[PATCH] krb5: fix warning variable may be used uninitialized

[Lukas Slebodnik]

ehlo,

scan-build found a warning.

src/providers/krb5/krb5_child.c:1988:9: error: ‘offline’ may be used
uninitialized in this function [-Werror=maybe-uninitialized]
     ret = k5c_setup(kr, offline);

gcc also reports this warning with enabled optimization >= "-O2".

It seems to me really weird, because compilator assumes that function
sss_atomic_read_s return value -1 and errno==0.
I tried to modify function sss_atomic_read_s, but I could not fix this warning.

Another possible solution is to initialize variable "offline" to some default
value.

I appreciate any comment.

Patch is attached.

LS

[Jakub Hrozek]

On Thu, Oct 10, 2013 at 12:09:17PM +0200, Lukas Slebodnik wrote:
> ehlo,
>
> scan-build found a warning.
>
> src/providers/krb5/krb5_child.c:1988:9: error: ‘offline’ may be used
> uninitialized in this function [-Werror=maybe-uninitialized]
>      ret = k5c_setup(kr, offline);
>
> gcc also reports this warning with enabled optimization >= "-O2".
>
> It seems to me really weird, because compilator assumes that function
> sss_atomic_read_s return value -1 and errno==0.
> I tried to modify function sss_atomic_read_s, but I could not fix this warning.
>
> Another possible solution is to initialize variable "offline" to some default
> value.
>
> I appreciate any comment.
>
> Patch is attached.
>
> LS

I think this is fine, the fix would protect us from a potential bug in
sss_atomic_write. Maybe even a better fix would be to modify
sss_atomic_write to return errno as a return value and the number of
bytes only on success?

But ticket #2112 is a different issue, we should remove the reference to
it from the commit message.

[Jakub Hrozek]

On Thu, Oct 10, 2013 at 05:00:14PM +0200, Jakub Hrozek wrote:
> On Thu, Oct 10, 2013 at 12:09:17PM +0200, Lukas Slebodnik wrote:
> > ehlo,
> >
> > scan-build found a warning.
> >
> > src/providers/krb5/krb5_child.c:1988:9: error: ‘offline’ may be used
> > uninitialized in this function [-Werror=maybe-uninitialized]
> >      ret = k5c_setup(kr, offline);
> >
> > gcc also reports this warning with enabled optimization >= "-O2".
> >
> > It seems to me really weird, because compilator assumes that function
> > sss_atomic_read_s return value -1 and errno==0.
> > I tried to modify function sss_atomic_read_s, but I could not fix this warning.
> >
> > Another possible solution is to initialize variable "offline" to some default
> > value.
> >
> > I appreciate any comment.
> >
> > Patch is attached.
> >
> > LS
>
> I think this is fine, the fix would protect us from a potential bug in
> sss_atomic_write. Maybe even a better fix would be to modify
> sss_atomic_write to return errno as a return value and the number of
> bytes only on success?
>
> But ticket #2112 is a different issue, we should remove the reference to
> it from the commit message.

The patch fixes a compilation warning so ACK and pushed to master.