Re: [PATCH 1/2] ARM: mm: Refactor __do_page_fault()

[Ard Biesheuvel]

On Tue, 1 Jun 2021 at 16:32, Russell King (Oracle)
<linux@xxxxxxxxxxxxxxx> wrote:
>
> On Sat, May 29, 2021 at 11:41:37AM +0800, Kefeng Wang wrote:
> > 1. cleanup access_error(), make vma flags set and check into
> >    __do_page_fault() and do_page_fault() directly.
> >
> > 2. drop fsr and task argument, instead, using vm_flags in
> >    __do_page_fault().
> >
> > 3. cleans up the multiple goto statements in __do_page_fault().
> >
> > 4. use current->mm directly in do_page_fault().
> >
> > Signed-off-by: Kefeng Wang <wangkefeng.wang@xxxxxxxxxx>
>
> This patch is a really good example of something that is very difficult
> to review and see that there are no unintended changes.
>
> Many people have complained about my patches, where I create a series of
> many patches where each patch does exactly _one_ simple transformation to
> the code. This is a good example _why_ I do that - a step by step single
> transformation approach is way easier to review.
>
> Sorry, but I'm not able to sensibly review this patch, and therefore
> I won't apply it. Please split it into smaller changes.
>

Agreed. If your commit message contains an enumeration of things the
patch does, it is a very strong hint that each of those things needs
to be a separate patch if at all possible.