[RFC] rtw88: fix subscript above array bounds compiler warning

[RFC] rtw88: fix subscript above array bounds compiler warning

[Stanislaw Gruszka]

My compiler complain about:

drivers/net/wireless/realtek/rtw88/phy.c: In function ‘rtw_phy_rf_power_2_rssi’:
drivers/net/wireless/realtek/rtw88/phy.c:430:26: warning: array subscript is above array bounds [-Warray-bounds]
  linear = db_invert_table[i][j];

According to comment power_db should be in range 1 ~ 96 .
Correct rtw_phy_power_2_db() to make max power 96 db
(still min is 0). This make the warning gone.

However power >= 20 check still looks somewhat suspicious to me.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
---
 drivers/net/wireless/realtek/rtw88/phy.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/net/wireless/realtek/rtw88/phy.c b/drivers/net/wireless/realtek/rtw88/phy.c
index 35a35dbca85f..a716a44d78b0 100644
--- a/drivers/net/wireless/realtek/rtw88/phy.c
+++ b/drivers/net/wireless/realtek/rtw88/phy.c
@@ -410,12 +410,12 @@ void rtw_phy_dynamic_mechanism(struct rtw_dev *rtwdev)
 
 static u8 rtw_phy_power_2_db(s8 power)
 {
-	if (power <= -100 || power >= 20)
+	if (power <= -96 || power >= 20)
 		return 0;
 	else if (power >= 0)
-		return 100;
+		return 96;
 	else
-		return 100 + power;
+		return 96 + power;
 }
 
 static u64 rtw_phy_db_2_linear(u8 power_db)
-- 
1.9.3

RE: [RFC] rtw88: fix subscript above array bounds compiler warning

[Tony Chuang]

> Subject: [RFC] rtw88: fix subscript above array bounds compiler warning
> 
> My compiler complain about:
> 
> drivers/net/wireless/realtek/rtw88/phy.c: In function
> ‘rtw_phy_rf_power_2_rssi’:
> drivers/net/wireless/realtek/rtw88/phy.c:430:26: warning: array subscript is
> above array bounds [-Warray-bounds]
>   linear = db_invert_table[i][j];
> 
> According to comment power_db should be in range 1 ~ 96 .
> Correct rtw_phy_power_2_db() to make max power 96 db
> (still min is 0). This make the warning gone.
> 
> However power >= 20 check still looks somewhat suspicious to me.
> 
> Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
> ---
>  drivers/net/wireless/realtek/rtw88/phy.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/wireless/realtek/rtw88/phy.c
> b/drivers/net/wireless/realtek/rtw88/phy.c
> index 35a35dbca85f..a716a44d78b0 100644
> --- a/drivers/net/wireless/realtek/rtw88/phy.c
> +++ b/drivers/net/wireless/realtek/rtw88/phy.c
> @@ -410,12 +410,12 @@ void rtw_phy_dynamic_mechanism(struct rtw_dev
> *rtwdev)
> 
>  static u8 rtw_phy_power_2_db(s8 power)
>  {
> -	if (power <= -100 || power >= 20)
> +	if (power <= -96 || power >= 20)
>  		return 0;
>  	else if (power >= 0)
> -		return 100;
> +		return 96;
>  	else
> -		return 100 + power;
> +		return 96 + power;
>  }
> 
>  static u64 rtw_phy_db_2_linear(u8 power_db)
> --

I think I should check with the radio team, that if the power from the
rx descriptor generated by hardware can possibly get >= 20

And also check what the actual logic they expected to deal with the power.
Thanks for reporting it.

Yan-Hsuan

Re: [RFC] rtw88: fix subscript above array bounds compiler warning

[Stanislaw Gruszka]

On Fri, May 03, 2019 at 12:01:05PM +0000, Tony Chuang wrote:
> > Subject: [RFC] rtw88: fix subscript above array bounds compiler warning
> > 
> > My compiler complain about:
> > 
> > drivers/net/wireless/realtek/rtw88/phy.c: In function
> > ‘rtw_phy_rf_power_2_rssi’:
> > drivers/net/wireless/realtek/rtw88/phy.c:430:26: warning: array subscript is
> > above array bounds [-Warray-bounds]
> >   linear = db_invert_table[i][j];
> > 
> > According to comment power_db should be in range 1 ~ 96 .
> > Correct rtw_phy_power_2_db() to make max power 96 db
> > (still min is 0). This make the warning gone.
> > 
> > However power >= 20 check still looks somewhat suspicious to me.
> > 
> > Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
> > ---
> >  drivers/net/wireless/realtek/rtw88/phy.c | 6 +++---
> >  1 file changed, 3 insertions(+), 3 deletions(-)
> > 
> > diff --git a/drivers/net/wireless/realtek/rtw88/phy.c
> > b/drivers/net/wireless/realtek/rtw88/phy.c
> > index 35a35dbca85f..a716a44d78b0 100644
> > --- a/drivers/net/wireless/realtek/rtw88/phy.c
> > +++ b/drivers/net/wireless/realtek/rtw88/phy.c
> > @@ -410,12 +410,12 @@ void rtw_phy_dynamic_mechanism(struct rtw_dev
> > *rtwdev)
> > 
> >  static u8 rtw_phy_power_2_db(s8 power)
> >  {
> > -	if (power <= -100 || power >= 20)
> > +	if (power <= -96 || power >= 20)
> >  		return 0;
> >  	else if (power >= 0)
> > -		return 100;
> > +		return 96;
> >  	else
> > -		return 100 + power;
> > +		return 96 + power;
> >  }
> > 
> >  static u64 rtw_phy_db_2_linear(u8 power_db)
> > --
> 
> I think I should check with the radio team, that if the power from the
> rx descriptor generated by hardware can possibly get >= 20
> 
> And also check what the actual logic they expected to deal with the power.
> Thanks for reporting it.

Yeah, this could be just teoretical issue as we can not get power
values >= 0 from HW. However I think compiler correctly complains, as
for power_db=100 we get i = ((100 - 1) >> 3) = 12 , what exceed by one
max first index of db_invert_table[][], which should be in range
0 - 11.

Stanislaw

RE: [RFC] rtw88: fix subscript above array bounds compiler warning

[Tony Chuang]

> Subject: Re: [RFC] rtw88: fix subscript above array bounds compiler warning
> 
> On Fri, May 03, 2019 at 12:01:05PM +0000, Tony Chuang wrote:
> > > Subject: [RFC] rtw88: fix subscript above array bounds compiler warning
> > >
> > > My compiler complain about:
> > >
> > > drivers/net/wireless/realtek/rtw88/phy.c: In function
> > > ‘rtw_phy_rf_power_2_rssi’:
> > > drivers/net/wireless/realtek/rtw88/phy.c:430:26: warning: array subscript is
> > > above array bounds [-Warray-bounds]
> > >   linear = db_invert_table[i][j];
> > >
> > > According to comment power_db should be in range 1 ~ 96 .
> > > Correct rtw_phy_power_2_db() to make max power 96 db
> > > (still min is 0). This make the warning gone.
> > >
> > > However power >= 20 check still looks somewhat suspicious to me.
> > >
> > > Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
> > > ---
> > >  drivers/net/wireless/realtek/rtw88/phy.c | 6 +++---
> > >  1 file changed, 3 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/drivers/net/wireless/realtek/rtw88/phy.c
> > > b/drivers/net/wireless/realtek/rtw88/phy.c
> > > index 35a35dbca85f..a716a44d78b0 100644
> > > --- a/drivers/net/wireless/realtek/rtw88/phy.c
> > > +++ b/drivers/net/wireless/realtek/rtw88/phy.c
> > > @@ -410,12 +410,12 @@ void rtw_phy_dynamic_mechanism(struct
> rtw_dev
> > > *rtwdev)
> > >
> > >  static u8 rtw_phy_power_2_db(s8 power)
> > >  {
> > > -	if (power <= -100 || power >= 20)
> > > +	if (power <= -96 || power >= 20)
> > >  		return 0;
> > >  	else if (power >= 0)
> > > -		return 100;
> > > +		return 96;
> > >  	else
> > > -		return 100 + power;
> > > +		return 96 + power;
> > >  }
> > >
> > >  static u64 rtw_phy_db_2_linear(u8 power_db)
> > > --
> >
> > I think I should check with the radio team, that if the power from the
> > rx descriptor generated by hardware can possibly get >= 20
> >
> > And also check what the actual logic they expected to deal with the power.
> > Thanks for reporting it.
> 
> Yeah, this could be just teoretical issue as we can not get power
> values >= 0 from HW. However I think compiler correctly complains, as
> for power_db=100 we get i = ((100 - 1) >> 3) = 12 , what exceed by one
> max first index of db_invert_table[][], which should be in range
> 0 - 11.
> 

I checked it. The power sum could actually be like 20 or something.
And the recommended modification is to restrict the value used for array
subscript between 1~96 before access the array. Such as:

@@ -578,6 +578,11 @@ static u64 rtw_phy_db_2_linear(u8 power_db)
        u8 i, j;
        u64 linear;

+       if (power_db > 96)
+               power_db = 96;
+       else if (power_db < 1)
+               return 1;
+
        /* 1dB ~ 96dB */
        i = (power_db - 1) >> 3;
        j = (power_db - 1) - (i << 3);


Yan-Hsuan