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: <d4b4fb72-79e7-4e06-a038-4645574a799b@bootlin.com>
Date: Thu, 25 Apr 2024 08:11:21 +0200
From: Bastien Curutchet <bastien.curutchet@...tlin.com>
To: Rodolfo Giometti <giometti@...eenne.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

Hi Rodolfo,


On 4/12/24 14:20, Bastien Curutchet wrote:
> Hi Rodolfo,
> 
> On 4/12/24 08:44, Rodolfo Giometti wrote:
>> On 11/04/24 14:44, Bastien Curutchet wrote:
>>>>
>>>> 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!
>>>>
>>>
>>> My understanding is that clear events are to be captured only when this
>>> capture_clear boolean is set. If it is not set, the PPS_CAPTURECLEAR
>>> flag is not added to pps_source_info->mode and get_irqf_trigger_flags()
>>> will return only one edge flag (rising or falling depending on
>>> assert-falling-edge DT property).
>>
>> Yes. You are right.
>>
>>> By the way, I see that the capture_clear is never set since the legacy
>>> platform data support has been dropped (commit ee89646619ba).
>>
>> I see, but it can be re-enabled in the future... In this scenario, I 
>> think we should add a DT entry to enable this special behavior. Maybe 
>> we can also add a warning as below: >
>> static irqreturn_t pps_gpio_irq_handler(int irq, void *data)
>> {
>>          ...
>>          if ((rising_edge && !info->assert_falling_edge) ||
>>                          (!rising_edge && info->assert_falling_edge))
>>                  pps_event(info->pps, &ts, PPS_CAPTUREASSERT, data);
>>          else if (info->capture_clear &&
>>                          ((rising_edge && info->assert_falling_edge) ||
>>                          (!rising_edge && !info->assert_falling_edge)))
>>                  pps_event(info->pps, &ts, PPS_CAPTURECLEAR, data);
>>      else
>>          dev_warn_ratelimited(dev, "no ASSERT or CAPTURE event? "
>>              "Maybe you need support-tiny-assert-pulse?");
>>
>>          return IRQ_HANDLED;
>> }
>>
> 
> I'm not sure a DT entry is needed. IMO there are two cases:
>   1) capture_clear is unset. We need to capture only assert events,
>      interrupt will be triggered by assert edge only so there is no need
>      to check GPIO state: we can use the bypass.
>   2) capture_clear is set. We need to capture assert and/or clear
>      events, interrupt will be triggered by both assert and clear edges
>      so we can't avoid the GPIO state checking to distinguish clear
>      events from assert events: we can't use the bypass.
> 
> So if we bypass the GPIO's state check when capture_clear is unset and
> leave current behavior when capture_clear is set:
>   - case 1) will be more efficient and we won't lose tiny pulses anymore
>   - case 2) is unchanged: we still might lose tiny pulses but as bypass
> can't be done here, I think that we can't do better.
> 
> I agree that adding warning when the handler is left without triggering
> a pps event can be useful, I can add it in a V3 version.
> 

Would this be OK for you ? If yes, I'll send a V3 version without DT
entry but with an additional warning.


Best regards,
Bastien