Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20200702140102.26129-1-pmenzel@molgen.mpg.de> <7c31d918-c967-5ebb-970e-7f6e913237e8@amd.com>
In-Reply-To: <7c31d918-c967-5ebb-970e-7f6e913237e8@amd.com>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Thu, 2 Jul 2020 12:42:37 -0700
Message-ID: <CAHk-=wimBucBdC+tzFEuFJU5X1UFTVWo+AHMvQTSkCcVnTT-ww@mail.gmail.com>
Subject: Re: [PATCH 1/2] moduleparams: Add hex type parameter
To:     =?UTF-8?Q?Christian_K=C3=B6nig?= <christian.koenig@amd.com>
Cc:     Paul Menzel <pmenzel@molgen.mpg.de>,
        Alex Deucher <alexander.deucher@amd.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        amd-gfx@lists.freedesktop.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Jul 2, 2020 at 7:42 AM Christian K=C3=B6nig <christian.koenig@amd.c=
om> wrote:
>
> I'm just not sure how well this is received upstream because it only
> covers u32
>
> On the other hand that is probably also the most used.

Not necessarily true. I'd argue that "unsigned long"  is equally
possible for some bit mask (or other hex-likely) type.

So don't call it just "hex". Call it "hexint" (the hex does imply
"unsigned", I feel - showing hex numbers with a sign sounds insane).

That way, if somebody ends up wanting it for unsigned long values,
we're not stuck.

Another option is to just say that hex values always have bit _sizes_.
So "hex32" and "hex64" would also make sense as names to me.

While at it, should the hex numbers always be padded out to the size?
The example Paul used doesn't have that issue (high bit being set).

Bbut often it may make sense to show a 32-bit hex number as "%#08x"
because it really makes things clearer when you're looking at high
bits, say.

It's really hard to tell the difference between "just bit 27 set" and
"just bit 31" set otherwise, and that's not all that uncommon when the
bitmasks are sparse.

             Linus