1 files changed, 558 insertions, 387 deletions

diff --git a/sys/compat/linux/linux_file.c b/sys/compat/linux/linux_file.c
index 2a5d673..3d16a94 100644
--- a/sys/compat/linux/linux_file.c
+++ b/sys/compat/linux/linux_file.c
@@ -149,175 +149,6 @@ linux_open(struct proc *p, struct linux_open_args *args)
     return error;
 }
 
-struct linux_flock {
-    short l_type;
-    short l_whence;
-    linux_off_t l_start;
-    linux_off_t l_len;
-    linux_pid_t l_pid;
-};
-
-static void
-linux_to_bsd_flock(struct linux_flock *linux_flock, struct flock *bsd_flock)
-{
-    switch (linux_flock->l_type) {
-    case LINUX_F_RDLCK:
-	bsd_flock->l_type = F_RDLCK;
-	break;
-    case LINUX_F_WRLCK:
-	bsd_flock->l_type = F_WRLCK;
-	break;
-    case LINUX_F_UNLCK:
-	bsd_flock->l_type = F_UNLCK;
-	break;
-    default:
-        bsd_flock->l_type = -1;
-        break;
-    }
-    bsd_flock->l_whence = linux_flock->l_whence;
-    bsd_flock->l_start = (off_t)linux_flock->l_start;
-    bsd_flock->l_len = (off_t)linux_flock->l_len;
-    bsd_flock->l_pid = (pid_t)linux_flock->l_pid;
-}
-
-static void
-bsd_to_linux_flock(struct flock *bsd_flock, struct linux_flock *linux_flock)
-{
-    switch (bsd_flock->l_type) {
-    case F_RDLCK:
-	linux_flock->l_type = LINUX_F_RDLCK;
-	break;
-    case F_WRLCK:
-	linux_flock->l_type = LINUX_F_WRLCK;
-	break;
-    case F_UNLCK:
-	linux_flock->l_type = LINUX_F_UNLCK;
-	break;
-    }
-    linux_flock->l_whence = bsd_flock->l_whence;
-    linux_flock->l_start = (linux_off_t)bsd_flock->l_start;
-    linux_flock->l_len = (linux_off_t)bsd_flock->l_len;
-    linux_flock->l_pid = (linux_pid_t)bsd_flock->l_pid;
-}
-
-int
-linux_fcntl(struct proc *p, struct linux_fcntl_args *args)
-{
-    int error, result;
-    struct fcntl_args /* {
-	int fd;
-	int cmd;
-	long arg;
-    } */ fcntl_args;
-    struct linux_flock linux_flock;
-    struct flock *bsd_flock;
-    struct filedesc *fdp;
-    struct file *fp;
-    caddr_t sg;
-
-    sg = stackgap_init();
-    bsd_flock = (struct flock *)stackgap_alloc(&sg, sizeof(struct flock));
-
-#ifdef DEBUG
-	if (ldebug(fcntl))
-		printf(ARGS(fcntl, "%d, %08x, *"), args->fd, args->cmd);
-#endif
-    fcntl_args.fd = args->fd;
-
-    switch (args->cmd) {
-    case LINUX_F_DUPFD:
-	fcntl_args.cmd = F_DUPFD;
-	fcntl_args.arg = args->arg;
-	return fcntl(p, &fcntl_args);
-
-    case LINUX_F_GETFD:
-	fcntl_args.cmd = F_GETFD;
-	return fcntl(p, &fcntl_args);
-
-    case LINUX_F_SETFD:
-	fcntl_args.cmd = F_SETFD;
-	fcntl_args.arg = args->arg;
-	return fcntl(p, &fcntl_args);
-
-    case LINUX_F_GETFL:
-	fcntl_args.cmd = F_GETFL;
-	error = fcntl(p, &fcntl_args);
-	result = p->p_retval[0];
-	p->p_retval[0] = 0;
-	if (result & O_RDONLY) p->p_retval[0] |= LINUX_O_RDONLY;
-	if (result & O_WRONLY) p->p_retval[0] |= LINUX_O_WRONLY;
-	if (result & O_RDWR) p->p_retval[0] |= LINUX_O_RDWR;
-	if (result & O_NDELAY) p->p_retval[0] |= LINUX_O_NONBLOCK;
-	if (result & O_APPEND) p->p_retval[0] |= LINUX_O_APPEND;
-	if (result & O_FSYNC) p->p_retval[0] |= LINUX_O_SYNC;
-	if (result & O_ASYNC) p->p_retval[0] |= LINUX_FASYNC;
-	return error;
-
-    case LINUX_F_SETFL:
-	fcntl_args.arg = 0;
-	if (args->arg & LINUX_O_NDELAY) fcntl_args.arg |= O_NONBLOCK;
-	if (args->arg & LINUX_O_APPEND) fcntl_args.arg |= O_APPEND;
-	if (args->arg & LINUX_O_SYNC) fcntl_args.arg |= O_FSYNC;
-	if (args->arg & LINUX_FASYNC) fcntl_args.arg |= O_ASYNC;
-	fcntl_args.cmd = F_SETFL;
-	return fcntl(p, &fcntl_args);
-
-    case LINUX_F_GETLK:
-	if ((error = copyin((caddr_t)args->arg, (caddr_t)&linux_flock,
-		   	    sizeof(struct linux_flock)))) 
-	    return error;
-	linux_to_bsd_flock(&linux_flock, bsd_flock);
-	fcntl_args.cmd = F_GETLK;
-	fcntl_args.arg = (long)bsd_flock;
-	error = fcntl(p, &fcntl_args);
-	if (error)
-	    return error;
-	bsd_to_linux_flock(bsd_flock, &linux_flock);
-	return copyout((caddr_t)&linux_flock, (caddr_t)args->arg,
-		       sizeof(struct linux_flock));
-
-    case LINUX_F_SETLK:
-	if ((error = copyin((caddr_t)args->arg, (caddr_t)&linux_flock,
-		   	    sizeof(struct linux_flock)))) 
-	    return error;
-	linux_to_bsd_flock(&linux_flock, bsd_flock);
-	fcntl_args.cmd = F_SETLK;
-	fcntl_args.arg = (long)bsd_flock;
-	return fcntl(p, &fcntl_args);
-
-    case LINUX_F_SETLKW:
-	if ((error = copyin((caddr_t)args->arg, (caddr_t)&linux_flock,
-		   	    sizeof(struct linux_flock)))) 
-	    return error;
-	linux_to_bsd_flock(&linux_flock, bsd_flock);
-	fcntl_args.cmd = F_SETLKW;
-	fcntl_args.arg = (long)bsd_flock;
-	return fcntl(p, &fcntl_args);
-
-    case LINUX_F_GETOWN:
-	fcntl_args.cmd = F_GETOWN;
-	return fcntl(p, &fcntl_args);
-
-    case LINUX_F_SETOWN:
-	/*
-	 * XXX some Linux applications depend on F_SETOWN having no
-	 * significant effect for pipes (SIGIO is not delivered for
-	 * pipes under Linux-2.2.35 at least).
-	 */
-	fdp = p->p_fd;
-	if ((u_int)args->fd >= fdp->fd_nfiles ||
-	  (fp = fdp->fd_ofiles[args->fd]) == NULL)
-	    return EBADF;
-	if (fp->f_type == DTYPE_PIPE)
-	    return EINVAL;
-
- 	fcntl_args.cmd = F_SETOWN;
-	fcntl_args.arg = args->arg;
-	return fcntl(p, &fcntl_args);
-    }
-    return EINVAL;
-}
-
 int
 linux_lseek(struct proc *p, struct linux_lseek_args *args)
 {
@@ -333,7 +164,7 @@ linux_lseek(struct proc *p, struct linux_lseek_args *args)
 #ifdef DEBUG
 	if (ldebug(lseek))
 		printf(ARGS(lseek, "%d, %ld, %d"),
-		    args->fdes, args->off, args->whence);
+		    args->fdes, (long)args->off, args->whence);
 #endif
     tmp_args.fd = args->fdes;
     tmp_args.offset = (off_t)args->off;
@@ -372,17 +203,6 @@ linux_llseek(struct proc *p, struct linux_llseek_args *args)
 }
 #endif /*!__alpha__*/
 
-
-struct linux_dirent {
-    long dino;
-    linux_off_t doff;
-    unsigned short dreclen;
-    char dname[LINUX_NAME_MAX + 1];
-};
-
-#define LINUX_RECLEN(de,namlen) \
-    ALIGN((((char *)&(de)->dname - (char *)de) + (namlen) + 1))
-
 #ifndef __alpha__
 int
 linux_readdir(struct proc *p, struct linux_readdir_args *args)
@@ -396,178 +216,250 @@ linux_readdir(struct proc *p, struct linux_readdir_args *args)
 }
 #endif /*!__alpha__*/
 
-int
-linux_getdents(struct proc *p, struct linux_getdents_args *args)
-{
-    register struct dirent *bdp;
-    struct vnode *vp;
-    caddr_t inp, buf;		/* BSD-format */
-    int len, reclen;		/* BSD-format */
-    caddr_t outp;		/* Linux-format */
-    int resid, linuxreclen=0;	/* Linux-format */
-    struct file *fp;
-    struct uio auio;
-    struct iovec aiov;
-    struct vattr va;
-    off_t off;
-    struct linux_dirent linux_dirent;
-    int buflen, error, eofflag, nbytes, justone;
-    u_long *cookies = NULL, *cookiep;
-    int ncookies;
+/*
+ * Note that linux_getdents(2) and linux_getdents64(2) have the same
+ * arguments. They only differ in the definition of struct dirent they
+ * operate on. We use this to common the code, with the exception of
+ * accessing struct dirent. Note that linux_readdir(2) is implemented
+ * by means of linux_getdents(2). In this case we never operate on
+ * struct dirent64 and thus don't need to handle it...
+ */
 
-#ifdef DEBUG
-	if (ldebug(getdents))
-		printf(ARGS(getdents, "%d, *, %d"), args->fd, args->count);
-#endif
-    if ((error = getvnode(p->p_fd, args->fd, &fp)) != 0) {
-	return (error);
-    }
+struct l_dirent {
+	l_long		d_ino;
+	l_off_t		d_off;
+	l_ushort	d_reclen;
+	char		d_name[LINUX_NAME_MAX + 1];
+};
 
-    if ((fp->f_flag & FREAD) == 0)
-	return (EBADF);
+struct l_dirent64 {
+	uint64_t	d_ino;
+	int64_t		d_off;
+	l_ushort	d_reclen;
+	u_char		d_type;
+	char		d_name[LINUX_NAME_MAX + 1];
+};
 
-    vp = (struct vnode *) fp->f_data;
+#define LINUX_RECLEN(de,namlen) \
+    ALIGN((((char *)&(de)->d_name - (char *)de) + (namlen) + 1))
 
-    if (vp->v_type != VDIR)
-	return (EINVAL);
+#define	LINUX_DIRBLKSIZ		512
 
-    if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p))) {
-	return error;
-    }
+static int
+getdents_common(struct proc *p, struct linux_getdents64_args *args,
+    int is64bit)
+{
+	register struct dirent *bdp;
+	struct vnode *vp;
+	caddr_t inp, buf;		/* BSD-format */
+	int len, reclen;		/* BSD-format */
+	caddr_t outp;			/* Linux-format */
+	int resid, linuxreclen=0;	/* Linux-format */
+	struct file *fp;
+	struct uio auio;
+	struct iovec aiov;
+	struct vattr va;
+	off_t off;
+	struct l_dirent linux_dirent;
+	struct l_dirent64 linux_dirent64;
+	int buflen, error, eofflag, nbytes, justone;
+	u_long *cookies = NULL, *cookiep;
+	int ncookies;
+
+	if ((error = getvnode(p->p_fd, args->fd, &fp)) != 0)
+		return (error);
+
+	if ((fp->f_flag & FREAD) == 0)
+		return (EBADF);
+
+	vp = (struct vnode *) fp->f_data;
+	if (vp->v_type != VDIR)
+		return (EINVAL);
+
+	if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)))
+		return (error);
+
+	nbytes = args->count;
+	if (nbytes == 1) {
+		/* readdir(2) case. Always struct dirent. */
+		if (is64bit)
+			return (EINVAL);
+		nbytes = sizeof(linux_dirent);
+		justone = 1;
+	} else
+		justone = 0;
+
+	off = fp->f_offset;
+
+	buflen = max(LINUX_DIRBLKSIZ, nbytes);
+	buflen = min(buflen, MAXBSIZE);
+	buf = malloc(buflen, M_TEMP, M_WAITOK);
+	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
 
-    nbytes = args->count;
-    if (nbytes == 1) {
-	nbytes = sizeof (struct linux_dirent);
-	justone = 1;
-    }
-    else
-	justone = 0;
-
-    off = fp->f_offset;
-#define	DIRBLKSIZ	512		/* XXX we used to use ufs's DIRBLKSIZ */
-    buflen = max(DIRBLKSIZ, nbytes);
-    buflen = min(buflen, MAXBSIZE);
-    buf = malloc(buflen, M_TEMP, M_WAITOK);
-    vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
 again:
-    aiov.iov_base = buf;
-    aiov.iov_len = buflen;
-    auio.uio_iov = &aiov;
-    auio.uio_iovcnt = 1;
-    auio.uio_rw = UIO_READ;
-    auio.uio_segflg = UIO_SYSSPACE;
-    auio.uio_procp = p;
-    auio.uio_resid = buflen;
-    auio.uio_offset = off;
-
-    if (cookies) {
-	free(cookies, M_TEMP);
-	cookies = NULL;
-    }
-
-    error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, &ncookies, &cookies);
-    if (error) {
-	goto out;
-    }
-
-    inp = buf;
-    outp = (caddr_t) args->dent;
-    resid = nbytes;
-    if ((len = buflen - auio.uio_resid) <= 0) {
-	goto eof;
-    }
-
-    cookiep = cookies;
-
-    if (cookies) {
-	/*
-	 * When using cookies, the vfs has the option of reading from
-	 * a different offset than that supplied (UFS truncates the
-	 * offset to a block boundary to make sure that it never reads
-	 * partway through a directory entry, even if the directory
-	 * has been compacted).
-	 */
-	while (len > 0 && ncookies > 0 && *cookiep <= off) {
-	    bdp = (struct dirent *) inp;
-	    len -= bdp->d_reclen;
-	    inp += bdp->d_reclen;
-	    cookiep++;
-	    ncookies--;
-	}
-    }
-
-    while (len > 0) {
-	if (cookiep && ncookies == 0)
-	    break;
-	bdp = (struct dirent *) inp;
-	reclen = bdp->d_reclen;
-	if (reclen & 3) {
-	    printf("linux_readdir: reclen=%d\n", reclen);
-	    error = EFAULT;
-	    goto out;
-	}
-  
-	if (bdp->d_fileno == 0) {
-	    inp += reclen;
-	    if (cookiep) {
-		off = *cookiep++;
-		ncookies--;
-	    } else
-		off += reclen;
-	    len -= reclen;
-	    continue;
-	}
-	linuxreclen = LINUX_RECLEN(&linux_dirent, bdp->d_namlen);
-	if (reclen > len || resid < linuxreclen) {
-	    outp++;
-	    break;
+	aiov.iov_base = buf;
+	aiov.iov_len = buflen;
+	auio.uio_iov = &aiov;
+	auio.uio_iovcnt = 1;
+	auio.uio_rw = UIO_READ;
+	auio.uio_segflg = UIO_SYSSPACE;
+	auio.uio_procp = p;
+	auio.uio_resid = buflen;
+	auio.uio_offset = off;
+
+	if (cookies) {
+		free(cookies, M_TEMP);
+		cookies = NULL;
 	}
-	linux_dirent.dino = (long) bdp->d_fileno;
-	if (justone) {
-	    /*
-	     * old linux-style readdir usage.
-	     */
-	    linux_dirent.doff = (linux_off_t) linuxreclen;
-	    linux_dirent.dreclen = (u_short) bdp->d_namlen;
-	} else {
-	    if (cookiep)
-		linux_dirent.doff = (linux_off_t)*cookiep;
-	    else
-		linux_dirent.doff = (linux_off_t)(off + reclen);
-	    linux_dirent.dreclen = (u_short) linuxreclen;
+
+	if ((error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, &ncookies,
+		 &cookies)))
+		goto out;
+
+	inp = buf;
+	outp = (caddr_t)args->dirent;
+	resid = nbytes;
+	if ((len = buflen - auio.uio_resid) <= 0)
+		goto eof;
+
+	cookiep = cookies;
+
+	if (cookies) {
+		/*
+		 * When using cookies, the vfs has the option of reading from
+		 * a different offset than that supplied (UFS truncates the
+		 * offset to a block boundary to make sure that it never reads
+		 * partway through a directory entry, even if the directory
+		 * has been compacted).
+		 */
+		while (len > 0 && ncookies > 0 && *cookiep <= off) {
+			bdp = (struct dirent *) inp;
+			len -= bdp->d_reclen;
+			inp += bdp->d_reclen;
+			cookiep++;
+			ncookies--;
+		}
 	}
-	strcpy(linux_dirent.dname, bdp->d_name);
-	if ((error = copyout((caddr_t)&linux_dirent, outp, linuxreclen))) {
-	    goto out;
+
+	while (len > 0) {
+		if (cookiep && ncookies == 0)
+			break;
+		bdp = (struct dirent *) inp;
+		reclen = bdp->d_reclen;
+		if (reclen & 3) {
+			error = EFAULT;
+			goto out;
+		}
+
+		if (bdp->d_fileno == 0) {
+			inp += reclen;
+			if (cookiep) {
+				off = *cookiep++;
+				ncookies--;
+			} else
+				off += reclen;
+
+			len -= reclen;
+			continue;
+		}
+
+		linuxreclen = (is64bit)
+		    ? LINUX_RECLEN(&linux_dirent64, bdp->d_namlen)
+		    : LINUX_RECLEN(&linux_dirent, bdp->d_namlen);
+
+		if (reclen > len || resid < linuxreclen) {
+			outp++;
+			break;
+		}
+
+		if (justone) {
+			/* readdir(2) case. */
+			linux_dirent.d_ino = (l_long)bdp->d_fileno;
+			linux_dirent.d_off = (l_off_t)linuxreclen;
+			linux_dirent.d_reclen = (l_ushort)bdp->d_namlen;
+			strcpy(linux_dirent.d_name, bdp->d_name);
+			error = copyout(&linux_dirent, outp, linuxreclen);
+		} else {
+			if (is64bit) {
+				linux_dirent64.d_ino = bdp->d_fileno;
+				linux_dirent64.d_off = (cookiep)
+				    ? (l_off_t)*cookiep
+				    : (l_off_t)(off + reclen);
+				linux_dirent64.d_reclen =
+				    (l_ushort)linuxreclen;
+				linux_dirent64.d_type = bdp->d_type;
+				strcpy(linux_dirent64.d_name, bdp->d_name);
+				error = copyout(&linux_dirent64, outp,
+				    linuxreclen);
+			} else {
+				linux_dirent.d_ino = bdp->d_fileno;
+				linux_dirent.d_off = (cookiep)
+				    ? (l_off_t)*cookiep
+				    : (l_off_t)(off + reclen);
+				linux_dirent.d_reclen = (l_ushort)linuxreclen;
+				strcpy(linux_dirent.d_name, bdp->d_name);
+				error = copyout(&linux_dirent, outp,
+				    linuxreclen);
+			}
+		}
+		if (error)
+			goto out;
+
+		inp += reclen;
+		if (cookiep) {
+			off = *cookiep++;
+			ncookies--;
+		} else
+			off += reclen;
+
+		outp += linuxreclen;
+		resid -= linuxreclen;
+		len -= reclen;
+		if (justone)
+			break;
 	}
-	inp += reclen;
-	if (cookiep) {
-	    off = *cookiep++;
-	    ncookies--;
-	} else
-	    off += reclen;
-	outp += linuxreclen;
-	resid -= linuxreclen;
-	len -= reclen;
-	if (justone)
-	    break;
-    }
 
-    if (outp == (caddr_t) args->dent)
-	goto again;
-    fp->f_offset = off;
+	if (outp == (caddr_t)args->dirent)
+		goto again;
 
-    if (justone)
-	nbytes = resid + linuxreclen;
+	fp->f_offset = off;
+	if (justone)
+		nbytes = resid + linuxreclen;
 
 eof:
-    p->p_retval[0] = nbytes - resid;
+	p->p_retval[0] = nbytes - resid;
+
 out:
-    if (cookies)
-	free(cookies, M_TEMP);
-    VOP_UNLOCK(vp, 0, p);
-    free(buf, M_TEMP);
-    return error;
+	if (cookies)
+		free(cookies, M_TEMP);
+
+	VOP_UNLOCK(vp, 0, p);
+	free(buf, M_TEMP);
+	return (error);
+}
+
+int
+linux_getdents(struct proc *p, struct linux_getdents_args *args)
+{
+
+#ifdef DEBUG
+	if (ldebug(getdents))
+		printf(ARGS(getdents, "%d, *, %d"), args->fd, args->count);
+#endif
+
+	return (getdents_common(p, (struct linux_getdents64_args*)args, 0));
+}
+
+int
+linux_getdents64(struct proc *p, struct linux_getdents64_args *args)
+{
+
+#ifdef DEBUG
+	if (ldebug(getdents64))
+		printf(ARGS(getdents64, "%d, *, %d"), args->fd, args->count);
+#endif
+
+	return (getdents_common(p, args, 1));
 }
 
 /*
@@ -649,50 +541,6 @@ linux_chmod(struct proc *p, struct linux_chmod_args *args)
 }
 
 int
-linux_chown(struct proc *p, struct linux_chown_args *args)
-{
-	struct chown_args bsd;
-	caddr_t sg;
-
-	sg = stackgap_init();
-	CHECKALTEXIST(p, &sg, args->path);
-
-#ifdef DEBUG
-	if (ldebug(chown))
-		printf(ARGS(chown, "%s, %d, %d"),
-		    args->path, args->uid, args->gid);
-#endif
-	bsd.path = args->path;
-	/* XXX size casts here */
-	bsd.uid = args->uid;
-	bsd.gid = args->gid;
-
-	return chown(p, &bsd);
-}
-
-int
-linux_lchown(struct proc *p, struct linux_lchown_args *args)
-{
-	struct lchown_args bsd;
-	caddr_t sg;
-
-	sg = stackgap_init();
-	CHECKALTEXIST(p, &sg, args->path);
-
-#ifdef DEBUG
-	if (ldebug(lchown))
-		printf(ARGS(lchown, "%s, %d, %d"),
-		    args->path, args->uid, args->gid);
-#endif
-	bsd.path = args->path;
-	/* XXX size casts here */
-	bsd.uid = args->uid;
-	bsd.gid = args->gid;
-
-	return lchown(p, &bsd);
-}
-
-int
 linux_mkdir(struct proc *p, struct linux_mkdir_args *args)
 {
 	struct mkdir_args bsd;
@@ -801,7 +649,8 @@ linux_truncate(struct proc *p, struct linux_truncate_args *args)
 
 #ifdef DEBUG
 	if (ldebug(truncate))
-		printf(ARGS(truncate, "%s, %ld"), args->path, args->length);
+		printf(ARGS(truncate, "%s, %ld"), args->path,
+		    (long)args->length);
 #endif
 	bsd.path = args->path;
 	bsd.length = args->length;
@@ -937,17 +786,17 @@ linux_mount(struct proc *p, struct linux_mount_args *args)
 }
 
 int
-linux_umount(struct proc *p, struct linux_umount_args *args)
+linux_oldumount(struct proc *p, struct linux_oldumount_args *args)
 {
-	struct linux_umount2_args args2;
+	struct linux_umount_args args2;
 
 	args2.path = args->path;
 	args2.flags = 0;
-	return (linux_umount2(p, &args2));
+	return (linux_umount(p, &args2));
 }
 
 int
-linux_umount2(struct proc *p, struct linux_umount2_args *args)
+linux_umount(struct proc *p, struct linux_umount_args *args)
 {
 	struct unmount_args bsd;
 
@@ -955,3 +804,325 @@ linux_umount2(struct proc *p, struct linux_umount2_args *args)
 	bsd.flags = args->flags;	/* XXX correct? */
 	return (unmount(p, &bsd));
 }
+
+/*
+ * fcntl family of syscalls
+ */
+
+struct l_flock {
+	l_short		l_type;
+	l_short		l_whence;
+	l_off_t		l_start;
+	l_off_t		l_len;
+	l_pid_t		l_pid;
+};
+
+static void
+linux_to_bsd_flock(struct l_flock *linux_flock, struct flock *bsd_flock)
+{
+	switch (linux_flock->l_type) {
+	case LINUX_F_RDLCK:
+		bsd_flock->l_type = F_RDLCK;
+		break;
+	case LINUX_F_WRLCK:
+		bsd_flock->l_type = F_WRLCK;
+		break;
+	case LINUX_F_UNLCK:
+		bsd_flock->l_type = F_UNLCK;
+		break;
+	default:
+		bsd_flock->l_type = -1;
+		break;
+	}
+	bsd_flock->l_whence = linux_flock->l_whence;
+	bsd_flock->l_start = (off_t)linux_flock->l_start;
+	bsd_flock->l_len = (off_t)linux_flock->l_len;
+	bsd_flock->l_pid = (pid_t)linux_flock->l_pid;
+}
+
+static void
+bsd_to_linux_flock(struct flock *bsd_flock, struct l_flock *linux_flock)
+{
+	switch (bsd_flock->l_type) {
+	case F_RDLCK:
+		linux_flock->l_type = LINUX_F_RDLCK;
+		break;
+	case F_WRLCK:
+		linux_flock->l_type = LINUX_F_WRLCK;
+		break;
+	case F_UNLCK:
+		linux_flock->l_type = LINUX_F_UNLCK;
+		break;
+	}
+	linux_flock->l_whence = bsd_flock->l_whence;
+	linux_flock->l_start = (l_off_t)bsd_flock->l_start;
+	linux_flock->l_len = (l_off_t)bsd_flock->l_len;
+	linux_flock->l_pid = (l_pid_t)bsd_flock->l_pid;
+}
+
+#if defined(__i386__)
+struct l_flock64 {
+	l_short		l_type;
+	l_short		l_whence;
+	l_loff_t	l_start;
+	l_loff_t	l_len;
+	l_pid_t		l_pid;
+};
+
+static void
+linux_to_bsd_flock64(struct l_flock64 *linux_flock, struct flock *bsd_flock)
+{
+	switch (linux_flock->l_type) {
+	case LINUX_F_RDLCK:
+		bsd_flock->l_type = F_RDLCK;
+		break;
+	case LINUX_F_WRLCK:
+		bsd_flock->l_type = F_WRLCK;
+		break;
+	case LINUX_F_UNLCK:
+		bsd_flock->l_type = F_UNLCK;
+		break;
+	default:
+		bsd_flock->l_type = -1;
+		break;
+	}
+	bsd_flock->l_whence = linux_flock->l_whence;
+	bsd_flock->l_start = (off_t)linux_flock->l_start;
+	bsd_flock->l_len = (off_t)linux_flock->l_len;
+	bsd_flock->l_pid = (pid_t)linux_flock->l_pid;
+}
+
+static void
+bsd_to_linux_flock64(struct flock *bsd_flock, struct l_flock64 *linux_flock)
+{
+	switch (bsd_flock->l_type) {
+	case F_RDLCK:
+		linux_flock->l_type = LINUX_F_RDLCK;
+		break;
+	case F_WRLCK:
+		linux_flock->l_type = LINUX_F_WRLCK;
+		break;
+	case F_UNLCK:
+		linux_flock->l_type = LINUX_F_UNLCK;
+		break;
+	}
+	linux_flock->l_whence = bsd_flock->l_whence;
+	linux_flock->l_start = (l_loff_t)bsd_flock->l_start;
+	linux_flock->l_len = (l_loff_t)bsd_flock->l_len;
+	linux_flock->l_pid = (l_pid_t)bsd_flock->l_pid;
+}
+#endif /* __i386__ */
+
+#if defined(__alpha__)
+#define	linux_fcntl64_args	linux_fcntl_args
+#endif
+
+static int
+fcntl_common(struct proc *p, struct linux_fcntl64_args *args)
+{
+	struct fcntl_args fcntl_args;
+	struct filedesc *fdp;
+	struct file *fp;
+	int error, result;
+
+	fcntl_args.fd = args->fd;
+
+	switch (args->cmd) {
+	case LINUX_F_DUPFD:
+		fcntl_args.cmd = F_DUPFD;
+		fcntl_args.arg = args->arg;
+		return (fcntl(p, &fcntl_args));
+
+	case LINUX_F_GETFD:
+		fcntl_args.cmd = F_GETFD;
+		return (fcntl(p, &fcntl_args));
+
+	case LINUX_F_SETFD:
+		fcntl_args.cmd = F_SETFD;
+		fcntl_args.arg = args->arg;
+		return (fcntl(p, &fcntl_args));
+
+	case LINUX_F_GETFL:
+		fcntl_args.cmd = F_GETFL;
+		error = fcntl(p, &fcntl_args);
+		result = p->p_retval[0];
+		p->p_retval[0] = 0;
+		if (result & O_RDONLY)
+			p->p_retval[0] |= LINUX_O_RDONLY;
+		if (result & O_WRONLY)
+			p->p_retval[0] |= LINUX_O_WRONLY;
+		if (result & O_RDWR)
+			p->p_retval[0] |= LINUX_O_RDWR;
+		if (result & O_NDELAY)
+			p->p_retval[0] |= LINUX_O_NONBLOCK;
+		if (result & O_APPEND)
+			p->p_retval[0] |= LINUX_O_APPEND;
+		if (result & O_FSYNC)
+			p->p_retval[0] |= LINUX_O_SYNC;
+		if (result & O_ASYNC)
+			p->p_retval[0] |= LINUX_FASYNC;
+		return (error);
+
+	case LINUX_F_SETFL:
+		fcntl_args.arg = 0;
+		if (args->arg & LINUX_O_NDELAY)
+			fcntl_args.arg |= O_NONBLOCK;
+		if (args->arg & LINUX_O_APPEND)
+			fcntl_args.arg |= O_APPEND;
+		if (args->arg & LINUX_O_SYNC)
+			fcntl_args.arg |= O_FSYNC;
+		if (args->arg & LINUX_FASYNC)
+			fcntl_args.arg |= O_ASYNC;
+		fcntl_args.cmd = F_SETFL;
+		return (fcntl(p, &fcntl_args));
+
+	case LINUX_F_GETOWN:
+		fcntl_args.cmd = F_GETOWN;
+		return (fcntl(p, &fcntl_args));
+
+	case LINUX_F_SETOWN:
+		/*
+		 * XXX some Linux applications depend on F_SETOWN having no
+		 * significant effect for pipes (SIGIO is not delivered for
+		 * pipes under Linux-2.2.35 at least).
+		 */
+		fdp = p->p_fd;
+		if ((u_int)args->fd >= fdp->fd_nfiles ||
+		    (fp = fdp->fd_ofiles[args->fd]) == NULL)
+			return (EBADF);
+		if (fp->f_type == DTYPE_PIPE)
+			return (EINVAL);
+
+		fcntl_args.cmd = F_SETOWN;
+		fcntl_args.arg = args->arg;
+		return (fcntl(p, &fcntl_args));
+	}
+
+	return (EINVAL);
+}
+
+int
+linux_fcntl(struct proc *p, struct linux_fcntl_args *args)
+{
+	struct linux_fcntl64_args args64;
+	struct fcntl_args fcntl_args;
+	struct l_flock linux_flock;
+	struct flock *bsd_flock;
+	int error;
+	caddr_t sg;
+
+	sg = stackgap_init();
+	bsd_flock = (struct flock *)stackgap_alloc(&sg, sizeof(bsd_flock));
+
+#ifdef DEBUG
+	if (ldebug(fcntl))
+		printf(ARGS(fcntl, "%d, %08x, *"), args->fd, args->cmd);
+#endif
+
+	switch (args->cmd) {
+	case LINUX_F_GETLK:
+		error = copyin((caddr_t)args->arg, &linux_flock,
+		    sizeof(linux_flock));
+		if (error)
+			return (error);
+		linux_to_bsd_flock(&linux_flock, bsd_flock);
+		fcntl_args.fd = args->fd;
+		fcntl_args.cmd = F_GETLK;
+		fcntl_args.arg = (long)bsd_flock;
+		error = fcntl(p, &fcntl_args);
+		if (error)
+			return (error);
+		bsd_to_linux_flock(bsd_flock, &linux_flock);
+		return (copyout(&linux_flock, (caddr_t)args->arg,
+		    sizeof(linux_flock)));
+
+	case LINUX_F_SETLK:
+		error = copyin((caddr_t)args->arg, &linux_flock,
+		    sizeof(linux_flock));
+		if (error)
+			return (error);
+		linux_to_bsd_flock(&linux_flock, bsd_flock);
+		fcntl_args.fd = args->fd;
+		fcntl_args.cmd = F_SETLK;
+		fcntl_args.arg = (long)bsd_flock;
+		return (fcntl(p, &fcntl_args));
+
+	case LINUX_F_SETLKW:
+		error = copyin((caddr_t)args->arg, &linux_flock,
+		    sizeof(linux_flock));
+		if (error)
+			return (error);
+		linux_to_bsd_flock(&linux_flock, bsd_flock);
+		fcntl_args.fd = args->fd;
+		fcntl_args.cmd = F_SETLKW;
+		fcntl_args.arg = (long)bsd_flock;
+		return (fcntl(p, &fcntl_args));
+	}
+
+	args64.fd = args->fd;
+	args64.cmd = args->cmd;
+	args64.arg = args->arg;
+	return (fcntl_common(p, &args64));
+}
+
+#if defined(__i386__)
+int
+linux_fcntl64(struct proc *p, struct linux_fcntl64_args *args)
+{
+	struct fcntl_args fcntl_args;
+	struct l_flock64 linux_flock;
+	struct flock *bsd_flock;
+	int error;
+	caddr_t sg;
+
+	sg = stackgap_init();
+	bsd_flock = (struct flock *)stackgap_alloc(&sg, sizeof(bsd_flock));
+
+#ifdef DEBUG
+	if (ldebug(fcntl64))
+		printf(ARGS(fcntl64, "%d, %08x, *"), args->fd, args->cmd);
+#endif
+
+	switch (args->cmd) {
+	case LINUX_F_GETLK:
+		error = copyin((caddr_t)args->arg, &linux_flock,
+		    sizeof(linux_flock));
+		if (error)
+			return (error);
+		linux_to_bsd_flock64(&linux_flock, bsd_flock);
+		fcntl_args.fd = args->fd;
+		fcntl_args.cmd = F_GETLK;
+		fcntl_args.arg = (long)bsd_flock;
+		error = fcntl(p, &fcntl_args);
+		if (error)
+			return (error);
+		bsd_to_linux_flock64(bsd_flock, &linux_flock);
+		return (copyout(&linux_flock, (caddr_t)args->arg,
+		    sizeof(linux_flock)));
+
+	case LINUX_F_SETLK:
+		error = copyin((caddr_t)args->arg, &linux_flock,
+		    sizeof(linux_flock));
+		if (error)
+			return (error);
+		linux_to_bsd_flock64(&linux_flock, bsd_flock);
+		fcntl_args.fd = args->fd;
+		fcntl_args.cmd = F_SETLK;
+		fcntl_args.arg = (long)bsd_flock;
+		return (fcntl(p, &fcntl_args));
+
+	case LINUX_F_SETLKW:
+		error = copyin((caddr_t)args->arg, &linux_flock,
+		    sizeof(linux_flock));
+		if (error)
+			return (error);
+		linux_to_bsd_flock64(&linux_flock, bsd_flock);
+		fcntl_args.fd = args->fd;
+		fcntl_args.cmd = F_SETLKW;
+		fcntl_args.arg = (long)bsd_flock;
+		return (fcntl(p, &fcntl_args));
+	}
+
+	return (fcntl_common(p, args));
+}
+#endif /* __i386__ */