Just out of curiosity, I tried to boot my desktop machine last night without switching on the router. I didn't need internet access for what I wanted to do just then. To my surprise, it never completed the POST. When I switched the router on, the bleep announcing the end of the POST sounded immediately and the elilo menu came up. Today I checked, and the behaviour is consistent. Is this normal for modern computers?

Not normal for most computers, but if it happens to be a chromebook or similar that depends on internet access for part of its OS functionality then yes.

I don't really understand why your desktop would do that though, unless there is some config that has to be done that requires networking -- such as an nfs mount or similar with the router your only network switch.

No, it looks like a traditional tower, although I gather the innards are those of a laptop. It was running Windows 8 when I first bought it. It boots with a uefi and I've noticed that one of the built-in boot options is PXE. I've never used that because I don't have a PXE server, but it might explain why there seems to be a requirement for a network connection.
No. I don't use nfs at all.

You might try changing the boot order so that it looks for the HD or SSD first before it tries PXE or any optical drives.

You say "desktop machine" so I assume you mean "wired network connection" system (and "elilo" means Slackware, right?).

Does disconnecting the network cable before powering up result in the same hanging POST?

I seem to recall something vaguely similar happening when mopping up after a power failure. Systems that are configured to automatically reboot upon reapplication of power occasionally have problems if the firewall system -- and router for the LAN -- isn't powered up and running when they come back up. SOP is to power-up/boot that system first---sort of like booting a system after all the peripherals (external disk enclosures, etc.) have been powered up and stabilized. But, it hasn't -- ever -- caused systems to hang in the POST. In the worst case, I've had to bounce the network layer on a system here and there but that is pretty rare. Even systems that mount filesystems on NFS hosts boot properly if the NFS host isn't serving filesystems yet and a "mount -a" on the client after the NFS host is back up fixes that. And even having to do that is a rarity. A patch+reboot of the NFS server only briefly affects access.

This is a head scratcher...

I have a Slackware 14.2-based server on our LAN that I can disconnect and power-cycle/reboot to see what happens. I'll let you know how it goes.

Later...

PXE is at the end of the list.

Correct.
I have no idea. I'll try that out and report back.
That's server stuff! I don't (as far as I know) have an automatic reboot configured. I don't think that's necessary in a home machine.
Well, I'm glad it's not just me.

Wonder if some grounding issue is present? I'd be careful if that is the case.

I tried it on my old Slackware "project" server. It booted right up; no hung-in-POST problem seen.

I was referring to systems here on our home LAN. Even the BIOS of my desktop system has an option for what to do after a power outage. If memory serves, it's called "Last State" or something like that. I usually have it set to go ahead and boot though, on some older (now decommissioned) systems I had it set to stay off when power was restored. (Those were systems that I had connected to external disk cabinets---the systems would come up before all the disks came online and booting was a problem. Well, booting wasn't the problem; it was the disk mounting.)

:^D

Is 'not on the list' an option? In other words, can you select under 'Boot Order' your boot disk, and set everything else to 'Disabled'? Some BIOSes allow this, but I've seen others that tend to want at least some 'fallback' option (even if its going to fail too).

The only other thing I could think of, but bear in mind I've *never* witnessed this behavior on Slackware, is the boot loader/OS is holding up the show waiting for network interface XYZ to come up or some update/package management process (which relies on the network) to complete, but I've only seen that happen on other linux distros (Ubuntu most frequently) - Slackware will happily break/fail all of those things on boot with no working network/no concern for the network.

I can now provide some additional info that might make things clearer.

When nothing is plugged into the ethernet port, the system comes up normally. When the router is plugged in but switched off, the UEFI program halts during the POST. If the router is then switched on, the boot process continues as normal.

There are four lights on the router that correspond to the four local ports; these come on briefly in sequence during the router's start-up process. I assume that this indicates some kind of testing procedure by the router itself. It is at this point that the POST completes. So the UEFI seems to be thinking, "Hmm! There is something plugged in here that should be sending me a signal, but I haven't received it. Maybe I'd better wait."

@obobskivich. This definitely has nothing to do with Slackware. It doesn't even have to do with elilo. It happens long before either is loaded.

@jefro. Your post seems rather ominous to me. Would you care to elaborate? I don't want to do anything dangerous.

It is possible your computer and router are at different ground levels which could be due to several reasons. Doubtful but not enough information to say if it is a safety hazard.

Might be a bad ethernet cable or connection. If the cable is near power wiring ac noise could be being picked when the router is off. Bad circuit or driver chip for that ethernet port. Try switching ports at the router to see if the problem goes away. It could also be a bad driver chip on the computer.

That sounds reasonable. There is indeed a tangle of cables there. But given how ports are arranged on the back of a tower machine, there is no way I know to keep data and power cables well away from each other.
No, I don't think it could be that, because the problem only appears when the router is switched off. I have never had any problems when it is on and that port is active.
Yup, that's possible. But again, I've never had any trouble with it in use. Presumably the driver would know whether something is plugged in there or not. Could it be the firmware on the chip (I assume there is some) that is at fault rather than the UEFI program?

Could be one or a combination of things. Just proposing a SWAG, when the router is off the state of the driver (physical not software) on that ethernet port is such a high impedance that basically the cable is acting like an antenna picking up EMI from somewhere. That EMI with whatever happens (circuit reset or diagnostic ) at POST time to the ethernet port and the noise is causing the computer to hang.

Additional troubling shooting would be to disconnect the cable from the router and see if the problem persists. If not it is probably the router. If it does try swapping cables with a known good cable to see if that fixes the problem. If that does not fix things then it might be the computer's ethernet port.

https://en.wikipedia.org/wiki/Differential_signaling

OK. The next experiment will be starting with the cable plugged into the computer but unplugged at the router end. But I need to know first if ethernet connections are hot-pluggable. I can't risk damaging anything.