Re: [PATCH] Use strcmp instead of strncmp

[Kurt Van Dijck <kurt.van.dijck@eia.be>]

On Wed, Nov 14, 2012 at 10:07:15AM +0100, Alexander Stein wrote:
> Hello,
> 
> On Wednesday 14 November 2012 09:55:37, Kurt Van Dijck wrote:
> > On Tue, Nov 13, 2012 at 09:26:50PM +0100, Marc Kleine-Budde wrote:
> > > On 11/13/2012 04:47 PM, Alexander Stein wrote:
> > > > This bug was detected by the clang warning:
> > > > libsocketcan.c:384:16: warning: argument to 'sizeof' in 'strncmp' call
> > > > is the same expression as the source; did you mean to provide an
> > > > explicit length? [-Wsizeof-pointer-memaccess]
> > > > sizeof(name)) != 0)
> > > > ~~~~~~~^~~~~~
> > strlen() requires a null terminator. Is that always present?
> > Unfortunately, I've no access to the libsocketcan code today...
> > > > 
> > > > Signed-off-by: Alexander Stein <alexander.stein@systec-electronic.com>
> > > > ---
> > > >  src/libsocketcan.c |    3 ++-
> > > >  1 files changed, 2 insertions(+), 1 deletions(-)
> > > > 
> > > > diff --git a/src/libsocketcan.c b/src/libsocketcan.c
> > > > index fedcbdc..3ad2a6c 100644
> > > > --- a/src/libsocketcan.c
> > > > +++ b/src/libsocketcan.c
> > > > @@ -364,6 +364,7 @@ static int do_get_nl_link(int fd, __u8 acquire, 
> const char *name, void *res)
> > > >  		     nl_msg = NLMSG_NEXT(nl_msg, u_msglen)) {
> > > >  			int type = nl_msg->nlmsg_type;
> > > >  			int len;
> > > > +			int namelen = strlen(name);
> > > >  
> > > >  			if (type == NLMSG_DONE) {
> > > >  				done++;
> > > > @@ -381,7 +382,7 @@ static int do_get_nl_link(int fd, __u8 acquire, 
> const char *name, void *res)
> > > >  
> > > >  			if (strncmp
> > > >  			    ((char *)RTA_DATA(tb[IFLA_IFNAME]), name,
> > > > -			     sizeof(name)) != 0)
> > > > +			     namelen) != 0)
> > > 
> > > What about using strcmp() instead. AFAIK it will compare until it finds
> > > the end of the string in one of the strings, so it should be no
> > > difference here?
> > if name contains a null terminator, using strcmp() is even better.
> 
> name is passed directly from public API functions like can_get_state. So the 
> user has to provide the null terminator. As this cannot be garuanteed 
I think we should assume that the user passes a null terminated string.
Otherwise, we must question if the user passes a valid address. I don't
like to go that way.

strcmp() will just work.

Kind regards,
Kurt