On Sunday 2009-11-01 14:07, Pierre Chifflier wrote:
>> >+ ip = (struct iphdr *) data;
>>
>> If data is not aligned, later dereferencing can fail :(
>
>I remember some arches (sparc, arm ?) will send a signal (SIGBUS I
>think) in this case.
>How can I prevent data from being unaligned ?
>Adding some __attribute__((aligned(4)) will be enough ?
__attribute__(aligned) only specifies the minimum requirement.
What one would need is __attribute__((packed)), and it needs
to be on struct iphdr - in /usr/include/... - and that is the
problem because you cannot change it.
And gcc does not allow saying (ignored the attribute)
struct iphdr __attribute__((packed)) *iph = data;
So.. tough luck.
>> >+
>> >+ while ( (data - packet_buf) < total_len) {
>> >+
>> >+ option_number = ntohs(*(u_int16_t*)(data));
>>
>> This can fail even if packet_buf _is_ aligned, all it takes
>> is that payload_len is an odd number.
>> Unfortunately, I see no way to easily address this without
>> lame bitshifting :(
>
>Yes, since this is a Key-Length-Value protocol it's hard to ensure there
>is no weird value. I tried to add checks (marker on start, etc.), but I
>see no easy way to address this;
+static inline uint32_t deref_get_32(const void *ptr)
+{
+ uint32_t ret;
+ memcpy(&ret, ptr, sizeof(ret));
+ return ret;
+}
Is what I use in one project; however, trying to minimize the
unalignedness - iptables is a "perfect" example, and netlink
may do the same tho I am not sure - seems to be the best
approach in serialized streams.
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
