Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220822141930.5f43b5e7@endymion.delvare>
In-Reply-To: <20220822141930.5f43b5e7@endymion.delvare>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Wed, 24 Aug 2022 10:31:35 -0700
Message-ID: <CAHk-=wiFRok=hU_BNEsqodjyGm=XV9LVZ1w=cm4ooEtWVpYLPg@mail.gmail.com>
Subject: Re: [GIT PULL] dmi update for v5.19
To:     Jean Delvare <jdelvare@suse.de>
Cc:     LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Aug 22, 2022 at 5:19 AM Jean Delvare <jdelvare@suse.de> wrote:
>
> Andy Shevchenko (1):
>       firmware: dmi: Use the proper accessor for the version field

I pulled this, but I kind of question it.

This replaces a single 32-bit memory access (and an optimized byte
swap) and a mask operation with three load-byte-and-shift operations.

It's not clear that the new code is better.

That said, I can't imagine it matters - but because I looked at it, I
note that the length check seems to be kind of iffy.

The code checks that the length of the block is < 32 before doing the
checksum on it, but shouldn't it also check for some minimum size?
Otherwise the dmi checksum is kind of pointless, isn't it?

It will access a minimum of 24 bytes for that dmi_base thing, so that
would be the most obvious minimum value. But maybe there is some
spec-defined size for that that only covers the header?

           Linus