u. is gone, pcb is at p_addr; aston => signotify

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

diff --git a/usr/src/sys/kern/vfs_vnops.c b/usr/src/sys/kern/vfs_vnops.c
index 7f2500b..71548ff 100644 (file)
--- a/usr/src/sys/kern/vfs_vnops.c
+++ b/usr/src/sys/kern/vfs_vnops.c
@@ -1,272 +1,388 @@
-/*     vfs_vnops.c     3.3     %G%     */
+/*
+ * Copyright (c) 1982, 1986, 1989 Regents of the University of California.
+ * All rights reserved.
+ *
+ * %sccs.include.redist.c%
+ *
+ *     @(#)vfs_vnops.c 7.29 (Berkeley) %G%
+ */

 
 

-#include "../h/param.h"
-#include "../h/systm.h"
-#include "../h/dir.h"
-#include "../h/user.h"
-#include "../h/filsys.h"
-#include "../h/file.h"
-#include "../h/conf.h"
-#include "../h/inode.h"
-#include "../h/reg.h"
-#include "../h/acct.h"
+#include "param.h"
+#include "systm.h"
+#include "kernel.h"
+#include "file.h"
+#include "stat.h"
+#include "buf.h"
+#include "proc.h"
+#include "mount.h"
+#include "namei.h"
+#include "vnode.h"
+#include "ioctl.h"
+#include "tty.h"
+
+struct         fileops vnops =
+       { vn_read, vn_write, vn_ioctl, vn_select, vn_close };

 
 /*
 
 /*

- * Convert a user supplied
- * file descriptor into a pointer
- * to a file structure.
- * Only task is to check range
- * of the descriptor.
+ * Common code for vnode open operations.
+ * Check permissions, and call the VOP_OPEN or VOP_CREATE routine.

  */
  */

-struct file *
-getf(f)
-register int f;
+vn_open(ndp, p, fmode, cmode)
+       register struct nameidata *ndp;
+       struct proc *p;
+       int fmode, cmode;

 {
 {

-       register struct file *fp;
+       register struct vnode *vp;
+       register struct ucred *cred = p->p_ucred;
+       struct vattr vat;
+       struct vattr *vap = &vat;
+       int error;

 
 

-       if(0 <= f && f < NOFILE) {
-               fp = u.u_ofile[f];
-               if(fp != NULL)
-                       return(fp);
+       if (fmode & FCREAT) {
+               ndp->ni_nameiop = CREATE | LOCKPARENT | LOCKLEAF;
+               if ((fmode & FEXCL) == 0)
+                       ndp->ni_nameiop |= FOLLOW;
+               if (error = namei(ndp, p))
+                       return (error);
+               if (ndp->ni_vp == NULL) {
+                       VATTR_NULL(vap);
+                       vap->va_type = VREG;
+                       vap->va_mode = cmode;
+                       if (error = VOP_CREATE(ndp, vap, p))
+                               return (error);
+                       fmode &= ~FTRUNC;
+                       vp = ndp->ni_vp;
+               } else {
+                       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 & FEXCL) {
+                               error = EEXIST;
+                               goto bad;
+                       }
+                       fmode &= ~FCREAT;
+               }
+       } else {
+               ndp->ni_nameiop = LOOKUP | FOLLOW | LOCKLEAF;
+               if (error = namei(ndp, p))
+                       return (error);
+               vp = ndp->ni_vp;
+       }
+       if (vp->v_type == VSOCK) {
+               error = EOPNOTSUPP;
+               goto bad;

        }
        }

-       u.u_error = EBADF;
-       return(NULL);
+       if ((fmode & FCREAT) == 0) {
+               if (fmode & FREAD) {
+                       if (error = VOP_ACCESS(vp, VREAD, cred, p))
+                               goto bad;
+               }
+               if (fmode & (FWRITE|FTRUNC)) {
+                       if (vp->v_type == VDIR) {
+                               error = EISDIR;
+                               goto bad;
+                       }
+                       if ((error = vn_writechk(vp)) ||
+                           (error = VOP_ACCESS(vp, VWRITE, cred, p)))
+                               goto bad;
+               }
+       }
+       if (fmode & FTRUNC) {
+               VATTR_NULL(vap);
+               vap->va_size = 0;
+               if (error = VOP_SETATTR(vp, vap, cred, p))
+                       goto bad;
+       }
+       VOP_UNLOCK(vp);
+       error = VOP_OPEN(vp, fmode, cred, p);
+       if (error)
+               vrele(vp);
+       return (error);
+
+bad:
+       vput(vp);
+       return (error);

 }
 
 /*
 }
 
 /*

- * Internal form of close.
- * Decrement reference count on
- * file structure.
- * Also make sure the pipe protocol
- * does not constipate.
- *
- * Decrement reference count on the inode following
- * removal to the referencing file structure.
- * Call device handler on last close.
+ * Check for write permissions on the specified vnode.
+ * The read-only status of the file system is checked.
+ * Also, prototype text segments cannot be written.

  */
  */

-closef(fp)
-register struct file *fp;
+vn_writechk(vp)
+       register struct vnode *vp;

 {
 {

-       register struct inode *ip;
-       int flag, mode;
-       dev_t dev;
-       register int (*cfunc)();
-       struct chan *cp;

 
 

-       if(fp == NULL)
-               return;
-       if (fp->f_count > 1) {
-               fp->f_count--;
-               return;
-       }
-       ip = fp->f_inode;
-       flag = fp->f_flag;
-       cp = fp->f_un.f_chan;
-       dev = (dev_t)ip->i_un.i_rdev;
-       mode = ip->i_mode;
-
-       plock(ip);
-       fp->f_count = 0;
-       if(flag & FPIPE) {
-               ip->i_mode &= ~(IREAD|IWRITE);
-               wakeup((caddr_t)ip+1);
-               wakeup((caddr_t)ip+2);
+       /*
+        * Disallow write attempts on read-only file systems;
+        * unless the file is a socket or a block or character
+        * device resident on the file system.
+        */
+       if (vp->v_mount->mnt_flag & MNT_RDONLY) {
+               switch (vp->v_type) {
+               case VREG: case VDIR: case VLNK:
+                       return (EROFS);
+               }

        }
        }

-       iput(ip);
+       /*
+        * 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) && !vnode_pager_uncache(vp))
+               return (ETXTBSY);
+       return (0);
+}

 
 

-       switch(mode&IFMT) {
+/*
+ * Vnode version of rdwri() for calls on file systems.
+ */
+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;

 
 

-       case IFCHR:
-       case IFMPC:
-               cfunc = cdevsw[major(dev)].d_close;
-               break;
+       if ((ioflg & IO_NODELOCKED) == 0)
+               VOP_LOCK(vp);
+       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);
+       return (error);
+}

 
 

-       case IFBLK:
-       case IFMPB:
-               cfunc = bdevsw[major(dev)].d_close;
-               break;
-       default:
-               return;
-       }
+vn_read(fp, uio, cred)
+       struct file *fp;
+       struct uio *uio;
+       struct ucred *cred;
+{
+       register struct vnode *vp = (struct vnode *)fp->f_data;
+       int count, error;

 
 

-       if (flag & FMP)
-               goto call;
+       VOP_LOCK(vp);
+       uio->uio_offset = fp->f_offset;
+       count = uio->uio_resid;
+       error = VOP_READ(vp, uio, (fp->f_flag & FNDELAY) ? IO_NDELAY : 0, cred);
+       fp->f_offset += count - uio->uio_resid;
+       VOP_UNLOCK(vp);
+       return (error);
+}

 
 

-       for(fp=file; fp < &file[NFILE]; fp++)
-               if (fp->f_count && fp->f_inode==ip)
-                       return;
+vn_write(fp, uio, cred)
+       struct file *fp;
+       struct uio *uio;
+       struct ucred *cred;
+{
+       register struct vnode *vp = (struct vnode *)fp->f_data;
+       int count, error, ioflag = 0;

 
 

-call:
-       (*cfunc)(dev, flag, cp);
+       if (vp->v_type == VREG && (fp->f_flag & FAPPEND))
+               ioflag |= IO_APPEND;
+       if (fp->f_flag & FNDELAY)
+               ioflag |= IO_NDELAY;
+       VOP_LOCK(vp);
+       uio->uio_offset = fp->f_offset;
+       count = uio->uio_resid;
+       error = VOP_WRITE(vp, uio, ioflag, cred);
+       if (ioflag & IO_APPEND)
+               fp->f_offset = uio->uio_offset;
+       else
+               fp->f_offset += count - uio->uio_resid;
+       VOP_UNLOCK(vp);
+       return (error);

 }
 
 /*
 }
 
 /*

- * openi called to allow handler
- * of special files to initialize and
- * validate before actual IO.
+ * Get stat info for a vnode.

  */
  */

-openi(ip, rw)
-register struct inode *ip;
+vn_stat(vp, sb, p)
+       struct vnode *vp;
+       register struct stat *sb;
+       struct proc *p;

 {
 {

-       dev_t dev;
-       register unsigned int maj;
+       struct vattr vattr;
+       register struct vattr *vap;
+       int error;
+       u_short mode;

 
 

-       dev = (dev_t)ip->i_un.i_rdev;
-       maj = major(dev);
-       switch(ip->i_mode&IFMT) {
-
-       case IFCHR:
-       case IFMPC:
-               if(maj >= nchrdev)
-                       goto bad;
-               (*cdevsw[maj].d_open)(dev, rw);
+       vap = &vattr;
+       error = VOP_GETATTR(vp, vap, p->p_ucred, p);
+       if (error)
+               return (error);
+       /*
+        * Copy from vattr table
+        */
+       sb->st_dev = vap->va_fsid;
+       sb->st_ino = vap->va_fileid;
+       mode = vap->va_mode;
+       switch (vp->v_type) {
+       case VREG:
+               mode |= S_IFREG;

                break;
                break;

-
-       case IFBLK:
-       case IFMPB:
-               if(maj >= nblkdev)
-                       goto bad;
-               (*bdevsw[maj].d_open)(dev, rw);
-       }
-       return;
-
-bad:
-       u.u_error = ENXIO;
+       case VDIR:
+               mode |= S_IFDIR;
+               break;
+       case VBLK:
+               mode |= S_IFBLK;
+               break;
+       case VCHR:
+               mode |= S_IFCHR;
+               break;
+       case VLNK:
+               mode |= S_IFLNK;
+               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_atime = vap->va_atime.tv_sec;
+       sb->st_spare1 = 0;
+       sb->st_mtime = vap->va_mtime.tv_sec;
+       sb->st_spare2 = 0;
+       sb->st_ctime = vap->va_ctime.tv_sec;
+       sb->st_spare3 = 0;
+       sb->st_blksize = vap->va_blocksize;
+       sb->st_flags = vap->va_flags;
+       sb->st_gen = vap->va_gen;
+       sb->st_blocks = vap->va_bytes / S_BLKSIZE;
+       return (0);

 }
 
 /*
 }
 
 /*

- * Check mode permission on inode pointer.
- * Mode is READ, WRITE or EXEC.
- * In the case of WRITE, the
- * read-only status of the file
- * system is checked.
- * Also in WRITE, prototype text
- * segments cannot be written.
- * The mode is shifted to select
- * the owner/group/other fields.
- * The super user is granted all
- * permissions.
+ * Vnode ioctl call

  */
  */

-access(ip, mode)
-register struct inode *ip;
+vn_ioctl(fp, com, data, p)
+       struct file *fp;
+       int com;
+       caddr_t data;
+       struct proc *p;

 {
 {

-       register m;
+       register struct vnode *vp = ((struct vnode *)fp->f_data);
+       struct vattr vattr;
+       int error;

 
 

-       m = mode;
-       if(m == IWRITE) {
-               if(getfs(ip->i_dev)->s_ronly != 0) {
-                       u.u_error = EROFS;
-                       return(1);
-               }
-               if (ip->i_flag&ITEXT)           /* try to free text */
-                       xrele(ip);
-               if(ip->i_flag & ITEXT) {
-                       u.u_error = ETXTBSY;
-                       return(1);
-               }
-       }
-       if(u.u_uid == 0)
-               return(0);
-       if(u.u_uid != ip->i_uid) {
-               m >>= 3;
-               if(u.u_gid != ip->i_gid)
-                       m >>= 3;
-       }
-       if((ip->i_mode&m) != 0)
-               return(0);
+       switch (vp->v_type) {

 
 

-       u.u_error = EACCES;
-       return(1);
-}
+       case VREG:
+       case VDIR:
+               if (com == FIONREAD) {
+                       if (error = VOP_GETATTR(vp, &vattr, p->p_ucred, p))
+                               return (error);
+                       *(off_t *)data = vattr.va_size - fp->f_offset;
+                       return (0);
+               }
+               if (com == FIONBIO || com == FIOASYNC)  /* XXX */
+                       return (0);                     /* XXX */
+               /* fall into ... */

 
 

-/*
- * Look up a pathname and test if
- * the resultant inode is owned by the
- * current user.
- * If not, try for super-user.
- * If permission is granted,
- * return inode pointer.
- */
-struct inode *
-owner()
-{
-       register struct inode *ip;
+       default:
+               return (ENOTTY);

 
 

-       ip = namei(uchar, 0);
-       if(ip == NULL)
-               return(NULL);
-       if(u.u_uid == ip->i_uid)
-               return(ip);
-       if(suser())
-               return(ip);
-       iput(ip);
-       return(NULL);
+       case VFIFO:
+       case VCHR:
+       case VBLK:
+               error = VOP_IOCTL(vp, com, data, fp->f_flag, p->p_ucred, p);
+               if (error == 0 && com == TIOCSCTTY) {
+                       p->p_session->s_ttyvp = vp;
+                       VREF(vp);
+               }
+               return (error);
+       }

 }
 
 /*
 }
 
 /*

- * Test if the current user is the
- * super user.
+ * Vnode select call

  */
  */

-suser()
+vn_select(fp, which, p)
+       struct file *fp;
+       int which;
+       struct proc *p;

 {
 
 {
 

-       if(u.u_uid == 0) {
-               u.u_acflag |= ASU;
-               return(1);
-       }
-       u.u_error = EPERM;
-       return(0);
+       return (VOP_SELECT(((struct vnode *)fp->f_data), which, fp->f_flag,
+               p->p_ucred, p));

 }
 
 /*
 }
 
 /*

- * Allocate a user file descriptor.
+ * Vnode close call

  */
  */

-ufalloc()
+vn_close(fp, p)
+       register struct file *fp;
+       struct proc *p;

 {
 {

-       register i;
+       struct vnode *vp = ((struct vnode *)fp->f_data);
+       int error;

 
 

-       for(i=0; i<NOFILE; i++)
-               if(u.u_ofile[i] == NULL) {
-                       u.u_r.r_val1 = i;
-                       u.u_pofile[i] = 0;
-                       return(i);
-               }
-       u.u_error = EMFILE;
-       return(-1);
+       /*
+        * Must delete vnode reference from this file entry
+        * before VOP_CLOSE, so that only other references
+        * will prevent close.
+        */
+       fp->f_data = (caddr_t) 0;
+       error = VOP_CLOSE(vp, fp->f_flag, fp->f_cred, p);
+       vrele(vp);
+       return (error);

 }
 
 }
 

-struct file *lastf = &file[0];

 /*
 /*

- * Allocate a user file descriptor
- * and a file structure.
- * Initialize the descriptor
- * to point at the file structure.
- *
- * no file -- if there are no available
- *     file structures.
+ * vn_fhtovp() - convert a fh to a vnode ptr (optionally locked)
+ *     - look up fsid in mount list (if not found ret error)
+ *     - get vp by calling VFS_FHTOVP() macro
+ *     - if lockflag lock it with VOP_LOCK()

  */
  */

-struct file *
-falloc()
+vn_fhtovp(fhp, lockflag, vpp)
+       fhandle_t *fhp;
+       int lockflag;
+       struct vnode **vpp;

 {
 {

-       register struct file *fp;
-       register i;
+       register struct mount *mp;

 
 

-       i = ufalloc();
-       if(i < 0)
-               return(NULL);
-       for(fp = lastf; fp < &file[NFILE]; fp++)
-               if(fp->f_count == 0)
-                       goto slot;
-       for(fp = &file[0]; fp < lastf; fp++)
-               if(fp->f_count == 0)
-                       goto slot;
-       printf("no file\n");
-       u.u_error = ENFILE;
-       return(NULL);
-slot:
-       u.u_ofile[i] = fp;
-       fp->f_count++;
-       fp->f_un.f_offset = 0;
-       lastf = fp + 1;
-       return(fp);
+       if ((mp = getvfs(&fhp->fh_fsid)) == NULL)
+               return (ESTALE);
+       if (VFS_FHTOVP(mp, &fhp->fh_fid, vpp))
+               return (ESTALE);
+       if (!lockflag)
+               VOP_UNLOCK(*vpp);
+       return (0);

 }
 }