wake-on-lan after power loss ( doesn't work).

[Kevin Lawton]

|| Okay - one step at a time.
|| The WOL function occurs before Windoze has even started to load - it
|| powers the machine back on.
|| WOL is enabled/disablled in the BIOS. If it is enabled, then it
|| should work - given the correct hardware.
|| Would be interested in knowing whereabouts in which version of
|| Windoze you found the option to turn OFF WOL ?
|| Kevin.
|
| The WOL function does indeed occur before Windows loads, which is the
| whole issue of the thread, that "something" is discriminating between
| different settings, not the BIOS, as that m'board BIOS isn't executed
| yet. It's not a certain version of Windows that allows changing the
| WOL feature, it's all(?) versions, an option given by the NIC driver,
| often accessible in the networking properties or perhaps Device
| Manager props for the NIC. As an example:
|
| http://69.36.189.159/usr_1034/nic_wake.gif
|
| My memory of it is a bit vague, but I believe this feature is
| implemented even if the NIC doesn't have a WOL header on it, then by
| PCI bus feature.

The above is a feature of the Linksys driver. Other drivers, for example
3com or RealTek, will alow you to set whether the NIC can bring the machine
out of standby mode via the 'Power Management' tab.
The reason why the above can be made a driver feature is that in standby
mode (also sometimes called 'susped to RAM') Windows is still loaded. Some
m/boards will have a small LED to indicate that the RAM is still powered
during standby. This helps to warn users not to remove or replace RAM strips
during this time.
Should system power be completely remove at this time then the contents of
RAM, and thus the 'standby' status will be lost.
The 'Magic Packet' --> Wake On Lan feature will just cause the machine to
power up appropriate to the way it was shut down. Just the same as pressing
the front panel power button. Power down into standby - and it comes out of
standby, Power down to 'hibernate' (AKA: 'suspend to disk') - and it comes
out of hibernation, power down fully - and it boots up.
Make any hardware changes between power-down and WOL and the results will
be less predictable.
AFAIK there is no detail in the PCI specification to provide for the WOL
signal - hence the seperate flying lead. What IS provided for via the PCI
bus is the +5v 'standby' current - and that is part of the ATX
specification.
The driver for the NIC will probably show the above feature regardless of
whether is has actually been implemented on the card or not.
Hope we've just about got the explanation nailed down now.
Kevin.

 

[kony]

| The WOL function does indeed occur before Windows loads, which is the
| whole issue of the thread, that "something" is discriminating between
| different settings, not the BIOS, as that m'board BIOS isn't executed
| yet. It's not a certain version of Windows that allows changing the
| WOL feature, it's all(?) versions, an option given by the NIC driver,
| often accessible in the networking properties or perhaps Device
| Manager props for the NIC. As an example:
|
| http://69.36.189.159/usr_1034/nic_wake.gif
|
| My memory of it is a bit vague, but I believe this feature is
| implemented even if the NIC doesn't have a WOL header on it, then by
| PCI bus feature.

The above is a feature of the Linksys driver. Other drivers, for example
3com or RealTek, will alow you to set whether the NIC can bring the machine
out of standby mode via the 'Power Management' tab.
The reason why the above can be made a driver feature is that in standby
mode (also sometimes called 'susped to RAM') Windows is still loaded.

No, that driver feature is NOT just a driver feature, it actually
controls how the "NIC" responds when the machine is OFF. I have used
that very same NIC (or another EXACTLY like it) to power on an old ATX
and it depends on which setting you choose. Again, this is why this
thread exists, because it is not as simple as you imply.

Some
m/boards will have a small LED to indicate that the RAM is still powered
during standby. This helps to warn users not to remove or replace RAM strips
during this time.
Should system power be completely remove at this time then the contents of
RAM, and thus the 'standby' status will be lost.

We are not talking about a sleep mode. We're not talking about
suspend to RAM, we're talking about OFF, ATX soft-off, with AC still
going to the power supply.

The 'Magic Packet' --> Wake On Lan feature will just cause the machine to
power up appropriate to the way it was shut down. Just the same as pressing
the front panel power button. Power down into standby - and it comes out of
standby, Power down to 'hibernate' (AKA: 'suspend to disk') - and it comes
out of hibernation, power down fully - and it boots up.
Make any hardware changes between power-down and WOL and the results will
be less predictable.
AFAIK there is no detail in the PCI specification to provide for the WOL
signal - hence the seperate flying lead. What IS provided for via the PCI
bus is the +5v 'standby' current - and that is part of the ATX
specification.

There needs be no specific PCI spec for WOL, because it (PCI) is more
versatile than that... not only the NIC, but any PCI device can do it,
if that device is designed to, and appropriate BIOS settings, set.

The driver for the NIC will probably show the above feature regardless of
whether is has actually been implemented on the card or not.
Hope we've just about got the explanation nailed down now.
Kevin.

Just about all NICs these days support it, I'd expect it's more common
for a card that can do it, to have a driver with that functionality
omitted.

 

[kony]

No, that driver feature is NOT just a driver feature, it actually
controls how the "NIC" responds when the machine is OFF. I have used
that very same NIC (or another EXACTLY like it) to power on an old ATX
and it depends on which setting you choose. Again, this is why this
thread exists, because it is not as simple as you imply.

Upon double-checking this, it appears that I'm either partially wrong,
or that it depends on the specific motherboard, or both. I just tried
an Asus A7S333 with Award BIOS (the system that the Linksys NIC from
my previously linked image, is installed in), and it will power-up the
system independant of the driver setting, but doesn't need the WOL
cable, only the setting in the BIOS for "Power Up On PCI Card" to be
enabled.

I still feel my memory is correct about that card providing power-on
ONLY when the driver had the setting enabled, I even remember the
motherboard it was used in at that time, an Intel AL440LX, which did
have a WOL cable installed unlike the A7S333.

 

[daytripper]

There needs be no specific PCI spec for WOL, because it (PCI) is more
versatile than that...

That's true because WOL has nothing to do with the PCI bus per se. One could
implement WOL on an ISA-only system, no big deal, as the implementations
pre-PME# use sidebanding from the pluggable agent to the host system board.

not only the NIC, but any PCI device can do it, if that device is designed to,

[snipped again]

I'm presuming you're referring to PME# here, but that exists only on platforms
that support PCI 2.2 at the minimum, and that actually implement PME# (not the
majority of systems to this point, fwiw)...

/daytripper