6407791 bringover into ZFS results in s. files newer than extracted source

6409927 failed DKIOCFLUSHWRITECACHE ioctls should not generate ereports
6410371 need to reserve more pool names

5 files changed, 88 insertions, 6 deletions

diff --git a/usr/src/lib/libzfs/common/libzfs_pool.c b/usr/src/lib/libzfs/common/libzfs_pool.c
index 2889725027..1fe6fa2d27 100644
--- a/usr/src/lib/libzfs/common/libzfs_pool.c
+++ b/usr/src/lib/libzfs/common/libzfs_pool.c
@@ -46,12 +46,30 @@
  * 'buf'.
  */
 static int
-zpool_name_valid(const char *pool, char *buf, size_t buflen)
+zpool_name_valid(const char *pool, boolean_t isopen, char *buf, size_t buflen)
 {
 	namecheck_err_t why;
 	char what;
+	int ret;
+
+	ret = pool_namecheck(pool, &why, &what);
+
+	/*
+	 * The rules for reserved pool names were extended at a later point.
+	 * But we need to support users with existing pools that may now be
+	 * invalid.  So we only check for this expanded set of names during a
+	 * create (or import), and only in userland.
+	 */
+	if (ret == 0 && !isopen &&
+	    (strncmp(pool, "mirror", 6) == 0 ||
+	    strncmp(pool, "raidz", 5) == 0 ||
+	    strncmp(pool, "spare", 5) == 0)) {
+		ret = -1;
+		why = NAME_ERR_RESERVED;
+	}
+
 
-	if (pool_namecheck(pool, &why, &what) != 0) {
+	if (ret != 0) {
 		if (buf != NULL) {
 			switch (why) {
 			case NAME_ERR_TOOLONG:
@@ -142,7 +160,7 @@ zpool_open_canfail(const char *pool)
 	/*
 	 * Make sure the pool name is valid.
 	 */
-	if (!zpool_name_valid(pool, NULL, 0)) {
+	if (!zpool_name_valid(pool, B_TRUE, NULL, 0)) {
 		zfs_error(dgettext(TEXT_DOMAIN, "cannot open '%s': invalid "
 		    "pool name"), pool);
 		return (NULL);
@@ -338,7 +356,7 @@ zpool_create(const char *pool, nvlist_t *nvroot, const char *altroot)
 	int err;
 	char reason[64];
 
-	if (!zpool_name_valid(pool, reason, sizeof (reason))) {
+	if (!zpool_name_valid(pool, B_FALSE, reason, sizeof (reason))) {
 		zfs_error(dgettext(TEXT_DOMAIN, "cannot create '%s': %s"),
 		    pool, reason);
 		return (-1);
@@ -688,7 +706,7 @@ zpool_import(nvlist_t *config, const char *newname, const char *altroot)
 	    &origname) == 0);
 
 	if (newname != NULL) {
-		if (!zpool_name_valid(newname, NULL, 0)) {
+		if (!zpool_name_valid(newname, B_FALSE, NULL, 0)) {
 			zfs_error(dgettext(TEXT_DOMAIN, "cannot import '%s': "
 			    "invalid pool name"), newname);
 			return (-1);
diff --git a/usr/src/uts/common/fs/zfs/sys/vdev_impl.h b/usr/src/uts/common/fs/zfs/sys/vdev_impl.h
index 8c094fba12..9ea2ca5373 100644
--- a/usr/src/uts/common/fs/zfs/sys/vdev_impl.h
+++ b/usr/src/uts/common/fs/zfs/sys/vdev_impl.h
@@ -173,6 +173,7 @@ struct vdev {
 	vdev_cache_t	vdev_cache;	/* physical block cache		*/
 	uint64_t	vdev_not_present; /* not present during import	*/
 	hrtime_t	vdev_last_try;	/* last reopen time		*/
+	boolean_t	vdev_nowritecache; /* true if flushwritecache failed */
 
 	/*
 	 * For DTrace to work in userland (libzpool) context, these fields must
diff --git a/usr/src/uts/common/fs/zfs/vdev_disk.c b/usr/src/uts/common/fs/zfs/vdev_disk.c
index 0b4c2c9696..2c50ba3837 100644
--- a/usr/src/uts/common/fs/zfs/vdev_disk.c
+++ b/usr/src/uts/common/fs/zfs/vdev_disk.c
@@ -174,6 +174,12 @@ vdev_disk_open(vdev_t *vd, uint64_t *psize, uint64_t *ashift)
 
 	*ashift = highbit(MAX(dkm.dki_lbsize, SPA_MINBLOCKSIZE)) - 1;
 
+	/*
+	 * Clear the nowritecache bit, so that on a vdev_reopen() we will
+	 * try again.
+	 */
+	vd->vdev_nowritecache = B_FALSE;
+
 	return (0);
 }
 
@@ -249,6 +255,11 @@ vdev_disk_io_start(zio_t *zio)
 
 		case DKIOCFLUSHWRITECACHE:
 
+			if (vd->vdev_nowritecache) {
+				zio->io_error = ENOTSUP;
+				break;
+			}
+
 			zio->io_dk_callback.dkc_callback = vdev_disk_ioctl_done;
 			zio->io_dk_callback.dkc_cookie = zio;
 
@@ -263,8 +274,17 @@ vdev_disk_io_start(zio_t *zio)
 				 * upon completion.
 				 */
 				return;
+			} else if (error == ENOTSUP) {
+				/*
+				 * If we get ENOTSUP, we know that no future
+				 * attempts will ever succeed.  In this case we
+				 * set a persistent bit so that we don't bother
+				 * with the ioctl in the future.
+				 */
+				vd->vdev_nowritecache = B_TRUE;
 			}
 			zio->io_error = error;
+
 			break;
 
 		default:
diff --git a/usr/src/uts/common/fs/zfs/zfs_fm.c b/usr/src/uts/common/fs/zfs/zfs_fm.c
index 007445c713..f0fdb8f6ea 100644
--- a/usr/src/uts/common/fs/zfs/zfs_fm.c
+++ b/usr/src/uts/common/fs/zfs/zfs_fm.c
@@ -122,6 +122,14 @@ zfs_ereport_post(const char *subclass, spa_t *spa, vdev_t *vd, zio_t *zio,
 	if (zio && zio_should_retry(zio))
 		return;
 
+	/*
+	 * If this is not a read or write zio, ignore the error.  This can occur
+	 * if the DKIOCFLUSHWRITECACHE ioctl fails.
+	 */
+	if (zio && zio->io_type != ZIO_TYPE_READ &&
+	    zio->io_type != ZIO_TYPE_WRITE)
+		return;
+
 	if ((ereport = fm_nvlist_create(NULL)) == NULL)
 		return;
 
diff --git a/usr/src/uts/common/fs/zfs/zfs_vnops.c b/usr/src/uts/common/fs/zfs/zfs_vnops.c
index 76e2069e63..67e886cac2 100644
--- a/usr/src/uts/common/fs/zfs/zfs_vnops.c
+++ b/usr/src/uts/common/fs/zfs/zfs_vnops.c
@@ -1815,13 +1815,22 @@ update:
 	return (error);
 }
 
-/* ARGSUSED */
 static int
 zfs_fsync(vnode_t *vp, int syncflag, cred_t *cr)
 {
 	znode_t	*zp = VTOZ(vp);
 	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
 
+	/*
+	 * Regardless of whether this is required for standards conformance,
+	 * this is the logical behavior when fsync() is called on a file with
+	 * dirty pages.  We use B_ASYNC since the ZIL transactions are already
+	 * going to be pushed out as part of the zil_commit().
+	 */
+	if (vn_has_cached_data(vp) && !(syncflag & FNODSYNC) &&
+	    (vp->v_type == VREG) && !(IS_SWAPVP(vp)))
+		(void) VOP_PUTPAGE(vp, (offset_t)0, (size_t)0, B_ASYNC, cr);
+
 	ZFS_ENTER(zfsvfs);
 	zil_commit(zfsvfs->z_log, zp->z_last_itx, FSYNC);
 	ZFS_EXIT(zfsvfs);
@@ -3358,6 +3367,27 @@ zfs_addmap(vnode_t *vp, offset_t off, struct as *as, caddr_t addr,
 	return (0);
 }
 
+/*
+ * The reason we push dirty pages as part of zfs_delmap() is so that we get a
+ * more accurate mtime for the associated file.  Since we don't have a way of
+ * detecting when the data was actually modified, we have to resort to
+ * heuristics.  If an explicit msync() is done, then we mark the mtime when the
+ * last page is pushed.  The problem occurs when the msync() call is omitted,
+ * which by far the most common case:
+ *
+ * 	open()
+ * 	mmap()
+ * 	<modify memory>
+ * 	munmap()
+ * 	close()
+ * 	<time lapse>
+ * 	putpage() via fsflush
+ *
+ * If we wait until fsflush to come along, we can have a modification time that
+ * is some arbitrary point in the future.  In order to prevent this in the
+ * common case, we flush pages whenever a (MAP_SHARED, PROT_WRITE) mapping is
+ * torn down.
+ */
 /* ARGSUSED */
 static int
 zfs_delmap(vnode_t *vp, offset_t off, struct as *as, caddr_t addr,
@@ -3367,6 +3397,11 @@ zfs_delmap(vnode_t *vp, offset_t off, struct as *as, caddr_t addr,
 
 	ASSERT3U(VTOZ(vp)->z_mapcnt, >=, pages);
 	atomic_add_64(&VTOZ(vp)->z_mapcnt, -pages);
+
+	if ((flags & MAP_SHARED) && (prot & PROT_WRITE) &&
+	    vn_has_cached_data(vp))
+		(void) VOP_PUTPAGE(vp, off, len, B_ASYNC, cr);
+
 	return (0);
 }