EEPROM is in a known state before
it read from for the first time (in e1000_probe()). The second assures
that the correct register is read to determine whether WOL should be
enabled (also in e1000_probe()). As such, neither fix affects how the
EEPROM can be written so as to disable WOL (if, in fact, that is the
problem). Just in case I missed something, I do plan on compiling 5.2.20
in its entirety and running it instead of the patched 4.4.12... :O)
As far as ethtool, this does not help either. I've looked through
e1000_ethtool.c and did exactly what the driver would have done if ethtool
was executed as you suggest (i.e. ethtool -s eth<x> wol d). Basically,
when WOL is to be disabled via ethtool, adapter->wol is set to 0. To avoid
having to use ethtool and just as a quick test, I simply hardcoded
adapter->wol to 0. This results in the driver executing the "disable WOL"
code in e1000_suspend() but, as I've already mentioned, the next time the
magic packet is sent out, the box wakes up anyway.
Thanks,
Paul
"Feldman, Scott"
<scott.feldman@in To: "Paul Burkacki" <PBurkacki@symantec.com>, <linux-net@vger.kernel.org>
tel.com> cc:
Subject: RE: Wake-on-LAN on PRO/1000 dual port adapter.
11/03/2003 02:25
PM
> I have a system whose motherboard (Supermicro X5DP8-G2) comes
> with a builtin Intel PRO/1000 MT Dual Port Server Adapter
> (82546EB). The system is running RedHat Linux kernel 2.4.18
> and the NIC driver is e1000 version 4.4.12.
Get the update driver from sf.net/projects/e1000. 5.2.20 is the current
version. There have been some WoL bug fixes since 4.4.12.
> I am wondering if it is possible to disable WOL on this NIC.
> From looking at the driver code, it's clear that the function
> e1000_suspend() is the one that sets up the NIC for what it
> should do in response to a WOL event. The driver writes WUC
> and WUFC registers with a 0 and then calls
> pci_enable_wake() for state 3 and 4 with a 0. However, this
> does not seem to disable WOL. Namely, even if I force
> execution of this code path, once the system shuts down and a
> magic packet is sent to the NIC, the system will come up again.
Use "ethtool -s eth<X> wol d" to disable wakeup on this interface.
Looks like the EEPROM for this LOM is programmed to enable WoL on
power-up. Verify with ethtool eth<x>.
-scott
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
