freebsd-dev/sys/kern/vfs_vnops.c
Kirk McKusick ad8ac923fa These changes appear to give us benefits with both small (32MB) and
large (1G) memory machine configurations.  I was able to run 'dbench 32'
on a 32MB system without bring the machine to a grinding halt.

    * buffer cache hash table now dynamically allocated.  This will
      have no effect on memory consumption for smaller systems and
      will help scale the buffer cache for larger systems.

    * minor enhancement to pmap_clearbit().  I noticed that
      all the calls to it used constant arguments.  Making
      it an inline allows the constants to propogate to
      deeper inlines and should produce better code.

    * removal of inherent vfs_ioopt support through the emplacement
      of appropriate #ifdef's, with John's permission.  If we do not
      find a use for it by the end of the year we will remove it entirely.

    * removal of getnewbufloops* counters & sysctl's - no longer
      necessary for debugging, getnewbuf() is now optimal.

    * buffer hash table functions removed from sys/buf.h and localized
      to vfs_bio.c

    * VFS_BIO_NEED_DIRTYFLUSH flag and support code added
      ( bwillwrite() ), allowing processes to block when too many dirty
      buffers are present in the system.

    * removal of a softdep test in bdwrite() that is no longer necessary
      now that bdwrite() no longer attempts to flush dirty buffers.

    * slight optimization added to bqrelse() - there is no reason
      to test for available buffer space on B_DELWRI buffers.

    * addition of reverse-scanning code to vfs_bio_awrite().
      vfs_bio_awrite() will attempt to locate clusterable areas
      in both the forward and reverse direction relative to the
      offset of the buffer passed to it.  This will probably not
      make much of a difference now, but I believe we will start
      to rely on it heavily in the future if we decide to shift
      some of the burden of the clustering closer to the actual
      I/O initiation.

    * Removal of the newbufcnt and lastnewbuf counters that Kirk
      added.  They do not fix any race conditions that haven't already
      been fixed by the gbincore() test done after the only call
      to getnewbuf().  getnewbuf() is a static, so there is no chance
      of it being misused by other modules.  ( Unless Kirk can think
      of a specific thing that this code fixes.  I went through it
      very carefully and didn't see anything ).

    * removal of VOP_ISLOCKED() check in flushbufqueues().  I do not
      think this check is necessary, the buffer should flush properly
      whether the vnode is locked or not. ( yes? ).

    * removal of extra arguments passed to getnewbuf() that are not
      necessary.

    * missed cluster_wbuild() that had to be a cluster_wbuild_wb() in
      vfs_cluster.c

    * vn_write() now calls bwillwrite() *PRIOR* to locking the vnode,
      which should greatly aid flushing operations in heavy load
      situations - both the pageout and update daemons will be able
      to operate more efficiently.

    * removal of b_usecount.  We may add it back in later but for now
      it is useless.  Prior implementations of the buffer cache never
      had enough buffers for it to be useful, and current implementations
      which make more buffers available might not benefit relative to
      the amount of sophistication required to implement a b_usecount.
      Straight LRU should work just as well, especially when most things
      are VMIO backed.  I expect that (even though John will not like
      this assumption) directories will become VMIO backed some point soon.

Submitted by:	Matthew Dillon <dillon@backplane.com>
Reviewed by:	Kirk McKusick <mckusick@mckusick.com>
1999-07-08 06:06:00 +00:00

569 lines
13 KiB
C

/*
* 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.
*
* 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_vnops.c 8.2 (Berkeley) 1/21/94
* $Id: vfs_vnops.c,v 1.69 1999/07/02 16:29:15 phk Exp $
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/fcntl.h>
#include <sys/file.h>
#include <sys/stat.h>
#include <sys/proc.h>
#include <sys/mount.h>
#include <sys/namei.h>
#include <sys/vnode.h>
#include <sys/filio.h>
#include <sys/ttycom.h>
static int vn_closefile __P((struct file *fp, struct proc *p));
static int vn_ioctl __P((struct file *fp, u_long com, caddr_t data,
struct proc *p));
static int vn_read __P((struct file *fp, struct uio *uio,
struct ucred *cred, int flags));
static int vn_poll __P((struct file *fp, int events, struct ucred *cred,
struct proc *p));
static int vn_write __P((struct file *fp, struct uio *uio,
struct ucred *cred, int flags));
struct fileops vnops =
{ vn_read, vn_write, vn_ioctl, vn_poll, vn_closefile };
/*
* Common code for vnode open operations.
* Check permissions, and call the VOP_OPEN or VOP_CREATE routine.
*/
int
vn_open(ndp, fmode, cmode)
register struct nameidata *ndp;
int fmode, cmode;
{
register struct vnode *vp;
register struct proc *p = ndp->ni_cnd.cn_proc;
register struct ucred *cred = p->p_ucred;
struct vattr vat;
struct vattr *vap = &vat;
int mode, error;
if (fmode & O_CREAT) {
ndp->ni_cnd.cn_nameiop = CREATE;
ndp->ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF;
if ((fmode & O_EXCL) == 0 && (fmode & O_NOFOLLOW) == 0)
ndp->ni_cnd.cn_flags |= FOLLOW;
error = namei(ndp);
if (error)
return (error);
if (ndp->ni_vp == NULL) {
VATTR_NULL(vap);
vap->va_type = VREG;
vap->va_mode = cmode;
if (fmode & O_EXCL)
vap->va_vaflags |= VA_EXCLUSIVE;
VOP_LEASE(ndp->ni_dvp, p, cred, LEASE_WRITE);
error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp,
&ndp->ni_cnd, vap);
vput(ndp->ni_dvp);
if (error)
return (error);
ASSERT_VOP_UNLOCKED(ndp->ni_dvp, "create");
ASSERT_VOP_LOCKED(ndp->ni_vp, "create");
fmode &= ~O_TRUNC;
vp = ndp->ni_vp;
} else {
VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd);
if (ndp->ni_dvp == ndp->ni_vp)
vrele(ndp->ni_dvp);
else
vput(ndp->ni_dvp);
ndp->ni_dvp = NULL;
vp = ndp->ni_vp;
if (fmode & O_EXCL) {
error = EEXIST;
goto bad;
}
fmode &= ~O_CREAT;
}
} else {
ndp->ni_cnd.cn_nameiop = LOOKUP;
ndp->ni_cnd.cn_flags =
((fmode & O_NOFOLLOW) ? NOFOLLOW : FOLLOW) | LOCKLEAF;
error = namei(ndp);
if (error)
return (error);
vp = ndp->ni_vp;
}
if (vp->v_type == VLNK) {
error = EMLINK;
goto bad;
}
if (vp->v_type == VSOCK) {
error = EOPNOTSUPP;
goto bad;
}
if ((fmode & O_CREAT) == 0) {
mode = 0;
if (fmode & (FWRITE | O_TRUNC)) {
if (vp->v_type == VDIR) {
error = EISDIR;
goto bad;
}
error = vn_writechk(vp);
if (error)
goto bad;
mode |= VWRITE;
}
if (fmode & FREAD)
mode |= VREAD;
if (mode) {
error = VOP_ACCESS(vp, mode, cred, p);
if (error)
goto bad;
}
}
if (fmode & O_TRUNC) {
VOP_UNLOCK(vp, 0, p); /* XXX */
VOP_LEASE(vp, p, cred, LEASE_WRITE);
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); /* XXX */
VATTR_NULL(vap);
vap->va_size = 0;
error = VOP_SETATTR(vp, vap, cred, p);
if (error)
goto bad;
}
error = VOP_OPEN(vp, fmode, cred, p);
if (error)
goto bad;
/*
* Make sure that a VM object is created for VMIO support.
*/
if (vp->v_type == VREG) {
if ((error = vfs_object_create(vp, p, cred)) != 0)
goto bad;
}
if (fmode & FWRITE)
vp->v_writecount++;
return (0);
bad:
vput(vp);
return (error);
}
/*
* Check for write permissions on the specified vnode.
* Prototype text segments cannot be written.
*/
int
vn_writechk(vp)
register struct vnode *vp;
{
/*
* If there's shared text associated with
* the vnode, try to free it up once. If
* we fail, we can't allow writing.
*/
if (vp->v_flag & VTEXT)
return (ETXTBSY);
return (0);
}
/*
* Vnode close call
*/
int
vn_close(vp, flags, cred, p)
register struct vnode *vp;
int flags;
struct ucred *cred;
struct proc *p;
{
int error;
if (flags & FWRITE)
vp->v_writecount--;
error = VOP_CLOSE(vp, flags, cred, p);
vrele(vp);
return (error);
}
/*
* Package up an I/O request on a vnode into a uio and do it.
*/
int
vn_rdwr(rw, vp, base, len, offset, segflg, ioflg, cred, aresid, p)
enum uio_rw rw;
struct vnode *vp;
caddr_t base;
int len;
off_t offset;
enum uio_seg segflg;
int ioflg;
struct ucred *cred;
int *aresid;
struct proc *p;
{
struct uio auio;
struct iovec aiov;
int error;
if ((ioflg & IO_NODELOCKED) == 0)
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
aiov.iov_base = base;
aiov.iov_len = len;
auio.uio_resid = len;
auio.uio_offset = offset;
auio.uio_segflg = segflg;
auio.uio_rw = rw;
auio.uio_procp = p;
if (rw == UIO_READ) {
error = VOP_READ(vp, &auio, ioflg, cred);
} else {
error = VOP_WRITE(vp, &auio, ioflg, cred);
}
if (aresid)
*aresid = auio.uio_resid;
else
if (auio.uio_resid && error == 0)
error = EIO;
if ((ioflg & IO_NODELOCKED) == 0)
VOP_UNLOCK(vp, 0, p);
return (error);
}
/*
* File table vnode read routine.
*/
static int
vn_read(fp, uio, cred, flags)
struct file *fp;
struct uio *uio;
struct ucred *cred;
int flags;
{
struct vnode *vp = (struct vnode *)fp->f_data;
struct proc *p = uio->uio_procp;
int error, ioflag;
ioflag = 0;
if (fp->f_flag & FNONBLOCK)
ioflag |= IO_NDELAY;
VOP_LEASE(vp, p, cred, LEASE_READ);
vn_lock(vp, LK_SHARED | LK_NOPAUSE | LK_RETRY, p);
if ((flags & FOF_OFFSET) == 0)
uio->uio_offset = fp->f_offset;
/*
* Sequential read heuristic.
* If we have been doing sequential input,
* a rewind operation doesn't turn off
* sequential input mode.
*/
if ((uio->uio_offset == 0 && fp->f_seqcount > 0) ||
uio->uio_offset == fp->f_nextread) {
int tmpseq = fp->f_seqcount;
/*
* XXX we assume that the filesystem block size is
* the default. Not true, but still gives us a pretty
* good indicator of how sequential the read operations
* are.
*/
tmpseq += (uio->uio_resid + BKVASIZE - 1) / BKVASIZE;
if (tmpseq >= 127)
tmpseq = 127;
fp->f_seqcount = tmpseq;
ioflag |= fp->f_seqcount << 16;
} else {
if (fp->f_seqcount > 1)
fp->f_seqcount = 1;
else
fp->f_seqcount = 0;
}
error = VOP_READ(vp, uio, ioflag, cred);
if ((flags & FOF_OFFSET) == 0)
fp->f_offset = uio->uio_offset;
fp->f_nextread = uio->uio_offset;
VOP_UNLOCK(vp, 0, p);
return (error);
}
/*
* File table vnode write routine.
*/
static int
vn_write(fp, uio, cred, flags)
struct file *fp;
struct uio *uio;
struct ucred *cred;
int flags;
{
struct vnode *vp;
struct proc *p = uio->uio_procp;
int error, ioflag;
vp = (struct vnode *)fp->f_data;
if (vp->v_type == VREG)
bwillwrite();
vp = (struct vnode *)fp->f_data; /* XXX needed? */
ioflag = IO_UNIT;
if (vp->v_type == VREG && (fp->f_flag & O_APPEND))
ioflag |= IO_APPEND;
if (fp->f_flag & FNONBLOCK)
ioflag |= IO_NDELAY;
if ((fp->f_flag & O_FSYNC) ||
(vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS)))
ioflag |= IO_SYNC;
VOP_LEASE(vp, p, cred, LEASE_WRITE);
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
if ((flags & FOF_OFFSET) == 0)
uio->uio_offset = fp->f_offset;
error = VOP_WRITE(vp, uio, ioflag, cred);
if ((flags & FOF_OFFSET) == 0)
fp->f_offset = uio->uio_offset;
VOP_UNLOCK(vp, 0, p);
return (error);
}
/*
* File table vnode stat routine.
*/
int
vn_stat(vp, sb, p)
struct vnode *vp;
register struct stat *sb;
struct proc *p;
{
struct vattr vattr;
register struct vattr *vap;
int error;
u_short mode;
vap = &vattr;
error = VOP_GETATTR(vp, vap, p->p_ucred, p);
if (error)
return (error);
/*
* Copy from vattr table
*/
if (vap->va_fsid != VNOVAL)
sb->st_dev = vap->va_fsid;
else
sb->st_dev = vp->v_mount->mnt_stat.f_fsid.val[0];
sb->st_ino = vap->va_fileid;
mode = vap->va_mode;
switch (vap->va_type) {
case VREG:
mode |= S_IFREG;
break;
case VDIR:
mode |= S_IFDIR;
break;
case VBLK:
mode |= S_IFBLK;
break;
case VCHR:
mode |= S_IFCHR;
break;
case VLNK:
mode |= S_IFLNK;
/* This is a cosmetic change, symlinks do not have a mode. */
if (vp->v_mount->mnt_flag & MNT_NOSYMFOLLOW)
sb->st_mode &= ~ACCESSPERMS; /* 0000 */
else
sb->st_mode |= ACCESSPERMS; /* 0777 */
break;
case VSOCK:
mode |= S_IFSOCK;
break;
case VFIFO:
mode |= S_IFIFO;
break;
default:
return (EBADF);
};
sb->st_mode = mode;
sb->st_nlink = vap->va_nlink;
sb->st_uid = vap->va_uid;
sb->st_gid = vap->va_gid;
sb->st_rdev = vap->va_rdev;
sb->st_size = vap->va_size;
sb->st_atimespec = vap->va_atime;
sb->st_mtimespec = vap->va_mtime;
sb->st_ctimespec = vap->va_ctime;
sb->st_blksize = vap->va_blocksize;
sb->st_flags = vap->va_flags;
if (suser_xxx(p->p_ucred, 0, 0))
sb->st_gen = 0;
else
sb->st_gen = vap->va_gen;
#if (S_BLKSIZE == 512)
/* Optimize this case */
sb->st_blocks = vap->va_bytes >> 9;
#else
sb->st_blocks = vap->va_bytes / S_BLKSIZE;
#endif
return (0);
}
/*
* File table vnode ioctl routine.
*/
static int
vn_ioctl(fp, com, data, p)
struct file *fp;
u_long com;
caddr_t data;
struct proc *p;
{
register struct vnode *vp = ((struct vnode *)fp->f_data);
struct vattr vattr;
int error;
switch (vp->v_type) {
case VREG:
case VDIR:
if (com == FIONREAD) {
error = VOP_GETATTR(vp, &vattr, p->p_ucred, p);
if (error)
return (error);
*(int *)data = vattr.va_size - fp->f_offset;
return (0);
}
if (com == FIONBIO || com == FIOASYNC) /* XXX */
return (0); /* XXX */
/* fall into ... */
default:
#if 0
return (ENOTTY);
#endif
case VFIFO:
case VCHR:
case VBLK:
error = VOP_IOCTL(vp, com, data, fp->f_flag, p->p_ucred, p);
if (error == 0 && com == TIOCSCTTY) {
/* Do nothing if reassigning same control tty */
if (p->p_session->s_ttyvp == vp)
return (0);
/* Get rid of reference to old control tty */
if (p->p_session->s_ttyvp)
vrele(p->p_session->s_ttyvp);
p->p_session->s_ttyvp = vp;
VREF(vp);
}
return (error);
}
}
/*
* File table vnode poll routine.
*/
static int
vn_poll(fp, events, cred, p)
struct file *fp;
int events;
struct ucred *cred;
struct proc *p;
{
return (VOP_POLL(((struct vnode *)fp->f_data), events, cred, p));
}
/*
* Check that the vnode is still valid, and if so
* acquire requested lock.
*/
int
#ifndef DEBUG_LOCKS
vn_lock(vp, flags, p)
#else
debug_vn_lock(vp, flags, p, filename, line)
#endif
struct vnode *vp;
int flags;
struct proc *p;
#ifdef DEBUG_LOCKS
const char *filename;
int line;
#endif
{
int error;
do {
if ((flags & LK_INTERLOCK) == 0)
simple_lock(&vp->v_interlock);
if (vp->v_flag & VXLOCK) {
vp->v_flag |= VXWANT;
simple_unlock(&vp->v_interlock);
tsleep((caddr_t)vp, PINOD, "vn_lock", 0);
error = ENOENT;
} else {
#ifdef DEBUG_LOCKS
vp->filename = filename;
vp->line = line;
#endif
error = VOP_LOCK(vp,
flags | LK_NOPAUSE | LK_INTERLOCK, p);
if (error == 0)
return (error);
}
flags &= ~LK_INTERLOCK;
} while (flags & LK_RETRY);
return (error);
}
/*
* File table vnode close routine.
*/
static int
vn_closefile(fp, p)
struct file *fp;
struct proc *p;
{
return (vn_close(((struct vnode *)fp->f_data), fp->f_flag,
fp->f_cred, p));
}