Re: [PATCH v4] EDAC/mc: Prefer strscpy over strcpy

From: Len Baker <len.baker@gmx.com>
To: Borislav Petkov <bp@alien8.de>, Joe Perches <joe@perches.com>
Cc: Len Baker <len.baker@gmx.com>,
	Mauro Carvalho Chehab <mchehab@kernel.org>,
	Tony Luck <tony.luck@intel.com>,
	James Morse <james.morse@arm.com>,
	Robert Richter <rric@kernel.org>,
	David Laight <David.Laight@aculab.com>,
	Kees Cook <keescook@chromium.org>,
	linux-hardening@vger.kernel.org, linux-edac@vger.kernel.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v4] EDAC/mc: Prefer strscpy over strcpy
Date: Sun, 29 Aug 2021 18:14:55 +0200	[thread overview]
Message-ID: <20210829150223.GB2185@titan> (raw)
In-Reply-To: <YSkmv8kz2z3OBHVe@zn.tnic>

Hi,

On Fri, Aug 27, 2021 at 07:54:07PM +0200, Borislav Petkov wrote:
> On Fri, Aug 27, 2021 at 07:36:33PM +0200, Len Baker wrote:
> > Well, the main purpose is to clean up the proliferation of str*cpy functions.
> > One task is to remove the strcpy uses: The first step (previous step) would
> > be to remove all the strcpy uses. Then, as a second step remove all the
> > strcpy implementations.
> >
> > I hope that this clarify your question.
>
> Yes, it does.
>
> Now lemme clarify why I'm asking: when your patch is committed to the
> kernel tree and someone reads its commit message months or even years
> from now - and those who do that are mostly maintainers trying to figure
> out why stuff was done the way it was - they will read:
>
> "This is a previous step in the path to remove the strcpy() function
> entirely from the kernel."
>
> and wonder what previous step that is what the following step is...
>
> So, long story short, your commit message should be complete on its own
> and understandable without any references to things which might not be
> as clear and self-evident in the future as they are now.
>
> Makes sense?

Ok, understood. Thanks for the advise and guidance.

>
> Also, if you're wondering if you should send the patch with the error
> checking of strscpy() added, as I requested, even if it might look
> superfluous now, yes you should.
>
> Even if it looks impossible now, we might change some of those defines
> in the future and forget to touch the logic which generates e->label and
> we might end up exhausting that string.
>
> So it would be a lot more robust if something would catch that change,
> albeit seemingly redundant now.
>
> I sincerely hope that clears up things.

Yes, it clears up things. However I think the same that Joe:

From Joe Perches:

[...]
I still think scnprintf is _way_ more common and intelligible as
a construct than this odd strscpy with required error checking.
[...]

So, I will send a new version for review with the commit message updated
and using the scnprintf. This way we can discuss using a real patch.
Anyway thanks for the review.

Regards,
Len