Re: [PATCH v2] x86: disable non-instrumented version of copy_mc when KMSAN is enabled

[Linus Torvalds]

On Wed, 6 Mar 2024 at 14:08, Tetsuo Handa
<penguin-kernel@xxxxxxxxxxxxxxxxxxx> wrote:
>
> Something like below one?

I'd rather leave the regular fallbacks (to memcpy and copy_to_user())
alone, and I'd just put the

        kmsan_memmove(dst, src, len - ret);

etc in the places that currently just call the MC copy functions.

The copy_mc_to_user() logic is already set up for that, since it has
to do the __uaccess_begin/end().

Changing copy_mc_to_kernel() to look visually the same would only
improve on this horror-show, I feel.

Obviously some kmsan person needs to validate your kmsan_memmove() thing, but

> Can we assume that 0 <= ret <= len is always true?

Yes. It had better be for other reasons.

                  Linus