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;
References: <1jk0cf6480.fsf@starbuckisacylon.baylibre.com>
 <20220327130801.15631-1-xiam0nd.tong@gmail.com>
User-agent: mu4e 1.6.10; emacs 27.1
From:   Jerome Brunet <jbrunet@baylibre.com>
To:     Xiaomeng Tong <xiam0nd.tong@gmail.com>
Cc:     alsa-devel@alsa-project.org, broonie@kernel.org,
        khilman@baylibre.com, lgirdwood@gmail.com,
        linux-amlogic@lists.infradead.org,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        martin.blumenstingl@googlemail.com, narmstrong@baylibre.com,
        perex@perex.cz, stable@vger.kernel.org, tiwai@suse.com
Subject: Re: [PATCH] soc: meson: fix a missing check on list iterator
Date:   Sun, 27 Mar 2022 18:05:26 +0200
In-reply-to: <20220327130801.15631-1-xiam0nd.tong@gmail.com>
Message-ID: <1j35j374us.fsf@starbuckisacylon.baylibre.com>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk


On Sun 27 Mar 2022 at 21:08, Xiaomeng Tong <xiam0nd.tong@gmail.com> wrote:

> On Sun, 27 Mar 2022 13:03:14 +0200, Jerome Brunet <jbrunet@baylibre.com> wrote:
>> On Sun 27 Mar 2022 at 16:18, Xiaomeng Tong <xiam0nd.tong@gmail.com> wrote:
>> 
>> > The bug is here:
>> > 	*dai_name = dai->driver->name;
>> >
>> > For for_each_component_dais(), just like list_for_each_entry,
>> > the list iterator 'runtime' will point to a bogus position
>> > containing HEAD if the list is empty or no element is found.
>> > This case must be checked before any use of the iterator,
>> > otherwise it will lead to a invalid memory access.
>> >
>> > To fix the bug, just move the assignment into loop and return
>> > 0 when element is found, otherwise return -EINVAL;
>> 
>> Except we already checked that the id is valid and know an element will
>> be be found once we enter the loop. No bug here and this patch does not
>> seem necessary to me.
>
> Yea, you should be right, it is not a bug here. id already be checked before
> enter the loop:
>
> if (id < 0 || id >= component->num_dai)
>                 return -EINVAL;
>
> but if component->num_dai is not correct due to miscaculation or others reason
> and the door is reopened, this patch can avoid a invalid memory
> access.

This is a speculation which just does not hold ATM. What this patch does
is adding dead code cause the last "return -EINVAL;" will never be
reached.

This no fix nor improvement.

> Anyway,
> it is a good choice to use the list iterator only inside the loop, as linus
> suggested[1]. and we are on the way to change all these use-after-iter cases.
>
> [1]https://lore.kernel.org/lkml/20220217184829.1991035-1-jakobkoschel@gmail.com/

You can make improvements as long as the code is kept clean an
maintainable. Dead code is not OK.