iio: invensense: fix integer overflow while multiplication

[PATCH] iio: invensense: fix integer overflow while multiplication

Posted by Karan Sanghavi 3 weeks ago

Typecast a variable to int64_t for 64-bit arithmetic multiplication

Signed-off-by: Karan Sanghavi <karansanghvi98@gmail.com>
---
 drivers/iio/common/inv_sensors/inv_sensors_timestamp.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
index f44458c380d9..d1d11d0b2458 100644
--- a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
+++ b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
@@ -105,8 +105,8 @@ static bool inv_update_chip_period(struct inv_sensors_timestamp *ts,
 
 static void inv_align_timestamp_it(struct inv_sensors_timestamp *ts)
 {
-	const int64_t period_min = ts->min_period * ts->mult;
-	const int64_t period_max = ts->max_period * ts->mult;
+	const int64_t period_min = (int64_t)ts->min_period * ts->mult;
+	const int64_t period_max = (int64_t)ts->max_period * ts->mult;
 	int64_t add_max, sub_max;
 	int64_t delta, jitter;
 	int64_t adjust;

---
base-commit: 81983758430957d9a5cb3333fe324fd70cf63e7e
change-id: 20241102-coverity1586045integeroverflow-cbbf357475d9

Best regards,
-- 
Karan Sanghavi <karansanghvi98@gmail.com>

Re: [PATCH] iio: invensense: fix integer overflow while multiplication

Posted by Jonathan Cameron 3 weeks ago

On Sun, 03 Nov 2024 08:43:14 +0000
Karan Sanghavi <karansanghvi98@gmail.com> wrote:

Hi Karan,

> Typecast a variable to int64_t for 64-bit arithmetic multiplication

The path to actually triggering this is non obvious as these
inputs are the result of rather complex code paths and per chip
constraints.  Have you identified a particular combination that overflows
or is this just based on the type?  I have no problem with applying this
as hardening against future uses but unless we have a path to trigger
it today it isn't a fix.

If you do have a path, this description should state what it is.

> 
> Signed-off-by: Karan Sanghavi <karansanghvi98@gmail.com>
If it's a real bug, needs a Fixes tag so we know how far to backport it.

> ---
>  drivers/iio/common/inv_sensors/inv_sensors_timestamp.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> index f44458c380d9..d1d11d0b2458 100644
> --- a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> +++ b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> @@ -105,8 +105,8 @@ static bool inv_update_chip_period(struct inv_sensors_timestamp *ts,
>  
>  static void inv_align_timestamp_it(struct inv_sensors_timestamp *ts)
>  {
> -	const int64_t period_min = ts->min_period * ts->mult;
> -	const int64_t period_max = ts->max_period * ts->mult;
> +	const int64_t period_min = (int64_t)ts->min_period * ts->mult;
> +	const int64_t period_max = (int64_t)ts->max_period * ts->mult;
>  	int64_t add_max, sub_max;
>  	int64_t delta, jitter;
>  	int64_t adjust;
> 
> ---
> base-commit: 81983758430957d9a5cb3333fe324fd70cf63e7e
> change-id: 20241102-coverity1586045integeroverflow-cbbf357475d9
> 
> Best regards,

Re: [PATCH] iio: invensense: fix integer overflow while multiplication

Posted by Karan Sanghavi 2 weeks, 4 days ago

On Sun, Nov 03, 2024 at 11:18:27AM +0000, Jonathan Cameron wrote:
> On Sun, 03 Nov 2024 08:43:14 +0000
> Karan Sanghavi <karansanghvi98@gmail.com> wrote:
> 
> Hi Karan,
> 
> > Typecast a variable to int64_t for 64-bit arithmetic multiplication
> 
> The path to actually triggering this is non obvious as these
> inputs are the result of rather complex code paths and per chip
> constraints.  Have you identified a particular combination that overflows
> or is this just based on the type?  I have no problem with applying this
> as hardening against future uses but unless we have a path to trigger
> it today it isn't a fix.
> 
> If you do have a path, this description should state what it is.
>
I found this in the coverity scan with CID:1586045 stating

overflow_before_widen: Potentially overflowing expression ts->min_period * ts->mult with type 
unsigned int (32 bits, unsigned) is evaluated using 32-bit arithmetic, and then used in a 
context that expects an expression of type int64_t const (64 bits, signed).

> > 
> > Signed-off-by: Karan Sanghavi <karansanghvi98@gmail.com>
> If it's a real bug, needs a Fixes tag so we know how far to backport it.
>
I thought that this is a fix to this coverity issue thus used fix in the
subject. 
If I have to mention the fix tag for this then can you please let me
know what should I mention in the fix tag.
> > ---
> >  drivers/iio/common/inv_sensors/inv_sensors_timestamp.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> > index f44458c380d9..d1d11d0b2458 100644
> > --- a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> > +++ b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> > @@ -105,8 +105,8 @@ static bool inv_update_chip_period(struct inv_sensors_timestamp *ts,
> >  
> >  static void inv_align_timestamp_it(struct inv_sensors_timestamp *ts)
> >  {
> > -	const int64_t period_min = ts->min_period * ts->mult;
> > -	const int64_t period_max = ts->max_period * ts->mult;
> > +	const int64_t period_min = (int64_t)ts->min_period * ts->mult;
> > +	const int64_t period_max = (int64_t)ts->max_period * ts->mult;
> >  	int64_t add_max, sub_max;
> >  	int64_t delta, jitter;
> >  	int64_t adjust;
> > 
> > ---
> > base-commit: 81983758430957d9a5cb3333fe324fd70cf63e7e
> > change-id: 20241102-coverity1586045integeroverflow-cbbf357475d9
> > 
> > Best regards,
>  
Thank you,
Karan.

Re: [PATCH] iio: invensense: fix integer overflow while multiplication

Posted by Karan Sanghavi 2 weeks, 5 days ago

On Sun, Nov 03, 2024 at 11:18:27AM +0000, Jonathan Cameron wrote:
> On Sun, 03 Nov 2024 08:43:14 +0000
> Karan Sanghavi <karansanghvi98@gmail.com> wrote:
> 
> Hi Karan,
> 
> > Typecast a variable to int64_t for 64-bit arithmetic multiplication
> 
> The path to actually triggering this is non obvious as these
> inputs are the result of rather complex code paths and per chip
> constraints.  Have you identified a particular combination that overflows
> or is this just based on the type?  I have no problem with applying this
> as hardening against future uses but unless we have a path to trigger
> it today it isn't a fix.
> 
> If you do have a path, this description should state what it is.
>

The above issue is discovered by Coverity with CID 1586045 and 1586044.
Link: https://scan7.scan.coverity.com/#/project-view/51946/11354?selectedIssue=1586045

Should I mention this path in the commit short message?

> > 
> > Signed-off-by: Karan Sanghavi <karansanghvi98@gmail.com>
> If it's a real bug, needs a Fixes tag so we know how far to backport it.
> 

What kind of Fixes tag should I provide here. 

> > ---
> >  drivers/iio/common/inv_sensors/inv_sensors_timestamp.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> > index f44458c380d9..d1d11d0b2458 100644
> > --- a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> > +++ b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> > @@ -105,8 +105,8 @@ static bool inv_update_chip_period(struct inv_sensors_timestamp *ts,
> >  
> >  static void inv_align_timestamp_it(struct inv_sensors_timestamp *ts)
> >  {
> > -	const int64_t period_min = ts->min_period * ts->mult;
> > -	const int64_t period_max = ts->max_period * ts->mult;
> > +	const int64_t period_min = (int64_t)ts->min_period * ts->mult;
> > +	const int64_t period_max = (int64_t)ts->max_period * ts->mult;
> >  	int64_t add_max, sub_max;
> >  	int64_t delta, jitter;
> >  	int64_t adjust;
> > 
> > ---
> > base-commit: 81983758430957d9a5cb3333fe324fd70cf63e7e
> > change-id: 20241102-coverity1586045integeroverflow-cbbf357475d9
> > 
> > Best regards,
>

Thank you,
Karan.

Re: [PATCH] iio: invensense: fix integer overflow while multiplication

Posted by Jonathan Cameron 2 weeks ago

On Mon, 4 Nov 2024 16:26:31 +0000
Karan Sanghavi <karansanghvi98@gmail.com> wrote:

> On Sun, Nov 03, 2024 at 11:18:27AM +0000, Jonathan Cameron wrote:
> > On Sun, 03 Nov 2024 08:43:14 +0000
> > Karan Sanghavi <karansanghvi98@gmail.com> wrote:
> > 
> > Hi Karan,
> >   
> > > Typecast a variable to int64_t for 64-bit arithmetic multiplication  
> > 
> > The path to actually triggering this is non obvious as these
> > inputs are the result of rather complex code paths and per chip
> > constraints.  Have you identified a particular combination that overflows
> > or is this just based on the type?  I have no problem with applying this
> > as hardening against future uses but unless we have a path to trigger
> > it today it isn't a fix.
> > 
> > If you do have a path, this description should state what it is.
> >  
> 
> The above issue is discovered by Coverity with CID 1586045 and 1586044.
> Link: https://scan7.scan.coverity.com/#/project-view/51946/11354?selectedIssue=1586045
> 
> Should I mention this path in the commit short message?

That wasn't what I meant.  I was after what combination of possible
inputs actually trigger this rather than (I suspect) local analysis coverity has
done.

> 
> > > 
> > > Signed-off-by: Karan Sanghavi <karansanghvi98@gmail.com>  
> > If it's a real bug, needs a Fixes tag so we know how far to backport it.
> >   
> 
> What kind of Fixes tag should I provide here. 
The patch that introduced the bug in the first place.  See submitting patches
docs for the format.

However, I suspect this is coverity firing a false positive be it a reasonable
one that we should tidy up. As such I'll queue this patch up, but not
as a fix that I'm rushing in, but just as general cleanup for next cycle.

If you find a path to trigger the overflow via userspace inputs etc
then I'm happy to move it over to being handled as an urgent fix.

Jonathan

> 
> > > ---
> > >  drivers/iio/common/inv_sensors/inv_sensors_timestamp.c | 4 ++--
> > >  1 file changed, 2 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> > > index f44458c380d9..d1d11d0b2458 100644
> > > --- a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> > > +++ b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> > > @@ -105,8 +105,8 @@ static bool inv_update_chip_period(struct inv_sensors_timestamp *ts,
> > >  
> > >  static void inv_align_timestamp_it(struct inv_sensors_timestamp *ts)
> > >  {
> > > -	const int64_t period_min = ts->min_period * ts->mult;
> > > -	const int64_t period_max = ts->max_period * ts->mult;
> > > +	const int64_t period_min = (int64_t)ts->min_period * ts->mult;
> > > +	const int64_t period_max = (int64_t)ts->max_period * ts->mult;
> > >  	int64_t add_max, sub_max;
> > >  	int64_t delta, jitter;
> > >  	int64_t adjust;
> > > 
> > > ---
> > > base-commit: 81983758430957d9a5cb3333fe324fd70cf63e7e
> > > change-id: 20241102-coverity1586045integeroverflow-cbbf357475d9
> > > 
> > > Best regards,  
> >  
> 
> Thank you,
> Karan.