2011-03-29 23:58:18

by Max Asbock

Subject: Should corrected machine check errors still taint the kernel?

I noticed that corrected machine check errors taint the kernel. And it
looks like they have done that forever.

Is that still the right thing to do?

The comment in add_taint() says:
"Can't trust the integrity of the kernel anymore."

I don't think that's necessarily the case after seeing a corrected
memory error.

thanks,
Max

2011-03-30 03:06:18

by Andi Kleen

Subject: Re: Should corrected machine check errors still taint the kernel?

On Tue, Mar 29, 2011 at 04:58:17PM -0700, Max Asbock wrote:
> I noticed that corrected machine check errors taint the kernel. And it
> looks like they have done that forever.
>
> Is that still the right thing to do?

Likely not.

>
> The comment in add_taint() says:
> "Can't trust the integrity of the kernel anymore."

The comment is definitely wrong agreed.

-Andi