linux-kernel - Re: [PATCH 2/5] firmware: arm_scmi: imx: Add i.MX95 CPU Protocol

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <Z5Dq0Yb6wsIJQYF4@pluto>
Date: Wed, 22 Jan 2025 12:55:45 +0000
From: Cristian Marussi <cristian.marussi@....com>
To: Dan Carpenter <dan.carpenter@...aro.org>
Cc: Cristian Marussi <cristian.marussi@....com>,
	"Peng Fan (OSS)" <peng.fan@....nxp.com>,
	Sudeep Holla <sudeep.holla@....com>,
	Shawn Guo <shawnguo@...nel.org>,
	Sascha Hauer <s.hauer@...gutronix.de>,
	Pengutronix Kernel Team <kernel@...gutronix.de>,
	Fabio Estevam <festevam@...il.com>, linux-kernel@...r.kernel.org,
	arm-scmi@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
	imx@...ts.linux.dev, Peng Fan <peng.fan@....com>
Subject: Re: [PATCH 2/5] firmware: arm_scmi: imx: Add i.MX95 CPU Protocol

On Wed, Jan 22, 2025 at 03:41:41PM +0300, Dan Carpenter wrote:
> On Wed, Jan 22, 2025 at 12:22:18PM +0000, Cristian Marussi wrote:
> > > > +struct scmi_msg_imx_cpu_attributes_out {
> > > > +	__le32 attributes;
> > > > +#define	CPU_MAX_NAME	16
> > > > +	u8 name[CPU_MAX_NAME];
> > > 
> > > char is always unsigned in the kernel these days but strings should
> > > still always be char.  Same thing in patch 1, there were a couple u8
> > > names.
> > > 
> > 

Hi Dan,

> > While it is certainly true that char is the way to go for strings and, as
> > such, it is used elsewhere to hold the resource names across all SCMI
> > protocols, in this context it is a field of structure representing
> > exactly the layout of message reply coming from the server, and defined
> > in the SCMI spec as a uint8 array, so, we have generally preferred to
> > used u8 to represent such fixed size array all across the SCMI stack
> > protocols implementation....
> > 
> > .... not saying that it is necessarily completelt right, but that is the
> > reason we are guilty :D
> 
> Fine.  I don't have intense emotions about this.
> 
> It does slightly bother me when we assume that the SCMI server NUL
> terminates these when we do things like:
> 
> 	dev_info(ph->dev, "i.MX CPU: name: %s\n", out->name);
> 

Hang on...I have not really done a proper review still on this series...
...and this printout above straight out of the message payload seems very
wrong to me too..

> But from a practical perspective we have to trust the SCMI server.
>

....nope we should NEVER trust the server...and instead assume it can
kill us (kernel) all the time :P

...despite what the spec says, we tend to assume tha the server can be
maliciously wrong (or just crappy), so in other protocols where we do used
an u8[] to describe the resource name field in a message, we have also always
(hopefully :D) taken care to use it ONLY after having processed that field like...

   strscpy(dom_info->name, attr->name, SCMI_SHORT_NAME_MAX_SIZE);

...to remove any possible bad outcome from a misbehaving SCMI fw server.

Thanks,
Cristian