linux-kernel - Re: [PATCH] gpio: aggregator: restore the set

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 for Android: free password hash cracker in your pocket

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

Message-ID: <1003b65c-7b4f-422c-a720-9f84e5642b1d@bootlin.com>
Date: Mon, 6 Oct 2025 14:26:01 +0200
From: Thomas Richard <thomas.richard@...tlin.com>
To: Geert Uytterhoeven <geert@...ux-m68k.org>
Cc: Linus Walleij <linus.walleij@...aro.org>,
 Bartosz Golaszewski <brgl@...ev.pl>,
 Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
 Bartosz Golaszewski <bartosz.golaszewski@...aro.org>,
 linux-gpio@...r.kernel.org, linux-kernel@...r.kernel.org,
 kernel test robot <oliver.sang@...el.com>
Subject: Re: [PATCH] gpio: aggregator: restore the set_config operation

On 10/6/25 9:42 AM, Geert Uytterhoeven wrote:
> Hi Thomas,
> 
> On Fri, 3 Oct 2025 at 16:30, Thomas Richard <thomas.richard@...tlin.com> wrote:
>> On 10/3/25 3:59 PM, Thomas Richard wrote:
>>>> Is there any specific reason why you are doing this unconditionally,
>>>> instead of only when any of its parents support .set_config(), like
>>>> was done before?
>>>>
>>> My idea was: it will be handled by the core, so the if statement is not
>>> needed. But if we conditionally add the operation we can save some time
>>> in case there is no chip supporting set_config().
>>
>> I just remembered the true reason why I'm doing this unconditionally.
>>
>> The user of the forwarder can override GPIO operations like I do in the
>> pinctrl-upboard driver [1].
>> And now we can add/remove GPIO desc at runtime, if set_config() is set
>> conditionally in gpiochip_fwd_desc_add() it will override the custom
>> set_config() operation.
>> So the only solution is to set the set_config() operation
>> unconditionally in devm_gpiochip_fwd_alloc().
> 
> OK, that makes sense, so
> Reviewed-by: Geert Uytterhoeven <geert+renesas@...der.be>
> 
> I do find this overriding a bit fragile.
> And in theory, such a driver could override chip->can_sleep to false,
> which might be overwritten again by gpiochip_fwd_desc_add()...

Yes, I agree.
Maybe we should not export gpiochip_fwd_get_gpiochip() so it will not be
possible to get the gpio_chip and override some properties. And we add
some helpers to override the GPIO operations. I can make a try.

Regards,

Thomas

-- 
Thomas Richard, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com