OS/2 is less forgiving of flaky hardware than most other OSes. Part
of the reason is that it tends to drive the hardware harder. Another
reason is that, when encountering an error like this, OS/2's reaction
is "Something serious has gone wrong; I'd better stop everything and
halt the system so that I don't inadvertently do serious damage to
data". OTOH, the Windows reaction is to try and suppress the error
and continue as long as possible, which tends to lead to system
instability and weird behaviour.
You probably _will_ see the symptoms of hardware failure in Linux
eventually (probably when doing something heavy like recompiling the
kernel). And some day, in Windows as well.