netdev - Re: [PATCH] ptp: ocp: don't allow on S390

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: <35952ae9-07a5-11aa-76ae-d698bcaa9804@infradead.org>
Date:   Wed, 25 Aug 2021 13:45:57 -0700
From:   Randy Dunlap <rdunlap@...radead.org>
To:     Jonathan Lemon <jonathan.lemon@...il.com>
Cc:     Arnd Bergmann <arnd@...nel.org>,
        Richard Cochran <richardcochran@...il.com>,
        Networking <netdev@...r.kernel.org>
Subject: Re: [PATCH] ptp: ocp: don't allow on S390

On 8/25/21 1:40 PM, Jonathan Lemon wrote:
> On Wed, Aug 25, 2021 at 10:29:51AM -0700, Randy Dunlap wrote:
>> On 8/25/21 10:08 AM, Jonathan Lemon wrote:
>>> On Wed, Aug 25, 2021 at 12:55:25PM +0200, Arnd Bergmann wrote:
>>>> On Tue, Aug 24, 2021 at 11:48 PM Randy Dunlap <rdunlap@...radead.org> wrote:
>>>>>
>>>>> On 8/20/21 8:31 AM, Richard Cochran wrote:
>>>>>> On Fri, Aug 20, 2021 at 12:45:42PM +0200, Arnd Bergmann wrote:
>>>>>>
>>>>>>> I would also suggest removing all the 'imply' statements, they
>>>>>>> usually don't do what the original author intended anyway.
>>>>>>> If there is a compile-time dependency with those drivers,
>>>>>>> it should be 'depends on', otherwise they can normally be
>>>>>>> left out.
>>>>>>
>>>>>> +1
>>>>>
>>>>> Hi,
>>>>>
>>>>> Removing the "imply" statements is simple enough and the driver
>>>>> still builds cleanly without them, so Yes, they aren't needed here.
>>>>>
>>>>> Removing the SPI dependency is also clean.
>>>>>
>>>>> The driver does use I2C, MTD, and SERIAL_8250 interfaces, so they
>>>>> can't be removed without some other driver changes, like using
>>>>> #ifdef/#endif (or #if IS_ENABLED()) blocks and some function stubs.
>>>>
>>>> If the SERIAL_8250 dependency is actually required, then using
>>>> 'depends on' for this is probably better than an IS_ENABLED() check.
>>>> The 'select' is definitely misplaced here, that doesn't even work when
>>>> the dependencies fo 8250 itself are not met, and it does force-enable
>>>> the entire TTY subsystem.
>>>
>>> So, something like the following (untested) patch?
>>> I admit to not fully understanding all the nuances around Kconfig.
>>
>> Hi,
>>
>> You can also remove the "select NET_DEVLINK". The driver builds fine
>> without it. And please drop the "default n" while at it.
> 
> I had to add this one because devlink is a dependency and the kbuild
> robot generated a config without it.

What kind of dependency is devlink?
The driver builds without NET_DEVLINK.


> The 'imply' statements were added because while the driver builds
> without them, the resources don't show up unless the platform
> modules are also present.  This was really confusing users, since
> they selected the OCP driver and then were not able to use the
> flash since the XILINX modules had not been selected.
> 
> Is there a better way of specifying these type of dependencies?

Documentation/  and/or one can add comments/docs in the Kconfig help
section.


-- 
~Randy