Re: [PATCH][CFT] [coredump] don't use __kernel_write() on kmap_local_page()

From: Linus Torvalds
Date: Mon Oct 03 2022 - 20:20:32 EST

On Mon, Oct 3, 2022 at 4:37 PM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> One variant would be to revert the original patch, put its
> (hopefully) fixed variant into -next and let it sit there for
> a while. Another is to put this incremental into -next and
> merge it into mainline once it gets a sane amount of testing.

Just do the incremental fix. It looks obvious enough ("oops, we need
to get the pos _after_ we've done any skip-lseeks on the core file")
that I think it would be just harder to follow a "revert and follow up
with a fix".

I don't think it needs a ton of extra testing, with Okajima having
already confirmed it fixes his problem case..