Re: [PATCH RESEND] staging: rtl8192u: ieee80211: Fix sparse endianness warnings

[Gaston Gonzalez]

 On Mon, Jun 01, 2015 at 08:24:17AM +0900, Greg KH wrote:
> On Sun, May 31, 2015 at 08:07:21PM -0300, Gaston Gonzalez wrote:
> > On Mon, Jun 01, 2015 at 07:17:39AM +0900, Greg KH wrote:
> > > On Sun, May 31, 2015 at 12:17:48PM -0300, Gaston Gonzalez wrote:
> > > > Fix the following sparse warnings:
> > > > 
> > > > drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:663:32: warning: incorrect type in assignment (different base types)
> > > > drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:663:32:    expected restricted __le16 [usertype] frame_ctl
> > > > drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:663:32:    got int
> > > > drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:664:50: warning: invalid assignment: |=
> > > > drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:664:50:    left side has type restricted __le16
> > > > drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c:664:50:    right side has type int
> > > > 
> > > > 
> > > > Signed-off-by: Gaston Gonzalez <gascoar@xxxxxxxxx>
> > > > ---
> > > >  drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c | 5 +++--
> > > >  1 file changed, 3 insertions(+), 2 deletions(-)
> > > 
> > > Still doesn't apply, but this might be a git bug, see below:
> > > 
> > > > diff --git a/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c b/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
> > > > index d2e8b12..0477ba1 100644
> > > > --- a/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
> > > > +++ b/drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
> > > > @@ -660,2 +660,2 @@ inline struct sk_buff *ieee80211_authentication_req(struct ieee80211_network *be
> > > >  	auth = (struct ieee80211_authentication *)
> > > >  		skb_put(skb, sizeof(struct ieee80211_authentication));
> > > > 
> > > > -	auth->header.frame_ctl = IEEE80211_STYPE_AUTH;
> > > > -	if (challengelen) auth->header.frame_ctl |= IEEE80211_FCTL_WEP;
> > > > +	auth->header.frame_ctl = cpu_to_le16(IEEE80211_STYPE_AUTH);
> > > > +	if (challengelen)
> > > > +		auth->header.frame_ctl |= cpu_to_le16(IEEE80211_FCTL_WEP);
> > > > 
> > > >  	auth->header.duration_id = 0x013a; //FIXME
> > > > 
> > > > --
> > > > 2.1.4
> > > 
> > > This patch is not created properly.  Did you edit it by hand?  If not,
> > > we need to report a bug as the diff is obviously wrong, look at the @@@
> > > line.
> > > 
> > > Can you update your version of git to see if it's fixed there?
> > > 
> > > Also, patch tells you that this is broken:
> > > 
> > > checking file drivers/staging/rtl8192u/ieee80211/ieee80211_softmac.c
> > > patch: **** malformed patch at line 132:                skb_put(skb, sizeof(struct ieee80211_authentication));
> > 
> > Greg,
> > 
> > The patch was prepared using git format-patch. The git version I'm using:
> > 
> > gaston22:\-$ git --version 
> > git version 2.1.4
> > 
> > I did deleted 4 whitespaces by hand after running checkpatch.pl on the patch.
> > Could that drive the issue? However, I could properly apply the patch using 'git
> > am' from an email and get the git log right.
> 
> Yes, but is there really a commit there?  That's what messed me up last
> time, git applied it, but with no actual diff (which is probably a bug
> too, just a different one...)
> 

Yes, there was a commit. In fact the patch prepare: git format-patch -1

> And if you edited it by hand, that might explain the issue, what did the
> "original" patch look like?

Attached the original patch I submitted:
0001-Regen-ieee80211_softmac.c-endianness-fix-patch.patch

Attached is also the patch without the hand-edited deletions:

ieee80211_softmac_fixes_v3.patch

My steps once having this were:

- run perl checkpatch -f 
- I got total: 4 errors, 0 warnings, 28 lines checked. The 4 warnings were extra white
  spaces. 
- Delete the spaces
- git send-email --annotate Patch

> 
> > I'll try with a newer version of git as you suggest.
> > Any other check I could do before the submission? I really don't like making
> > noise :-(
> 
> Let's see the original version first before blaming git.  Odds are you
> got something wrong when editing the patch by hand.
> 

Ok.
> thanks,
> 
> greg k-h

thanks,

Gaston