[PATCH] iwd: Use same signal strength calculation as wpa_supplicant

[PATCH] iwd: Use same signal strength calculation as wpa_supplicant

[Jonathan Liu]

Fixes the signal strength reported by connman being lower when using iwd
compared to wpa_supplicant.
---
 plugins/iwd.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/plugins/iwd.c b/plugins/iwd.c
index ac3d1e17..2a245fe2 100644
--- a/plugins/iwd.c
+++ b/plugins/iwd.c
@@ -1128,7 +1128,9 @@ static unsigned char calculate_strength(int strength)
 	 * ConnMan expects it in the range from 100 (strongest) to 0
 	 * (weakest).
 	 */
-	res = (unsigned char)((strength + 10000) / 100);
+	res = (unsigned char)(120 + strength / 100);
+	if (res > 100)
+		res = 100;
 
 	return res;
 }
-- 
2.35.1

Re: [PATCH] iwd: Use same signal strength calculation as wpa_supplicant

[Marcel Holtmann]

Hi Jonathan,

> Fixes the signal strength reported by connman being lower when using iwd
> compared to wpa_supplicant.
> ---
> plugins/iwd.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/plugins/iwd.c b/plugins/iwd.c
> index ac3d1e17..2a245fe2 100644
> --- a/plugins/iwd.c
> +++ b/plugins/iwd.c
> @@ -1128,7 +1128,9 @@ static unsigned char calculate_strength(int strength)
> 	 * ConnMan expects it in the range from 100 (strongest) to 0
> 	 * (weakest).
> 	 */
> -	res = (unsigned char)((strength + 10000) / 100);
> +	res = (unsigned char)(120 + strength / 100);
> +	if (res > 100)
> +		res = 100;

I think it would be better to fix wpa_supplicant plugin to adjust to iwd. Since the iwd signal strength calculation should not be seconded guessed. That will get you into trouble at some point.

Regards

Marcel

Re: [PATCH] iwd: Use same signal strength calculation as wpa_supplicant

[Daniel Wagner]

Hi Jonathan,

On Mon, Feb 14, 2022 at 01:13:57PM +1100, Jonathan Liu wrote:
> Fixes the signal strength reported by connman being lower when using iwd
> compared to wpa_supplicant.
> ---
>  plugins/iwd.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/plugins/iwd.c b/plugins/iwd.c
> index ac3d1e17..2a245fe2 100644
> --- a/plugins/iwd.c
> +++ b/plugins/iwd.c
> @@ -1128,7 +1128,9 @@ static unsigned char calculate_strength(int strength)
>  	 * ConnMan expects it in the range from 100 (strongest) to 0
>  	 * (weakest).
>  	 */
> -	res = (unsigned char)((strength + 10000) / 100);
> +	res = (unsigned char)(120 + strength / 100);
> +	if (res > 100)
> +		res = 100;

Can you explain your math here? The original code maps the reported
signal strength linearly to the ConnMan's range.

Daniel

Re: [PATCH] iwd: Use same signal strength calculation as wpa_supplicant

[Jonathan Liu]

Hi Daniel,

It is the same mapping of dBm to 0-100 range as
https://git.kernel.org/pub/scm/network/connman/connman.git/tree/plugins/wifi.c#n2805
For wpa_supplicant,
g_supplicant_network_get_signal(supplicant_network) returns the dBm
value.
For iwd, you need to divide the number by 100 to get dBm value.

Regards,
Jonathan

On Mon, 14 Feb 2022 at 18:36, Daniel Wagner <wagi@monom.org> wrote:
>
> Hi Jonathan,
>
> On Mon, Feb 14, 2022 at 01:13:57PM +1100, Jonathan Liu wrote:
> > Fixes the signal strength reported by connman being lower when using iwd
> > compared to wpa_supplicant.
> > ---
> >  plugins/iwd.c | 4 +++-
> >  1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/plugins/iwd.c b/plugins/iwd.c
> > index ac3d1e17..2a245fe2 100644
> > --- a/plugins/iwd.c
> > +++ b/plugins/iwd.c
> > @@ -1128,7 +1128,9 @@ static unsigned char calculate_strength(int strength)
> >        * ConnMan expects it in the range from 100 (strongest) to 0
> >        * (weakest).
> >        */
> > -     res = (unsigned char)((strength + 10000) / 100);
> > +     res = (unsigned char)(120 + strength / 100);
> > +     if (res > 100)
> > +             res = 100;
>
> Can you explain your math here? The original code maps the reported
> signal strength linearly to the ConnMan's range.
>
> Daniel
>

Re: [PATCH] iwd: Use same signal strength calculation as wpa_supplicant

[Marcel Holtmann]

Hi Jonathan,

> It is the same mapping of dBm to 0-100 range as
> https://git.kernel.org/pub/scm/network/connman/connman.git/tree/plugins/wifi.c#n2805
> For wpa_supplicant,
> g_supplicant_network_get_signal(supplicant_network) returns the dBm
> value.
> For iwd, you need to divide the number by 100 to get dBm value.

from iwd documentation.

                        int16 SignalStrength                                     
                                                                                 
                                Network's maximum signal strength expressed
                                in 100 * dBm.  The value is the range of 0
                                (strongest signal) to -10000 (weakest signal)

Anyway, the signal strength in % expression is a total mess since every hardware reports different value and none of them are normalized.

It is better to fix the GSupplicant code to do exactly what iwd does.

Regards

Marcel

Re: [PATCH] iwd: Use same signal strength calculation as wpa_supplicant

[Daniel Wagner]

On Mon, Feb 14, 2022 at 10:31:03PM +1100, Jonathan Liu wrote:
> It is the same mapping of dBm to 0-100 range as
> https://git.kernel.org/pub/scm/network/connman/connman.git/tree/plugins/wifi.c#n2805
> For wpa_supplicant,
> g_supplicant_network_get_signal(supplicant_network) returns the dBm
> value.
> For iwd, you need to divide the number by 100 to get dBm value.

Please add this information to the commit message.

Daniel

Re: [PATCH] iwd: Use same signal strength calculation as wpa_supplicant

[Jonathan Liu]

Hi Daniel and Marcel,

On Tue, 15 Feb 2022 at 02:23, Daniel Wagner <wagi@monom.org> wrote:
>
> On Mon, Feb 14, 2022 at 10:31:03PM +1100, Jonathan Liu wrote:
> > It is the same mapping of dBm to 0-100 range as
> > https://git.kernel.org/pub/scm/network/connman/connman.git/tree/plugins/wifi.c#n2805
> > For wpa_supplicant,
> > g_supplicant_network_get_signal(supplicant_network) returns the dBm
> > value.
> > For iwd, you need to divide the number by 100 to get dBm value.
>
> Please add this information to the commit message.

Can we first reach a consensus about whether to
1. Change the iwd signal calculation to match wpa_supplicant
2. Change the wpa_supplicant signal calculation to match iwd
3. Do nothing to either plugin to avoid changing behavior of existing
devices/applications or Linux distributions using connman

If we change the wpa_supplicant calculation then I imagine people
would complain about reduced signal strength after updating (e.g. in
OSMC and many other devices/applications or Linux distributions using
connman).

The wpa_supplicant calculation regards a signal level of -20 dBm or
higher as 100%.
The iwd calculation regards a signal level of -20 dBm as 80% and 0 dBm as 100%.

Thanks.

Regards,
Jonathan

Re: [PATCH] iwd: Use same signal strength calculation as wpa_supplicant

[Daniel Wagner]

On Tue, Feb 15, 2022 at 08:41:35AM +1100, Jonathan Liu wrote:
> Can we first reach a consensus about whether to
> 1. Change the iwd signal calculation to match wpa_supplicant

Yes.

> 2. Change the wpa_supplicant signal calculation to match iwd
> 3. Do nothing to either plugin to avoid changing behavior of existing
> devices/applications or Linux distributions using connman

No, people will complain and think it's an regression.

> If we change the wpa_supplicant calculation then I imagine people
> would complain about reduced signal strength after updating (e.g. in
> OSMC and many other devices/applications or Linux distributions using
> connman).

For sure.

> The wpa_supplicant calculation regards a signal level of -20 dBm or
> higher as 100%.
> The iwd calculation regards a signal level of -20 dBm as 80% and 0 dBm as 100%.

The signal strength is used to sort the services internally. IIRC, this
might trigger the connect algorithm for wpa_supplicant but for iwd it
shouldn't as we rely on iwd to do the right thing (see the 'native'
connect mode). So we should not change the wpa_supplicant signal
strenght math (check the git history on this topic).

Daniel