Re: signed vs unsigned bug in ath9k

Re: signed vs unsigned bug in ath9k

[Dan Carpenter]

On Mon, Dec 07, 2009 at 07:55:42AM -0800, Luis R. Rodriguez wrote:
> On Mon, Dec 07, 2009 at 03:07:36AM -0800, Dan Carpenter wrote:
> > drivers/net/wireless/ath/ath9k/recv.c
> >    205          sta = ieee80211_find_sta(sc->hw, hdr->addr2);
> >    206          if (sta) {
> >    207                  an = (struct ath_node *) sta->drv_priv;
> >    208                  if (ds->ds_rxstat.rs_rssi != ATH9K_RSSI_BAD &&
> > 
> > ds->ds_rxstat.rs_rssi is a signed 8 bit so it will never be == ATH9K_RSSI_BAD.
> 
> ATH9K_RSSI_BAD is -128 and the minimum value for int8_t is -128 so why is it a bug?
> 

It could be that someone fixed this already in the net tree?  In mainline it's still
positive 128.

#define ATH9K_RSSI_BAD                  0x80

regards,
dan carpenter

> >    209                     !ds->ds_rxstat.rs_moreaggr)
> >    210                          ATH_RSSI_LPF(an->last_rssi, ds->ds_rxstat.rs_rssi);
> >    211                  last_rssi = an->last_rssi;
> >    212          }
> > 
> > I would normally just change the declaration to unsigned but it looks like
> > someone may have chosen to have it signed on purpose.
> 
> Yeah it comes from hardware.
> 
>   Luis

Re: signed vs unsigned bug in ath9k

[John W. Linville]

On Tue, Dec 08, 2009 at 10:20:32AM +0200, Dan Carpenter wrote:
> On Mon, Dec 07, 2009 at 07:55:42AM -0800, Luis R. Rodriguez wrote:
> > On Mon, Dec 07, 2009 at 03:07:36AM -0800, Dan Carpenter wrote:
> > > drivers/net/wireless/ath/ath9k/recv.c
> > >    205          sta = ieee80211_find_sta(sc->hw, hdr->addr2);
> > >    206          if (sta) {
> > >    207                  an = (struct ath_node *) sta->drv_priv;
> > >    208                  if (ds->ds_rxstat.rs_rssi != ATH9K_RSSI_BAD &&
> > > 
> > > ds->ds_rxstat.rs_rssi is a signed 8 bit so it will never be == ATH9K_RSSI_BAD.
> > 
> > ATH9K_RSSI_BAD is -128 and the minimum value for int8_t is -128 so why is it a bug?
> > 
> 
> It could be that someone fixed this already in the net tree?  In mainline it's still
> positive 128.
> 
> #define ATH9K_RSSI_BAD                  0x80

I queued a patch to fix that for 2.6.33.

Hth!

John
-- 
John W. Linville		Someday the world will need a hero, and you
linville@tuxdriver.com			might be all we have.  Be ready.

Re: signed vs unsigned bug in ath9k

[Dan Carpenter]

On Tue, Dec 08, 2009 at 09:42:10AM -0500, John W. Linville wrote:
> On Tue, Dec 08, 2009 at 10:20:32AM +0200, Dan Carpenter wrote:
> > On Mon, Dec 07, 2009 at 07:55:42AM -0800, Luis R. Rodriguez wrote:
> > > On Mon, Dec 07, 2009 at 03:07:36AM -0800, Dan Carpenter wrote:
> > > > drivers/net/wireless/ath/ath9k/recv.c
> > > >    205          sta = ieee80211_find_sta(sc->hw, hdr->addr2);
> > > >    206          if (sta) {
> > > >    207                  an = (struct ath_node *) sta->drv_priv;
> > > >    208                  if (ds->ds_rxstat.rs_rssi != ATH9K_RSSI_BAD &&
> > > > 
> > > > ds->ds_rxstat.rs_rssi is a signed 8 bit so it will never be == ATH9K_RSSI_BAD.
> > > 
> > > ATH9K_RSSI_BAD is -128 and the minimum value for int8_t is -128 so why is it a bug?
> > > 
> > 
> > It could be that someone fixed this already in the net tree?  In mainline it's still
> > positive 128.
> > 
> > #define ATH9K_RSSI_BAD                  0x80
> 
> I queued a patch to fix that for 2.6.33.
> 

Grand.  Sorry for the noise.  We static checker people are all finding the same
bugs...

I have been trying to buy enough wifi cards to download the latest git, but 
apparently the English gentleman who generates them has been on a week
long drinking binge.  I have been here asking about it on five days out of the 
last seven...  :/

regards,
dan carpenter

> Hth!
> 
> John
> -- 
> John W. Linville		Someday the world will need a hero, and you
> linville@tuxdriver.com			might be all we have.  Be ready.