BSD 4_4_Lite1 release

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

diff --git a/usr/src/sys/kern/vfs_bio.c b/usr/src/sys/kern/vfs_bio.c
index 3acf1ff..ec5c962 100644 (file)
--- a/usr/src/sys/kern/vfs_bio.c
+++ b/usr/src/sys/kern/vfs_bio.c
@@ -1,768 +1,339 @@
-/*
- * Copyright (c) 1982, 1986, 1989 Regents of the University of California.
- * All rights reserved.
+/*-
+ * Copyright (c) 1982, 1986, 1989, 1993
+ *     The Regents of the University of California.  All rights reserved.
+ * (c) UNIX System Laboratories, Inc.
+ * All or some portions of this file are derived from material licensed
+ * to the University of California by American Telephone and Telegraph
+ * Co. or Unix System Laboratories, Inc. and are reproduced herein with
+ * the permission of UNIX System Laboratories, Inc.
+ *
+ * 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.
+ * 3. All advertising materials mentioning features or use of this software
+ *    must display the following acknowledgement:
+ *     This product includes software developed by the University of
+ *     California, Berkeley and its contributors.
+ * 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.

  *
  *

- * Redistribution and use in source and binary forms are permitted
- * provided that the above copyright notice and this paragraph are
- * duplicated in all such forms and that any documentation,
- * advertising materials, and other materials related to such
- * distribution and use acknowledge that the software was developed
- * by the University of California, Berkeley.  The name of the
- * University may not 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.
+ * 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.

  *
  *

- *     @(#)vfs_bio.c   7.10 (Berkeley) %G%
+ *     from: @(#)vfs_bio.c     8.6 (Berkeley) 1/11/94

  */
 
  */
 

-#include "param.h"
-#include "user.h"
-#include "buf.h"
-#include "vnode.h"
-#include "trace.h"
-#include "ucred.h"
+#include <sys/param.h>
+#include <sys/systm.h>
+#include <sys/proc.h>
+#include <sys/buf.h>
+#include <sys/vnode.h>
+#include <sys/mount.h>
+#include <sys/trace.h>
+#include <sys/malloc.h>
+#include <sys/resourcevar.h>

 
 /*
 
 /*

- * Read in (if necessary) the block and return a buffer pointer.
+ * Definitions for the buffer hash lists.

  */
  */

-bread(vp, blkno, size, cred, bpp)
-       struct vnode *vp;
-       daddr_t blkno;
-       int size;
-       struct ucred *cred;
-       struct buf **bpp;
-#ifdef SECSIZE
-       long secsize;
-#endif SECSIZE
-{
-       register struct buf *bp;
-
-       if (size == 0)
-               panic("bread: size 0");
-#ifdef SECSIZE
-       bp = getblk(dev, blkno, size, secsize);
-#else SECSIZE
-       *bpp = bp = getblk(vp, blkno, size);
-#endif SECSIZE
-       if (bp->b_flags&(B_DONE|B_DELWRI)) {
-               trace(TR_BREADHIT, pack(vp->v_mount->m_fsid[0], size), blkno);
-               return (0);
-       }
-       bp->b_flags |= B_READ;
-       if (bp->b_bcount > bp->b_bufsize)
-               panic("bread");
-       if (bp->b_rcred == NOCRED && cred != NOCRED) {
-               crhold(cred);
-               bp->b_rcred = cred;
-       }
-       VOP_STRATEGY(bp);
-       trace(TR_BREADMISS, pack(vp->v_mount->m_fsid[0], size), blkno);
-       u.u_ru.ru_inblock++;            /* pay for read */
-       return (biowait(bp));
-}
+#define        BUFHASH(dvp, lbn)       \
+       (&bufhashtbl[((int)(dvp) / sizeof(*(dvp)) + (int)(lbn)) & bufhash])
+LIST_HEAD(bufhashhdr, buf) *bufhashtbl, invalhash;
+u_long bufhash;

 
 /*
 
 /*

- * Read in the block, like bread, but also start I/O on the
- * read-ahead block (which is not allocated to the caller)
+ * Insq/Remq for the buffer hash lists.

  */
  */

-breada(vp, blkno, size, rablkno, rabsize, cred, bpp)
-       struct vnode *vp;
-       daddr_t blkno; int size;
-#ifdef SECSIZE
-       long secsize;
-#endif SECSIZE
-       daddr_t rablkno; int rabsize;
-       struct ucred *cred;
-       struct buf **bpp;
-{
-       register struct buf *bp, *rabp;
+#define        binshash(bp, dp)        LIST_INSERT_HEAD(dp, bp, b_hash)
+#define        bremhash(bp)            LIST_REMOVE(bp, b_hash)

 
 

-       bp = NULL;
-       /*
-        * If the block isn't in core, then allocate
-        * a buffer and initiate i/o (getblk checks
-        * for a cache hit).
-        */
-       if (!incore(vp, blkno)) {
-               *bpp = bp = getblk(vp, blkno, size);
-#endif SECSIZE
-               if ((bp->b_flags&(B_DONE|B_DELWRI)) == 0) {
-                       bp->b_flags |= B_READ;
-                       if (bp->b_bcount > bp->b_bufsize)
-                               panic("breada");
-                       if (bp->b_rcred == NOCRED && cred != NOCRED) {
-                               crhold(cred);
-                               bp->b_rcred = cred;
-                       }
-                       VOP_STRATEGY(bp);
-                       trace(TR_BREADMISS, pack(vp->v_mount->m_fsid[0], size),
-                           blkno);
-                       u.u_ru.ru_inblock++;            /* pay for read */
-               } else
-                       trace(TR_BREADHIT, pack(vp->v_mount->m_fsid[0], size),
-                           blkno);
-       }
+/*
+ * Definitions for the buffer free lists.
+ */
+#define        BQUEUES         4               /* number of free buffer queues */

 
 

-       /*
-        * If there's a read-ahead block, start i/o
-        * on it also (as above).
-        */
-       if (rablkno && !incore(vp, rablkno)) {
-               rabp = getblk(vp, rablkno, rabsize);
-#endif SECSIZE
-               if (rabp->b_flags & (B_DONE|B_DELWRI)) {
-                       brelse(rabp);
-                       trace(TR_BREADHITRA,
-                           pack(vp->v_mount->m_fsid[0], rabsize), blkno);
-               } else {
-                       rabp->b_flags |= B_READ|B_ASYNC;
-                       if (rabp->b_bcount > rabp->b_bufsize)
-                               panic("breadrabp");
-                       if (bp->b_rcred == NOCRED && cred != NOCRED) {
-                               crhold(cred);
-                               bp->b_rcred = cred;
-                       }
-                       VOP_STRATEGY(rabp);
-                       trace(TR_BREADMISSRA,
-                           pack(vp->v_mount->m_fsid[0], rabsize), rablock);
-                       u.u_ru.ru_inblock++;            /* pay in advance */
-               }
-       }
+#define        BQ_LOCKED       0               /* super-blocks &c */
+#define        BQ_LRU          1               /* lru, useful buffers */
+#define        BQ_AGE          2               /* rubbish */
+#define        BQ_EMPTY        3               /* buffer headers with no memory */

 
 

-       /*
-        * If block was in core, let bread get it.
-        * If block wasn't in core, then the read was started
-        * above, and just wait for it.
-        */
-       if (bp == NULL)
-#ifdef SECSIZE
-               return (bread(dev, blkno, size, secsize));
-#else SECSIZE
-               return (bread(vp, blkno, size, cred, bpp));
-       return (biowait(bp));
-}
+TAILQ_HEAD(bqueues, buf) bufqueues[BQUEUES];
+int needbuffer;

 
 /*
 
 /*

- * Write the buffer, waiting for completion.
- * Then release the buffer.
+ * Insq/Remq for the buffer free lists.

  */
  */

-bwrite(bp)
-       register struct buf *bp;
+#define        binsheadfree(bp, dp)    TAILQ_INSERT_HEAD(dp, bp, b_freelist)
+#define        binstailfree(bp, dp)    TAILQ_INSERT_TAIL(dp, bp, b_freelist)
+
+void
+bremfree(bp)
+       struct buf *bp;

 {
 {

-       register int flag;
-       int error;
-
-       flag = bp->b_flags;
-       bp->b_flags &= ~(B_READ | B_DONE | B_ERROR | B_DELWRI);
-       if ((flag&B_DELWRI) == 0)
-               u.u_ru.ru_oublock++;            /* noone paid yet */
-       trace(TR_BWRITE,
-           pack(bp->b_vp->v_mount->m_fsid[0], bp->b_bcount), bp->b_blkno);
-       if (bp->b_bcount > bp->b_bufsize)
-               panic("bwrite");
-       VOP_STRATEGY(bp);
+       struct bqueues *dp = NULL;

 
        /*
 
        /*

-        * If the write was synchronous, then await i/o completion.
-        * If the write was "delayed", then we put the buffer on
-        * the q of blocks awaiting i/o completion status.
+        * We only calculate the head of the freelist when removing
+        * the last element of the list as that is the only time that
+        * it is needed (e.g. to reset the tail pointer).
+        *
+        * NB: This makes an assumption about how tailq's are implemented.

         */
         */

-       if ((flag&B_ASYNC) == 0) {
-               error = biowait(bp);
-               brelse(bp);
-       } else if (flag & B_DELWRI) {
-               bp->b_flags |= B_AGE;
-               error = 0;
+       if (bp->b_freelist.tqe_next == NULL) {
+               for (dp = bufqueues; dp < &bufqueues[BQUEUES]; dp++)
+                       if (dp->tqh_last == &bp->b_freelist.tqe_next)
+                               break;
+               if (dp == &bufqueues[BQUEUES])
+                       panic("bremfree: lost tail");

        }
        }

-       return (error);
+       TAILQ_REMOVE(dp, bp, b_freelist);

 }
 
 /*
 }
 
 /*

- * Release the buffer, marking it so that if it is grabbed
- * for another purpose it will be written out before being
- * given up (e.g. when writing a partial block where it is
- * assumed that another write for the same block will soon follow).
- * This can't be done for magtape, since writes must be done
- * in the same order as requested.
+ * Initialize buffers and hash links for buffers.

  */
  */

-bdwrite(bp)
+void
+bufinit()
+{

        register struct buf *bp;
        register struct buf *bp;

+       struct bqueues *dp;
+       register int i;
+       int base, residual;
+
+       for (dp = bufqueues; dp < &bufqueues[BQUEUES]; dp++)
+               TAILQ_INIT(dp);
+       bufhashtbl = hashinit(nbuf, M_CACHE, &bufhash);
+       base = bufpages / nbuf;
+       residual = bufpages % nbuf;
+       for (i = 0; i < nbuf; i++) {
+               bp = &buf[i];
+               bzero((char *)bp, sizeof *bp);
+               bp->b_dev = NODEV;
+               bp->b_rcred = NOCRED;
+               bp->b_wcred = NOCRED;
+               bp->b_vnbufs.le_next = NOLIST;
+               bp->b_data = buffers + i * MAXBSIZE;
+               if (i < residual)
+                       bp->b_bufsize = (base + 1) * CLBYTES;
+               else
+                       bp->b_bufsize = base * CLBYTES;
+               bp->b_flags = B_INVAL;
+               dp = bp->b_bufsize ? &bufqueues[BQ_AGE] : &bufqueues[BQ_EMPTY];
+               binsheadfree(bp, dp);
+               binshash(bp, &invalhash);
+       }
+}
+
+bread(a1, a2, a3, a4, a5)
+       struct vnode *a1;
+       daddr_t a2;
+       int a3;
+       struct ucred *a4;
+       struct buf **a5;

 {
 
 {
 

-       if ((bp->b_flags&B_DELWRI) == 0)
-               u.u_ru.ru_oublock++;            /* noone paid yet */
-#ifdef notdef

        /*
        /*

-        * This does not work for buffers associated with
-        * vnodes that are remote - they have no dev.
-        * Besides, we don't use bio with tapes, so rather
-        * than develop a fix, we just ifdef this out for now.
+        * Body deleted.

         */
         */

-       if (bdevsw[major(bp->b_dev)].d_flags & B_TAPE)
-               bawrite(bp);
-       else {
-               bp->b_flags |= B_DELWRI | B_DONE;
-               brelse(bp);
-       }
-#endif
-       bp->b_flags |= B_DELWRI | B_DONE;
-       brelse(bp);
+       return (EIO);

 }
 
 }
 

-/*
- * Release the buffer, start I/O on it, but don't wait for completion.
- */
-bawrite(bp)
-       register struct buf *bp;
+breadn(a1, a2, a3, a4, a5, a6, a7, a8)
+       struct vnode *a1;
+       daddr_t a2; int a3;
+       daddr_t a4[]; int a5[];
+       int a6;
+       struct ucred *a7;
+       struct buf **a8;

 {
 
 {
 

-       bp->b_flags |= B_ASYNC;
-       (void) bwrite(bp);
+       /*
+        * Body deleted.
+        */
+       return (EIO);

 }
 
 }
 

-/*
- * Release the buffer, with no I/O implied.
- */
-brelse(bp)
-       register struct buf *bp;
+bwrite(a1)
+       struct buf *a1;

 {
 {

-       register struct buf *flist;
-       register s;

 
 

-       trace(TR_BRELSE,
-           pack(bp->b_vp->v_mount->m_fsid[0], bp->b_bufsize), bp->b_blkno);

        /*
        /*

-        * If someone's waiting for the buffer, or
-        * is waiting for a buffer wake 'em up.
+        * Body deleted.

         */
         */

-       if (bp->b_flags&B_WANTED)
-               wakeup((caddr_t)bp);
-       if (bfreelist[0].b_flags&B_WANTED) {
-               bfreelist[0].b_flags &= ~B_WANTED;
-               wakeup((caddr_t)bfreelist);
-       }
-       if (bp->b_flags & B_NOCACHE) {
-               bp->b_flags |= B_INVAL;
-       }
-       if (bp->b_flags&B_ERROR)
-               if (bp->b_flags & B_LOCKED)
-                       bp->b_flags &= ~B_ERROR;        /* try again later */
-               else
-                       brelvp(bp);                     /* no assoc */
+       return (EIO);
+}
+
+int
+vn_bwrite(ap)
+       struct vop_bwrite_args *ap;
+{
+       return (bwrite(ap->a_bp));
+}
+
+bdwrite(a1)
+       struct buf *a1;
+{

 
        /*
 
        /*

-        * Stick the buffer back on a free list.
+        * Body deleted.

         */
         */

-       s = splbio();
-       if (bp->b_bufsize <= 0) {
-               /* block has no buffer ... put at front of unused buffer list */
-               flist = &bfreelist[BQ_EMPTY];
-               binsheadfree(bp, flist);
-       } else if (bp->b_flags & (B_ERROR|B_INVAL)) {
-               /* block has no info ... put at front of most free list */
-               flist = &bfreelist[BQ_AGE];
-               binsheadfree(bp, flist);
-       } else {
-               if (bp->b_flags & B_LOCKED)
-                       flist = &bfreelist[BQ_LOCKED];
-               else if (bp->b_flags & B_AGE)
-                       flist = &bfreelist[BQ_AGE];
-               else
-                       flist = &bfreelist[BQ_LRU];
-               binstailfree(bp, flist);
-       }
-       bp->b_flags &= ~(B_WANTED|B_BUSY|B_ASYNC|B_AGE|B_NOCACHE);
-       splx(s);
+       return;

 }
 
 }
 

-/*
- * See if the block is associated with some buffer
- * (mainly to avoid getting hung up on a wait in breada)
- */
-incore(vp, blkno)
-       struct vnode *vp;
-       daddr_t blkno;
+bawrite(a1)
+       struct buf *a1;

 {
 {

-       register struct buf *bp;
-       register struct buf *dp;

 
 

-       dp = BUFHASH(vp, blkno);
-       for (bp = dp->b_forw; bp != dp; bp = bp->b_forw)
-               if (bp->b_blkno == blkno && bp->b_vp == vp &&
-                   (bp->b_flags & B_INVAL) == 0)
-                       return (1);
-       return (0);
+       /*
+        * Body deleted.
+        */
+       return;

 }
 
 }
 

-baddr(vp, blkno, size, cred, bpp)
-       struct vnode *vp;
-       daddr_t blkno;
-       int size;
-       struct ucred *cred;
-       struct buf **bpp;
-#ifdef SECSIZE
-       long secsize;
-#endif SECSIZE
+brelse(a1)
+       struct buf *a1;

 {
 
 {
 

-       if (incore(vp, blkno))
-               return (bread(vp, blkno, size, cred, bpp));
-       *bpp = 0;
-#endif SECSIZE
-       return (0);
+       /*
+        * Body deleted.
+        */
+       return;

 }
 
 }
 

-/*
- * Assign a buffer for the given block.  If the appropriate
- * block is already associated, return it; otherwise search
- * for the oldest non-busy buffer and reassign it.
- *
- * If we find the buffer, but it is dirty (marked DELWRI) and
- * its size is changing, we must write it out first. When the
- * buffer is shrinking, the write is done by brealloc to avoid
- * losing the unwritten data. When the buffer is growing, the
- * write is done by getblk, so that bread will not read stale
- * disk data over the modified data in the buffer.
- *
- * We use splx here because this routine may be called
- * on the interrupt stack during a dump, and we don't
- * want to lower the ipl back to 0.
- */

 struct buf *
 struct buf *

-#ifdef SECSIZE
-getblk(dev, blkno, size, secsize)
-#else SECSIZE
-getblk(vp, blkno, size)
-       register struct vnode *vp;
-       daddr_t blkno;
-       int size;
-#ifdef SECSIZE
-       long secsize;
-#endif SECSIZE
+incore(a1, a2)
+       struct vnode *a1;
+       daddr_t a2;

 {
 {

-       register struct buf *bp, *dp;
-       int s;

 
 

-       if (size > MAXBSIZE)
-               panic("getblk: size too big");

        /*
        /*

-        * To prevent overflow of 32-bit ints when converting block
-        * numbers to byte offsets, blknos > 2^32 / DEV_BSIZE are set
-        * to the maximum number that can be converted to a byte offset
-        * without overflow. This is historic code; what bug it fixed,
-        * or whether it is still a reasonable thing to do is open to
-        * dispute. mkm 9/85
+        * Body deleted.

         */
         */

-       if ((unsigned)blkno >= 1 << (sizeof(int)*NBBY-DEV_BSHIFT))
-               blkno = 1 << ((sizeof(int)*NBBY-DEV_BSHIFT) + 1);
-       /*
-        * Search the cache for the block.  If we hit, but
-        * the buffer is in use for i/o, then we wait until
-        * the i/o has completed.
-        */
-       dp = BUFHASH(vp, blkno);
-loop:
-       for (bp = dp->b_forw; bp != dp; bp = bp->b_forw) {
-               if (bp->b_blkno != blkno || bp->b_vp != vp ||
-                   bp->b_flags&B_INVAL)
-                       continue;
-               s = splbio();
-               if (bp->b_flags&B_BUSY) {
-                       bp->b_flags |= B_WANTED;
-                       sleep((caddr_t)bp, PRIBIO+1);
-                       splx(s);
-                       goto loop;
-               }
-               splx(s);
-               notavail(bp);
-               if (bp->b_bcount != size) {
-                       if (bp->b_bcount < size && (bp->b_flags&B_DELWRI)) {
-                               bp->b_flags &= ~B_ASYNC;
-                               (void) bwrite(bp);
-                               goto loop;
-                       }
-                       if (brealloc(bp, size) == 0)
-                               goto loop;
-               }
-               if (bp->b_bcount != size && brealloc(bp, size) == 0)
-                       goto loop;
-               bp->b_flags |= B_CACHE;
-               return (bp);
-       }
-       bp = getnewbuf();
-       bfree(bp);
-       bremhash(bp);
-       if (bp->b_vp)
-               brelvp(bp);
-       VREF(vp);
-       bp->b_vp = vp;
-       bp->b_dev = vp->v_rdev;
-#ifdef SECSIZE
-       bp->b_blksize = secsize;
-#endif SECSIZE
-       bp->b_blkno = blkno;
-       bp->b_error = 0;
-       bp->b_resid = 0;
-       binshash(bp, dp);
-       if (brealloc(bp, size) == 0)
-               goto loop;
-       return (bp);
+       return (0);

 }
 
 }
 

-/*
- * get an empty block,
- * not assigned to any particular device
- */

 struct buf *
 struct buf *

-geteblk(size)
-       int size;
+getblk(a1, a2, a3, a4, a5)
+       struct vnode *a1;
+       daddr_t a2;
+       int a3, a4, a5;

 {
 {

-       register struct buf *bp, *flist;
-
-       if (size > MAXBSIZE)
-               panic("geteblk: size too big");
-loop:
-       bp = getnewbuf();
-       bp->b_flags |= B_INVAL;
-       bfree(bp);
-       bremhash(bp);
-       flist = &bfreelist[BQ_AGE];
-       brelvp(bp);
-#ifdef SECSIZE
-       bp->b_blksize = DEV_BSIZE;
-#endif SECSIZE
-       bp->b_error = 0;
-       bp->b_resid = 0;
-       binshash(bp, flist);
-       if (brealloc(bp, size) == 0)
-               goto loop;
-       return (bp);
+
+       /*
+        * Body deleted.
+        */
+       return ((struct buf *)0);

 }
 
 }
 

-/*
- * Allocate space associated with a buffer.
- * If can't get space, buffer is released
- */
-brealloc(bp, size)
-       register struct buf *bp;
-       int size;
+struct buf *
+geteblk(a1)
+       int a1;

 {
 {

-       daddr_t start, last;
-       register struct buf *ep;
-       struct buf *dp;
-       int s;

 
        /*
 
        /*

-        * First need to make sure that all overlapping previous I/O
-        * is dispatched with.
+        * Body deleted.

         */
         */

-       if (size == bp->b_bcount)
-               return (1);
-       if (size < bp->b_bcount) { 
-               if (bp->b_flags & B_DELWRI) {
-                       (void) bwrite(bp);
-                       return (0);
-               }
-               if (bp->b_flags & B_LOCKED)
-                       panic("brealloc");
-               return (allocbuf(bp, size));
-       }
-       bp->b_flags &= ~B_DONE;
-       if (bp->b_vp == (struct vnode *)0)
-               return (allocbuf(bp, size));
+       return ((struct buf *)0);
+}
+
+allocbuf(a1, a2)
+       struct buf *a1;
+       int a2;
+{

 
 

-       trace(TR_BREALLOC,
-           pack(bp->b_vp->v_mount->m_fsid[0], size), bp->b_blkno);

        /*
        /*

-        * Search cache for any buffers that overlap the one that we
-        * are trying to allocate. Overlapping buffers must be marked
-        * invalid, after being written out if they are dirty. (indicated
-        * by B_DELWRI) A disk block must be mapped by at most one buffer
-        * at any point in time. Care must be taken to avoid deadlocking
-        * when two buffer are trying to get the same set of disk blocks.
+        * Body deleted.

         */
         */

-       start = bp->b_blkno;
-#ifdef SECSIZE
-       last = start + size/bp->b_blksize - 1;
-#else SECSIZE
-       last = start + btodb(size) - 1;
-#endif SECSIZE
-       dp = BUFHASH(bp->b_vp, bp->b_blkno);
-loop:
-       for (ep = dp->b_forw; ep != dp; ep = ep->b_forw) {
-               if (ep == bp || ep->b_vp != bp->b_vp ||
-                   (ep->b_flags & B_INVAL))
-                       continue;
-               /* look for overlap */
-               if (ep->b_bcount == 0 || ep->b_blkno > last ||
-#ifdef SECSIZE
-                   ep->b_blkno + ep->b_bcount/ep->b_blksize <= start)
-#else SECSIZE
-                   ep->b_blkno + btodb(ep->b_bcount) <= start)
-#endif SECSIZE
-                       continue;
-               s = splbio();
-               if (ep->b_flags&B_BUSY) {
-                       ep->b_flags |= B_WANTED;
-                       sleep((caddr_t)ep, PRIBIO+1);
-                       splx(s);
-                       goto loop;
-               }
-               splx(s);
-               notavail(ep);
-               if (ep->b_flags & B_DELWRI) {
-                       (void) bwrite(ep);
-                       goto loop;
-               }
-               ep->b_flags |= B_INVAL;
-               brelse(ep);
-       }
-       return (allocbuf(bp, size));
+       return (0);

 }
 
 }
 

-/*
- * Find a buffer which is available for use.
- * Select something from a free list.
- * Preference is to AGE list, then LRU list.
- */

 struct buf *
 struct buf *

-getnewbuf()
+getnewbuf(a1, a2)
+       int a1, a2;

 {
 {

-       register struct buf *bp, *dp;
-       register struct ucred *cred;
-       int s;
-
-loop:
-       s = splbio();
-       for (dp = &bfreelist[BQ_AGE]; dp > bfreelist; dp--)
-               if (dp->av_forw != dp)
-                       break;
-       if (dp == bfreelist) {          /* no free blocks */
-               dp->b_flags |= B_WANTED;
-               sleep((caddr_t)dp, PRIBIO+1);
-               splx(s);
-               goto loop;
-       }
-       splx(s);
-       bp = dp->av_forw;
-       notavail(bp);
-       if (bp->b_flags & B_DELWRI) {
-               (void) bawrite(bp);
-               goto loop;
-       }
-       trace(TR_BRELSE,
-           pack(bp->b_vp->v_mount->m_fsid[0], bp->b_bufsize), bp->b_blkno);
-       brelvp(bp);
-       if (bp->b_rcred != NOCRED) {
-               cred = bp->b_rcred;
-               bp->b_rcred = NOCRED;
-               crfree(cred);
-       }
-       if (bp->b_wcred != NOCRED) {
-               cred = bp->b_wcred;
-               bp->b_wcred = NOCRED;
-               crfree(cred);
-       }
-       bp->b_flags = B_BUSY;
-       return (bp);
+
+       /*
+        * Body deleted.
+        */
+       return ((struct buf *)0);

 }
 
 }
 

-/*
- * Wait for I/O completion on the buffer; return errors
- * to the user.
- */
-biowait(bp)
-       register struct buf *bp;
+biowait(a1)
+       struct buf *a1;

 {
 {

-       int s;

 
 

-       s = splbio();
-       while ((bp->b_flags & B_DONE) == 0)
-               sleep((caddr_t)bp, PRIBIO);
-       splx(s);

        /*
        /*

-        * Pick up the device's error number and pass it to the user;
-        * if there is an error but the number is 0 set a generalized code.
+        * Body deleted.

         */
         */

-       if ((bp->b_flags & B_ERROR) == 0)
-               return (0);
-       if (bp->b_error)
-               return (bp->b_error);

        return (EIO);
 }
 
        return (EIO);
 }
 

-/*
- * Mark I/O complete on a buffer.
- * If someone should be called, e.g. the pageout
- * daemon, do so.  Otherwise, wake up anyone
- * waiting for it.
- */
-biodone(bp)
-       register struct buf *bp;
+void
+biodone(a1)
+       struct buf *a1;

 {
 
 {
 

-       if (bp->b_flags & B_DONE)
-               panic("dup biodone");
-       bp->b_flags |= B_DONE;
-       if ((bp->b_flags & B_READ) == 0)
-               bp->b_dirtyoff = bp->b_dirtyend = 0;
-       if (bp->b_flags & B_CALL) {
-               bp->b_flags &= ~B_CALL;
-               (*bp->b_iodone)(bp);
-               return;
-       }
-       if (bp->b_flags&B_ASYNC)
-               brelse(bp);
-       else {
-               bp->b_flags &= ~B_WANTED;
-               wakeup((caddr_t)bp);
-       }
+       /*
+        * Body deleted.
+        */
+       return;

 }
 
 }
 

-/*
- * Ensure that no part of a specified block is in an incore buffer.
-#ifdef SECSIZE
- * "size" is given in device blocks (the units of b_blkno).
-#endif SECSIZE
-#ifdef SECSIZE
- * "size" is given in device blocks (the units of b_blkno).
-#endif SECSIZE
- */
-blkflush(vp, blkno, size)
-       struct vnode *vp;
-       daddr_t blkno;
-#ifdef SECSIZE
-       int size;
-#else SECSIZE
-       long size;
-#endif SECSIZE
+int
+count_lock_queue()

 {
 {

-       register struct buf *ep;
-       struct buf *dp;
-       daddr_t start, last;
-       int s, error, allerrors = 0;
-
-       start = blkno;
-#ifdef SECSIZE
-       last = start + size - 1;
-#else SECSIZE
-       last = start + btodb(size) - 1;
-#endif SECSIZE
-       dp = BUFHASH(vp, blkno);
-loop:
-       for (ep = dp->b_forw; ep != dp; ep = ep->b_forw) {
-               if (ep->b_vp != vp || (ep->b_flags & B_INVAL))
-                       continue;
-               /* look for overlap */
-               if (ep->b_bcount == 0 || ep->b_blkno > last ||
-#ifdef SECSIZE
-                   ep->b_blkno + ep->b_bcount / ep->b_blksize <= start)
-#else SECSIZE
-                   ep->b_blkno + btodb(ep->b_bcount) <= start)
-#endif SECSIZE
-                       continue;
-               s = splbio();
-               if (ep->b_flags&B_BUSY) {
-                       ep->b_flags |= B_WANTED;
-                       sleep((caddr_t)ep, PRIBIO+1);
-                       splx(s);
-                       goto loop;
-               }
-               if (ep->b_flags & B_DELWRI) {
-                       splx(s);
-                       notavail(ep);
-                       if (error = bwrite(ep))
-                               allerrors = error;
-                       goto loop;
-               }
-               splx(s);
-       }
-       return (allerrors);
-}

 
 

-/*
- * Make sure all write-behind blocks associated
- * with mount point are flushed out (from sync).
- */
-bflush(mountp)
-       struct mount *mountp;
-{
-       register struct buf *bp;
-       register struct buf *flist;
-       int s;
-
-loop:
-       s = splbio();
-       for (flist = bfreelist; flist < &bfreelist[BQ_EMPTY]; flist++) {
-               for (bp = flist->av_forw; bp != flist; bp = bp->av_forw) {
-                       if ((bp->b_flags & B_BUSY))
-                               continue;
-                       if ((bp->b_flags & B_DELWRI) == 0)
-                               continue;
-                       if (bp->b_vp && bp->b_vp->v_mount == mountp) {
-                               notavail(bp);
-                               (void) bawrite(bp);
-                               splx(s);
-                               goto loop;
-                       }
-               }
-       }
-       splx(s);
+       /*
+        * Body deleted.
+        */
+       return (0);

 }
 
 }
 

+#ifdef DIAGNOSTIC

 /*
 /*

- * Invalidate in core blocks belonging to closed or umounted filesystem
- *
- * We walk through the buffer pool and invalidate any buffers for the
- * indicated mount point. Normally this routine is preceeded by a bflush
- * call, so that on a quiescent filesystem there will be no dirty
- * buffers when we are done. We return the count of dirty buffers when
- * we are finished.
+ * Print out statistics on the current allocation of the buffer pool.
+ * Can be enabled to print out on every ``sync'' by setting "syncprt"
+ * in vfs_syscalls.c using sysctl.

  */
  */

-binval(mountp)
-       struct mount *mountp;
+void
+vfs_bufstats()

 {
 {

+       int s, i, j, count;

        register struct buf *bp;
        register struct buf *bp;

-       register struct bufhd *hp;
-       int s, dirty = 0;
-#define dp ((struct buf *)hp)
-
-loop:
-       s = splbio();
-       for (hp = bufhash; hp < &bufhash[BUFHSZ]; hp++) {
-               for (bp = dp->b_forw; bp != dp; bp = bp->b_forw) {
-                       if (bp->b_vp == NULL || bp->b_vp->v_mount != mountp)
-                               continue;
-                       if (bp->b_flags & B_BUSY) {
-                               bp->b_flags |= B_WANTED;
-                               sleep((caddr_t)bp, PRIBIO+1);
-                               splx(s);
-                               goto loop;
-                       }
-                       notavail(bp);
-                       if (bp->b_flags & B_DELWRI) {
-                               (void) bawrite(bp);
-                               dirty++;
-                               continue;
-                       }
-                       bp->b_flags |= B_INVAL;
-                       brelvp(bp);
-                       brelse(bp);
+       register struct bqueues *dp;
+       int counts[MAXBSIZE/CLBYTES+1];
+       static char *bname[BQUEUES] = { "LOCKED", "LRU", "AGE", "EMPTY" };
+
+       for (dp = bufqueues, i = 0; dp < &bufqueues[BQUEUES]; dp++, i++) {
+               count = 0;
+               for (j = 0; j <= MAXBSIZE/CLBYTES; j++)
+                       counts[j] = 0;
+               s = splbio();
+               for (bp = dp->tqh_first; bp; bp = bp->b_freelist.tqe_next) {
+                       counts[bp->b_bufsize/CLBYTES]++;
+                       count++;

                }
                }

+               splx(s);
+               printf("%s: total-%d", bname[i], count);
+               for (j = 0; j <= MAXBSIZE/CLBYTES; j++)
+                       if (counts[j] != 0)
+                               printf(", %d-%d", j * CLBYTES, counts[j]);
+               printf("\n");

        }
        }

-       return (dirty);
-}
-
-brelvp(bp)
-       struct buf *bp;
-{
-       struct vnode *vp;
-
-       if (bp->b_vp == (struct vnode *) 0)
-               return;
-       vp = bp->b_vp;
-       bp->b_vp = (struct vnode *) 0;
-       vrele(vp);

 }
 }

+#endif /* DIAGNOSTIC */