Message-ID: <eaa0c99a-3b41-4444-906c-d2f005d326b9@linux.intel.com>
Date: Mon, 22 Apr 2024 16:24:36 -0500
From: Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
To: Krzysztof Kozlowski <krzk@...nel.org>, Liam Girdwood
 <lgirdwood@...il.com>, Peter Ujfalusi <peter.ujfalusi@...ux.intel.com>,
 Bard Liao <yung-chuan.liao@...ux.intel.com>,
 Ranjani Sridharan <ranjani.sridharan@...ux.intel.com>,
 Daniel Baluta <daniel.baluta@....com>,
 Kai Vehmanen <kai.vehmanen@...ux.intel.com>, Mark Brown
 <broonie@...nel.org>, Jaroslav Kysela <perex@...ex.cz>,
 Takashi Iwai <tiwai@...e.com>, Shawn Guo <shawnguo@...nel.org>,
 Sascha Hauer <s.hauer@...gutronix.de>,
 Pengutronix Kernel Team <kernel@...gutronix.de>,
 Fabio Estevam <festevam@...il.com>, Matthias Brugger
 <matthias.bgg@...il.com>,
 AngeloGioacchino Del Regno <angelogioacchino.delregno@...labora.com>
Cc: sound-open-firmware@...a-project.org, linux-sound@...r.kernel.org,
 linux-kernel@...r.kernel.org, imx@...ts.linux.dev,
 linux-arm-kernel@...ts.infradead.org, linux-mediatek@...ts.infradead.org
Subject: Re: [PATCH 00/14] ASoC: Constify local snd_sof_dsp_ops



>>> There are multiple reasons and benefits for const, like compiler
>>> optimization, code readability (meaning) up to security improvements,
>>> e.g. by some GCC plugins or marking rodata as really non-writeable, so
>>> closing some ways of exploits. There are many opportunities here, even
>>> if they are not yet enabled.
>>
>> Possibly, but the SOF core does not know if the structure it uses is
>> rodata or not. Using the 'const' identifier would be misleading.
> 
> How so? If core does not modify structure, it should take it via ops,
> just like 100 other widely known structures (see checkpatch). Why is
> this different?

I don't understand "it should take it via ops"

We are already fetching the structure with private_data.

>>>> that's a different interpretation to the 'software' view you're
>>>> describing. "this structure will not modified by this function" is not
>>>> the same thing as "this structure CANNOT be modified".
>>>
>>> Yes, but can we please discuss specific patchset then? Patches which
>>> change pointers to const have one "interpretation". Patches which modify
>>> static or global data have another.
>>
>> Just look at sound/soc/sof/intel/mtl.c... The core will sometimes use a
> 
> That's a driver (or specific implementation), not core.

You are making an assumption on what the SOF core is. The core is used
by ACPI or PCI drivers as a library. The structures are all allocated in
ACPI/PCI drivers and passed to the core library.

>> constant structure and sometimes not, depending on the PCI ID reported
>> by hardware. This was intentional to override common defaults and make
>> the differences limited in scope between hardware generations.
> 
> 
>>
>> int sof_mtl_ops_init(struct snd_sof_dev *sdev)
>> {
>> 	struct sof_ipc4_fw_data *ipc4_data;
>>
>> 	/* common defaults */
>> 	memcpy(&sof_mtl_ops, &sof_hda_common_ops, sizeof(struct
>> snd_sof_dsp_ops)); <<<< THE BASELINE IS CONSTANT
> 
> Yes, I saw it and such users are not changed. They won't receive any
> safety. But all others are getting safer.


Maybe there's a misunderstanding on what the 'SOF core' is. This is just
a helper library that are used by the PCI drivers. The core has zero
knowledge on anything really.

> I really do not understand what is the problem here. In entire Linux all
> of such changes are welcomed with open arms. So what is different here?
Adding 'const' at the SOF core level does not mean that we can treat
structures as rodata. It only means that the structure is not modified
by the core library. Not the same thing.

Powered by blists - more mailing lists