Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <c1d10d21-f3ca-7633-8b65-5d3916cfa648@samsung.com>
Date:   Mon, 12 Dec 2022 18:52:51 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
        Thunderbird/102.4.2
Subject: Re: [PATCH] drivers/firmware_loader: remove list entry before
 deallocation
To:     Luis Chamberlain <mcgrof@kernel.org>
Cc:     linux-kernel@vger.kernel.org, russell.h.weight@intel.com,
        gregkh@linuxfoundation.org, rafael@kernel.org,
        ming.lei@canonical.com, tiwai@suse.de
Content-Language: en-US
From:   =?UTF-8?Q?Micha=c5=82_Lach?= <michal.lach@samsung.com>
In-Reply-To: <Y5bEYfQOMyA4XQEW@bombadil.infradead.org>
Content-Transfer-Encoding: 8bit
Content-Type: text/plain; charset="utf-8"
CMS-TYPE: 201P
References: <CGME20221123111806eucas1p23fdcdbe6e5f4a9e714db428fcd6552b9@eucas1p2.samsung.com>
        <20221123111455.94972-1-michal.lach@samsung.com>
        <Y5bEYfQOMyA4XQEW@bombadil.infradead.org>
Precedence: bulk

On 12/12/22 07:04, Luis Chamberlain wrote:
> Hey Michal! Thanks for your patch! I have a few doubts though!

:-)

> To account for not having to deal with specific drivers we have the
> Linux kernel selftests. And so you can test the firmware loader with all
> sorts of crazy situations which any driver could use and try to see
> if you can re-recreate the issue.
> 
> The kernel selftests driver for the firmware loader is in
> lib/test_firmware.c and you can use thetools/testing/selftests/firmware/fw_run_tests.sh
> to run all the tests. To test the firmware fallback alone you can use
> just fw_fallback.sh.
> 
> If you want to just virtualize this and you can also use kdevops [0] and
> enable the firmware loader selftest and use:;
> 
> make menuconfig          #  enable selftests and just the firmware test
> make linux               #  build linux, pick linux-next
> make selftests
> make selftests-firmware
> 
> But this may be something more you can use later once you get your flow
> going. Just compiling the kernel and running the selftest manually with
> fw_fallback.sh should suffice.

Thanks a lot, I had no idea that there is something like this.

> OK so this proves the bug can happen but I'd like to see the full trace
> and the exact kernel version showing that this can happen on a recent
> kernel. Without that I'm not seeing how this can trigger yet.

Unfortunately I cannot provide a trace. The kernel version was 5.15.41 (-stable).

Keeping that in mind, I will try to reproduce this behaviour with in-tree
code and provide proof.

With regards,
Michał