Re: [PATCH] Fix a potential abuse of seq_printf() format string in drivers

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Thu, Nov 21, 2024 at 08:38:27PM -0800, Kees Cook wrote:
> 
> 
> On November 20, 2024 11:28:35 AM PST, Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> >On Wed, Nov 20, 2024 at 10:12:40AM -0800, Kees Cook wrote:
> >> On Wed, Nov 20, 2024 at 08:35:38AM +0100, Linus Walleij wrote:
> >> > On Wed, Nov 20, 2024 at 6:31 AM David Wang <00107082@163.com> wrote:
> >> > 
> >> > > Using device name as format string of seq_printf() is proned to
> >> > > "Format string attack", opens possibility for exploitation.
> >> > > Seq_puts() is safer and more efficient.
> >> > >
> >> > > Signed-off-by: David Wang <00107082@163.com>
> >> > 
> >> > Okay better get Kees' eye on this, he looks after string vulnerabilities.
> >> > (But I think you're right.)
> >> 
> >> Agreed, this may lead to kernel memory content exposures. seq_puts()
> >> looks right.
> >> 
> >> Reviewed-by: Kees Cook <kees@kernel.org>
> >
> >Wait, userspace "shouldn't" be controlling a device name, but odds are
> >there are some paths/subsystems that do this, ugh.
> >
> >> To defend against this, it might be interesting to detect
> >> single-argument seq_printf() usage and aim it at seq_puts()
> >> automatically...
> >
> >Yeah, that would be good to squash this type of issue.
> >
> >> > >  drivers/gpio/gpio-aspeed-sgpio.c            | 2 +-
> >> > >  drivers/gpio/gpio-aspeed.c                  | 2 +-
> >> > >  drivers/gpio/gpio-ep93xx.c                  | 2 +-
> >> > >  drivers/gpio/gpio-hlwd.c                    | 2 +-
> >> > >  drivers/gpio/gpio-mlxbf2.c                  | 2 +-
> >> > >  drivers/gpio/gpio-omap.c                    | 2 +-
> >> > >  drivers/gpio/gpio-pca953x.c                 | 2 +-
> >> > >  drivers/gpio/gpio-pl061.c                   | 2 +-
> >> > >  drivers/gpio/gpio-tegra.c                   | 2 +-
> >> > >  drivers/gpio/gpio-tegra186.c                | 2 +-
> >> > >  drivers/gpio/gpio-tqmx86.c                  | 2 +-
> >> > >  drivers/gpio/gpio-visconti.c                | 2 +-
> >> > >  drivers/gpio/gpio-xgs-iproc.c               | 2 +-
> >> > >  drivers/irqchip/irq-gic.c                   | 2 +-
> >> > >  drivers/irqchip/irq-mvebu-pic.c             | 2 +-
> >> > >  drivers/irqchip/irq-versatile-fpga.c        | 2 +-
> >> > >  drivers/pinctrl/bcm/pinctrl-iproc-gpio.c    | 2 +-
> >> > >  drivers/pinctrl/mvebu/pinctrl-armada-37xx.c | 2 +-
> >> > >  drivers/pinctrl/pinctrl-mcp23s08.c          | 2 +-
> >> > >  drivers/pinctrl/pinctrl-stmfx.c             | 2 +-
> >> > >  drivers/pinctrl/pinctrl-sx150x.c            | 2 +-
> >> > >  drivers/pinctrl/renesas/pinctrl-rzg2l.c     | 2 +-
> >> > 
> >> > Can you split this in three patches per-subsystem?
> >> > One for gpio, one for irqchip and one for pinctrl?
> >> > 
> >> > Then send to each subsystem maintainer and CC kees on
> >> > each.
> >> > 
> >> > I'm just the pinctrl maintainer. The rest can be found with
> >> > scripts/get_maintainer.pl.
> >> 
> >> Oof. That's a lot of work for a mechanical change like this. Perhaps
> >> Greg KH can take it directly to the drivers tree instead?
> >
> >I can take it all, as-is, right now, if you want me to.  Just let me
> >know.
> 
> Yes, please do. I will send a patch for making seq_printf more defensive separately.

Great, now queued up, let's make sure 0-day is ok with it...

greg k-h