2019-12-10 20:25:51

by Arnd Bergmann

Subject: [PATCH] netfilter: nf_flow_table: fix big-endian integer overflow

In some configurations, gcc reports an integer overflow:

net/netfilter/nf_flow_table_offload.c: In function 'nf_flow_rule_match':
net/netfilter/nf_flow_table_offload.c:80:21: error: unsigned conversion from 'int' to '__be16' {aka 'short unsigned int'} changes value from '327680' to '0' [-Werror=overflow]
mask->tcp.flags = TCP_FLAG_RST | TCP_FLAG_FIN;
^~~~~~~~~~~~

From what I can tell, we want the upper 16 bits of these constants,
so they need to be shifted in cpu-endian mode.

Fixes: c29f74e0df7a ("netfilter: nf_flow_table: hardware offload support")
Signed-off-by: Arnd Bergmann <[email protected]>
---
I'm not sure if this is the correct fix, please check carefully
---
net/netfilter/nf_flow_table_offload.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/netfilter/nf_flow_table_offload.c b/net/netfilter/nf_flow_table_offload.c
index c54c9a6cc981..24543c45d501 100644
--- a/net/netfilter/nf_flow_table_offload.c
+++ b/net/netfilter/nf_flow_table_offload.c
@@ -77,7 +77,7 @@ static int nf_flow_rule_match(struct nf_flow_match *match,
switch (tuple->l4proto) {
case IPPROTO_TCP:
key->tcp.flags = 0;
- mask->tcp.flags = TCP_FLAG_RST | TCP_FLAG_FIN;
+ mask->tcp.flags = cpu_to_be16(be32_to_cpu(TCP_FLAG_RST | TCP_FLAG_FIN) >> 16);
match->dissector.used_keys |= BIT(FLOW_DISSECTOR_KEY_TCP);
break;
case IPPROTO_UDP:
--
2.20.0

2019-12-11 22:05:28

by Pablo Neira Ayuso

Subject: Re: [PATCH] netfilter: nf_flow_table: fix big-endian integer overflow

Hi Arnd,

On Tue, Dec 10, 2019 at 09:24:28PM +0100, Arnd Bergmann wrote:
> In some configurations, gcc reports an integer overflow:
>
> net/netfilter/nf_flow_table_offload.c: In function 'nf_flow_rule_match':
> net/netfilter/nf_flow_table_offload.c:80:21: error: unsigned conversion from 'int' to '__be16' {aka 'short unsigned int'} changes value from '327680' to '0' [-Werror=overflow]
> mask->tcp.flags = TCP_FLAG_RST | TCP_FLAG_FIN;
> ^~~~~~~~~~~~
>
> From what I can tell, we want the upper 16 bits of these constants,
> so they need to be shifted in cpu-endian mode.
>
> Fixes: c29f74e0df7a ("netfilter: nf_flow_table: hardware offload support")
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
> I'm not sure if this is the correct fix, please check carefully

Thanks. I posted this one:

https://patchwork.ozlabs.org/patch/1206384/

The flow dissector matching on tcp flags seems also broken on big-endian.

2019-12-20 01:11:50

by Pablo Neira Ayuso

Subject: Re: [PATCH] netfilter: nf_flow_table: fix big-endian integer overflow

On Tue, Dec 10, 2019 at 09:24:28PM +0100, Arnd Bergmann wrote:
> In some configurations, gcc reports an integer overflow:
>
> net/netfilter/nf_flow_table_offload.c: In function 'nf_flow_rule_match':
> net/netfilter/nf_flow_table_offload.c:80:21: error: unsigned conversion from 'int' to '__be16' {aka 'short unsigned int'} changes value from '327680' to '0' [-Werror=overflow]
> mask->tcp.flags = TCP_FLAG_RST | TCP_FLAG_FIN;
> ^~~~~~~~~~~~
>
> From what I can tell, we want the upper 16 bits of these constants,
> so they need to be shifted in cpu-endian mode.
>
> Fixes: c29f74e0df7a ("netfilter: nf_flow_table: hardware offload support")
> Signed-off-by: Arnd Bergmann <[email protected]>

Applied, thanks.