linux-kernel - Re: [PATCH 1/1] pps: clients: gpio: Bypass edge's direction check when not needed

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: <c1e93cfe-371c-4855-9a13-b4b453bb9e88@enneenne.com>
Date: Thu, 11 Apr 2024 09:20:44 +0200
From: Rodolfo Giometti <giometti@...eenne.com>
To: Bastien Curutchet <bastien.curutchet@...tlin.com>
Cc: linux-kernel@...r.kernel.org,
 Thomas Petazzoni <thomas.petazzoni@...tlin.com>, herve.codina@...tlin.com,
 christophercordahi@...ometrics.ca
Subject: Re: [PATCH 1/1] pps: clients: gpio: Bypass edge's direction check
 when not needed

On 10/04/24 18:05, Bastien Curutchet wrote:
> 
> 
> On 4/10/24 17:24, Rodolfo Giometti wrote:
>> On 10/04/24 16:46, Bastien Curutchet wrote:
>>> Hi Rodolfo,
>>>
>>> On 4/10/24 16:23, Rodolfo Giometti wrote:
>>>> On 10/04/24 13:35, Bastien Curutchet wrote:
>>>>> In the IRQ handler, the GPIO's state is read to verify the direction of
>>>>> the edge that triggered the interruption before generating the PPS event.
>>>>> If a pulse is too short, the GPIO line can reach back its original state
>>>>> before this verification and the PPS event is lost.
>>>>>
>>>>> This check is needed when info->capture_clear is set because it needs
>>>>> interruptions on both rising and falling edges. When info->capture_clear
>>>>> is not set, interruption is triggered by one edge only so this check can
>>>>> be omitted.
>>>>>
>>>>> Bypass the edge's direction verification when info->capture_clear is not
>>>>> set.
>>>>>
>>>>> Signed-off-by: Bastien Curutchet <bastien.curutchet@...tlin.com>
>>>>> ---
>>>>>   drivers/pps/clients/pps-gpio.c | 9 +++++++++
>>>>>   1 file changed, 9 insertions(+)
>>>>>
>>>>> diff --git a/drivers/pps/clients/pps-gpio.c b/drivers/pps/clients/pps-gpio.c
>>>>> index 2f4b11b4dfcd..c2a96e3e3836 100644
>>>>> --- a/drivers/pps/clients/pps-gpio.c
>>>>> +++ b/drivers/pps/clients/pps-gpio.c
>>>>> @@ -52,6 +52,15 @@ static irqreturn_t pps_gpio_irq_handler(int irq, void 
>>>>> *data)
>>>>>       info = data;
>>>>> +    if (!info->capture_clear) {
>>>>> +        /*
>>>>> +         * If capture_clear is unset, IRQ is triggered by one edge only.
>>>>> +         * So the check on edge direction is not needed here
>>>>> +         */
>>>>> +        pps_event(info->pps, &ts, PPS_CAPTUREASSERT, data);
>>>>> +        return IRQ_HANDLED;
>>>>> +    }
>>>>> +
>>>>>       rising_edge = gpiod_get_value(info->gpio_pin);
>>>>>       if ((rising_edge && !info->assert_falling_edge) ||
>>>>>               (!rising_edge && info->assert_falling_edge))
>>>>
>>>> Apart the code duplication, which are the real benefits of doing so?
>>>>
>>>
>>> It prevents from losing a PPS event when the pulse is so short (or the
>>> kernel so busy) that the trailing edge of the pulse occurs before the
>>> interrupt handler can read the state of the GPIO pin.
>>
>> Have you a real case when this happens?
>>
> 
> Yes, on my use case, a GPS provides a tiny pulse (~10 us) that is
> sometimes missed when CPU is very busy.

I see...

>> In any cases we should avoid code duplication... so I think we should do 
>> something as below:
>>
>> diff --git a/drivers/pps/clients/pps-gpio.c b/drivers/pps/clients/pps-gpio.c
>> index 2f4b11b4dfcd..f05fb15ed7f4 100644
>> --- a/drivers/pps/clients/pps-gpio.c
>> +++ b/drivers/pps/clients/pps-gpio.c
>> @@ -52,7 +52,9 @@ static irqreturn_t pps_gpio_irq_handler(int irq, void *data)
>>
>>          info = data;
>>
>> -       rising_edge = gpiod_get_value(info->gpio_pin);
>> +       rising_edge = info->capture_clear ? \
>> +                       gpiod_get_value(info->gpio_pin) : \
>> +                       !info->assert_falling_edge;
>>          if ((rising_edge && !info->assert_falling_edge) ||
>>                          (!rising_edge && info->assert_falling_edge))
>>                  pps_event(info->pps, &ts, PPS_CAPTUREASSERT, data);
>>
>> Please, review and test it before resubmitting. :)
>>
> 
> I'll try this and send a V2 after my tests, thank you.

OK, thanks.

However we should think very well about this modification since it could be the 
case where we have a device sending both assert and clear events but we wish to 
catch just the asserts... in this case we will get doubled asserts!

Maybe, can we add a special flag within the DTS (something as 
"support-tiny-pulses" or something like that) to specify that we are in this 
special condition and then checking this setting against capture_clear flag?

Ciao,

Rodolfo

-- 
GNU/Linux Solutions                  e-mail: giometti@...eenne.com
Linux Device Driver                          giometti@...ux.it
Embedded Systems                     phone:  +39 349 2432127
UNIX programming