linux-kernel - Re: [PATCH v3] watchdog: Add driver for Gunyah Watchdog

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: <321f5289-64c0-48f1-91b5-c45e82396ca9@linaro.org>
Date: Tue, 28 Oct 2025 17:30:48 +0100
From: Neil Armstrong <neil.armstrong@...aro.org>
To: Pavan Kondeti <pavan.kondeti@....qualcomm.com>,
 Krzysztof Kozlowski <krzk@...nel.org>
Cc: Hrishabh Rajput <hrishabh.rajput@....qualcomm.com>,
 Bjorn Andersson <andersson@...nel.org>,
 Konrad Dybcio <konradybcio@...nel.org>,
 Wim Van Sebroeck <wim@...ux-watchdog.org>, Guenter Roeck
 <linux@...ck-us.net>, Rob Herring <robh@...nel.org>,
 Krzysztof Kozlowski <krzk+dt@...nel.org>, Conor Dooley
 <conor+dt@...nel.org>, linux-arm-msm@...r.kernel.org,
 linux-watchdog@...r.kernel.org, devicetree@...r.kernel.org,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3] watchdog: Add driver for Gunyah Watchdog

On 10/28/25 13:27, Pavan Kondeti wrote:
> On Tue, Oct 28, 2025 at 12:07:40PM +0100, Krzysztof Kozlowski wrote:
>> On 28/10/2025 12:04, Krzysztof Kozlowski wrote:
>>> On 28/10/2025 11:58, Hrishabh Rajput wrote:
>>>>
>>>> On 10/28/2025 3:10 PM, Krzysztof Kozlowski wrote:
>>>>> On 28/10/2025 10:35, Hrishabh Rajput via B4 Relay wrote:
>>>>>> +
>>>>>> +static int __init gunyah_wdt_init(void)
>>>>>> +{
>>>>>> +	struct arm_smccc_res res;
>>>>>> +	struct device_node *np;
>>>>>> +	int ret;
>>>>>> +
>>>>>> +	/* Check if we're running on a Qualcomm device */
>>>>>> +	np = of_find_compatible_node(NULL, NULL, "qcom,smem");
>>>>> I don't think you implemented my feedback. This again is executed on
>>>>> every platform, e.g. on Samsung, pointlessly.
>>>>>
>>>>> Implement previous feedback.
>>>>
>>>> Do you want us to add platform device from another driver which is
>>>> probed only on Qualcomm devices (like socinfo from previous discussion)
>>>> and get rid of the module init function entirely? As keeping anything in
>>>> the module init will get it executed on all platforms.
>>>
>>> Instead of asking the same can you read previous discussion? What is
>>> unclear here:
>>> https://lore.kernel.org/all/3b901f9d-dbfa-4f93-a8d2-3e89bd9783c9@kernel.org/
>>> ?
>>>
>>>>
>>>>
>>>> With this patch version, we have tried to reduce the code execution on
>>>> non-Qualcomm devices (also tried the alternative as mentioned in the
>>>> cover letter). Adding platform device from another driver as described
>>>> above would eliminate it entirely, please let us know if you want us to
>>>> do that.
>>>
>>> Why do I need to repeat the same as last time?
>>
>>
>> Now I see that you completely ignored previous discussion and sent THE
>> SAME approach.
> 
> Our intention is not to waste reviewers time at all. It is just a
> misunderstanding on what your comment is about. Let me elaborate further
> not to defend our approach here but to get a clarity so that we don't
> end up in the same situation when v4 is posted.
> 
> https://lore.kernel.org/all/b94d8ca3-af58-4a78-9a5a-12e3db0bf75f@kernel.org/
> 
> You mentioned here
> 
> ```
> To me socinfo feels even better. That way only, really only qcom devices
> will execute this SMC.
> ```
> 
> We interpreted this comment as `avoid executing this SMC on non qcom
> devices`. That is exactly what we have done in the current patch. since
> `smem` is not present on non qcom devices, we don't make a SMC. In fact
> we don't even create platform device/driver.
> 
> Please help us understand if the better approach is to just register
> platform driver here and let qcom specific code add the platform device.
> 
> Also, please help me understand why would non qcom platform who care
> about performance load all modules that can be built w/ ARM64. There
> will be many init calls and platform drivers registerd but they never
> get probed at all since their platform does not support. I am not
> defending our aproach, but trying to understand the rationale behind our
> approach vs alternatives.

+static int __init gunyah_wdt_init(void)

will be called on ___all____ arm64 systems which uses the vanilla arm64 defconfig,
while we could say the first call of "of_find_compatible_node()" would fail on all
non-qcom platforms this is still unacceptable.

The solution is to attach the wdt init to something only probed on qcom
platforms (not the module init, the _probe_ which is mapped to a DT compatible)
and very generic like socinfo which could accept HYP stuff.

You could also setup the HYP WDT from the qcom scm driver since the
communication is smc based.

Neil

> 
>>
>> NAK. It is waste of our time if you keep ignoring reviewers and force us
>> to re-iterate the same over and over again.
>>
> Thanks for your time and valuable feedback. I am told getting negative
> feedback is better than no feedback from reviewers in my upstream training :-)
> 
> We will incorporate your feedback in the next version based on your
> answer to the above question.
> 
> Thanks,
> Pavan