BSD 4_3_Reno release

[unix-history] / usr / src / sys / kern / uipc_mbuf.c

diff --git a/usr/src/sys/kern/uipc_mbuf.c b/usr/src/sys/kern/uipc_mbuf.c
index 91e9178..ba17d3f 100644 (file)
--- a/usr/src/sys/kern/uipc_mbuf.c
+++ b/usr/src/sys/kern/uipc_mbuf.c
@@ -1,124 +1,143 @@
-/*     uipc_mbuf.c     1.27    82/01/24        */
-
-#include "../h/param.h"
-#include "../h/dir.h"
-#include "../h/user.h"
-#include "../h/proc.h"
-#include "../h/pte.h"
-#include "../h/cmap.h"
-#include "../h/map.h"
-#include "../h/mbuf.h"
-#include "../net/in_systm.h"           /* XXX */
-#include "../h/vm.h"
+/*
+ * Copyright (c) 1982, 1986, 1988 Regents of the University of California.
+ * All rights reserved.
+ *
+ * Redistribution is only permitted until one year after the first shipment
+ * of 4.4BSD by the Regents.  Otherwise, redistribution and use in source and
+ * binary forms are permitted provided that: (1) source distributions retain
+ * this entire copyright notice and comment, and (2) distributions including
+ * binaries display the following acknowledgement:  This product includes
+ * software developed by the University of California, Berkeley and its
+ * contributors'' in the documentation or other materials provided with the
+ * distribution and in all advertising materials mentioning features or use
+ * of this software.  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 AS IS'' AND WITHOUT ANY EXPRESS OR IMPLIED
+ * WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ *     @(#)uipc_mbuf.c 7.17 (Berkeley) 7/25/90
+ */
+
+#include "param.h"
+#include "user.h"
+#include "proc.h"
+#include "cmap.h"
+#include "malloc.h"
+#include "map.h"
+#define MBTYPES
+#include "mbuf.h"
+#include "vm.h"
+#include "kernel.h"
+#include "syslog.h"
+#include "domain.h"
+#include "protosw.h"
+#include "machine/pte.h"

 
 mbinit()
 {
 
 mbinit()
 {

+       int s;

 
 

-COUNT(MBINIT);
-       if (m_reserve(32) == 0)
-               goto bad;
-       if (m_clalloc(4, MPG_MBUFS) == 0)
-               goto bad;
-       if (m_clalloc(32, MPG_CLUSTERS) == 0)
+#if MCLBYTES < 4096
+#define NCL_INIT       (4096/CLBYTES)
+#else
+#define NCL_INIT       1
+#endif
+       s = splimp();
+       if (m_clalloc(NCL_INIT, M_DONTWAIT) == 0)

                goto bad;
                goto bad;

+       splx(s);

        return;
 bad:
        panic("mbinit");
 }
 
        return;
 bad:
        panic("mbinit");
 }
 

-caddr_t
-m_clalloc(ncl, how)
+/*
+ * Allocate some number of mbuf clusters
+ * and place on cluster free list.
+ * Must be called at splimp.
+ */
+/* ARGSUSED */
+m_clalloc(ncl, canwait)

        register int ncl;
        register int ncl;

-       int how;

 {
        int npg, mbx;
 {
        int npg, mbx;

-       register struct mbuf *m;
+       register caddr_t p;

        register int i;
        register int i;

-       int s;
+       static int logged;

 
 

-COUNT(M_CLALLOC);

        npg = ncl * CLSIZE;
        npg = ncl * CLSIZE;

-       s = splimp();           /* careful: rmalloc isn't reentrant */
-       mbx = rmalloc(mbmap, npg);
-       splx(s);
-       if (mbx == 0)
+       mbx = rmalloc(mbmap, (long)npg);
+       if (mbx == 0) {
+               if (logged == 0) {
+                       logged++;
+                       log(LOG_ERR, "mbuf map full\n");
+               }

                return (0);
                return (0);

-       m = cltom(mbx / CLSIZE);
-       if (memall(&Mbmap[mbx], ncl * CLSIZE, proc, CSYS) == 0)
+       }
+       p = cltom(mbx * NBPG / MCLBYTES);
+       if (memall(&Mbmap[mbx], npg, proc, CSYS) == 0) {
+               rmfree(mbmap, (long)npg, (long)mbx);

                return (0);
                return (0);

-       vmaccess(&Mbmap[mbx], (caddr_t)m, npg);
-       switch (how) {
-
-       case MPG_CLUSTERS:
-               s = splimp();
-               for (i = 0; i < ncl; i++) {
-                       m->m_off = 0;
-                       m->m_next = mclfree;
-                       mclfree = m;
-                       m += CLBYTES / sizeof (*m);
-                       nmclfree++;
-               }
-               mbstat.m_clusters += ncl;
-               splx(s);
-               break;
-
-       case MPG_MBUFS:
-               for (i = ncl * CLBYTES / sizeof (*m); i > 0; i--) {
-                       m->m_off = 0;
-                       m->m_free = 0;
-                       (void) m_free(m);
-                       m++;
-               }
-               mbstat.m_clusters += ncl;
-               break;

        }
        }

-       return ((caddr_t)m);
-}
-
-m_pgfree(addr, n)
-       caddr_t addr;
-       int n;
-{
-
-COUNT(M_PGFREE);
+       vmaccess(&Mbmap[mbx], p, npg);
+       ncl = ncl * CLBYTES / MCLBYTES;
+       for (i = 0; i < ncl; i++) {
+               ((union mcluster *)p)->mcl_next = mclfree;
+               mclfree = (union mcluster *)p;
+               p += MCLBYTES;
+               mbstat.m_clfree++;
+       }
+       mbstat.m_clusters += ncl;
+       return (1);

 }
 
 }
 

-m_expand()
+/*
+ * When MGET failes, ask protocols to free space when short of memory,
+ * then re-attempt to allocate an mbuf.
+ */
+struct mbuf *
+m_retry(i, t)
+       int i, t;

 {
 {

+       register struct mbuf *m;

 
 

-COUNT(M_EXPAND);
-       if (mbstat.m_bufs >= mbstat.m_hiwat)
-               return (0);
-       if (m_clalloc(1, MPG_MBUFS) == 0)
-               goto steal;
-       return (1);
-steal:
-       /* should ask protocols to free code */
-       return (0);
+       m_reclaim();
+#define m_retry(i, t)  (struct mbuf *)0
+       MGET(m, i, t);
+#undef m_retry
+       return (m);

 }
 
 }
 

-/* NEED SOME WAY TO RELEASE SPACE */
-

 /*
 /*

- * Space reservation routines
+ * As above; retry an MGETHDR.

  */
  */

-m_reserve(mbufs)
-       int mbufs;
+struct mbuf *
+m_retryhdr(i, t)
+       int i, t;

 {
 {

+       register struct mbuf *m;

 
 

-       if (mbstat.m_lowat + (mbufs>>1) > (NMBCLUSTERS-32) * CLBYTES) 
-               return (0);
-       mbstat.m_hiwat += mbufs;
-       mbstat.m_lowat = mbstat.m_hiwat >> 1;
-       return (1);
+       m_reclaim();
+#define m_retryhdr(i, t) (struct mbuf *)0
+       MGETHDR(m, i, t);
+#undef m_retryhdr
+       return (m);

 }
 
 }
 

-m_release(mbufs)
-       int mbufs;
+m_reclaim()

 {
 {

-
-       mbstat.m_hiwat -= mbufs;
-       mbstat.m_lowat = mbstat.m_hiwat >> 1;
+       register struct domain *dp;
+       register struct protosw *pr;
+       int s = splimp();
+
+       for (dp = domains; dp; dp = dp->dom_next)
+               for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
+                       if (pr->pr_drain)
+                               (*pr->pr_drain)();
+       splx(s);
+       mbstat.m_drain++;

 }
 
 /*
 }
 
 /*


@@ -127,27 +146,34 @@ m_release(mbufs)
  * for critical paths.
  */
 struct mbuf *
  * for critical paths.
  */
 struct mbuf *

-m_get(canwait)
-       int canwait;
+m_get(canwait, type)
+       int canwait, type;
+{
+       register struct mbuf *m;
+
+       MGET(m, canwait, type);
+       return (m);
+}
+
+struct mbuf *
+m_gethdr(canwait, type)
+       int canwait, type;

 {
        register struct mbuf *m;
 
 {
        register struct mbuf *m;
 

-COUNT(M_GET);
-       MGET(m, canwait);
+       MGETHDR(m, canwait, type);

        return (m);
 }
 
 struct mbuf *
        return (m);
 }
 
 struct mbuf *

-m_getclr(canwait)
-       int canwait;
+m_getclr(canwait, type)
+       int canwait, type;

 {
        register struct mbuf *m;
 
 {
        register struct mbuf *m;
 

-COUNT(M_GETCLR);
-       m = m_get(canwait);
+       MGET(m, canwait, type);

        if (m == 0)
                return (0);
        if (m == 0)
                return (0);

-       m->m_off = MMINOFF;

        bzero(mtod(m, caddr_t), MLEN);
        return (m);
 }
        bzero(mtod(m, caddr_t), MLEN);
        return (m);
 }


@@ -158,65 +184,80 @@ m_free(m)
 {
        register struct mbuf *n;
 
 {
        register struct mbuf *n;
 

-COUNT(M_FREE);

        MFREE(m, n);
        return (n);
 }
 
        MFREE(m, n);
        return (n);
 }
 

-/*ARGSUSED*/
-struct mbuf *
-m_more(type)
-       int type;
-{
-       register struct mbuf *m;
-
-COUNT(M_MORE);
-       if (!m_expand()) {
-               mbstat.m_drops++;
-               return (NULL);
-       }
-#define m_more(x) (panic("m_more"), (struct mbuf *)0)
-       MGET(m, type);
-#undef m_more
-       return (m);
-}
-

 m_freem(m)
        register struct mbuf *m;
 {
        register struct mbuf *n;
 m_freem(m)
        register struct mbuf *m;
 {
        register struct mbuf *n;

-       register int s;

 
 

-COUNT(M_FREEM);

        if (m == NULL)
                return;
        if (m == NULL)
                return;

-       s = splimp();

        do {
                MFREE(m, n);
        } while (m = n);
        do {
                MFREE(m, n);
        } while (m = n);

-       splx(s);

 }
 
 /*
  * Mbuffer utility routines.
  */
 }
 
 /*
  * Mbuffer utility routines.
  */

+
+/*
+ * Lesser-used path for M_PREPEND:
+ * allocate new mbuf to prepend to chain,
+ * copy junk along.
+ */
+struct mbuf *
+m_prepend(m, len, how)
+       register struct mbuf *m;
+       int len, how;
+{
+       struct mbuf *mn;
+
+       MGET(mn, how, m->m_type);
+       if (mn == (struct mbuf *)NULL) {
+               m_freem(m);
+               return ((struct mbuf *)NULL);
+       }
+       if (m->m_flags & M_PKTHDR) {
+               M_COPY_PKTHDR(mn, m);
+               m->m_flags &= ~M_PKTHDR;
+       }
+       mn->m_next = m;
+       m = mn;
+       if (len < MHLEN)
+               MH_ALIGN(m, len);
+       m->m_len = len;
+       return (m);
+}
+
+/*
+ * Make a copy of an mbuf chain starting "off0" bytes from the beginning,
+ * continuing for "len" bytes.  If len is M_COPYALL, copy to end of mbuf.
+ * The wait parameter is a choice of M_WAIT/M_DONTWAIT from caller.
+ */
+int MCFail;
+

 struct mbuf *
 struct mbuf *

-m_copy(m, off, len)
+m_copym(m, off0, len, wait)

        register struct mbuf *m;
        register struct mbuf *m;

-       int off;
+       int off0, wait;

        register int len;
 {
        register struct mbuf *n, **np;
        register int len;
 {
        register struct mbuf *n, **np;

-       struct mbuf *top, *p;
-COUNT(M_COPY);
+       register int off = off0;
+       struct mbuf *top;
+       int copyhdr = 0;

 
 

-       if (len == 0)
-               return (0);

        if (off < 0 || len < 0)
        if (off < 0 || len < 0)

-               panic("m_copy");
+               panic("m_copym");
+       if (off == 0 && m->m_flags & M_PKTHDR)
+               copyhdr = 1;

        while (off > 0) {
                if (m == 0)
        while (off > 0) {
                if (m == 0)

-                       panic("m_copy");
+                       panic("m_copym");

                if (off < m->m_len)
                        break;
                off -= m->m_len;
                if (off < m->m_len)
                        break;
                off -= m->m_len;


@@ -227,64 +268,117 @@ COUNT(M_COPY);
        while (len > 0) {
                if (m == 0) {
                        if (len != M_COPYALL)
        while (len > 0) {
                if (m == 0) {
                        if (len != M_COPYALL)

-                               panic("m_copy");
+                               panic("m_copym");

                        break;
                }
                        break;
                }

-               MGET(n, 1);
+               MGET(n, wait, m->m_type);

                *np = n;
                if (n == 0)
                        goto nospace;
                *np = n;
                if (n == 0)
                        goto nospace;

+               if (copyhdr) {
+                       M_COPY_PKTHDR(n, m);
+                       if (len == M_COPYALL)
+                               n->m_pkthdr.len -= off0;
+                       else
+                               n->m_pkthdr.len = len;
+                       copyhdr = 0;
+               }

                n->m_len = MIN(len, m->m_len - off);
                n->m_len = MIN(len, m->m_len - off);

-               if (m->m_off > MMAXOFF) {
-                       p = mtod(m, struct mbuf *);
-                       n->m_off = ((int)p - (int)n) + off;
-                       mclrefcnt[mtocl(p)]++;
-               } else {
-                       n->m_off = MMINOFF;
+               if (m->m_flags & M_EXT) {
+                       n->m_data = m->m_data + off;
+                       mclrefcnt[mtocl(m->m_ext.ext_buf)]++;
+                       n->m_ext = m->m_ext;
+                       n->m_flags |= M_EXT;
+               } else

                        bcopy(mtod(m, caddr_t)+off, mtod(n, caddr_t),
                            (unsigned)n->m_len);
                        bcopy(mtod(m, caddr_t)+off, mtod(n, caddr_t),
                            (unsigned)n->m_len);

-               }

                if (len != M_COPYALL)
                        len -= n->m_len;
                off = 0;
                m = m->m_next;
                np = &n->m_next;
        }
                if (len != M_COPYALL)
                        len -= n->m_len;
                off = 0;
                m = m->m_next;
                np = &n->m_next;
        }

+       if (top == 0)
+               MCFail++;

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

+       MCFail++;

        return (0);
 }
 
        return (0);
 }
 

+/*
+ * Copy data from an mbuf chain starting "off" bytes from the beginning,
+ * continuing for "len" bytes, into the indicated buffer.
+ */
+m_copydata(m, off, len, cp)
+       register struct mbuf *m;
+       register int off;
+       register int len;
+       caddr_t cp;
+{
+       register unsigned count;
+
+       if (off < 0 || len < 0)
+               panic("m_copydata");
+       while (off > 0) {
+               if (m == 0)
+                       panic("m_copydata");
+               if (off < m->m_len)
+                       break;
+               off -= m->m_len;
+               m = m->m_next;
+       }
+       while (len > 0) {
+               if (m == 0)
+                       panic("m_copydata");
+               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.
+ */

 m_cat(m, n)
        register struct mbuf *m, *n;
 {
 m_cat(m, n)
        register struct mbuf *m, *n;
 {

-

        while (m->m_next)
                m = m->m_next;
        while (m->m_next)
                m = m->m_next;

-       while (n)
-               if (m->m_off + m->m_len + n->m_len <= MMAXOFF) {
-                       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);
-               } else {
+       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;
                        m->m_next = n;

-                       m = n;
-                       n = m->m_next;
+                       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);
+       }

 }
 
 }
 

-m_adj(mp, len)
+m_adj(mp, req_len)

        struct mbuf *mp;
        struct mbuf *mp;

-       register int len;

 {
 {

-       register struct mbuf *m, *n;
+       register int len = req_len;
+       register struct mbuf *m;
+       register count;

 
 

-COUNT(M_ADJ);

        if ((m = mp) == NULL)
                return;
        if (len >= 0) {
        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;
                while (m != NULL && len > 0) {
                        if (m->m_len <= len) {
                                len -= m->m_len;


@@ -292,66 +386,123 @@ COUNT(M_ADJ);
                                m = m->m_next;
                        } else {
                                m->m_len -= len;
                                m = m->m_next;
                        } else {
                                m->m_len -= len;

-                               m->m_off += len;
-                               break;
+                               m->m_data += len;
+                               len = 0;

                        }
                }
                        }
                }

+               m = mp;
+               if (mp->m_flags & M_PKTHDR)
+                       m->m_pkthdr.len -= (req_len - len);

        } else {
        } else {

-               /* a 2 pass algorithm might be better */
+               /*
+                * 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;
                len = -len;

-               while (len > 0 && m->m_len != 0) {
-                       while (m != NULL && m->m_len != 0) {
-                               n = m;
-                               m = m->m_next;
-                       }
-                       if (n->m_len <= len) {
-                               len -= n->m_len;
-                               n->m_len = 0;
-                               m = mp;
-                       } else {
-                               n->m_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)->m_flags & M_PKTHDR)
+                               m->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;

                                break;
                        }
                                break;
                        }

+                       count -= m->m_len;

                }
                }

+               while (m = m->m_next)
+                       m->m_len = 0;

        }
 }
 
        }
 }
 

+/*
+ * 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.
+ */
+int MPFail;
+

 struct mbuf *
 struct mbuf *

-m_pullup(m0, len)
-       struct mbuf *m0;
+m_pullup(n, len)
+       register struct mbuf *n;

        int len;
 {
        int len;
 {

-       register struct mbuf *m, *n;
-       int cnt;
-
-       if (len > MLEN)
-               goto bad;
-       MGET(m, 0);
-       if (m == 0)
-               goto bad;
-       m->m_off = MMINOFF;
-       m->m_len = 0;
-       n = m0;
+       register struct mbuf *m;
+       register 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 == 0)
+                       goto bad;
+               m->m_len = 0;
+               if (n->m_flags & M_PKTHDR) {
+                       M_COPY_PKTHDR(m, n);
+                       n->m_flags &= ~M_PKTHDR;
+               }
+       }
+       space = &m->m_dat[MLEN] - (m->m_data + m->m_len);

        do {
        do {

-               cnt = MIN(MLEN - m->m_len, len);
-               if (cnt > n->m_len)
-                       cnt = n->m_len;
-               bcopy(mtod(n, caddr_t), mtod(m, caddr_t)+m->m_len, cnt);
-               len -= cnt;
-               m->m_len += cnt;
-               n->m_off += cnt;
-               n->m_len -= cnt;
+               count = min(min(max(len, max_protohdr), space), n->m_len);
+               bcopy(mtod(n, caddr_t), mtod(m, caddr_t) + m->m_len,
+                 (unsigned)count);
+               len -= count;
+               m->m_len += count;
+               n->m_len -= count;
+               space -= count;

                if (n->m_len)
                if (n->m_len)

-                       break;
-               n = m_free(n);
-       } while (n);
-       if (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:
                (void) m_free(m);
                goto bad;
        }
        m->m_next = n;
        return (m);
 bad:

-       m_freem(m0);
+       m_freem(n);
+       MPFail++;

        return (0);
 }
        return (0);
 }