linux-kernel - Re: [PATCH 3/3] gpio: realtek-otto: use of instead of device handlers

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CAMRc=MfeSASpCKa_ppsreJ_dUZuGgYZtfQNxTNXKWAS6ZreJ7Q@mail.gmail.com>
Date: Thu, 18 Dec 2025 10:58:35 +0100
From: Bartosz Golaszewski <brgl@...nel.org>
To: Rosen Penev <rosenp@...il.com>
Cc: linux-gpio@...r.kernel.org, Linus Walleij <linusw@...nel.org>, 
	open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 3/3] gpio: realtek-otto: use of instead of device handlers

On Thu, Dec 18, 2025 at 10:32 AM Rosen Penev <rosenp@...il.com> wrote:
>
> On Thu, Dec 18, 2025 at 12:31 AM Bartosz Golaszewski <brgl@...nel.org> wrote:
> >
> > On Wed, Dec 17, 2025 at 9:26 PM Rosen Penev <rosenp@...il.com> wrote:
> > > >
> > > > Yeah, no, there's almost *never* a point in using OF-specific accessors. NAK.
> > >
> > > Argument made to me on netdev before is it's pointless overhead for an
> > > OF only driver.
> >
> > Would you mind posting a lore link? I'll gladly chime in.
> >
> > Drivers are OF-only until they aren't. Vide: lots of discussions
> > currently about supporting ARM laptops and servers with mixed DT-ACPI
> > setup.
> Hrm I must have misremembered or I can't find it. In any case, these
> devices will never support ACPI. They're all embedded devices with
> uboot.
> >

If I had a euro everytime someone was wrong saying "never"... :)

> > The overhead of going through the fwnode pointer is absolutely
> > negligible while using generic accessors allows taking secondary
> > fwnodes into account.
> >
> > Please don't tell me people are going around the kernel converting
> > drivers to using of_ routines?
> There's one special case I know of, of_get_mac_address vs
> device_get_mac_address. The former supports NVMEM. The latter does
> not.
> >

Yes, that is one of those special cases but a better fix would be to
make NVMEM core fwnode-agnostic. While nvmem remains OF-specific, we
can live with using of_get_mac_address().

Bart