Re: [PATCH] thermal: Add a sanity check for invalid state at stats update

[Takashi Iwai <tiwai@suse.de>]

On Tue, 07 Apr 2020 15:30:51 +0200,
Amit Kucheria wrote:
> 
> On Mon, Mar 30, 2020 at 7:39 PM Takashi Iwai <tiwai@suse.de> wrote:
> >
> > The thermal sysfs handler keeps the statistics table with the fixed
> > size that was determined from the initial max_states() call, and the
> > table entry is updated at each sysfs cur_state write call.  And, when
> > the driver's set_cur_state() ops accepts the value given from
> > user-space, the thermal sysfs core blindly applies it to the
> > statistics table entry, which may overflow and cause an Oops.
> > Although it's rather a bug in the driver's ops implementations, we
> > shouldn't crash but rather give a proper warning instead.
> >
> > This patch adds a sanity check for avoiding such an OOB access and
> > warns with a stack trace to show the suspicious device in question.
> 
> Hi Takashi,
> 
> Instead of this warning, I think we should reject such input when
> writing to cur_state.
> 
> See attached patch. If you think this OK, I'll submit it.

Actually the input value itself is correct, the problem is rather
about the max_states that may vary depending on other driver.  So IMO,
we don't want to refuse the input completely.

Please see the thread:
  https://lore.kernel.org/linux-acpi/s5h5zeiwd01.wl-tiwai@suse.de/


thanks,

Takashi


> 
> Regards,
> Amit
> 
> > Signed-off-by: Takashi Iwai <tiwai@suse.de>
> > ---
> >
> > We've hit some crash by stress tests, and this patch at least works
> > around the crash itself.  While the actual bug fix of the buggy driver
> > is still being investigated, I submit the hardening in the core side
> > at first.
> >
> >  drivers/thermal/thermal_sysfs.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> >
> > diff --git a/drivers/thermal/thermal_sysfs.c b/drivers/thermal/thermal_sysfs.c
> > index aa99edb4dff7..a23c4e701d63 100644
> > --- a/drivers/thermal/thermal_sysfs.c
> > +++ b/drivers/thermal/thermal_sysfs.c
> > @@ -772,6 +772,11 @@ void thermal_cooling_device_stats_update(struct thermal_cooling_device *cdev,
> >
> >         spin_lock(&stats->lock);
> >
> > +       if (dev_WARN_ONCE(&cdev->device, new_state >= stats->max_states,
> > +                         "new state %ld exceeds max_state %ld",
> > +                         new_state, stats->max_states))
> > +               goto unlock;
> > +
> >         if (stats->state == new_state)
> >                 goto unlock;
> >
> > --
> > 2.16.4
> >
> From 54266260d483ab4476510dd4461a1cafc611e17d Mon Sep 17 00:00:00 2001
> Message-Id: <54266260d483ab4476510dd4461a1cafc611e17d.1586266224.git.amit.kucheria@linaro.org>
> From: Amit Kucheria <amit.kucheria@linaro.org>
> Date: Tue, 7 Apr 2020 18:48:14 +0530
> Subject: [PATCH] thermal: Reject invalid cur_state input from userspace
> 
> We don't check if the cur_state value input in sysfs is greater than the
> maximum cooling state that the cooling device supports. This can cause
> access to unallocated memory in case THERMAL_STATISTICS in enabled and
> could also crash cooling devices that don't check for an invalid state in
> their set_cur_state() callback.
> 
> Return an error if the state being requested in greater than the maximum
> cooling state the device supports.
> 
> Reported-by: Takashi Iwai <tiwai@suse.de>
> Signed-off-by: Amit Kucheria <amit.kucheria@linaro.org>
> ---
>  drivers/thermal/thermal_sysfs.c | 9 ++++++++-
>  1 file changed, 8 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/thermal/thermal_sysfs.c b/drivers/thermal/thermal_sysfs.c
> index 7e1d11bdd258..8033e5a9386a 100644
> --- a/drivers/thermal/thermal_sysfs.c
> +++ b/drivers/thermal/thermal_sysfs.c
> @@ -703,7 +703,7 @@ cur_state_store(struct device *dev, struct device_attribute *attr,
>  		const char *buf, size_t count)
>  {
>  	struct thermal_cooling_device *cdev = to_cooling_device(dev);
> -	unsigned long state;
> +	unsigned long state, max_state;
>  	int result;
>  
>  	if (sscanf(buf, "%ld\n", &state) != 1)
> @@ -712,6 +712,13 @@ cur_state_store(struct device *dev, struct device_attribute *attr,
>  	if ((long)state < 0)
>  		return -EINVAL;
>  
> +	result = cdev->ops->get_max_state(cdev, &max_state);
> +	if (result)
> +		return result;
> +
> +	if (state >= max_state)
> +		return -EINVAL;
> +
>  	mutex_lock(&cdev->lock);
>  
>  	result = cdev->ops->set_cur_state(cdev, state);
> -- 
> 2.20.1
>