As far as I know, Linus' point of view is that this "feature" is a
bug. Just to let those that have the bug know, it is printed. Feel
free to ignore the message every time you see it, or to remove the
printk if it annoys you.
If Linux works around these bugs without any verbosity, there is no
drive for anyone to fix it. You had to add "-f no-unroll-loops" to
your compile flags until gcc got fixed. Linus wants to make a
difference in this manner, and he is succeeding.
*I* added that message to the kernel last fall when I first learned of this
Pentium Pro erratum and we were trying to determine what it's consequences
were. Now that it is well understood, there is no advantage to printing a
kernel message every time it is encountered. If you have pre-stepping 9
Pentium Pro CPUs, you will eventually get this message, and with the right
combination of hardware you will see it frequently.
This bug has already been fixed as much as it ever will be. If it's necessary
to have any information on this available to users, then /proc/cpuinfo would
now be the place for it to appear. There are workarounds for all sorts of chip
problems in the kernel and we don't find it necessary to print a message every
time one is activated. The only effect of this kernel message now is to waste
people's time repeatedly answering the question about what it means.
And my personal kernels already have this patch installed, naturally.
Leonard