Re: [PATCH v7 01/25] kallsyms: avoid hardcoding the buffer size

From: Jarkko Sakkinen <jarkko@kernel.org>
To: Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Cc: Miguel Ojeda <ojeda@kernel.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	rust-for-linux <rust-for-linux@vger.kernel.org>,
	linux-kernel <linux-kernel@vger.kernel.org>,
	Boqun Feng <boqun.feng@gmail.com>,
	Kees Cook <keescook@chromium.org>
Subject: Re: [PATCH v7 01/25] kallsyms: avoid hardcoding the buffer size
Date: Thu, 26 May 2022 07:54:03 +0300	[thread overview]
Message-ID: <Yo8H6/CgjmvtC3Qz@iki.fi> (raw)
In-Reply-To: <CANiq72=K4eC5EX9M8YRzL5Uy9onoTzha9rki-XZ3enSPy0p+UQ@mail.gmail.com>

On Tue, May 24, 2022 at 06:21:14PM +0200, Miguel Ojeda wrote:
> On Mon, May 23, 2022 at 9:46 PM Jarkko Sakkinen <jarkko@kernel.org> wrote:
> >
> > "Declare KSY_NAME_LEN, which describes the maximum length for a kernel
> > symbol read by kallsyms from the input.  In read_symbol(), define the
> > buffer to be of length "KSY_NAME_LEN + 1", which includes the terminator
> > character."
> >
> > would be better.
> 
> Note that the patch is not declaring `KSYM_NAME_LEN`, but a new
> constant for a fairly arbitrarily sized for an input buffer.
> 
> I am all for detailed commit messages, and I agree this can be
> expanded. However, I think the first sentence of what you wrote should
> be part of the docs of the constant, and the second one sounds like it
> could be a comment on the code. Something like "Introduce
> KSYM_NAME_LEN_BUFFER in place of the previously hardcoded size of the
> input buffer (...)" would be better for a reviewer.

Inline comment would be sufficient a remainder, and actually a better
idea.

> > You must split this then into two patches:
> 
> Note that the size is not really being increased in a meaningful way
> -- the important bit is the introduction of the relationship between
> constants. The changes are all meant as a replacement for the
> previously hardcoded constant, so I don't think the split is a "must",
> but we can do it.
> 
> We can even split this into 3 patches: clean up the unneeded `sizeof`,
> replace (and, importantly, document) the hardcoded constant, and
> finally introduce the relationship.
> 
> Thanks for taking a look!
> 
> Cheers,
> Miguel

BR, Jarkko