linux-kernel - Re: [PATCH v3 3/4] pinctrl: Implementation of the generic scmi-pinctrl driver

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] [day] [month] [year] [list]

Message-ID: <87o7k6h344.fsf@epam.com>
Date:   Thu, 20 Jul 2023 18:03:40 +0000
From:   Oleksii Moisieiev <Oleksii_Moisieiev@...m.com>
To:     Andy Shevchenko <andy.shevchenko@...il.com>
CC:     "sudeep.holla@....com" <sudeep.holla@....com>,
        Cristian Marussi <cristian.marussi@....com>,
        Rob Herring <robh+dt@...nel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Conor Dooley <conor+dt@...nel.org>,
        Linus Walleij <linus.walleij@...aro.org>,
        "linux-arm-kernel@...ts.infradead.org" 
        <linux-arm-kernel@...ts.infradead.org>,
        "devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-gpio@...r.kernel.org" <linux-gpio@...r.kernel.org>
Subject: Re: [PATCH v3 3/4] pinctrl: Implementation of the generic
 scmi-pinctrl driver

Hi Andy,

Andy Shevchenko <andy.shevchenko@...il.com> writes:

> On Thu, Jul 20, 2023 at 4:40 PM Oleksii Moisieiev
> <Oleksii_Moisieiev@...m.com> wrote:
>> andy.shevchenko@...il.com writes:
>> > Tue, Jun 06, 2023 at 04:22:28PM +0000, Oleksii Moisieiev kirjoitti:
>
> ...
>
>> >> +    devm_kfree(pmx->dev, pmx->functions[selector].groups);
>> >
>> > Red Flag. Please, elaborate.
>>
>> Thank you for the review.
>> I did some research regarding this and now I'm confused. Could you
>> please explain to me why it's a red flag?
>> IIUC devm_alloc/free functions are the calls to the resource-managed
>> alloc/free command, which is bound to the device.
>> pinctrl-scmi driver does devm_pinctrl_register_and_init which does
>> devres_alloc and doesn't open devres_group like
>> scmi_alloc_init_protocol_instance (thanks to Cristian detailed
>> explanation).
>>
>> As was mentioned in Documentation/driver-api/driver-model/devres.rst:
>>
>> ```
>> No matter what, all devres entries are released on driver detach.  On
>> release, the associated release function is invoked and then the
>> devres entry is freed.
>> ```
>
> Precisely. So, why do you intervene in this?
>
>> Also there is devm_pinctrl_get call listed in the managed interfaces.
>>
>> My understanding is that all resources, bound to the particular device
>> will be freed on driver detach.
>>
>> Also I found some examples of using devm_alloc/free like from dt_node_to_map
>> call in pinctrl-simple.c driver.
>>
>> I agree that I need to implement .remove callback with proper cleanup,
>> but why can't I use devm_* here?
>
> You can use devm_*(), but what's the point if you call release
> yourself? That's quite a red flag usually shows a bigger issue
> (misunderstanding of the objects lifetimes and their interaction).
>

The idea was to follow the way of how pinctrl subsystem manages
resources. It assumes that functions, groups and pins should be
registered using helper functions
pinmux_generic_add_function, pinmux_generic_remove_function,
pinconf_generic_add_group, pinconf_generic_remove_group, etc. Which has
data as the input parameter and should be freed on pinctrl_unregister
call. So pins, groups and functions should live until pinctrl_unregister
is called (from remove callback or from devm_pinctrl_dev_release)

Unfortunately, I can't use this helpers because pins, funcs and groups should
have selector which is understandable by SCMI.

pinctrl_scmi_get_function_groups returns pointer to the allocated
resources to the caller, so I'm allocating managed resources to be sure
that they should be freed on detach.
devm_kfree is called only if scmi_get_group_name call was failed while
converting group_ids to group_names. I count that as a lack of memory,
so I clean allocated groups to give caller a chance to free additional
memory and repeat the call.

So IMHO devm_* fits here good. What do you think?

Sorry for being annoying but I'm trying to understand...

-- 
Thanks,
Oleksii