Sometimes you never know...

[Chuck(G)]

I was powering up some equipment that I need for my next job. One of those systems is an "old reliable" Socket 754 board (NForce 3A). Hit the power, BIOS signon appears and then--nothing--keyboard dead, can't get into setup. Try this about 4 times--this is a rackmount unit, so not easy to handle. Pull all the connections off, dismount the 4U chassis and take it to the test bench, where it powers right up and boots.
Take it back to the rack, connect everything and nope--BIOS logon with the "Press DEL to enter Setup" message and then nothing.

Pull all the connections, and then re-insert them one at a time between booting.

Finally, the culprit was a USB 10 port hub that prevented booting. Huh? looking carefully, the USB plug for a device was partially out of the type A socket--you could barely notice it. Pull that connection, plug in the hub, everything works. Plug the device back into the hub--and everything works.

Ah well, I had to replace the CR2032 cell anyway...

 

[GiGaBiTe]

I always liked how a single malfunctioning USB device could take down the entire system.

Can't tell you how many late night emergency service calls I've been on where a rogue flash drive, keyboard or USB hub that went crazy and took down the entire system.

Especially great are those tiny USB flash drives that someone stuck into the back of their computer long ago and forgot about, for it to lose its mind months/years later and prevent the system from booting, or lock it up.

USB was, is, and always will be a perpetual dumpster fire. I miss Firewire.

 

[Chuck(G)]

I was gobsmacked that POST couldn't even get as far as the memory self-test or setup.

Anyone who's tried to implement a USB composite device from scratch can tell you what a goshawful mess the "standard" is.

 

[GiGaBiTe]

I'd love to know what the USB controller is doing to hang up the entire system. Is it blasting out interrupts and creating an interrupt storm? Or is it that the BIOS is trying to enumerate bootable devices and the USB controller is stuck telling the system it's busy and waits forever?

Pretty ridiculous that no sanity checking was ever implemented in USB to say "OK this device on the bus is clearly malfunctioning, I'm going to disable and ignore it until it's removed."

 

[kb2syd]

My little (Dell) linux computer I use to run my ham shack has a strange USB issue. I use a powered USB 3.0 hub. If I lose power or the unit comes unplugged, the CMOS battery drains. If I then try to boot it fails with a flashing orange light. No sound, no display, nothing. If I disconnect the power to the USB hub the machine will boot.

 

[GiGaBiTe]

The flashing orange power light is a diagnostic code you can lookup on Dell's website. It will flash a number of times, then pause briefly and then do the flashing again, repeating until the machine is turned off.

If you count the number of flashes, it should tell you what it's mad about.

 

[kb2syd]

The flashing orange power light is a diagnostic code you can lookup on Dell's website. It will flash a number of times, then pause briefly and then do the flashing again, repeating until the machine is turned off.

If you count the number of flashes, it should tell you what it's mad about.

I know. Decodes to "Replace Motherboard". What it means is that it is seeing power from two places, doesn't know which one to select and freaks out. If I remove the power from the external hub it starts right up.

 

[luckybob]

I was gobsmacked that POST couldn't even get as far as the memory self-test or setup.

Anyone who's tried to implement a USB composite device from scratch can tell you what a goshawful mess the "standard" is.

I've used a Supermicro motherboard for years. it would NEVER complete a warm-reset from Windows. You had to pull the power to reboot. Made windows updates an utter nightmare.

retired the system to the garage a few months ago. Works fine. I eventually fingered out, it was because of the USB mouse I was using.

 

[Pepinno]

USB was designed by Intel. Intel makes microchips. Somehow, Intel decided that USB peripherals had an utter need to be able to directly access the CPU, it seems.