From: Stefano Brivio <sbri...@redhat.com>
Date: Fri, 25 Aug 2017 11:02:06 +0200
> On Fri, 25 Aug 2017 09:52:17 +0200
> Steffen Klassert <steffen.klass...@secunet.com> wrote:
>
>> On Fri, Aug 25, 2017 at 09:05:42AM +0200, Steffen Klassert wrote:
>> > rt_cookie might be used uninitialized, fix this by
>> > initializing it.
>> >
>> > Fixes: c5cff8561d2d ("ipv6: add rcu grace period before freeing fib6_node")
>> > Signed-off-by: Steffen Klassert <steffen.klass...@secunet.com>
>> > ---
>> > net/ipv6/route.c | 2 +-
>> > 1 file changed, 1 insertion(+), 1 deletion(-)
>> >
>> > diff --git a/net/ipv6/route.c b/net/ipv6/route.c
>> > index a9d3564..48c8c92 100644
>> > --- a/net/ipv6/route.c
>> > +++ b/net/ipv6/route.c
>> > @@ -1289,7 +1289,7 @@ static void rt6_dst_from_metrics_check(struct
>> > rt6_info *rt)
>> >
>> > static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie)
>> > {
>> > - u32 rt_cookie;
>> > + u32 rt_cookie = 0;
>> >
>> > if (!rt6_get_cookie_safe(rt, &rt_cookie) || rt_cookie != cookie)
>> > return NULL;
>>
>> The compiler warning seems to be a false positive, as
>> rt_cookie != cookie is only checked if rt6_get_cookie_safe
>> returns true in which case rt_cookie is initialized.
>>
>> Please disregard this patch.
>
> ...or not? I was thinking of sending a similar patch with
> uninitialized_var(rt_cookie), but it seems we have similar cases
> where we just initialize to zero instead.
>
> I wonder which approach is considered the most acceptable nowadays. I
> would be in favour of uninitialized_var() as it doesn't change the
> binary output, but https://lwn.net/Articles/529954/ also contains some
> valid criticism. Ideas?
Generally speaking I guess initializing to zero is Ok to do.
As far as which approach is better, I don't have any strong opinion.
So I will probably just apply Steffen's patch.
