Really interesting that it was a 486 bug, given the provenance listed in the article. Windows 3.0 was, indeed, released after the 80486 was. I am not sure why the reset busy code was repeated 3 times, I assume the bit must have been somewhat sticky.
"If an unmasked exception occurs when the numeric exception bit in CR0 is clear and the IGNNE# pin is active, the performance of the FPU will be retarded as long as the exception remains pending."
"If an unmasked exception occurs when the numeric exception bit in CR0 is clear and the IGNNE# pin is active, the performance of the FPU will be retarded as long as the exception remains pending."
https://www.cs.earlham.edu/~dusko/cs63/prepentium.html
I wonder if that has anything to do with it all.