vfs: check userland buffers before reading them.

[haiku.git] / src / libs / compat / freebsd11_network / fbsd_mbuf.c

/*-
 * Copyright (c) 1982, 1986, 1988, 1991, 1993
 *      The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *      @(#)uipc_mbuf.c 8.2 (Berkeley) 1/4/94
 */
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>

/*
 * Copy data from an mbuf chain starting "off" bytes from the beginning,
 * continuing for "len" bytes, into the indicated buffer.
 */
void
m_copydata(const struct mbuf *m, int off, int len, caddr_t cp)
{
        u_int count;

        KASSERT(off >= 0, ("m_copydata, negative off %d", off));
        KASSERT(len >= 0, ("m_copydata, negative len %d", len));
        while (off > 0) {
                KASSERT(m != NULL, ("m_copydata, offset > size of mbuf chain"));
                if (off < m->m_len)
                        break;
                off -= m->m_len;
                m = m->m_next;
        }
        while (len > 0) {
                KASSERT(m != NULL, ("m_copydata, length > size of mbuf chain"));
                count = min(m->m_len - off, len);
                bcopy(mtod(m, caddr_t) + off, cp, count);
                len -= count;
                cp += count;
                off = 0;
                m = m->m_next;
        }
}

/*
 * Concatenate mbuf chain n to m.
 * Both chains must be of the same type (e.g. MT_DATA).
 * Any m_pkthdr is not updated.
 */
void
m_cat(struct mbuf *m, struct mbuf *n)
{
        while (m->m_next)
                m = m->m_next;
        while (n) {
                if (m->m_flags & M_EXT ||
                    m->m_data + m->m_len + n->m_len >= &m->m_dat[MLEN]) {
                        /* just join the two chains */
                        m->m_next = n;
                        return;
                }
                /* splat the data from one into the other */
                bcopy(mtod(n, caddr_t), mtod(m, caddr_t) + m->m_len,
                    (u_int)n->m_len);
                m->m_len += n->m_len;
                n = m_free(n);
        }
}

u_int
m_length(struct mbuf *m0, struct mbuf **last)
{
        struct mbuf *m;
        u_int len;

        len = 0;
        for (m = m0; m != NULL; m = m->m_next) {
                len += m->m_len;
                if (m->m_next == NULL)
                        break;
        }
        if (last != NULL)
                *last = m;
        return (len);
}

u_int
m_fixhdr(struct mbuf *m0)
{
        u_int len;

        len = m_length(m0, NULL);
        m0->m_pkthdr.len = len;
        return (len);
}

/*
 * Duplicate "from"'s mbuf pkthdr in "to".
 * "from" must have M_PKTHDR set, and "to" must be empty.
 * In particular, this does a deep copy of the packet tags.
 */
int
m_dup_pkthdr(struct mbuf *to, struct mbuf *from, int how)
{
        MBUF_CHECKSLEEP(how);
        to->m_flags = (from->m_flags & M_COPYFLAGS) | (to->m_flags & M_EXT);
        if ((to->m_flags & M_EXT) == 0)
                to->m_data = to->m_pktdat;
        to->m_pkthdr = from->m_pkthdr;
        SLIST_INIT(&to->m_pkthdr.tags);
        return (m_tag_copy_chain(to, from, MBTOM(how)));
}

/*
 * Defragment a mbuf chain, returning the shortest possible
 * chain of mbufs and clusters.  If allocation fails and
 * this cannot be completed, NULL will be returned, but
 * the passed in chain will be unchanged.  Upon success,
 * the original chain will be freed, and the new chain
 * will be returned.
 *
 * If a non-packet header is passed in, the original
 * mbuf (chain?) will be returned unharmed.
 */
struct mbuf *
m_defrag(struct mbuf *m0, int how)
{
        struct mbuf *m_new = NULL, *m_final = NULL;
        int progress = 0, length;

        MBUF_CHECKSLEEP(how);
        if (!(m0->m_flags & M_PKTHDR))
                return (m0);

        m_fixhdr(m0); /* Needed sanity check */

        if (m0->m_pkthdr.len > MHLEN)
                m_final = m_getcl(how, MT_DATA, M_PKTHDR);
        else
                m_final = m_gethdr(how, MT_DATA);

        if (m_final == NULL)
                goto nospace;

        if (m_dup_pkthdr(m_final, m0, how) == 0)
                goto nospace;

        m_new = m_final;

        while (progress < m0->m_pkthdr.len) {
                length = m0->m_pkthdr.len - progress;
                if (length > MCLBYTES)
                        length = MCLBYTES;

                if (m_new == NULL) {
                        if (length > MLEN)
                                m_new = m_getcl(how, MT_DATA, 0);
                        else
                                m_new = m_get(how, MT_DATA);
                        if (m_new == NULL)
                                goto nospace;
                }

                m_copydata(m0, progress, length, mtod(m_new, caddr_t));
                progress += length;
                m_new->m_len = length;
                if (m_new != m_final)
                        m_cat(m_final, m_new);
                m_new = NULL;
        }

        m_freem(m0);
        m0 = m_final;
        return (m0);
nospace:
        if (m_final)
                m_freem(m_final);
        return (NULL);
}

void
m_adj(struct mbuf *mp, int req_len)
{
        int len = req_len;
        struct mbuf *m;
        int count;

        if ((m = mp) == NULL)
                return;
        if (len >= 0) {
                /*
                 * Trim from head.
                 */
                while (m != NULL && len > 0) {
                        if (m->m_len <= len) {
                                len -= m->m_len;
                                m->m_len = 0;
                                m = m->m_next;
                        } else {
                                m->m_len -= len;
                                m->m_data += len;
                                len = 0;
                        }
                }
                m = mp;
                if (mp->m_flags & M_PKTHDR)
                        m->m_pkthdr.len -= (req_len - len);
        } else {
                /*
                 * Trim from tail.  Scan the mbuf chain,
                 * calculating its length and finding the last mbuf.
                 * If the adjustment only affects this mbuf, then just
                 * adjust and return.  Otherwise, rescan and truncate
                 * after the remaining size.
                 */
                len = -len;
                count = 0;
                for (;;) {
                        count += m->m_len;
                        if (m->m_next == (struct mbuf *)0)
                                break;
                        m = m->m_next;
                }
                if (m->m_len >= len) {
                        m->m_len -= len;
                        if (mp->m_flags & M_PKTHDR)
                                mp->m_pkthdr.len -= len;
                        return;
                }
                count -= len;
                if (count < 0)
                        count = 0;
                /*
                 * Correct length for chain is "count".
                 * Find the mbuf with last data, adjust its length,
                 * and toss data from remaining mbufs on chain.
                 */
                m = mp;
                if (m->m_flags & M_PKTHDR)
                        m->m_pkthdr.len = count;
                for (; m; m = m->m_next) {
                        if (m->m_len >= count) {
                                m->m_len = count;
                                if (m->m_next != NULL) {
                                        m_freem(m->m_next);
                                        m->m_next = NULL;
                                }
                                break;
                        }
                        count -= m->m_len;
                }
        }
}

/*
 * Rearange an mbuf chain so that len bytes are contiguous
 * and in the data area of an mbuf (so that mtod and dtom
 * will work for a structure of size len).  Returns the resulting
 * mbuf chain on success, frees it and returns null on failure.
 * If there is room, it will add up to max_protohdr-len extra bytes to the
 * contiguous region in an attempt to avoid being called next time.
 */
struct mbuf *
m_pullup(struct mbuf *n, int len)
{
        struct mbuf *m;
        int count;
        int space;

        /*
         * If first mbuf has no cluster, and has room for len bytes
         * without shifting current data, pullup into it,
         * otherwise allocate a new mbuf to prepend to the chain.
         */
        if ((n->m_flags & M_EXT) == 0 &&
            n->m_data + len < &n->m_dat[MLEN] && n->m_next) {
                if (n->m_len >= len)
                        return (n);
                m = n;
                n = n->m_next;
                len -= m->m_len;
        } else {
                if (len > MHLEN)
                        goto bad;
                MGET(m, M_DONTWAIT, n->m_type);
                if (m == NULL)
                        goto bad;
                m->m_len = 0;
                if (n->m_flags & M_PKTHDR)
                        M_MOVE_PKTHDR(m, n);
        }
        space = &m->m_dat[MLEN] - (m->m_data + m->m_len);
        do {
                count = min(min(max(len, max_protohdr), space), n->m_len);
                bcopy(mtod(n, caddr_t), mtod(m, caddr_t) + m->m_len,
                  (u_int)count);
                len -= count;
                m->m_len += count;
                n->m_len -= count;
                space -= count;
                if (n->m_len)
                        n->m_data += count;
                else
                        n = m_free(n);
        } while (len > 0 && n);
        if (len > 0) {
                (void) m_free(m);
                goto bad;
        }
        m->m_next = n;
        return (m);
bad:
        m_freem(n);
        return (NULL);
}

/*
 * Lesser-used path for M_PREPEND:
 * allocate new mbuf to prepend to chain,
 * copy junk along.
 */
struct mbuf *
m_prepend(struct mbuf *m, int len, int how)
{
        struct mbuf *mn;

        if (m->m_flags & M_PKTHDR)
                MGETHDR(mn, how, m->m_type);
        else
                MGET(mn, how, m->m_type);
        if (mn == NULL) {
                m_freem(m);
                return (NULL);
        }
        if (m->m_flags & M_PKTHDR)
                M_MOVE_PKTHDR(mn, m);
        mn->m_next = m;
        m = mn;
        if (len < MHLEN)
                MH_ALIGN(m, len);
        m->m_len = len;
        return (m);
}

/*
 * "Move" mbuf pkthdr from "from" to "to".
 * "from" must have M_PKTHDR set, and "to" must be empty.
 */
void
m_move_pkthdr(struct mbuf *to, struct mbuf *from)
{
#ifdef MAC
        /*
         * XXXMAC: It could be this should also occur for non-MAC?
         */
        if (to->m_flags & M_PKTHDR)
                m_tag_delete_chain(to, NULL);
#endif
        to->m_flags = (from->m_flags & M_COPYFLAGS) | (to->m_flags & M_EXT);
        if ((to->m_flags & M_EXT) == 0)
                to->m_data = to->m_pktdat;
        to->m_pkthdr = from->m_pkthdr;          /* especially tags */
        SLIST_INIT(&from->m_pkthdr.tags);       /* purge tags from src */
        from->m_flags &= ~M_PKTHDR;
}

/*
 * Routine to copy from device local memory into mbufs.
 * Note that `off' argument is offset into first mbuf of target chain from
 * which to begin copying the data to.
 */
struct mbuf *
m_devget(char *buf, int totlen, int off, struct ifnet *ifp,
         void (*copy)(char *from, caddr_t to, u_int len))
{
        struct mbuf *m;
        struct mbuf *top = NULL, **mp = &top;
        int len;

        if (off < 0 || off > MHLEN)
                return (NULL);

        while (totlen > 0) {
                if (top == NULL) {      /* First one, must be PKTHDR */
                        if (totlen + off >= MINCLSIZE) {
                                m = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR);
                                len = MCLBYTES;
                        } else {
                                m = m_gethdr(M_DONTWAIT, MT_DATA);
                                len = MHLEN;

                                /* Place initial small packet/header at end of mbuf */
                                if (m && totlen + off + max_linkhdr <= MLEN) {
                                        m->m_data += max_linkhdr;
                                        len -= max_linkhdr;
                                }
                        }
                        if (m == NULL)
                                return NULL;
                        m->m_pkthdr.rcvif = ifp;
                        m->m_pkthdr.len = totlen;
                } else {
                        if (totlen + off >= MINCLSIZE) {
                                m = m_getcl(M_DONTWAIT, MT_DATA, 0);
                                len = MCLBYTES;
                        } else {
                                m = m_get(M_DONTWAIT, MT_DATA);
                                len = MLEN;
                        }
                        if (m == NULL) {
                                m_freem(top);
                                return NULL;
                        }
                }
                if (off) {
                        m->m_data += off;
                        len -= off;
                        off = 0;
                }
                m->m_len = len = min(totlen, len);
                if (copy)
                        copy(buf, mtod(m, caddr_t), (u_int)len);
                else
                        bcopy(buf, mtod(m, caddr_t), (u_int)len);
                buf += len;
                *mp = m;
                mp = &m->m_next;
                totlen -= len;
        }
        return (top);
}

/*
 * Copy data from a buffer back into the indicated mbuf chain,
 * starting "off" bytes from the beginning, extending the mbuf
 * chain if necessary.
 */
void
m_copyback(struct mbuf *m0, int off, int len, caddr_t cp)
{
        int mlen;
        struct mbuf *m = m0, *n;
        int totlen = 0;

        if (m0 == NULL)
                return;
        while (off > (mlen = m->m_len)) {
                off -= mlen;
                totlen += mlen;
                if (m->m_next == NULL) {
                        n = m_get(M_DONTWAIT, m->m_type);
                        if (n == NULL)
                                goto out;
                        bzero(mtod(n, caddr_t), MLEN);
                        n->m_len = min(MLEN, len + off);
                        m->m_next = n;
                }
                m = m->m_next;
        }
        while (len > 0) {
                mlen = min (m->m_len - off, len);
                bcopy(cp, off + mtod(m, caddr_t), (u_int)mlen);
                cp += mlen;
                len -= mlen;
                mlen += off;
                off = 0;
                totlen += mlen;
                if (len == 0)
                        break;
                if (m->m_next == NULL) {
                        n = m_get(M_DONTWAIT, m->m_type);
                        if (n == NULL)
                                break;
                        n->m_len = min(MLEN, len);
                        m->m_next = n;
                }
                m = m->m_next;
        }
out:    if (((m = m0)->m_flags & M_PKTHDR) && (m->m_pkthdr.len < totlen))
                m->m_pkthdr.len = totlen;
}

/*
 * Append the specified data to the indicated mbuf chain,
 * Extend the mbuf chain if the new data does not fit in
 * existing space.
 *
 * Return 1 if able to complete the job; otherwise 0.
 */
int
m_append(struct mbuf *m0, int len, c_caddr_t cp)
{
        struct mbuf *m, *n;
        int remainder, space;

        for (m = m0; m->m_next != NULL; m = m->m_next)
                ;
        remainder = len;
        space = M_TRAILINGSPACE(m);
        if (space > 0) {
                /*
                 * Copy into available space.
                 */
                if (space > remainder)
                        space = remainder;
                bcopy(cp, mtod(m, caddr_t) + m->m_len, space);
                m->m_len += space;
                cp += space, remainder -= space;
        }
        while (remainder > 0) {
                /*
                 * Allocate a new mbuf; could check space
                 * and allocate a cluster instead.
                 */
                n = m_get(M_DONTWAIT, m->m_type);
                if (n == NULL)
                        break;
                n->m_len = min(MLEN, remainder);
                bcopy(cp, mtod(n, caddr_t), n->m_len);
                cp += n->m_len, remainder -= n->m_len;
                m->m_next = n;
                m = n;
        }
        if (m0->m_flags & M_PKTHDR)
                m0->m_pkthdr.len += len - remainder;
        return (remainder == 0);
}

/*
 * Defragment an mbuf chain, returning at most maxfrags separate
 * mbufs+clusters.  If this is not possible NULL is returned and
 * the original mbuf chain is left in it's present (potentially
 * modified) state.  We use two techniques: collapsing consecutive
 * mbufs and replacing consecutive mbufs by a cluster.
 *
 * NB: this should really be named m_defrag but that name is taken
 */
struct mbuf *
m_collapse(struct mbuf *m0, int how, int maxfrags)
{
        struct mbuf *m, *n, *n2, **prev;
        u_int curfrags;

        /*
         * Calculate the current number of frags.
         */
        curfrags = 0;
        for (m = m0; m != NULL; m = m->m_next)
                curfrags++;
        /*
         * First, try to collapse mbufs.  Note that we always collapse
         * towards the front so we don't need to deal with moving the
         * pkthdr.  This may be suboptimal if the first mbuf has much
         * less data than the following.
         */
        m = m0;
again:
        for (;;) {
                n = m->m_next;
                if (n == NULL)
                        break;
                if ((m->m_flags & M_RDONLY) == 0 &&
                    n->m_len < M_TRAILINGSPACE(m)) {
                        bcopy(mtod(n, void *), mtod(m, char *) + m->m_len,
                                n->m_len);
                        m->m_len += n->m_len;
                        m->m_next = n->m_next;
                        m_free(n);
                        if (--curfrags <= maxfrags)
                                return m0;
                } else
                        m = n;
        }
        KASSERT(maxfrags > 1,
                ("maxfrags %u, but normal collapse failed", maxfrags));
        /*
         * Collapse consecutive mbufs to a cluster.
         */
        prev = &m0->m_next;             /* NB: not the first mbuf */
        while ((n = *prev) != NULL) {
                if ((n2 = n->m_next) != NULL &&
                    n->m_len + n2->m_len < MCLBYTES) {
                        m = m_getcl(how, MT_DATA, 0);
                        if (m == NULL)
                                goto bad;
                        bcopy(mtod(n, void *), mtod(m, void *), n->m_len);
                        bcopy(mtod(n2, void *), mtod(m, char *) + n->m_len,
                                n2->m_len);
                        m->m_len = n->m_len + n2->m_len;
                        m->m_next = n2->m_next;
                        *prev = m;
                        m_free(n);
                        m_free(n2);
                        if (--curfrags <= maxfrags)     /* +1 cl -2 mbufs */
                                return m0;
                        /*
                         * Still not there, try the normal collapse
                         * again before we allocate another cluster.
                         */
                        goto again;
                }
                prev = &n->m_next;
        }
        /*
         * No place where we can collapse to a cluster; punt.
         * This can occur if, for example, you request 2 frags
         * but the packet requires that both be clusters (we
         * never reallocate the first mbuf to avoid moving the
         * packet header).
         */
bad:
        return NULL;
}

/*
 * Attach the the cluster from *m to *n, set up m_ext in *n
 * and bump the refcount of the cluster.
 */
static void
mb_dupcl(struct mbuf *n, struct mbuf *m)
{
        KASSERT((m->m_flags & M_EXT) == M_EXT, ("%s: M_EXT not set", __func__));
        KASSERT((n->m_flags & M_EXT) == 0, ("%s: M_EXT set", __func__));

        n->m_ext.ext_buf = m->m_ext.ext_buf;
        n->m_ext.ext_size = m->m_ext.ext_size;
        n->m_ext.ext_type = m->m_ext.ext_type;
        n->m_flags |= M_EXT;
}

/*
 * Partition an mbuf chain in two pieces, returning the tail --
 * all but the first len0 bytes.  In case of failure, it returns NULL and
 * attempts to restore the chain to its original state.
 *
 * Note that the resulting mbufs might be read-only, because the new
 * mbuf can end up sharing an mbuf cluster with the original mbuf if
 * the "breaking point" happens to lie within a cluster mbuf. Use the
 * M_WRITABLE() macro to check for this case.
 */
struct mbuf *
m_split(struct mbuf *m0, int len0, int wait)
{
        struct mbuf *m, *n;
        u_int len = len0, remain;

        MBUF_CHECKSLEEP(wait);
        for (m = m0; m && len > m->m_len; m = m->m_next)
                len -= m->m_len;
        if (m == NULL)
                return (NULL);
        remain = m->m_len - len;
        if (m0->m_flags & M_PKTHDR) {
                MGETHDR(n, wait, m0->m_type);
                if (n == NULL)
                        return (NULL);
                n->m_pkthdr.rcvif = m0->m_pkthdr.rcvif;
                n->m_pkthdr.len = m0->m_pkthdr.len - len0;
                m0->m_pkthdr.len = len0;
                if (m->m_flags & M_EXT)
                        goto extpacket;
                if (remain > MHLEN) {
                        /* m can't be the lead packet */
                        MH_ALIGN(n, 0);
                        n->m_next = m_split(m, len, wait);
                        if (n->m_next == NULL) {
                                (void) m_free(n);
                                return (NULL);
                        } else {
                                n->m_len = 0;
                                return (n);
                        }
                } else
                        MH_ALIGN(n, remain);
        } else if (remain == 0) {
                n = m->m_next;
                m->m_next = NULL;
                return (n);
        } else {
                MGET(n, wait, m->m_type);
                if (n == NULL)
                        return (NULL);
                M_ALIGN(n, remain);
        }
extpacket:
        if (m->m_flags & M_EXT) {
                n->m_data = m->m_data + len;
                mb_dupcl(n, m);
        } else {
                bcopy(mtod(m, caddr_t) + len, mtod(n, caddr_t), remain);
        }
        n->m_len = remain;
        m->m_len = len;
        n->m_next = m->m_next;
        m->m_next = NULL;
        return (n);
}

/*
 * Copy a packet header mbuf chain into a completely new chain, including
 * copying any mbuf clusters.  Use this instead of m_copypacket() when
 * you need a writable copy of an mbuf chain.
 */
struct mbuf *
m_dup(struct mbuf *m, int how)
{
        struct mbuf **p, *top = NULL;
        int remain, moff, nsize;

        MBUF_CHECKSLEEP(how);
        /* Sanity check */
        if (m == NULL)
                return (NULL);
        M_ASSERTPKTHDR(m);

        /* While there's more data, get a new mbuf, tack it on, and fill it */
        remain = m->m_pkthdr.len;
        moff = 0;
        p = &top;
        while (remain > 0 || top == NULL) {     /* allow m->m_pkthdr.len == 0 */
                struct mbuf *n;

                /* Get the next new mbuf */
                if (remain >= MINCLSIZE) {
                        n = m_getcl(how, m->m_type, 0);
                        nsize = MCLBYTES;
                } else {
                        n = m_get(how, m->m_type);
                        nsize = MLEN;
                }
                if (n == NULL)
                        goto nospace;

                if (top == NULL) {              /* First one, must be PKTHDR */
                        if (!m_dup_pkthdr(n, m, how)) {
                                m_free(n);
                                goto nospace;
                        }
                        if ((n->m_flags & M_EXT) == 0)
                                nsize = MHLEN;
                }
                n->m_len = 0;

                /* Link it into the new chain */
                *p = n;
                p = &n->m_next;

                /* Copy data from original mbuf(s) into new mbuf */
                while (n->m_len < nsize && m != NULL) {
                        int chunk = min(nsize - n->m_len, m->m_len - moff);

                        bcopy(m->m_data + moff, n->m_data + n->m_len, chunk);
                        moff += chunk;
                        n->m_len += chunk;
                        remain -= chunk;
                        if (moff == m->m_len) {
                                m = m->m_next;
                                moff = 0;
                        }
                }

                /* Check correct total mbuf length */
                KASSERT((remain > 0 && m != NULL) || (remain == 0 && m == NULL),
                        ("%s: bogus m_pkthdr.len", __func__));
        }
        return (top);

nospace:
        m_freem(top);
        return (NULL);
}

/*
 * Create a writable copy of the mbuf chain.  While doing this
 * we compact the chain with a goal of producing a chain with
 * at most two mbufs.  The second mbuf in this chain is likely
 * to be a cluster.  The primary purpose of this work is to create
 * a writable packet for encryption, compression, etc.  The
 * secondary goal is to linearize the data so the data can be
 * passed to crypto hardware in the most efficient manner possible.
 */
struct mbuf *
m_unshare(struct mbuf *m0, int how)
{
        struct mbuf *m, *mprev;
        struct mbuf *n, *mfirst, *mlast;
        int len, off;

        mprev = NULL;
        for (m = m0; m != NULL; m = mprev->m_next) {
                /*
                 * Regular mbufs are ignored unless there's a cluster
                 * in front of it that we can use to coalesce.  We do
                 * the latter mainly so later clusters can be coalesced
                 * also w/o having to handle them specially (i.e. convert
                 * mbuf+cluster -> cluster).  This optimization is heavily
                 * influenced by the assumption that we're running over
                 * Ethernet where MCLBYTES is large enough that the max
                 * packet size will permit lots of coalescing into a
                 * single cluster.  This in turn permits efficient
                 * crypto operations, especially when using hardware.
                 */
                if ((m->m_flags & M_EXT) == 0) {
                        if (mprev && (mprev->m_flags & M_EXT) &&
                            m->m_len <= M_TRAILINGSPACE(mprev)) {
                                /* XXX: this ignores mbuf types */
                                memcpy(mtod(mprev, caddr_t) + mprev->m_len,
                                       mtod(m, caddr_t), m->m_len);
                                mprev->m_len += m->m_len;
                                mprev->m_next = m->m_next;      /* unlink from chain */
                                m_free(m);                      /* reclaim mbuf */
#if 0
                                newipsecstat.ips_mbcoalesced++;
#endif
                        } else {
                                mprev = m;
                        }
                        continue;
                }
                /*
                 * Writable mbufs are left alone (for now).
                 */
                if (M_WRITABLE(m)) {
                        mprev = m;
                        continue;
                }

                /*
                 * Not writable, replace with a copy or coalesce with
                 * the previous mbuf if possible (since we have to copy
                 * it anyway, we try to reduce the number of mbufs and
                 * clusters so that future work is easier).
                 */
                KASSERT(m->m_flags & M_EXT, ("m_flags 0x%x", m->m_flags));
                /* NB: we only coalesce into a cluster or larger */
                if (mprev != NULL && (mprev->m_flags & M_EXT) &&
                    m->m_len <= M_TRAILINGSPACE(mprev)) {
                        /* XXX: this ignores mbuf types */
                        memcpy(mtod(mprev, caddr_t) + mprev->m_len,
                               mtod(m, caddr_t), m->m_len);
                        mprev->m_len += m->m_len;
                        mprev->m_next = m->m_next;      /* unlink from chain */
                        m_free(m);                      /* reclaim mbuf */
#if 0
                        newipsecstat.ips_clcoalesced++;
#endif
                        continue;
                }

                /*
                 * Allocate new space to hold the copy...
                 */
                /* XXX why can M_PKTHDR be set past the first mbuf? */
                if (mprev == NULL && (m->m_flags & M_PKTHDR)) {
                        /*
                         * NB: if a packet header is present we must
                         * allocate the mbuf separately from any cluster
                         * because M_MOVE_PKTHDR will smash the data
                         * pointer and drop the M_EXT marker.
                         */
                        MGETHDR(n, how, m->m_type);
                        if (n == NULL) {
                                m_freem(m0);
                                return (NULL);
                        }
                        M_MOVE_PKTHDR(n, m);
                        MCLGET(n, how);
                        if ((n->m_flags & M_EXT) == 0) {
                                m_free(n);
                                m_freem(m0);
                                return (NULL);
                        }
                } else {
                        n = m_getcl(how, m->m_type, m->m_flags);
                        if (n == NULL) {
                                m_freem(m0);
                                return (NULL);
                        }
                }
                /*
                 * ... and copy the data.  We deal with jumbo mbufs
                 * (i.e. m_len > MCLBYTES) by splitting them into
                 * clusters.  We could just malloc a buffer and make
                 * it external but too many device drivers don't know
                 * how to break up the non-contiguous memory when
                 * doing DMA.
                 */
                len = m->m_len;
                off = 0;
                mfirst = n;
                mlast = NULL;
                for (;;) {
                        int cc = min(len, MCLBYTES);
                        memcpy(mtod(n, caddr_t), mtod(m, caddr_t) + off, cc);
                        n->m_len = cc;
                        if (mlast != NULL)
                                mlast->m_next = n;
                        mlast = n;
#if 0
                        newipsecstat.ips_clcopied++;
#endif

                        len -= cc;
                        if (len <= 0)
                                break;
                        off += cc;

                        n = m_getcl(how, m->m_type, m->m_flags);
                        if (n == NULL) {
                                m_freem(mfirst);
                                m_freem(m0);
                                return (NULL);
                        }
                }
                n->m_next = m->m_next;
                if (mprev == NULL)
                        m0 = mfirst;            /* new head of chain */
                else
                        mprev->m_next = mfirst; /* replace old mbuf */
                m_free(m);                      /* release old mbuf */
                mprev = mfirst;
        }
        return (m0);
}

/*
 * Set the m_data pointer of a newly-allocated mbuf
 * to place an object of the specified size at the
 * end of the mbuf, longword aligned.
 */
void
m_align(struct mbuf *m, int len)
{
        int adjust;

        if (m->m_flags & M_EXT)
                adjust = m->m_ext.ext_size - len;
        else if (m->m_flags & M_PKTHDR)
                adjust = MHLEN - len;
        else
                adjust = MLEN - len;
        m->m_data += adjust &~ (sizeof(long)-1);
}

/*
 * Copy an entire packet, including header (which must be present).
 * An optimization of the common case `m_copym(m, 0, M_COPYALL, how)'.
 * Note that the copy is read-only, because clusters are not copied,
 * only their reference counts are incremented.
 * Preserve alignment of the first mbuf so if the creator has left
 * some room at the beginning (e.g. for inserting protocol headers)
 * the copies still have the room available.
 */
struct mbuf *
m_copypacket(struct mbuf *m, int how)
{
        struct mbuf *top, *n, *o;

        MBUF_CHECKSLEEP(how);
        MGET(n, how, m->m_type);
        top = n;
        if (n == NULL)
                goto nospace;

        if (!m_dup_pkthdr(n, m, how))
                goto nospace;
        n->m_len = m->m_len;
        if (m->m_flags & M_EXT) {
                n->m_data = m->m_data;
                mb_dupcl(n, m);
        } else {
                n->m_data = n->m_pktdat + (m->m_data - m->m_pktdat );
                bcopy(mtod(m, char *), mtod(n, char *), n->m_len);
        }

        m = m->m_next;
        while (m) {
                MGET(o, how, m->m_type);
                if (o == NULL)
                        goto nospace;

                n->m_next = o;
                n = n->m_next;

                n->m_len = m->m_len;
                if (m->m_flags & M_EXT) {
                        n->m_data = m->m_data;
                        mb_dupcl(n, m);
                } else {
                        bcopy(mtod(m, char *), mtod(n, char *), n->m_len);
                }

                m = m->m_next;
        }
        return top;
nospace:
        m_freem(top);
        return (NULL);
}