linux-kernel - Re: [PATCH RFC v2 4/6] gpio: aggregator: handle runtime registration of gpio_desc in gpiochip

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

Message-ID: <6bf787ab-9ed4-43a5-8796-262e234edb42@bootlin.com>
Date: Wed, 9 Apr 2025 16:50:11 +0200
From: Thomas Richard <thomas.richard@...tlin.com>
To: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
Cc: Linus Walleij <linus.walleij@...aro.org>,
 Bartosz Golaszewski <brgl@...ev.pl>,
 Geert Uytterhoeven <geert+renesas@...der.be>, linux-gpio@...r.kernel.org,
 linux-kernel@...r.kernel.org, thomas.petazzoni@...tlin.com,
 DanieleCleri@...on.eu, GaryWang@...on.com.tw
Subject: Re: [PATCH RFC v2 4/6] gpio: aggregator: handle runtime registration
 of gpio_desc in gpiochip_fwd

On 3/17/25 18:13, Andy Shevchenko wrote:
> On Mon, Mar 17, 2025 at 04:38:02PM +0100, Thomas Richard wrote:
>> Add request() callback to check if the GPIO descriptor was well registered
>> in the gpiochip_fwd before to use it. This is done to handle the case
>> where GPIO descriptor is added at runtime in the forwarder.
>>
>> If at least one GPIO descriptor was not added before the forwarder
>> registration, we assume the forwarder can sleep as if a GPIO is added at
>> runtime it may sleep.
> 
> Hmm... This should rather be reformatted each time a new descriptor is added,
> no?
I used the easiest solution.

Otherwise to switch to can_sleep mode:
- gpio_fwd_add_gpio_desc() inits the mutex if a sleeping GPIO line is added.
- gpio_fwd_add_gpio_desc() locks the mutex if the spinlock is locked
(gpio_fwd_get_multiple_locked() or gpio_fwd_set_multiple_locked() was
called).
- gpio_fwd_add_gpio_desc() set can_sleep mode to true
- gpio_fwd_get_multiple_locked() or gpio_fwd_set_multiple_locked() shall
unlock the spinlock and the mutex at end.

I think it is a bit over-engineered, and I probably do not handle all
corner cases.

Regards,

Thomas