[PATCH iproute2] lib/libnetlink: fix response seq check

[PATCH iproute2] lib/libnetlink: fix response seq check

[vlad]

From: Vlad Dumitrescu <vladum@google.com>

Taking a one-iovec example, with rtnl->seq at 42. iovlen == 1, seq
becomes 43 on line 604, and a message is sent with nlmsg_seq == 43. If
a response with nlmsg_seq of 42 is received, the condition being fixed
in this patch would incorrectly accept it.

Fixes: 72a2ff3916e5 ("lib/libnetlink: Add a new function rtnl_talk_iov")
Signed-off-by: Vlad Dumitrescu <vladum@google.com>
---
 lib/libnetlink.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/libnetlink.c b/lib/libnetlink.c
index f18dceac..4d2416bf 100644
--- a/lib/libnetlink.c
+++ b/lib/libnetlink.c
@@ -647,7 +647,7 @@ static int __rtnl_talk_iov(struct rtnl_handle *rtnl, struct iovec *iov,
 
 			if (nladdr.nl_pid != 0 ||
 			    h->nlmsg_pid != rtnl->local.nl_pid ||
-			    h->nlmsg_seq > seq || h->nlmsg_seq < seq - iovlen) {
+			    h->nlmsg_seq > seq || h->nlmsg_seq < seq - iovlen + 1) {
 				/* Don't forget to skip that message. */
 				status -= NLMSG_ALIGN(len);
 				h = (struct nlmsghdr *)((char *)h + NLMSG_ALIGN(len));
-- 
2.19.0.605.g01d371f741-goog

Re: [PATCH iproute2] lib/libnetlink: fix response seq check

[Vlad Dumitrescu]

Hi,

On Fri, Sep 28, 2018 at 10:14 AM <vlad@dumitrescu.ro> wrote:
>
> From: Vlad Dumitrescu <vladum@google.com>
>
> Taking a one-iovec example, with rtnl->seq at 42. iovlen == 1, seq
> becomes 43 on line 604, and a message is sent with nlmsg_seq == 43. If
> a response with nlmsg_seq of 42 is received, the condition being fixed
> in this patch would incorrectly accept it.
>
> Fixes: 72a2ff3916e5 ("lib/libnetlink: Add a new function rtnl_talk_iov")
> Signed-off-by: Vlad Dumitrescu <vladum@google.com>
> ---
>  lib/libnetlink.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/lib/libnetlink.c b/lib/libnetlink.c
> index f18dceac..4d2416bf 100644
> --- a/lib/libnetlink.c
> +++ b/lib/libnetlink.c
> @@ -647,7 +647,7 @@ static int __rtnl_talk_iov(struct rtnl_handle *rtnl, struct iovec *iov,
>
>                         if (nladdr.nl_pid != 0 ||
>                             h->nlmsg_pid != rtnl->local.nl_pid ||
> -                           h->nlmsg_seq > seq || h->nlmsg_seq < seq - iovlen) {
> +                           h->nlmsg_seq > seq || h->nlmsg_seq < seq - iovlen + 1) {
>                                 /* Don't forget to skip that message. */
>                                 status -= NLMSG_ALIGN(len);
>                                 h = (struct nlmsghdr *)((char *)h + NLMSG_ALIGN(len));
> --
> 2.19.0.605.g01d371f741-goog

Did anybody get a chance to review this? I'm not 100% sure I'm fixing
the right thing.

Thanks,
Vlad

Re: [PATCH iproute2] lib/libnetlink: fix response seq check

[Stephen Hemminger]

On Wed, 3 Oct 2018 16:01:40 -0700
Vlad Dumitrescu <vlad@dumitrescu.ro> wrote:

> Hi,
> 
> On Fri, Sep 28, 2018 at 10:14 AM <vlad@dumitrescu.ro> wrote:
> >
> > From: Vlad Dumitrescu <vladum@google.com>
> >
> > Taking a one-iovec example, with rtnl->seq at 42. iovlen == 1, seq
> > becomes 43 on line 604, and a message is sent with nlmsg_seq == 43. If
> > a response with nlmsg_seq of 42 is received, the condition being fixed
> > in this patch would incorrectly accept it.
> >
> > Fixes: 72a2ff3916e5 ("lib/libnetlink: Add a new function rtnl_talk_iov")
> > Signed-off-by: Vlad Dumitrescu <vladum@google.com>
> > ---
> >  lib/libnetlink.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/lib/libnetlink.c b/lib/libnetlink.c
> > index f18dceac..4d2416bf 100644
> > --- a/lib/libnetlink.c
> > +++ b/lib/libnetlink.c
> > @@ -647,7 +647,7 @@ static int __rtnl_talk_iov(struct rtnl_handle *rtnl, struct iovec *iov,
> >
> >                         if (nladdr.nl_pid != 0 ||
> >                             h->nlmsg_pid != rtnl->local.nl_pid ||
> > -                           h->nlmsg_seq > seq || h->nlmsg_seq < seq - iovlen) {
> > +                           h->nlmsg_seq > seq || h->nlmsg_seq < seq - iovlen + 1) {
> >                                 /* Don't forget to skip that message. */
> >                                 status -= NLMSG_ALIGN(len);
> >                                 h = (struct nlmsghdr *)((char *)h + NLMSG_ALIGN(len));
> > --
> > 2.19.0.605.g01d371f741-goog  
> 
> Did anybody get a chance to review this? I'm not 100% sure I'm fixing
> the right thing.
> 
> Thanks,
> Vlad

Could you give an example where this failed.
Better yet one of the tests.