Received-SPF: pass (google.com: domain of linux-wireless+bounces-7634-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Message-ID: <812f01e5-58e8-4128-b329-a2456c7e6b86@bootlin.com>
Date: Tue, 14 May 2024 10:59:38 +0200
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [bug report] wifi: wilc1000: convert list management to RCU
To: Dan Carpenter <dan.carpenter@linaro.org>
Cc: linux-wireless@vger.kernel.org
References: <3b46ec7c-baee-49fd-b760-3bc12fb12eaf@moroto.mountain>
 <1a982ba8-c5f3-415a-bb39-fae9973b100e@bootlin.com>
 <8ee9481f-7677-4a0e-be11-0c81d4bbef9f@suswa.mountain>
From: =?UTF-8?Q?Alexis_Lothor=C3=A9?= <alexis.lothore@bootlin.com>
Content-Language: en-US
In-Reply-To: <8ee9481f-7677-4a0e-be11-0c81d4bbef9f@suswa.mountain>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

On 5/14/24 10:54, Dan Carpenter wrote:
> On Mon, May 13, 2024 at 04:16:05PM +0200, Alexis Lothoré wrote:
>> Hello Dan,
>> sorry for the late reply
>>
>> On 5/9/24 15:24, Dan Carpenter wrote:
>>> Hello Alexis Lothoré,
>>>
>>> Commit f236464f1db7 ("wifi: wilc1000: convert list management to
>>> RCU") from Apr 10, 2024 (linux-next), leads to the following Smatch
>>> static checker warning:
>>>
>>> 	drivers/net/wireless/microchip/wilc1000/mon.c:236 wilc_wfi_init_mon_interface()
>>> 	warn: sleeping in atomic context
>>
>> I am struggling to reproduce the warning in smatch. I executed those basic steps:
>> - clone and build latest smatch
>> - checkout linux-next master branch
>> - export ARCH and CROSS_COMPILE
>> - load a defconfig enabling all features in wilc driver
>> - run ~/src/smatch/smatch_scripts/build_kernel_data.sh
>> - then run ~/src/smatch/smatch_scripts/test_kernel.sh
>> Yet I do no see any mention to wilc in the resulting smatch_warns.txt file. The
>> outcome is the same if I run smatch only on the wilc driver:
>> - ~/src/smatch/smatch_scripts/kchecker drivers/net/wireless/microchip/
>>
>> Am I missing something obvious ?
> 
> Yeah.  I'm sorry, I need to write a blog entry about this.  These checks
> rely on the cross function database, and you need to rebuild it a bunch
> of times.  It's a simple process but very time consuming.
> 
> Instead of that, it's better to do run time testing using
> CONFIG_DEBUG_ATOMIC_SLEEP=y

Yes, that's what I have been doing in the mean time, and it allowed me to at
least reproduce the second bug you raised, but I just wanted to make sure to
have suppressed those issues once I will have written the proper fixes. I'll
continue with the runtime checkers like CONFIG_DEBUG_ATOMIC_SLEEP :)

Thanks,

Alexis

> 
> regards,
> dan carpenter
> 

-- 
Alexis Lothoré, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com