1 \documentstyle[12pt,twoside]{article}
2 \def\TITLE{IPv6 Flow Labels}
5 \Large\bf IPv6 Flow Labels in Linux-2.2.
10 { \large Alexey~N.~Kuznetsov } \\
11 \em Institute for Nuclear Research, Moscow \\
12 \verb|kuznet@ms2.inr.ac.ru| \\
20 \section{Introduction.}
22 Every IPv6 packet carries 28 bits of flow information. RFC2460 splits
23 these bits to two fields: 8 bits of traffic class (or DS field, if you
24 prefer this term) and 20 bits of flow label. Currently there exist
25 no well-defined API to manage IPv6 flow information. In this document
26 I describe an attempt to design the API for Linux-2.2 IPv6 stack.
30 The API must solve the following tasks:
34 \item To allow user to set traffic class bits.
36 \item To allow user to read traffic class bits of received packets.
37 This feature is not so useful as the first one, however it will be
38 necessary f.e.\ to implement ECN [RFC2481] for datagram oriented services
39 or to implement receiver side of SRP or another end-to-end protocol
40 using traffic class bits.
42 \item To assign flow labels to packets sent by user.
44 \item To get flow labels of received packets. I do not know
45 any applications of this feature, but it is possible that receiver will
46 want to use flow labels to distinguish sub-flows.
48 \item To allocate flow labels in the way, compliant to RFC2460. Namely:
52 Flow labels must be uniformly distributed (pseudo-)random numbers,
53 so that any subset of 20 bits can be used as hash key.
56 Flows with coinciding source address and flow label must have identical
57 destination address and not-fragmentable extensions headers (i.e.\
58 hop by hop options and all the headers up to and including routing header,
62 There is a hole in specs: some hop-by-hop options can be
63 defined only on per-packet base (f.e.\ jumbo payload option).
64 Essentially, it means that such options cannot present in packets
68 NB notes here and below reflect only my personal opinion,
69 they should be read with smile or should not be read at all :-).
74 Flow labels have finite lifetime and source is not allowed to reuse
75 flow label for another flow within the maximal lifetime has expired,
76 so that intermediate nodes will be able to invalidate flow state before
77 the label is taken over by another flow.
78 Flow state, including lifetime, is propagated along datagram path
79 by some application specific methods
80 (f.e.\ in RSVP PATH messages or in some hop-by-hop option).
87 \section{Sending/receiving flow information.}
89 \paragraph{Discussion.}
90 \addcontentsline{toc}{subsection}{Discussion}
91 It was proposed (Where? I do not remember any explicit statement)
92 to solve the first four tasks using
93 \verb|sin6_flowinfo| field added to \verb|struct| \verb|sockaddr_in6|
97 This method is difficult to consider as reasonable, because it
98 puts additional overhead to all the services, despite of only
99 very small subset of them (none, to be more exact) really use it.
100 It contradicts both to IETF spirit and the letter. Before RFC2553
101 one justification existed, IPv6 address alignment left 4 byte
102 hole in \verb|sockaddr_in6| in any case. Now it has no justification.
105 We have two problems with this method. The first one is common for all OSes:
106 if \verb|recvmsg()| initializes \verb|sin6_flowinfo| to flow info
107 of received packet, we loose one very important property of BSD socket API,
108 namely, we are not allowed to use received address for reply directly
109 and have to mangle it, even if we are not interested in flowinfo subtleties.
112 RFC2553 adds new requirement: to clear \verb|sin6_flowinfo|.
113 Certainly, it is not solution but rather attempt to force applications
114 to make unnecessary work. Well, as usually, one mistake in design
115 is followed by attempts to patch the hole and more mistakes...
118 Another problem is Linux specific. Historically Linux IPv6 did not
119 initialize \verb|sin6_flowinfo| at all, so that, if kernel does not
120 support flow labels, this field is not zero, but a random number.
121 Some applications also did not take care about it.
124 Following RFC2553 such applications can be considered as broken,
125 but I still think that they are right: clearing all the address
126 before filling known fields is robust but stupid solution.
127 Useless wasting CPU cycles and
128 memory bandwidth is not a good idea. Such patches are acceptable
129 as temporary hacks, but not as standard of the future.
133 \paragraph{Implementation.}
134 \addcontentsline{toc}{subsection}{Implementation}
135 By default Linux IPv6 does not read \verb|sin6_flowinfo| field
136 assuming that common applications are not obliged to initialize it
137 and are permitted to consider it as pure alignment padding.
138 In order to tell kernel that application
139 is aware of this field, it is necessary to set socket option
140 \verb|IPV6_FLOWINFO_SEND|.
144 setsockopt(sock, SOL_IPV6, IPV6_FLOWINFO_SEND,
145 (void*)&on, sizeof(on));
148 Linux kernel never fills \verb|sin6_flowinfo| field, when passing
149 message to user space, though the kernels which support flow labels
150 initialize it to zero. If user wants to get received flowinfo, he
151 will set option \verb|IPV6_FLOWINFO| and after this he will receive
152 flowinfo as ancillary data object of type \verb|IPV6_FLOWINFO|
157 setsockopt(sock, SOL_IPV6, IPV6_FLOWINFO, (void*)&on, sizeof(on));
160 Flowinfo received and latched by a connected TCP socket also may be fetched
161 with \verb|getsockopt()| \verb|IPV6_PKTOPTIONS| together with
162 another optional information.
164 Besides that, in the spirit of RFC2292 the option \verb|IPV6_FLOWINFO|
165 may be used as alternative way to send flowinfo with \verb|sendmsg()| or
166 to latch it with \verb|IPV6_PKTOPTIONS|.
168 \paragraph{Note about IPv6 options and destination address.}
169 \addcontentsline{toc}{subsection}{IPv6 options and destination address}
170 If \verb|sin6_flowinfo| does contain not zero flow label,
171 destination address in \verb|sin6_addr| and non-fragmentable
172 extension headers are ignored. Instead, kernel uses the values
173 cached at flow setup (see below). However, for connected sockets
174 kernel prefers the values set at connection time.
177 \addcontentsline{toc}{subsection}{Example}
178 After setting socket option \verb|IPV6_FLOWINFO|
179 flowlabel and DS field are received as ancillary data object
180 of type \verb|IPV6_FLOWINFO| and level \verb|SOL_IPV6|.
181 In the cases when it is convenient to use \verb|recvfrom(2)|,
182 it is possible to replace library variant with your own one,
186 #include <sys/socket.h>
187 #include <netinet/in6.h>
189 size_t recvfrom(int fd, char *buf, size_t len, int flags,
190 struct sockaddr *addr, int *addrlen)
195 struct iovec iov = { buf, len };
196 struct msghdr msg = { addr, *addrlen,
201 cc = recvmsg(fd, &msg, flags);
204 ((struct sockaddr_in6*)addr)->sin6_flowinfo = 0;
205 *addrlen = msg.msg_namelen;
206 for (c=CMSG_FIRSTHDR(&msg); c; c = CMSG_NEXTHDR(&msg, c)) {
207 if (c->cmsg_level != SOL_IPV6 ||
208 c->cmsg_type != IPV6_FLOWINFO)
210 ((struct sockaddr_in6*)addr)->sin6_flowinfo = *(__u32*)CMSG_DATA(c);
218 \section{Flow label management.}
220 \paragraph{Discussion.}
221 \addcontentsline{toc}{subsection}{Discussion}
222 Requirements of RFC2460 are pretty tough. Particularly, lifetimes
223 longer than boot time require to store allocated labels at stable
224 storage, so that the full implementation necessarily includes user space flow
225 label manager. There are at least three different approaches:
228 \item {\bf ``Cooperative''. } We could leave flow label allocation wholly
229 to user space. When user needs label he requests manager directly. The approach
230 is valid, but as any ``cooperative'' approach it suffers of security problems.
233 One idea is to disallow not privileged user to allocate flow
234 labels, but instead to pass the socket to manager via \verb|SCM_RIGHTS|
235 control message, so that it will allocate label and assign it to socket
236 itself. Hmm... the idea is interesting.
239 \item {\bf ``Indirect''.} Kernel redirects requests to user level daemon
240 and does not install label until the daemon acknowledged the request.
241 The approach is the most promising, it is especially pleasant to recognize
242 parallel with IPsec API [RFC2367,Craig]. Actually, it may share API with
245 \item {\bf ``Stupid''.} To allocate labels in kernel space. It is the simplest
246 method, but it suffers of two serious flaws: the first,
247 we cannot lease labels with lifetimes longer than boot time, the second,
248 it is sensitive to DoS attacks. Kernel have to remember all the obsolete
249 labels until their expiration and malicious user may fastly eat all the
254 Certainly, I choose the most ``stupid'' method. It is the cheapest one
255 for implementor (i.e.\ me), and taking into account that flow labels
256 still have no serious applications it is not useful to work on more
257 advanced API, especially, taking into account that eventually we
258 will get it for no fee together with IPsec.
261 \paragraph{Implementation.}
262 \addcontentsline{toc}{subsection}{Implementation}
263 Socket option \verb|IPV6_FLOWLABEL_MGR| allows to
264 request flow label manager to allocate new flow label, to reuse
265 already allocated one or to delete old flow label.
266 Its argument is \verb|struct| \verb|in6_flowlabel_req|:
269 struct in6_flowlabel_req
271 struct in6_addr flr_dst;
278 __u32 __flr_reserved;
279 /* Options in format of IPV6_PKTOPTIONS */
285 \item \verb|dst| is IPv6 destination address associated with the label.
287 \item \verb|label| is flow label value in network byte order. If it is zero,
288 kernel will allocate new pseudo-random number. Otherwise, kernel will try
289 to lease flow label ordered by user. In this case, it is user task to provide
290 necessary flow label randomness.
292 \item \verb|action| is requested operation. Currently, only three operations
296 #define IPV6_FL_A_GET 0 /* Get flow label */
297 #define IPV6_FL_A_PUT 1 /* Release flow label */
298 #define IPV6_FL_A_RENEW 2 /* Update expire time */
301 \item \verb|flags| are optional modifiers. Currently
302 only \verb|IPV6_FL_A_GET| has modifiers:
305 #define IPV6_FL_F_CREATE 1 /* Allowed to create new label */
306 #define IPV6_FL_F_EXCL 2 /* Do not create new label */
310 \item \verb|share| defines who is allowed to reuse the same flow label.
313 #define IPV6_FL_S_NONE 0 /* Not defined */
314 #define IPV6_FL_S_EXCL 1 /* Label is private */
315 #define IPV6_FL_S_PROCESS 2 /* May be reused by this process */
316 #define IPV6_FL_S_USER 3 /* May be reused by this user */
317 #define IPV6_FL_S_ANY 255 /* Anyone may reuse it */
320 \item \verb|linger| is time in seconds. After the last user releases flow
321 label, it will not be reused with different destination and options at least
322 during this time. If \verb|share| is not \verb|IPV6_FL_S_EXCL| the label
323 still can be shared by another sockets. Current implementation does not allow
324 unprivileged user to set linger longer than 60 sec.
326 \item \verb|expires| is time in seconds. Flow label will be kept at least
327 for this time, but it will not be destroyed before user released it explicitly
328 or closed all the sockets using it. Current implementation does not allow
329 unprivileged user to set timeout longer than 60 sec. Proviledged applications
330 MAY set longer lifetimes, but in this case they MUST save allocated
331 labels at stable storage and restore them back after reboot before the first
332 application allocates new flow.
336 This structure is followed by optional extension headers associated
337 with this flow label in format of \verb|IPV6_PKTOPTIONS|. Only
338 \verb|IPV6_HOPOPTS|, \verb|IPV6_RTHDR| and, if \verb|IPV6_RTHDR| presents,
339 \verb|IPV6_DSTOPTS| are allowed.
342 \addcontentsline{toc}{subsection}{Example}
343 The function \verb|get_flow_label| allocates
347 int get_flow_label(int fd, struct sockaddr_in6 *dst, __u32 fl)
350 struct in6_flowlabel_req freq;
352 memset(&freq, 0, sizeof(freq));
353 freq.flr_label = htonl(fl);
354 freq.flr_action = IPV6_FL_A_GET;
355 freq.flr_flags = IPV6_FL_F_CREATE | IPV6_FL_F_EXCL;
356 freq.flr_share = IPV6_FL_S_EXCL;
357 memcpy(&freq.flr_dst, &dst->sin6_addr, 16);
358 if (setsockopt(fd, SOL_IPV6, IPV6_FLOWLABEL_MGR,
359 &freq, sizeof(freq)) == -1) {
360 perror ("can't lease flowlabel");
363 dst->sin6_flowinfo |= freq.flr_label;
365 if (setsockopt(fd, SOL_IPV6, IPV6_FLOWINFO_SEND,
366 &on, sizeof(on)) == -1) {
367 perror ("can't send flowinfo");
369 freq.flr_action = IPV6_FL_A_PUT;
370 setsockopt(fd, SOL_IPV6, IPV6_FLOWLABEL_MGR,
371 &freq, sizeof(freq));
378 A bit more complicated example using routing header can be found
379 in \verb|ping6| utility (\verb|iputils| package). Linux rsvpd backend
380 contains an example of using operation \verb|IPV6_FL_A_RENEW|.
382 \paragraph{Listing flow labels.}
383 \addcontentsline{toc}{subsection}{Listing flow labels}
384 List of currently allocated
385 flow labels may be read from \verb|/proc/net/ip6_flowlabel|.
388 Label S Owner Users Linger Expires Dst Opt
389 A1BE5 1 0 0 6 3 3ffe2400000000010a0020fffe71fb30 0
393 \item \verb|Label| is hexadecimal flow label value.
394 \item \verb|S| is sharing style.
395 \item \verb|Owner| is ID of creator, it is zero, pid or uid, depending on
397 \item \verb|Users| is number of applications using the label now.
398 \item \verb|Linger| is \verb|linger| of this label in seconds.
399 \item \verb|Expires| is time until expiration of the label in seconds. It may
400 be negative, if the label is in use.
401 \item \verb|Dst| is IPv6 destination address.
402 \item \verb|Opt| is length of options, associated with the label. Option
403 data are not accessible.
407 \paragraph{Flow labels and RSVP.}
408 \addcontentsline{toc}{subsection}{Flow labels and RSVP}
409 RSVP daemon supports IPv6 flow labels
410 without any modifications to standard ISI RAPI. Sender must allocate
411 flow label, fill corresponding sender template and submit it to local rsvp
412 daemon. rsvpd will check the label and start to announce it in PATH
413 messages. Rsvpd on sender node will renew the flow label, so that it will not
414 be reused before path state expires and all the intermediate
415 routers and receiver purge flow state.
417 \verb|rtap| utility is modified to parse flow labels. F.e.\ if user allocated
418 flow label \verb|0xA1234|, he may write:
421 RTAP> sender 3ffe:2400::1/FL0xA1234 <Tspec>
424 Receiver makes reservation with command:
426 RTAP> reserve ff 3ffe:2400::1/FL0xA1234 <Flowspec>