VOP_RDWR(9)

NAME

     VOP_READ, VOP_WRITE -- read or write a file

SYNOPSIS

     #include <sys/param.h>
     #include <sys/vnode.h>
     #include <sys/uio.h>

     int
     VOP_READ(struct vnode *vp, struct uio *uio, int ioflag,
	 struct ucred *cred);

     int
     VOP_WRITE(struct vnode *vp, struct uio *uio, int ioflag,
	 struct ucred *cred);

DESCRIPTION

     These entry points read or write the contents of a file

     The arguments are:

     vp      the vnode of the file

     uio     the location of the data to be read or written

     ioflag  various flags

     cnp     the credentials of the caller

     The ioflag argument is used to give directives and hints to the file sys-
     tem.  When attempting a read, the high 16 bits are used to provide a
     read-ahead hint (in units of file system blocks) that the file system
     should attempt.  The low 16 bits are a bit mask which can contain the
     following flags:

     IO_UNIT	    do I/O as atomic unit

     IO_APPEND	    append write to end

     IO_SYNC	    do I/O synchronously

     IO_NODELOCKED  underlying node already locked

     IO_NDELAY	    FNDELAY flag set in file table

     IO_VMIO	    data already in VMIO space

LOCKS

     The file should be locked on entry and will still be locked on exit.

RETURN VALUES

     Zero is returned on success, otherwise an error code is returned.

PSEUDOCODE

     int
     vop_read(struct vnode *vp, struct uio *uio, int ioflag, struct ucred *cred)
     {
	 struct buf *bp;
	 off_t bytesinfile;
	 daddr_t lbn, nextlbn;
	 long size, xfersize, blkoffset;
	 int error;

	 size = block size of file system;

	 for (error = 0, bp = NULL; uio->uio_resid > 0; bp = NULL) {
	     bytesinfile = size of file - uio->uio_offset;
	     if (bytesinfile <= 0)
		 break;

	     lbn = uio->uio_offset / size;
	     blkoffset = uio->uio_offset - lbn * size;

	     xfersize = size - blkoffset;
	     if (uio->uio_resid < xfersize)
		 xfersize = uio->uio_resid;
	     if (bytesinfile < xfersize)
		 xfersize = bytesinfile;

	     error = bread(vp, lbn, size, NOCRED, &bp);
	     if (error) {
		 brelse(bp);
		 bp = NULL;
		 break;
	     }

	     /*
	      * We should only get non-zero b_resid when an I/O error
	      * has occurred, which should cause us to break above.
	      * However, if the short read did not cause an error,
	      * then we want to ensure that we do not uiomove bad
	      * or uninitialized data.
	      */
	     size -= bp->b_resid;
	     if (size < xfersize) {
		 if (size == 0)
		     break;
		 xfersize = size;
	     }

	     error = uiomove((char *)bp->b_data + blkoffset, (int)xfersize, uio);
	     if (error)
		 break;

	     bqrelse(bp);
	 }
	 if (bp != NULL)
	     bqrelse(bp);

	 return (error);
     }

     int
     vop_write(struct vnode *vp, struct uio *uio, int ioflag, struct ucred *cred)
     {
	 struct buf *bp;
	 off_t bytesinfile;
	 daddr_t lbn, nextlbn;
	 off_t osize;
	 long size, resid, xfersize, blkoffset;
	 int flags;
	 int error;

	 osize = size of file;
	 size = block size of file system;
	 resid = uio->uio_resid;
	 if (ioflag & IO_SYNC)
	     flags = B_SYNC;
	 else
	     flags = 0;

	 for (error = 0; uio->uio_resid > 0;) {
	     lbn = uio->uio_offset / size;
	     blkoffset = uio->uio_offset - lbn * size;

	     xfersize = size - blkoffset;
	     if (uio->uio_resid < xfersize)
		 xfersize = uio->uio_resid;

	     if (uio->uio_offset + xfersize > size of file)
		 vnode_pager_setsize(vp, uio->uio_offset + xfersize);

	     if (size > xfersize)
		 flags |= B_CLRBUF;
	     else
		 flags &= ~B_CLRBUF;

	     error = find_block_in_file(vp, lbn, blkoffset + xfersize,
					cred, &bp, flags);
	     if (error)
		 break;

	     if (uio->uio_offset + xfersize > size of file)
		 set size of file to uio->uio_offset + xfersize;

	     error = uiomove((char *)bp->b_data + blkoffset, (int) xfersize, uio);
	     /* XXX ufs does not check the error here.	Why? */

	     if (ioflag & IO_VMIO)
		 bp->b_flags |= B_RELBUF; /* ??? */

	     if (ioflag & IO_SYNC)
		 bwrite(bp);
	     else if (xfersize + blkoffset == size)
		 bawrite(bp);
	     else
		 bdwrite(bp);

	     if (error || xfersize == 0)
		 break;
	 }

	 if (error) {
	     if (ioflag & IO_UNIT) {
		 /* call private routine to truncate file. */
		 your_truncate(vp, osize, ioflag & IO_SYNC, cred, uio->uio_td);
		 uio->uio_offset -= resid - uio->uio_resid;
		 uio->uio_resid = resid;
	     }
	 } else if (resid > uio->uio_resid && (ioflag & IO_SYNC)) {
	     struct timeval tv;
	     error = VOP_UPDATE(vp, &tv, &tv, 1); /* XXX what does this do? */
	 }

	 return (error);
     }

ERRORS

     [EFBIG]		An attempt was made to write a file that exceeds the
			process's file size limit or the maximum file size.

     [ENOSPC]		The file system is full.

     [EPERM]		An append-only flag is set on the file, but the caller
			is attempting to write before the current end of file.

SEE ALSO

     uiomove(9), vnode(9)

AUTHORS

     This man page was written by Doug Rabson.