On Mon, Jun 26, 2017 at 11:20 PM, tiggersWelt.net (Support)
<supp...@tiggerswelt.net> wrote:
> Good evening Denys,
>
> thank you for your effort!
>
> Am 26.06.2017 um 20:08 schrieb Denys Vlasenko:
>> On Wed, Jun 21, 2017 at 4:29 PM, tiggersWelt.net (Support)
>> <supp...@tiggerswelt.net> wrote:
>>> static const char opt_req[] = {
>>> (D6_OPT_ORO >> 8), (D6_OPT_ORO & 0xff),
>>> - 0, 6,
>>> + 0, 10,
>>> (D6_OPT_DNS_SERVERS >> 8), (D6_OPT_DNS_SERVERS & 0xff),
>>> (D6_OPT_DOMAIN_LIST >> 8), (D6_OPT_DOMAIN_LIST & 0xff),
>>> - (D6_OPT_CLIENT_FQDN >> 8), (D6_OPT_CLIENT_FQDN & 0xff)
>>> + (D6_OPT_CLIENT_FQDN >> 8), (D6_OPT_CLIENT_FQDN & 0xff),
>>> + (D6_OPT_TZ_POSIX >> 8), (D6_OPT_TZ_POSIX & 0xff),
>>> + (D6_OPT_TZ_NAME >> 8), (D6_OPT_TZ_NAME & 0xff)
>>> };
>>
>> Please explain what does this do, and why.
>
> It's the content of the DHCP6 Option Request Option (ORO) as of RFC 3315,
> 22.7. It actually tells the
> server what options are supported and wanted on client-side.
>
> I introduced this option in my last patch (see commit
> ed898ed2ddc1e3627555488671155420a32d0167) as some
> servers required this to deliver the listed options. Now it was only enhanced
> by the new options -
> haven't tested without it, but I'll give it a try tomorrow.
Let's unbreak -O OPT instead? IPv4 client adds list of -O options
to outgoing packets using add_client_options(), like this:
static void add_client_options(struct dhcp_packet *packet)
{
...
end = udhcp_end_option(packet->options);
len = 0;
for (i = 1; i < DHCP_END; i++) {
if (client_config.opt_mask[i >> 3] & (1 << (i & 7))) {
packet->options[end + OPT_DATA + len] = i;
len++;
}
}
...
}
static NOINLINE int send_discover(uint32_t xid, uint32_t requested)
{
struct dhcp_packet packet;
/* Fill in: op, htype, hlen, cookie, chaddr fields,
* random xid field (we override it below),
* client-id option (unless -C), message type option:
*/
init_packet(&packet, DHCPDISCOVER);
packet.xid = xid;
if (requested)
udhcp_add_simple_option(&packet, DHCP_REQUESTED_IP, requested);
/* Add options: maxsize,
* optionally: hostname, fqdn, vendorclass,
* "param req" option according to -O, options specified with -x
*/
add_client_options(&packet);
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
bb_error_msg("sending %s", "discover");
return raw_bcast_from_client_config_ifindex(&packet, INADDR_ANY);
}
IPv6 client does not do it:
static NOINLINE int send_d6_discover(uint32_t xid, struct in6_addr
*requested_ipv6)
{
struct d6_packet packet;
uint8_t *opt_ptr;
unsigned len;
/* Fill in: msg type, client id */
opt_ptr = init_d6_packet(&packet, D6_MSG_SOLICIT, xid);
/* Create new IA_NA, optionally with included IAADDR with
requested IP */
free(client6_data.ia_na);
len = requested_ipv6 ? 2+2+4+4+4 + 2+2+16+4+4 : 2+2+4+4+4;
client6_data.ia_na = xzalloc(len);
client6_data.ia_na->code = D6_OPT_IA_NA;
client6_data.ia_na->len = len - 4;
*(uint32_t*)client6_data.ia_na->data = rand(); /* IAID */
if (requested_ipv6) {
struct d6_option *iaaddr =
(void*)(client6_data.ia_na->data + 4+4+4);
iaaddr->code = D6_OPT_IAADDR;
iaaddr->len = 16+4+4;
memcpy(iaaddr->data, requested_ipv6, 16);
}
opt_ptr = mempcpy(opt_ptr, client6_data.ia_na, len);
/* Request additional options */
opt_ptr = mempcpy(opt_ptr, &opt_req, sizeof(opt_req));
opt_ptr = mempcpy(opt_ptr, &opt_fqdn_req, sizeof(opt_fqdn_req));
/* Add options:
* "param req" option according to -O, options specified with -x
*/
opt_ptr = add_d6_client_options(opt_ptr);
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
bb_error_msg("sending %s", "discover");
return d6_mcast_from_client_config_ifindex(&packet, opt_ptr);
}
but add_d6_client_options() is a dummy function: currently it simply
returns opt_ptr.
What it needs to do is look at client_config.opt_mask, and if
there are requested opts, append OPTION_ORO at opt_ptr:
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| OPTION_ORO | option-len |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| requested-option-code-1 | requested-option-code-2 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| ... |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
(And opt_req[] needs to die then).
Can you try making a patch?
_______________________________________________
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
