The mouse of death...

[01000101]

Today, my girlfriend and I decided to rearrange our room. After about 4-5 hours of labor, we finally got things organized and was time for me to set back up all the computer/network stuff. I got everything on the rackmount plugged in and re-setup my desk the way I wanted... then when I turned on my computer (this computer, running ubuntu 8.04.1), I was greeted by a frozen NVIDIA splash screen that would NOT go away (not even from ctrl-alt-f1/2/3/del/bkspace). So I rebooted into my previous kernel that I had from the install cd, and this time, it was frozen at a black screen after the boot scripts had ran.

Then I set the xorg.conf file back to the defaults/low-res and got the same issue. No matter what I tried, it kept freezing during bootup. I opened my computer and checked out things to make sure nothing was out of place or unusual, and I double checked all the cords again. Then after about 2 hours of pulling my hair out, I re-connected the cables for my KVM switch, and found that if I unplugged my primary-input mouse, that it booted all the way. Then when I plugged it back in, it froze. It was then that I realized that one of the pins were bent by just a bit, resulting in it going in about 3/4 of the way (which looks normal). So I fixed it with some precision with a butter knife, and it booted up fine and with mouse enabled.

Why would this cause such a HUGE disruption to ubuntu? Windows would have just lacked mouse support, but this was a rediculous reaction.

[piranha]

Maybe Linux like to have rodents as pets?

I think that it got confused. Because it was used to having a mouse, and then it didn't.
I know there isn't any computer science base for this really, but my computer gets confused sometimes too.

-JL

[AJ]

Why would this cause such a HUGE disruption to ubuntu? Windows would have just lacked mouse support, but this was a rediculous reaction.

It sounds to me like someone should have put a timeout in to a KBC polling routine but didn't - I guess if the software is polling a flag but due to a controller short this flag never gets set/unset, something goes in to an infinite loop. It strikes me that this is somewhere in ring 0 code (thus other interrupts are disabled so no other tasks can run) suggesting a kernel issue(?). I guess the hang is too late in the boot process for tasking not to have been initialised.

I wonder if it is worth reporting the error - whether it is in the Linux kernel or in ubuntu specifically, I'm sure the devteam would want to avoid this situation for other people.

Cheers,
Adam