Factor out non-portable vnode_t usage

On FreeBSD file offset state is maintained in struct file. A given
vnode can be referenced from many different struct file *. As a
consequence, FreeBSD's SPL doesn't support vn_rdwr with the FAPPEND
flag.

This change replaces the non-portable vnode_t with the portable
file_t in the common code.

Signed-off-by: Matt Macy <mmacy@FreeBSD.org>

include/os/linux/spl/sys/kobj.h

@@ -35,8 +35,6 @@ typedef struct _buf buf_t;
 
 extern struct _buf *kobj_open_file(const char *name);
 extern void kobj_close_file(struct _buf *file);
-extern int kobj_read_file(struct _buf *file, char *buf, unsigned size,
-    unsigned off);
 extern int kobj_get_filesize(struct _buf *file, uint64_t *size);
 
 #endif /* SPL_KOBJ_H */

include/os/linux/spl/sys/vnode.h

@@ -169,9 +169,6 @@ extern int vn_open(const char *path, uio_seg_t seg, int flags, int mode,
     vnode_t **vpp, int x1, void *x2);
 extern int vn_openat(const char *path, uio_seg_t seg, int flags, int mode,
     vnode_t **vpp, int x1, void *x2, vnode_t *vp, int fd);
-extern int vn_rdwr(uio_rw_t uio, vnode_t *vp, void *addr, ssize_t len,
-    offset_t off, uio_seg_t seg, int x1, rlim64_t x2,
-    void *x3, ssize_t *residp);
 extern int vn_close(vnode_t *vp, int flags, int x1, int x2, void *x3, void *x4);
 extern int vn_seek(vnode_t *vp, offset_t o, offset_t *op, void *ct);
 

include/sys/Makefile.am

@@ -104,6 +104,7 @@ COMMON_H = \
 	$(top_srcdir)/include/sys/zfs_context.h \
 	$(top_srcdir)/include/sys/zfs_debug.h \
 	$(top_srcdir)/include/sys/zfs_delay.h \
+	$(top_srcdir)/include/sys/zfs_file.h \
 	$(top_srcdir)/include/sys/zfs_fuid.h \
 	$(top_srcdir)/include/sys/zfs_project.h \
 	$(top_srcdir)/include/sys/zfs_ratelimit.h \

include/sys/dmu.h

@@ -1071,7 +1071,7 @@ void dmu_traverse_objset(objset_t *os, uint64_t txg_start,
     dmu_traverse_cb_t cb, void *arg);
 
 int dmu_diff(const char *tosnap_name, const char *fromsnap_name,
-    struct vnode *vp, offset_t *offp);
+    file_t *fp, offset_t *offp);
 
 /* CRC64 table */
 #define	ZFS_CRC64_POLY	0xC96C5795D7870F42ULL	/* ECMA-182, reflected form */

include/sys/dmu_recv.h

@@ -62,7 +62,7 @@ typedef struct dmu_recv_cookie {
 	nvlist_t *drc_begin_nvl;
 
 	objset_t *drc_os;
-	vnode_t *drc_vp; /* The vnode to read the stream from */
+	file_t *drc_fp; /* The file to read the stream from */
 	uint64_t drc_voff; /* The current offset in the stream */
 	uint64_t drc_bytes_read;
 	/*
@@ -82,10 +82,11 @@ typedef struct dmu_recv_cookie {
 int dmu_recv_begin(char *tofs, char *tosnap, dmu_replay_record_t *drr_begin,
     boolean_t force, boolean_t resumable, nvlist_t *localprops,
     nvlist_t *hidden_args, char *origin, dmu_recv_cookie_t *drc,
-    vnode_t *vp, offset_t *voffp);
+    file_t *fp, offset_t *voffp);
 int dmu_recv_stream(dmu_recv_cookie_t *drc, int cleanup_fd,
     uint64_t *action_handlep, offset_t *voffp);
 int dmu_recv_end(dmu_recv_cookie_t *drc, void *owner);
 boolean_t dmu_objset_is_receiving(objset_t *os);
 
+
 #endif /* _DMU_RECV_H */

include/sys/zfs_context.h

@@ -524,6 +524,12 @@ typedef struct vnode {
 	int		v_dump_fd;
 } vnode_t;
 
+typedef struct file {
+	int f_fd;
+	loff_t f_pos;
+	loff_t f_offset;
+} file_t;
+
 extern char *vn_dumpdir;
 #define	AV_SCANSTAMP_SZ	32		/* length of anti-virus scanstamp */
 
@@ -617,8 +623,6 @@ extern int vn_open(char *path, int x1, int oflags, int mode, vnode_t **vpp,
     int x2, int x3);
 extern int vn_openat(char *path, int x1, int oflags, int mode, vnode_t **vpp,
     int x2, int x3, vnode_t *vp, int fd);
-extern int vn_rdwr(int uio, vnode_t *vp, void *addr, ssize_t len,
-    offset_t offset, int x1, int x2, rlim64_t x3, void *x4, ssize_t *residp);
 extern void vn_close(vnode_t *vp);
 
 #define	vn_remove(path, x1, x2)		remove(path)
@@ -760,8 +764,6 @@ typedef struct ace_object {
 #define	ACE_SYSTEM_ALARM_OBJECT_ACE_TYPE	0x08
 
 extern struct _buf *kobj_open_file(char *name);
-extern int kobj_read_file(struct _buf *file, char *buf, unsigned size,
-    unsigned off);
 extern void kobj_close_file(struct _buf *file);
 extern int kobj_get_filesize(struct _buf *file, uint64_t *size);
 extern int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr);

include/sys/zfs_file.h

@@ -0,0 +1,29 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License (the "License").
+ * You may not use this file except in compliance with the License.
+ *
+ * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+ * or http://www.opensolaris.org/os/licensing.
+ * See the License for the specific language governing permissions
+ * and limitations under the License.
+ *
+ * When distributing Covered Code, include this CDDL HEADER in each
+ * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+ * If applicable, add the following below this CDDL HEADER, with the
+ * fields enclosed by brackets "[]" replaced with your own identifying
+ * information: Portions Copyright [yyyy] [name of copyright owner]
+ *
+ * CDDL HEADER END
+ */
+
+#ifndef	_SYS_ZFS_FILE_H
+#define	_SYS_ZFS_FILE_H
+
+int zfs_file_write(file_t *, const void *, size_t, loff_t *, ssize_t *);
+int zfs_file_read(file_t *, void *, size_t, loff_t *, ssize_t *);
+int zfs_file_seek(file_t *, loff_t *, int);
+
+#endif /* _SYS_ZFS_FILE_H */

lib/libzpool/Makefile.am

@@ -144,6 +144,7 @@ KERNEL_C = \
 	zfeature.c \
 	zfs_byteswap.c \
 	zfs_debug.c \
+	zfs_file_os.c \
 	zfs_fm.c \
 	zfs_fuid.c \
 	zfs_sa.c \

lib/libzpool/kernel.c

@@ -584,58 +584,6 @@ vn_openat(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2,
 	return (ret);
 }
 
-/*ARGSUSED*/
-int
-vn_rdwr(int uio, vnode_t *vp, void *addr, ssize_t len, offset_t offset,
-    int x1, int x2, rlim64_t x3, void *x4, ssize_t *residp)
-{
-	ssize_t rc, done = 0, split;
-
-	if (uio == UIO_READ) {
-		rc = pread64(vp->v_fd, addr, len, offset);
-		if (vp->v_dump_fd != -1 && rc != -1) {
-			int status;
-			status = pwrite64(vp->v_dump_fd, addr, rc, offset);
-			ASSERT(status != -1);
-		}
-	} else {
-		/*
-		 * To simulate partial disk writes, we split writes into two
-		 * system calls so that the process can be killed in between.
-		 */
-		int sectors = len >> SPA_MINBLOCKSHIFT;
-		split = (sectors > 0 ? rand() % sectors : 0) <<
-		    SPA_MINBLOCKSHIFT;
-		rc = pwrite64(vp->v_fd, addr, split, offset);
-		if (rc != -1) {
-			done = rc;
-			rc = pwrite64(vp->v_fd, (char *)addr + split,
-			    len - split, offset + split);
-		}
-	}
-
-#ifdef __linux__
-	if (rc == -1 && errno == EINVAL) {
-		/*
-		 * Under Linux, this most likely means an alignment issue
-		 * (memory or disk) due to O_DIRECT, so we abort() in order to
-		 * catch the offender.
-		 */
-		abort();
-	}
-#endif
-	if (rc == -1)
-		return (errno);
-
-	done += rc;
-
-	if (residp)
-		*residp = len - done;
-	else if (done != len)
-		return (EIO);
-	return (0);
-}
-
 void
 vn_close(vnode_t *vp)
 {
@@ -879,18 +827,6 @@ kobj_open_file(char *name)
 	return (file);
 }
 
-int
-kobj_read_file(struct _buf *file, char *buf, unsigned size, unsigned off)
-{
-	ssize_t resid = 0;
-
-	if (vn_rdwr(UIO_READ, (vnode_t *)file->_fd, buf, size, (offset_t)off,
-	    UIO_SYSSPACE, 0, 0, 0, &resid) != 0)
-		return (-1);
-
-	return (size - resid);
-}
-
 void
 kobj_close_file(struct _buf *file)
 {

module/os/linux/spl/spl-generic.c

@@ -520,6 +520,27 @@ ddi_copyout(const void *from, void *to, size_t len, int flags)
 }
 EXPORT_SYMBOL(ddi_copyout);
 
+static ssize_t
+spl_kernel_read(struct file *file, void *buf, size_t count, loff_t *pos)
+{
+#if defined(HAVE_KERNEL_READ_PPOS)
+	return (kernel_read(file, buf, count, pos));
+#else
+	mm_segment_t saved_fs;
+	ssize_t ret;
+
+	saved_fs = get_fs();
+	set_fs(KERNEL_DS);
+
+	ret = vfs_read(file, (void __user *)buf, count, pos);
+
+	set_fs(saved_fs);
+
+	return (ret);
+#endif
+}
+
+
 /*
  * Read the unique system identifier from the /etc/hostid file.
  *
@@ -566,6 +587,7 @@ hostid_read(uint32_t *hostid)
 	struct _buf *file;
 	uint32_t value = 0;
 	int error;
+	loff_t off;
 
 	file = kobj_open_file(spl_hostid_path);
 	if (file == (struct _buf *)-1)
@@ -582,11 +604,12 @@ hostid_read(uint32_t *hostid)
 		return (EINVAL);
 	}
 
+	off = 0;
 	/*
 	 * Read directly into the variable like eglibc does.
 	 * Short reads are okay; native behavior is preserved.
 	 */
-	error = kobj_read_file(file, (char *)&value, sizeof (value), 0);
+	error = spl_kernel_read(file->vp->v_file, &value, sizeof (value), &off);
 	if (error < 0) {
 		kobj_close_file(file);
 		return (EIO);

module/os/linux/spl/spl-kobj.c

@@ -56,19 +56,6 @@ kobj_close_file(struct _buf *file)
 } /* kobj_close_file() */
 EXPORT_SYMBOL(kobj_close_file);
 
-int
-kobj_read_file(struct _buf *file, char *buf, unsigned size, unsigned off)
-{
-	ssize_t resid;
-
-	if (vn_rdwr(UIO_READ, file->vp, buf, size, (offset_t)off,
-	    UIO_SYSSPACE, 0, 0, 0, &resid) != 0)
-		return (-1);
-
-	return (size - resid);
-} /* kobj_read_file() */
-EXPORT_SYMBOL(kobj_read_file);
-
 int
 kobj_get_filesize(struct _buf *file, uint64_t *size)
 {

module/os/linux/spl/spl-vnode.c

@@ -73,46 +73,6 @@ spl_filp_fsync(struct file *fp, int sync)
 #endif /* HAVE_2ARGS_VFS_FSYNC */
 }
 
-static ssize_t
-spl_kernel_write(struct file *file, const void *buf, size_t count, loff_t *pos)
-{
-#if defined(HAVE_KERNEL_WRITE_PPOS)
-	return (kernel_write(file, buf, count, pos));
-#else
-	mm_segment_t saved_fs;
-	ssize_t ret;
-
-	saved_fs = get_fs();
-	set_fs(KERNEL_DS);
-
-	ret = vfs_write(file, (__force const char __user *)buf, count, pos);
-
-	set_fs(saved_fs);
-
-	return (ret);
-#endif
-}
-
-static ssize_t
-spl_kernel_read(struct file *file, void *buf, size_t count, loff_t *pos)
-{
-#if defined(HAVE_KERNEL_READ_PPOS)
-	return (kernel_read(file, buf, count, pos));
-#else
-	mm_segment_t saved_fs;
-	ssize_t ret;
-
-	saved_fs = get_fs();
-	set_fs(KERNEL_DS);
-
-	ret = vfs_read(file, (void __user *)buf, count, pos);
-
-	set_fs(saved_fs);
-
-	return (ret);
-#endif
-}
-
 vtype_t
 vn_mode_to_vtype(mode_t mode)
 {
@@ -282,42 +242,6 @@ vn_openat(const char *path, uio_seg_t seg, int flags, int mode,
 } /* vn_openat() */
 EXPORT_SYMBOL(vn_openat);
 
-int
-vn_rdwr(uio_rw_t uio, vnode_t *vp, void *addr, ssize_t len, offset_t off,
-    uio_seg_t seg, int ioflag, rlim64_t x2, void *x3, ssize_t *residp)
-{
-	struct file *fp = vp->v_file;
-	loff_t offset = off;
-	int rc;
-
-	ASSERT(uio == UIO_WRITE || uio == UIO_READ);
-	ASSERT(seg == UIO_SYSSPACE);
-	ASSERT((ioflag & ~FAPPEND) == 0);
-
-	if (ioflag & FAPPEND)
-		offset = fp->f_pos;
-
-	if (uio & UIO_WRITE)
-		rc = spl_kernel_write(fp, addr, len, &offset);
-	else
-		rc = spl_kernel_read(fp, addr, len, &offset);
-
-	fp->f_pos = offset;
-
-	if (rc < 0)
-		return (-rc);
-
-	if (residp) {
-		*residp = len - rc;
-	} else {
-		if (rc != len)
-			return (EIO);
-	}
-
-	return (0);
-} /* vn_rdwr() */
-EXPORT_SYMBOL(vn_rdwr);
-
 int
 vn_close(vnode_t *vp, int flags, int x1, int x2, void *x3, void *x4)
 {

module/os/linux/zfs/Makefile.in

@@ -26,6 +26,7 @@ $(MODULE)-objs += ../os/linux/zfs/zfs_acl.o
 $(MODULE)-objs += ../os/linux/zfs/zfs_ctldir.o
 $(MODULE)-objs += ../os/linux/zfs/zfs_debug.o
 $(MODULE)-objs += ../os/linux/zfs/zfs_dir.o
+$(MODULE)-objs += ../os/linux/zfs/zfs_file_os.o
 $(MODULE)-objs += ../os/linux/zfs/zfs_ioctl_os.o
 $(MODULE)-objs += ../os/linux/zfs/zfs_onexit_os.o
 $(MODULE)-objs += ../os/linux/zfs/zfs_sysfs.o