* [PATCH] netfilter: nf_tables: prevent shift wrap in nft_chain_parse_hook()
@ 2019-04-02 13:30 Dan Carpenter
2019-04-05 8:34 ` Dan Carpenter
2019-04-06 5:26 ` [PATCH v2] " Dan Carpenter
0 siblings, 2 replies; 5+ messages in thread
From: Dan Carpenter @ 2019-04-02 13:30 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: Jozsef Kadlecsik, Florian Westphal, netfilter-devel, coreteam,
kernel-janitors
I believe that "hook->num" can be up to UINT_MAX so the shift has to
be unsigned or shifting more than 31 is undefined. It's possible that
this leads to array overflow in nf_tables_addchain():
ops->hook = hook.type->hooks[ops->hooknum];
Fixes: fe19c04ca137 ("netfilter: nf_tables: remove nhooks field from struct nft_af_info")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
net/netfilter/nf_tables_api.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index ef7772e976cc..b6e48fe7c776 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -1545,7 +1545,7 @@ static int nft_chain_parse_hook(struct net *net,
if (IS_ERR(type))
return PTR_ERR(type);
}
- if (!(type->hook_mask & (1 << hook->num)))
+ if (!(type->hook_mask & (1U << hook->num)))
return -EOPNOTSUPP;
if (type->type == NFT_CHAIN_T_NAT &&
--
2.17.1
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH] netfilter: nf_tables: prevent shift wrap in nft_chain_parse_hook()
2019-04-02 13:30 [PATCH] netfilter: nf_tables: prevent shift wrap in nft_chain_parse_hook() Dan Carpenter
@ 2019-04-05 8:34 ` Dan Carpenter
2019-04-05 10:01 ` Florian Westphal
2019-04-06 5:26 ` [PATCH v2] " Dan Carpenter
1 sibling, 1 reply; 5+ messages in thread
From: Dan Carpenter @ 2019-04-05 8:34 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: Jozsef Kadlecsik, Florian Westphal, netfilter-devel, coreteam,
kernel-janitors
On Tue, Apr 02, 2019 at 04:30:38PM +0300, Dan Carpenter wrote:
> I believe that "hook->num" can be up to UINT_MAX so the shift has to
> be unsigned or shifting more than 31 is undefined. It's possible that
> this leads to array overflow in nf_tables_addchain():
>
> ops->hook = hook.type->hooks[ops->hooknum];
>
> Fixes: fe19c04ca137 ("netfilter: nf_tables: remove nhooks field from struct nft_af_info")
> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
> ---
> net/netfilter/nf_tables_api.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
> index ef7772e976cc..b6e48fe7c776 100644
> --- a/net/netfilter/nf_tables_api.c
> +++ b/net/netfilter/nf_tables_api.c
> @@ -1545,7 +1545,7 @@ static int nft_chain_parse_hook(struct net *net,
> if (IS_ERR(type))
> return PTR_ERR(type);
> }
> - if (!(type->hook_mask & (1 << hook->num)))
> + if (!(type->hook_mask & (1U << hook->num)))
Ugh... No this doesn't work. It's still undefined behavoir even if we
make it unsigned. It has to be:
if (hook->num > 31 || !(type->hook_mask & (1 << hook->num)))
which seems like a slightly ugly magic number but it fixes the buffer
overflow.
regards,
dan carpenter
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH v2] netfilter: nf_tables: prevent shift wrap in nft_chain_parse_hook()
2019-04-02 13:30 [PATCH] netfilter: nf_tables: prevent shift wrap in nft_chain_parse_hook() Dan Carpenter
2019-04-05 8:34 ` Dan Carpenter
@ 2019-04-06 5:26 ` Dan Carpenter
2019-04-13 12:58 ` Pablo Neira Ayuso
1 sibling, 1 reply; 5+ messages in thread
From: Dan Carpenter @ 2019-04-06 5:26 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: Jozsef Kadlecsik, Florian Westphal, netfilter-devel, coreteam,
kernel-janitors
I believe that "hook->num" can be up to UINT_MAX. Shifting more than
31 bits would is undefined in C but in practice it would lead to shift
wrapping. That would lead to an array overflow in nf_tables_addchain():
ops->hook = hook.type->hooks[ops->hooknum];
Fixes: fe19c04ca137 ("netfilter: nf_tables: remove nhooks field from struct nft_af_info")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
v2: In my original fix I thought making shifts unsigned was defined in
C but it's not. My original patch didn't fix or change runtime at all.
Thanks to Florian Westphal for his help on this.
net/netfilter/nf_tables_api.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index ef7772e976cc..1606eaa5ae0d 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -1545,7 +1545,7 @@ static int nft_chain_parse_hook(struct net *net,
if (IS_ERR(type))
return PTR_ERR(type);
}
- if (!(type->hook_mask & (1 << hook->num)))
+ if (hook->num > NF_MAX_HOOKS || !(type->hook_mask & (1 << hook->num)))
return -EOPNOTSUPP;
if (type->type == NFT_CHAIN_T_NAT &&
--
2.17.1
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH v2] netfilter: nf_tables: prevent shift wrap in nft_chain_parse_hook()
2019-04-06 5:26 ` [PATCH v2] " Dan Carpenter
@ 2019-04-13 12:58 ` Pablo Neira Ayuso
0 siblings, 0 replies; 5+ messages in thread
From: Pablo Neira Ayuso @ 2019-04-13 12:58 UTC (permalink / raw)
To: Dan Carpenter
Cc: Jozsef Kadlecsik, Florian Westphal, netfilter-devel, coreteam,
kernel-janitors
On Sat, Apr 06, 2019 at 08:26:52AM +0300, Dan Carpenter wrote:
> I believe that "hook->num" can be up to UINT_MAX. Shifting more than
> 31 bits would is undefined in C but in practice it would lead to shift
> wrapping. That would lead to an array overflow in nf_tables_addchain():
>
> ops->hook = hook.type->hooks[ops->hooknum];
Applied, thanks.
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2019-04-13 12:58 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2019-04-02 13:30 [PATCH] netfilter: nf_tables: prevent shift wrap in nft_chain_parse_hook() Dan Carpenter
2019-04-05 8:34 ` Dan Carpenter
2019-04-05 10:01 ` Florian Westphal
2019-04-06 5:26 ` [PATCH v2] " Dan Carpenter
2019-04-13 12:58 ` Pablo Neira Ayuso
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).