Updated: 2022/Sep/29
Please read Privacy Policy. It's for your privacy.
ALTQ(9) Kernel Developer's Manual ALTQ(9) NAME ALTQ - kernel interfaces for manipulating output queues on network interfaces SYNOPSIS #include <sys/types.h> #include <sys/socket.h> #include <net/if.h> void IFQ_ENQUEUE(struct ifaltq *ifq, struct mbuf *m, struct altq_pktattr *pattr, int err); void IFQ_DEQUEUE(struct ifaltq *ifq, struct mbuf *m); void IFQ_POLL(struct ifaltq *ifq, struct mbuf *m); void IFQ_PURGE(struct ifaltq *ifq); void IFQ_CLASSIFY(struct ifaltq *ifq, struct mbuf *m, int af, struct altq_pktattr *pattr); void IFQ_IS_EMPTY(struct ifaltq *ifq); void IFQ_SET_MAXLEN(struct ifaltq *ifq, int len); void IFQ_INC_LEN(struct ifaltq *ifq); void IFQ_DEC_LEN(struct ifaltq *ifq); void IFQ_INC_DROPS(struct ifaltq *ifq); void IFQ_SET_READY(struct ifaltq *ifq); DESCRIPTION The ALTQ system is a framework to manage queueing disciplines on network interfaces. ALTQ introduces new macros to manipulate output queues. The output queue macros are used to abstract queue operations and not to touch the internal fields of the output queue structure. The macros are independent from the ALTQ implementation, and compatible with the traditional ifqueue macros for ease of transition. IFQ_ENQUEUE() enqueues a packet m to the queue ifq. The underlying queueing discipline may discard the packet. err is set to 0 on success, or ENOBUFS if the packet is discarded. m will be freed by the device driver on success or by the queueing discipline on failure, so that the caller should not touch m after calling IFQ_ENQUEUE(). IFQ_DEQUEUE() dequeues a packet from the queue. The dequeued packet is returned in m, or m is set to NULL if no packet is dequeued. The caller must always check m since a non-empty queue could return NULL under rate- limiting. IFQ_POLL() returns the next packet without removing it from the queue. It is guaranteed by the underlying queueing discipline that IFQ_DEQUEUE() immediately after IFQ_POLL() returns the same packet. IFQ_PURGE() discards all the packets in the queue. The purge operation is needed since a non-work conserving queue cannot be emptied by a dequeue loop. IFQ_CLASSIFY() classifies a packet to a scheduling class, and returns the result in pattr. IFQ_IS_EMPTY() can be used to check if the queue is empty. Note that IFQ_DEQUEUE() could still return NULL if the queueing discipline is non- work conserving. IFQ_SET_MAXLEN() sets the queue length limit to the default FIFO queue. IFQ_INC_LEN() and IFQ_DEC_LEN() increment or decrement the current queue length in packets. IFQ_INC_DROPS() increments the drop counter and is equal to IF_DROP(). It is defined for naming consistency. IFQ_SET_READY() sets a flag to indicate this driver is converted to use the new macros. ALTQ can be enabled only on interfaces with this flag. COMPATIBILITY ifaltq structure In order to keep compatibility with the existing code, the new output queue structure ifaltq has the same fields. The traditional IF_XXX() macros and the code directly referencing the fields within if_snd still work with ifaltq. (Once we finish conversions of all the drivers, we no longer need these fields.) ##old-style## ##new-style## | struct ifqueue { | struct ifaltq { struct mbuf *ifq_head; | struct mbuf *ifq_head; struct mbuf *ifq_tail; | struct mbuf *ifq_tail; int ifq_len; | int ifq_len; int ifq_maxlen; | int ifq_maxlen; uint64_t ifq_drops; | uint64_t ifq_drops; }; | /* altq related fields */ | ...... | }; | The new structure replaces struct ifqueue in struct ifnet. ##old-style## ##new-style## | struct ifnet { | struct ifnet { .... | .... | struct ifqueue if_snd; | struct ifaltq if_snd; | .... | .... }; | }; | The (simplified) new IFQ_XXX() macros looks like: #ifdef ALTQ #define IFQ_DEQUEUE(ifq, m) \ if (ALTQ_IS_ENABLED((ifq)) \ ALTQ_DEQUEUE((ifq), (m)); \ else \ IF_DEQUEUE((ifq), (m)); #else #define IFQ_DEQUEUE(ifq, m) IF_DEQUEUE((ifq), (m)); #endif Enqueue operation The semantics of the enqueue operation are changed. In the new style, enqueue and packet drop are combined since they cannot be easily separated in many queueing disciplines. The new enqueue operation corresponds to the following macro that is written with the old macros. #define IFQ_ENQUEUE(ifq, m, pattr, err) \ do { \ if (ALTQ_IS_ENABLED((ifq))) \ ALTQ_ENQUEUE((ifq), (m), (pattr), (err)); \ else { \ if (IF_QFULL((ifq))) { \ m_freem((m)); \ (err) = ENOBUFS; \ } else { \ IF_ENQUEUE((ifq), (m)); \ (err) = 0; \ } \ } \ if ((err)) \ (ifq)->ifq_drops++; \ } while (false) IFQ_ENQUEUE() does the following: - queue a packet - drop (and free) a packet if the enqueue operation fails If the enqueue operation fails, err is set to ENOBUFS. m is freed by the queueing discipline. The caller should not touch mbuf after calling IFQ_ENQUEUE() so that the caller may need to copy m_pkthdr.len or m_flags field beforehand for statistics. The caller should not use senderr() since mbuf was already freed. The new style if_output() looks as follows: ##old-style## ##new-style## | int | int ether_output(ifp, m0, dst, rt0) | ether_output(ifp, m0, dst, rt0) { | { ...... | ...... | | mflags = m->m_flags; | len = m->m_pkthdr.len; s = splimp(); | s = splimp(); if (IF_QFULL(&ifp->if_snd)) { | IFQ_ENQUEUE(&ifp->if_snd, m, | NULL, error); IF_DROP(&ifp->if_snd); | if (error != 0) { splx(s); | splx(s); senderr(ENOBUFS); | return (error); } | } IF_ENQUEUE(&ifp->if_snd, m); | ifp->if_obytes += | ifp->if_obytes += len; m->m_pkthdr.len; | if (m->m_flags & M_MCAST) | if (mflags & M_MCAST) ifp->if_omcasts++; | ifp->if_omcasts++; | if ((ifp->if_flags & IFF_OACTIVE) | if ((ifp->if_flags & IFF_OACTIVE) == 0) | == 0) (*ifp->if_start)(ifp); | (*ifp->if_start)(ifp); splx(s); | splx(s); return (error); | return (error); | bad: | bad: if (m) | if (m) m_freem(m); | m_freem(m); return (error); | return (error); } | } | Classifier The classifier mechanism is currently implemented in if_output(). struct altq_pktattr is used to store the classifier result, and it is passed to the enqueue function. (We will change the method to tag the classifier result to mbuf in the future.) int ether_output(ifp, m0, dst, rt0) { ...... struct altq_pktattr pktattr; ...... /* classify the packet before prepending link-headers */ IFQ_CLASSIFY(&ifp->if_snd, m, dst->sa_family, &pktattr); /* prepend link-level headers */ ...... IFQ_ENQUEUE(&ifp->if_snd, m, &pktattr, error); ...... } HOW TO CONVERT THE EXISTING DRIVERS First, make sure the corresponding if_output() is already converted to the new style. Look for if_snd in the driver. You will probably need to make changes to the lines that include if_snd. Empty check operation If the code checks ifq_head to see whether the queue is empty or not, use IFQ_IS_EMPTY(). ##old-style## ##new-style## | if (ifp->if_snd.ifq_head != NULL) | if (IFQ_IS_EMPTY(&ifp->if_snd) == 0) | Note that IFQ_POLL() can be used for the same purpose, but IFQ_POLL() could be costly for a complex scheduling algorithm since IFQ_POLL() needs to run the scheduling algorithm to select the next packet. On the other hand, IFQ_IS_EMPTY() checks only if there is any packet stored in the queue. Another difference is that even when IFQ_IS_EMPTY() is false, IFQ_DEQUEUE() could still return NULL if the queue is under rate- limiting. Dequeue operation Replace IF_DEQUEUE() by IFQ_DEQUEUE(). Always check whether the dequeued mbuf is NULL or not. Note that even when IFQ_IS_EMPTY() is false, IFQ_DEQUEUE() could return NULL due to rate-limiting. ##old-style## ##new-style## | IF_DEQUEUE(&ifp->if_snd, m); | IFQ_DEQUEUE(&ifp->if_snd, m); | if (m == NULL) | return; | A driver is supposed to call if_start() from transmission complete interrupts in order to trigger the next dequeue. Poll-and-dequeue operation If the code polls the packet at the head of the queue and actually uses the packet before dequeueing it, use IFQ_POLL() and IFQ_DEQUEUE(). ##old-style## ##new-style## | m = ifp->if_snd.ifq_head; | IFQ_POLL(&ifp->if_snd, m); if (m != NULL) { | if (m != NULL) { | /* use m to get resources */ | /* use m to get resources */ if (something goes wrong) | if (something goes wrong) return; | return; | IF_DEQUEUE(&ifp->if_snd, m); | IFQ_DEQUEUE(&ifp->if_snd, m); | /* kick the hardware */ | /* kick the hardware */ } | } | It is guaranteed that IFQ_DEQUEUE() immediately after IFQ_POLL() returns the same packet. Note that they need to be guarded by splimp() if called from outside of if_start(). Eliminating IF_PREPEND If the code uses IF_PREPEND(), you have to eliminate it since the prepend operation is not possible for many queueing disciplines. A common use of IF_PREPEND() is to cancel the previous dequeue operation. You have to convert the logic into poll-and-dequeue. ##old-style## ##new-style## | IF_DEQUEUE(&ifp->if_snd, m); | IFQ_POLL(&ifp->if_snd, m); if (m != NULL) { | if (m != NULL) { | if (something_goes_wrong) { | if (something_goes_wrong) { IF_PREPEND(&ifp->if_snd, m); | return; | return; } | } | | /* at this point, the driver | * is committed to send this | * packet. | */ | IFQ_DEQUEUE(&ifp->if_snd, m); | /* kick the hardware */ | /* kick the hardware */ } | } | Purge operation Use IFQ_PURGE() to empty the queue. Note that a non-work conserving queue cannot be emptied by a dequeue loop. ##old-style## ##new-style## | while (ifp->if_snd.ifq_head != NULL) {| IFQ_PURGE(&ifp->if_snd); IF_DEQUEUE(&ifp->if_snd, m); | m_freem(m); | } | | Attach routine Use IFQ_SET_MAXLEN() to set ifq_maxlen to len. Add IFQ_SET_READY() to show this driver is converted to the new style. (This is used to distinguish new-style drivers.) ##old-style## ##new-style## | ifp->if_snd.ifq_maxlen = qsize; | IFQ_SET_MAXLEN(&ifp->if_snd, qsize); | IFQ_SET_READY(&ifp->if_snd); if_attach(ifp); | if_attach(ifp); | Other issues The new macros for statistics: ##old-style## ##new-style## | IF_DROP(&ifp->if_snd); | IFQ_INC_DROPS(&ifp->if_snd); | ifp->if_snd.ifq_len++; | IFQ_INC_LEN(&ifp->if_snd); | ifp->if_snd.ifq_len--; | IFQ_DEC_LEN(&ifp->if_snd); | Some drivers instruct the hardware to invoke transmission complete interrupts only when it thinks necessary. Rate-limiting breaks its assumption. How to convert drivers using multiple ifqueues Some (pseudo) devices (such as slip) have another ifqueue to prioritize packets. It is possible to eliminate the second queue since ALTQ provides more flexible mechanisms but the following shows how to keep the original behavior. struct sl_softc { struct ifnet sc_if; /* network-visible interface */ ... struct ifqueue sc_fastq; /* interactive output queue */ ... }; The driver doesn't compile in the new model since it has the following line (if_snd is no longer a type of struct ifqueue). struct ifqueue *ifq = &ifp->if_snd; A simple way is to use the original IF_XXX() macros for sc_fastq and use the new IFQ_XXX() macros for if_snd. The enqueue operation looks like: ##old-style## ##new-style## | struct ifqueue *ifq = &ifp->if_snd; | struct ifqueue *ifq = NULL; | if (ip->ip_tos & IPTOS_LOWDELAY) | if ((ip->ip_tos & IPTOS_LOWDELAY) && ifq = &sc->sc_fastq; | !ALTQ_IS_ENABLED(&sc->sc_if.if_snd)) { | ifq = &sc->sc_fastq; if (IF_QFULL(ifq)) { | if (IF_QFULL(ifq)) { IF_DROP(ifq); | IF_DROP(ifq); m_freem(m); | m_freem(m); splx(s); | error = ENOBUFS; sc->sc_if.if_oerrors++; | } else { return (ENOBUFS); | IF_ENQUEUE(ifq, m); } | error = 0; IF_ENQUEUE(ifq, m); | } | } else | IFQ_ENQUEUE(&sc->sc_if.if_snd, | m, NULL, error); | | if (error) { | splx(s); | sc->sc_if.if_oerrors++; | return (error); | } if ((sc->sc_oqlen = | if ((sc->sc_oqlen = sc->sc_ttyp->t_outq.c_cc) == 0) | sc->sc_ttyp->t_outq.c_cc) == 0) slstart(sc->sc_ttyp); | slstart(sc->sc_ttyp); splx(s); | splx(s); | The dequeue operations looks like: ##old-style## ##new-style## | s = splimp(); | s = splimp(); IF_DEQUEUE(&sc->sc_fastq, m); | IF_DEQUEUE(&sc->sc_fastq, m); if (m == NULL) | if (m == NULL) IF_DEQUEUE(&sc->sc_if.if_snd, m); | IFQ_DEQUEUE(&sc->sc_if.if_snd, m); splx(s); | splx(s); | QUEUEING DISCIPLINES Queueing disciplines need to maintain ifq_len (used by IFQ_IS_EMPTY()). Queueing disciplines also need to guarantee the same mbuf is returned if IFQ_DEQUEUE() is called immediately after IFQ_POLL(). SEE ALSO pf(4), altq.conf(5), pf.conf(5), altqd(8), tbrconfig(8) HISTORY The ALTQ system first appeared in March 1997 and found its home in the KAME project (http://www.kame.net). It was imported into NetBSD 1.6. NetBSD 10.99 October 24, 2022 NetBSD 10.99