[gPXE-devel] [PATCH] Re-enable management mode checks on e1000e

[Geoff Lywood]

Yes, management mode is a side-band mechanism. It is vaguely like IPMI, where there is an always-on component (like the BMC) that can be used to remotely administer the machine. It allows you to remotely perform power operations, and it can do serial port redirection and IDE redirection. I have access to a laptop that implements IAMT, but I don't know which type of PHY it uses, so I don't know if testing on it will be of any use.

In the BIOS setup screens you can specify the IP address of this side-band channel, or (with proper driver support) it can share the IP address of the OS network interface, listening on a dedicated port.

I think the e1000e driver doesn't necessarily need support for all of the fancy features like passing the DHCP options to the management controller, but I agree with Michael that it should at least play nice and try not to reset things that the management controller is using.

Geoff


> -----Original Message-----
> From: Michael Brown [mailto:mbrown at fensystems.co.uk]
> Sent: Thursday, May 27, 2010 5:55 PM
> To: gpxe-devel at etherboot.org
> Cc: Geoff Lywood; Joshua Oreman
> Subject: Re: [gPXE-devel] [PATCH] Re-enable management mode checks on
> e1000e
> 
> On Friday 28 May 2010 01:20:54 Geoff Lywood wrote:
> > 1. The driver avoids powering off the PHY if management mode is enabled.
> >  Presumably, powering off the PHY has ill effects on the other (IAMT)
> >  driver using the PHY. I'm not sure if we ever call into this path,
> though,
> >  so we may be able to remove it.
> >
> > 2. e1000e_copper_link_setup_gg82563_80003es2lan checks if management
> mode
> >  is enabled before writing some PHY registers. I think that this check
> >  might really be needed, because it implies that the PHY registers could
> >  have different values if they were set up by the IAMT side of the
> >  interface, and we don't want to clobber what is there.
> 
> Does "management mode" on e1000 mean that the NIC is in use for sideband
> management (IPMI, iLO, etc.)?
> 
> If so, then I agree that we really do need to be making this check (or,
> alternatively and arguably preferably, just never doing things like
> powering
> off the PHY anyway).
> 
> Michael