[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAHNKnsSDX_vRJ7ot1SBfGxcFfd+EoYtdz-fyF66U=t8egmfu9g@mail.gmail.com>
Date: Fri, 6 Aug 2021 12:13:15 +0300
From: Sergey Ryazanov <ryazanov.s.a@...il.com>
To: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
Cc: Loic Poulain <loic.poulain@...aro.org>,
"David S. Miller" <davem@...emloft.net>,
Network Development <netdev@...r.kernel.org>,
open list <linux-kernel@...r.kernel.org>,
Johannes Berg <johannes@...solutions.net>,
Jakub Kicinski <kuba@...nel.org>
Subject: Re: [PATCH v1 1/1] wwan: core: Avoid returning error pointer from wwan_create_dev()
Hello Andy,
On Fri, Aug 6, 2021 at 12:08 PM Andy Shevchenko
<andriy.shevchenko@...ux.intel.com> wrote:
> On Thu, Aug 05, 2021 at 09:53:57PM +0200, Loic Poulain wrote:
>> On Thu, 5 Aug 2021 at 20:38, Andy Shevchenko
>> <andriy.shevchenko@...ux.intel.com> wrote:
>>>
>>> wwan_create_dev() is expected to return either valid pointer or NULL,
>>> In some cases it might return the error pointer. Prevent this by converting
>>> it to NULL after wwan_dev_get_by_parent().
>>
>> wwan_create_dev is called both from wwan_register_ops() and
>> wwan_create_port(), one using IS_ERR and the other using NULL testing,
>> they should be aligned as well.
>
> Ah, good catch!
>
> I just sent v2, but eventually I have decided to switch to error pointer since
> it seems the most used pattern in the code.
I agree that returning the error pointer is a good solution here.
Thank you for the fix.
--
Sergey
Powered by blists - more mailing lists