Re: Pentium DEATH and Cyrix COMA in user-mode

Alan Cox (alan@lxorguk.ukuu.org.uk)
Mon, 10 Nov 1997 09:02:00 +0000 (GMT)


> The pentium bug is caused by the chip executing an invalid opcode - its
> really a very different bug from the cyrix bug.... I doubt
> enabling/disabling anything is going to help with the pentium bug.

They are very similar it appears. One is "lock cmpxchg8" which causes a double
lock cycle on the bus (I believe its to the bus), the other is a dual issue
of an implicitly locked operation (again dual lock).

The other cyrix bug some folks see (divide by zero) does appear to be a
kernel bug. It appears the 6x86 can dual issue RDTSC and give the same clock
cycle back for both calls. Valid but Linux kind of assumes it isnt....

Alan