diff --git a/servers/pfs/read.c b/servers/pfs/read.c
index 691c0fddc..d80647b58 100644
--- a/servers/pfs/read.c
+++ b/servers/pfs/read.c
@@ -32,23 +32,34 @@ int fs_readwrite(message *fs_m_in, message *fs_m_out)
   /* Get the values from the request message */
   rw_flag = (fs_m_in->m_type == REQ_READ ? READING : WRITING);
   gid = (cp_grant_id_t) fs_m_in->REQ_GRANT;
-  position = fs_m_in->REQ_SEEK_POS_LO;
   nrbytes = (unsigned) fs_m_in->REQ_NBYTES;
 
   /* We can't read beyond the max file position */
   if (nrbytes > MAX_FILE_POS) return(EFBIG);
 
-  if (rw_flag == WRITING) {
-	  /* Check in advance to see if file will grow too big. */
-	  /* Casting nrbytes to signed is safe, because it's guaranteed not to
-	     be beyond max signed value (i.e., MAX_FILE_POS). */
-	  if (position > PIPE_BUF - (signed) nrbytes) return(EFBIG);
-  }
-
   /* Mark inode in use */
   if ((get_inode(rip->i_dev, rip->i_num)) == NULL) return(err_code);
   if ((bp = get_block(rip->i_dev, rip->i_num)) == NULL) return(err_code);
 
+  if (rw_flag == WRITING) {
+	/* Check in advance to see if file will grow too big. */
+	/* Casting nrbytes to signed is safe, because it's guaranteed not to
+	 * be beyond max signed value (i.e., MAX_FILE_POS).
+	 */
+	position = rip->i_size;
+	if (position > PIPE_BUF - (signed) nrbytes) {
+		put_inode(rip);
+		put_block(rip->i_dev, rip->i_num);
+		return(EFBIG);
+	}
+  } else {
+	position = bp->b_bytes;
+	if (nrbytes > rip->i_size - position) {
+		/* There aren't that many bytes to read */
+		nrbytes = rip->i_size - position;
+	}
+  }
+
   if (rw_flag == READING) {
 	/* Copy a chunk from the block buffer to user space. */
 	r = sys_safecopyto(VFS_PROC_NR, gid, (vir_bytes) 0,
@@ -78,10 +89,12 @@ int fs_readwrite(message *fs_m_in, message *fs_m_out)
 	}
   }
 
-  bp->b_bytes = position;
+  if (rw_flag == READING)
+	bp->b_bytes = position;
   if (rw_flag == READING) rip->i_update |= ATIME;
   if (rw_flag == WRITING) rip->i_update |= CTIME | MTIME;
   fs_m_out->RES_NBYTES = (size_t) cum_io;
+
   put_inode(rip);
   put_block(rip->i_dev, rip->i_num);
 
diff --git a/servers/vfs/open.c b/servers/vfs/open.c
index 0bc7690ff..cc386f6a8 100644
--- a/servers/vfs/open.c
+++ b/servers/vfs/open.c
@@ -248,8 +248,6 @@ int common_open(char path[PATH_MAX], int oflags, mode_t omode)
 			r = map_vnode(vp, PFS_PROC_NR);
 			if (r == OK) {
 				if (vp->v_ref_count == 1) {
-					vp->v_pipe_rd_pos = 0;
-					vp->v_pipe_wr_pos = 0;
 					if (vp->v_size != 0)
 						r = truncate_vnode(vp, 0);
 				}
diff --git a/servers/vfs/pipe.c b/servers/vfs/pipe.c
index 494f229df..0ad1c471b 100644
--- a/servers/vfs/pipe.c
+++ b/servers/vfs/pipe.c
@@ -108,8 +108,6 @@ int do_pipe()
   vp->v_inode_nr = res.inode_nr;
   vp->v_mapinode_nr = res.inode_nr;
   vp->v_mode = res.fmode;
-  vp->v_pipe_rd_pos= 0;
-  vp->v_pipe_wr_pos= 0;
   vp->v_fs_count = 1;
   vp->v_mapfs_count = 1;
   vp->v_ref_count = 1;
@@ -175,13 +173,13 @@ endpoint_t map_to_fs_e;
 /*===========================================================================*
  *				pipe_check				     *
  *===========================================================================*/
-int pipe_check(vp, rw_flag, oflags, bytes, position, notouch)
-register struct vnode *vp;	/* the inode of the pipe */
-int rw_flag;			/* READING or WRITING */
-int oflags;			/* flags set by open or fcntl */
-register int bytes;		/* bytes to be read or written (all chunks) */
-u64_t position;			/* current file position */
-int notouch;			/* check only */
+int pipe_check(
+struct vnode *vp,	/* the inode of the pipe */
+int rw_flag,		/* READING or WRITING */
+int oflags,		/* flags set by open or fcntl */
+int bytes,		/* bytes to be read or written (all chunks) */
+int notouch		/* check only */
+)
 {
 /* Pipes are a little different.  If a process reads from an empty pipe for
  * which a writer still exists, suspend the reader.  If the pipe is empty
@@ -191,13 +189,15 @@ int notouch;			/* check only */
   off_t pos;
   int r = OK;
 
-  if (ex64hi(position) != 0)
-	panic("pipe_check: position too large in pipe");
-  pos = ex64lo(position);
+  /* Reads start at the beginning; writes append to pipes */
+  if (rw_flag == READING)
+	pos = 0;
+  else
+	pos = vp->v_size;
 
   /* If reading, check for empty pipe. */
   if (rw_flag == READING) {
-	if (pos >= vp->v_size) {
+	if (vp->v_size == 0) {
 		/* Process is reading from an empty pipe. */
 		if (find_filp(vp, W_BIT) != NULL) {
 			/* Writer exists */
diff --git a/servers/vfs/proto.h b/servers/vfs/proto.h
index 4b1f2f220..b98d7420d 100644
--- a/servers/vfs/proto.h
+++ b/servers/vfs/proto.h
@@ -187,7 +187,7 @@ int do_pipe(void);
 int map_vnode(struct vnode *vp, endpoint_t fs_e);
 void unpause(endpoint_t proc_e);
 int pipe_check(struct vnode *vp, int rw_flag, int oflags, int bytes,
-	u64_t position, int notouch);
+	int notouch);
 void release(struct vnode *vp, int op, int count);
 void revive(endpoint_t proc_e, int returned);
 void suspend(int why);
diff --git a/servers/vfs/read.c b/servers/vfs/read.c
index 59cbd2b7a..65722ab41 100644
--- a/servers/vfs/read.c
+++ b/servers/vfs/read.c
@@ -264,7 +264,7 @@ size_t req_size;
   int r, oflags, partial_pipe = 0;
   size_t size, cum_io, cum_io_incr;
   struct vnode *vp;
-  u64_t position, new_pos;
+  u64_t  position, new_pos;
 
   /* Must make sure we're operating on locked filp and vnode */
   assert(tll_locked_by_me(&f->filp_vno->v_lock));
@@ -272,12 +272,12 @@ size_t req_size;
 
   oflags = f->filp_flags;
   vp = f->filp_vno;
-  position = cvu64((rw_flag == READING) ? vp->v_pipe_rd_pos :
-							vp->v_pipe_wr_pos);
+  position = cvu64(0);	/* Not actually used */
+
   /* fp->fp_cum_io_partial is only nonzero when doing partial writes */
   cum_io = fp->fp_cum_io_partial;
 
-  r = pipe_check(vp, rw_flag, oflags, req_size, position, 0);
+  r = pipe_check(vp, rw_flag, oflags, req_size, 0);
   if (r <= 0) {
 	if (r == SUSPEND) pipe_suspend(f, buf, req_size);
 	return(r);
@@ -287,16 +287,8 @@ size_t req_size;
   if (size < req_size) partial_pipe = 1;
 
   /* Truncate read request at size. */
-  if((rw_flag == READING) &&
-	cmp64ul(add64ul(position, size), vp->v_size) > 0) {
-	/* Position always should fit in an off_t (LONG_MAX). */
-	off_t pos32;
-
-	assert(cmp64ul(position, LONG_MAX) <= 0);
-	pos32 = cv64ul(position);
-	assert(pos32 >= 0);
-	assert(pos32 <= LONG_MAX);
-	size = vp->v_size - pos32;
+  if (rw_flag == READING && size > vp->v_size) {
+	size = vp->v_size;
   }
 
   if (vp->v_mapfs_e == 0)
@@ -309,7 +301,6 @@ size_t req_size;
 	if (ex64hi(new_pos))
 		panic("rw_pipe: bad new pos");
 
-	position = new_pos;
 	cum_io += cum_io_incr;
 	buf += cum_io_incr;
 	req_size -= cum_io_incr;
@@ -317,27 +308,19 @@ size_t req_size;
 
   /* On write, update file size and access time. */
   if (rw_flag == WRITING) {
-	if (cmp64ul(position, vp->v_size) > 0) {
-		if (ex64hi(position) != 0) {
+	if (cmp64ul(new_pos, vp->v_size) > 0) {
+		if (ex64hi(new_pos) != 0) {
 			panic("read_write: file size too big for v_size");
 		}
-		vp->v_size = ex64lo(position);
+		vp->v_size = ex64lo(new_pos);
 	}
   } else {
-	if (cmp64ul(position, vp->v_size) >= 0) {
-		/* Reset pipe pointers */
+	if (cmp64ul(new_pos, vp->v_size) >= 0) {
+		/* Pipe emtpy; reset size */
 		vp->v_size = 0;
-		vp->v_pipe_rd_pos= 0;
-		vp->v_pipe_wr_pos= 0;
-		position = cvu64(0);
 	}
   }
 
-  if (rw_flag == READING)
-	vp->v_pipe_rd_pos= cv64ul(position);
-  else
-	vp->v_pipe_wr_pos= cv64ul(position);
-
   if (r == OK) {
 	if (partial_pipe) {
 		/* partial write on pipe with */
diff --git a/servers/vfs/select.c b/servers/vfs/select.c
index 089822931..8c3a13ce7 100644
--- a/servers/vfs/select.c
+++ b/servers/vfs/select.c
@@ -451,7 +451,8 @@ static int select_request_pipe(struct filp *f, int *ops, int block)
   orig_ops = *ops;
 
   if ((*ops & (SEL_RD|SEL_ERR))) {
-	err = pipe_check(f->filp_vno, READING, 0, 1, f->filp_pos, 1);
+	/* Check if we can read 1 byte */
+	err = pipe_check(f->filp_vno, READING, 0, 1, 1 /* Check only */);
 
 	if (err != SUSPEND)
 		r |= SEL_RD;
@@ -466,7 +467,8 @@ static int select_request_pipe(struct filp *f, int *ops, int block)
   }
 
   if ((*ops & (SEL_WR|SEL_ERR))) {
-	err = pipe_check(f->filp_vno, WRITING, 0, 1, f->filp_pos, 1);
+	/* Check if we can write 1 byte */
+	err = pipe_check(f->filp_vno, WRITING, 0, 1, 1 /* Check only */);
 
 	if (err != SUSPEND)
 		r |= SEL_WR;
diff --git a/servers/vfs/vnode.h b/servers/vfs/vnode.h
index bcea1f063..2f01d7379 100644
--- a/servers/vfs/vnode.h
+++ b/servers/vfs/vnode.h
@@ -16,8 +16,6 @@ EXTERN struct vnode {
 #if 0
   int v_ref_check;		/* for consistency checks */
 #endif
-  off_t v_pipe_rd_pos;
-  off_t v_pipe_wr_pos;
   endpoint_t v_bfs_e;		/* endpoint number for the FS proces in case
 				   of a block special file */
   dev_t v_dev;                  /* device number on which the corresponding