Michael Buesch wrote:
> On Sunday 05 August 2007, Larry Finger wrote:
>> The link quality output from wireless extensions is too small by the ratio
>> of 100/BCM43xx_RX_MAX_SSI (60) for bcm43xx-mac80211. This patch puts the
>> quantity on the proper scale.
>> Signed-off-by: Larry Finger <[EMAIL PROTECTED]>
>> ---
>> Index: wireless-dev/drivers/net/wireless/bcm43xx-mac80211/bcm43xx_xmit.c
>> ===================================================================
>> --- wireless-dev.orig/drivers/net/wireless/bcm43xx-mac80211/bcm43xx_xmit.c
>> +++ wireless-dev/drivers/net/wireless/bcm43xx-mac80211/bcm43xx_xmit.c
>> @@ -537,7 +537,8 @@ void bcm43xx_rx(struct bcm43xx_wldev *de
>>                                            (phystat0 & 
>>                                            (phystat3 & 
>>      status.noise = dev->stats.link_noise;
>> -    status.signal = jssi; /* this looks wrong, but is what mac80211 wants */
>> +    /* the next line looks wrong, but is what mac80211 wants */
>> +    status.signal = (jssi * 100) / BCM43xx_RX_MAX_SSI;
> So signal is in percent?
> Where is this actually documented. I cannot find a hint on what
> the values of all these things are supposed to be.

Yes, it is clear as mud, with the additional complications of mac80211 mixing 
the definitions of 
signal and rssi (as far as I'm concerned). The scale is set by the following 
code snippet in 

         hw->max_signal = 100;
         hw->max_rssi = -110;
         hw->max_noise = -110;

In this code, "signal" is put on a scale of 0 to 100, and rssi and noise on a 
scale of -110 to 0 and 
are assumed to be dBm. Of course, rssi should be a positive number and signal 
should be in dBm, but 
my renaming of signal => quality and rssi => signal was shot down, so we are 

An alternative to the patch above would be to set hw->max_signal = 
BCM43xx_RX_MAX_SSI. In that case, 
the line of iwconfig output that reads "Link Quality=83/100  Signal level=-34 
dBm  Noise level=-71 
dBm" would have a "Link Quality" of 50/60 instead of 83/100.

The bottom line is that it is an arbitrary quantity on an arbitrary scale. Is 
it better for it to be 
  XX/100 than YY/60? I think so, but YMMV.

Bcm43xx-dev mailing list

Reply via email to