Merge tag 'nfs-for-4.8-1' of git://git.linux-nfs.org/projects/trondmy…

…/linux-nfs

Pull NFS client updates from Trond Myklebust:
 "Highlights include:

  Stable bugfixes:
   - nfs: don't create zero-length requests

   - several LAYOUTGET bugfixes

  Features:
   - several performance related features

   - more aggressive caching when we can rely on close-to-open
     cache consistency

   - remove serialisation of O_DIRECT reads and writes

   - optimise several code paths to not flush to disk unnecessarily.

     However allow for the idiosyncracies of pNFS for those layout
     types that need to issue a LAYOUTCOMMIT before the metadata can
     be updated on the server.

   - SUNRPC updates to the client data receive path

   - pNFS/SCSI support RH/Fedora dm-mpath device nodes

   - pNFS files/flexfiles can now use unprivileged ports when
     the generic NFS mount options allow it.

  Bugfixes:
   - Don't use RDMA direct data placement together with data
     integrity or privacy security flavours

   - Remove the RDMA ALLPHYSICAL memory registration mode as
     it has potential security holes.

   - Several layout recall fixes to improve NFSv4.1 protocol
     compliance.

   - Fix an Oops in the pNFS files and flexfiles connection
     setup to the DS

   - Allow retry of operations that used a returned delegation
      stateid

   - Don't mark the inode as revalidated if a LAYOUTCOMMIT is
     outstanding

   - Fix writeback races in nfs4_copy_range() and
     nfs42_proc_deallocate()"

* tag 'nfs-for-4.8-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (104 commits)
  pNFS: Actively set attributes as invalid if LAYOUTCOMMIT is outstanding
  NFSv4: Clean up lookup of SECINFO_NO_NAME
  NFSv4.2: Fix warning "variable ‘stateids’ set but not used"
  NFSv4: Fix warning "no previous prototype for ‘nfs4_listxattr’"
  SUNRPC: Fix a compiler warning in fs/nfs/clnt.c
  pNFS: Remove redundant smp_mb() from pnfs_init_lseg()
  pNFS: Cleanup - do layout segment initialisation in one place
  pNFS: Remove redundant stateid invalidation
  pNFS: Remove redundant pnfs_mark_layout_returned_if_empty()
  pNFS: Clear the layout metadata if the server changed the layout stateid
  pNFS: Cleanup - don't open code pnfs_mark_layout_stateid_invalid()
  NFS: pnfs_mark_matching_lsegs_return() should match the layout sequence id
  pNFS: Do not set plh_return_seq for non-callback related layoutreturns
  pNFS: Ensure layoutreturn acts as a completion for layout callbacks
  pNFS: Fix CB_LAYOUTRECALL stateid verification
  pNFS: Always update the layout barrier seqid on LAYOUTGET
  pNFS: Always update the layout stateid if NFS_LAYOUT_INVALID_STID is set
  pNFS: Clear the layout return tracking on layout reinitialisation
  pNFS: LAYOUTRETURN should only update the stateid if the layout is valid
  nfs: don't create zero-length requests
  ...

fs/nfs/Makefile

@@ -6,7 +6,7 @@ obj-$(CONFIG_NFS_FS) += nfs.o
 
 CFLAGS_nfstrace.o += -I$(src)
 nfs-y 			:= client.o dir.o file.o getroot.o inode.o super.o \
-			   direct.o pagelist.o read.o symlink.o unlink.o \
+			   io.o direct.o pagelist.o read.o symlink.o unlink.o \
 			   write.o namespace.o mount_clnt.o nfstrace.o
 nfs-$(CONFIG_ROOT_NFS)	+= nfsroot.o
 nfs-$(CONFIG_SYSCTL)	+= sysctl.o

fs/nfs/blocklayout/dev.c

@@ -65,8 +65,8 @@ nfs4_block_decode_volume(struct xdr_stream *xdr, struct pnfs_block_volume *b)
 		if (!p)
 			return -EIO;
 		b->simple.nr_sigs = be32_to_cpup(p++);
-		if (!b->simple.nr_sigs) {
-			dprintk("no signature\n");
+		if (!b->simple.nr_sigs || b->simple.nr_sigs > PNFS_BLOCK_MAX_UUIDS) {
+			dprintk("Bad signature count: %d\n", b->simple.nr_sigs);
 			return -EIO;
 		}
 
@@ -89,7 +89,8 @@ nfs4_block_decode_volume(struct xdr_stream *xdr, struct pnfs_block_volume *b)
 			memcpy(&b->simple.sigs[i].sig, p,
 				b->simple.sigs[i].sig_len);
 
-			b->simple.len += 8 + 4 + b->simple.sigs[i].sig_len;
+			b->simple.len += 8 + 4 + \
+				(XDR_QUADLEN(b->simple.sigs[i].sig_len) << 2);
 		}
 		break;
 	case PNFS_BLOCK_VOLUME_SLICE:
@@ -104,7 +105,12 @@ nfs4_block_decode_volume(struct xdr_stream *xdr, struct pnfs_block_volume *b)
 		p = xdr_inline_decode(xdr, 4);
 		if (!p)
 			return -EIO;
+
 		b->concat.volumes_count = be32_to_cpup(p++);
+		if (b->concat.volumes_count > PNFS_BLOCK_MAX_DEVICES) {
+			dprintk("Too many volumes: %d\n", b->concat.volumes_count);
+			return -EIO;
+		}
 
 		p = xdr_inline_decode(xdr, b->concat.volumes_count * 4);
 		if (!p)
@@ -116,8 +122,13 @@ nfs4_block_decode_volume(struct xdr_stream *xdr, struct pnfs_block_volume *b)
 		p = xdr_inline_decode(xdr, 8 + 4);
 		if (!p)
 			return -EIO;
+
 		p = xdr_decode_hyper(p, &b->stripe.chunk_size);
 		b->stripe.volumes_count = be32_to_cpup(p++);
+		if (b->stripe.volumes_count > PNFS_BLOCK_MAX_DEVICES) {
+			dprintk("Too many volumes: %d\n", b->stripe.volumes_count);
+			return -EIO;
+		}
 
 		p = xdr_inline_decode(xdr, b->stripe.volumes_count * 4);
 		if (!p)
@@ -224,18 +235,20 @@ bl_parse_simple(struct nfs_server *server, struct pnfs_block_dev *d,
 		struct pnfs_block_volume *volumes, int idx, gfp_t gfp_mask)
 {
 	struct pnfs_block_volume *v = &volumes[idx];
+	struct block_device *bdev;
 	dev_t dev;
 
 	dev = bl_resolve_deviceid(server, v, gfp_mask);
 	if (!dev)
 		return -EIO;
 
-	d->bdev = blkdev_get_by_dev(dev, FMODE_READ | FMODE_WRITE, NULL);
-	if (IS_ERR(d->bdev)) {
+	bdev = blkdev_get_by_dev(dev, FMODE_READ | FMODE_WRITE, NULL);
+	if (IS_ERR(bdev)) {
 		printk(KERN_WARNING "pNFS: failed to open device %d:%d (%ld)\n",
-			MAJOR(dev), MINOR(dev), PTR_ERR(d->bdev));
-		return PTR_ERR(d->bdev);
+			MAJOR(dev), MINOR(dev), PTR_ERR(bdev));
+		return PTR_ERR(bdev);
 	}
+	d->bdev = bdev;
 
 
 	d->len = i_size_read(d->bdev->bd_inode);
@@ -287,44 +300,71 @@ bl_validate_designator(struct pnfs_block_volume *v)
 	}
 }
 
+/*
+ * Try to open the udev path for the WWN.  At least on Debian the udev
+ * by-id path will always point to the dm-multipath device if one exists.
+ */
+static struct block_device *
+bl_open_udev_path(struct pnfs_block_volume *v)
+{
+	struct block_device *bdev;
+	const char *devname;
+
+	devname = kasprintf(GFP_KERNEL, "/dev/disk/by-id/wwn-0x%*phN",
+				v->scsi.designator_len, v->scsi.designator);
+	if (!devname)
+		return ERR_PTR(-ENOMEM);
+
+	bdev = blkdev_get_by_path(devname, FMODE_READ | FMODE_WRITE, NULL);
+	if (IS_ERR(bdev)) {
+		pr_warn("pNFS: failed to open device %s (%ld)\n",
+			devname, PTR_ERR(bdev));
+	}
+
+	kfree(devname);
+	return bdev;
+}
+
+/*
+ * Try to open the RH/Fedora specific dm-mpath udev path for this WWN, as the
+ * wwn- links will only point to the first discovered SCSI device there.
+ */
+static struct block_device *
+bl_open_dm_mpath_udev_path(struct pnfs_block_volume *v)
+{
+	struct block_device *bdev;
+	const char *devname;
+
+	devname = kasprintf(GFP_KERNEL,
+			"/dev/disk/by-id/dm-uuid-mpath-%d%*phN",
+			v->scsi.designator_type,
+			v->scsi.designator_len, v->scsi.designator);
+	if (!devname)
+		return ERR_PTR(-ENOMEM);
+
+	bdev = blkdev_get_by_path(devname, FMODE_READ | FMODE_WRITE, NULL);
+	kfree(devname);
+	return bdev;
+}
+
 static int
 bl_parse_scsi(struct nfs_server *server, struct pnfs_block_dev *d,
 		struct pnfs_block_volume *volumes, int idx, gfp_t gfp_mask)
 {
 	struct pnfs_block_volume *v = &volumes[idx];
+	struct block_device *bdev;
 	const struct pr_ops *ops;
-	const char *devname;
 	int error;
 
 	if (!bl_validate_designator(v))
 		return -EINVAL;
 
-	switch (v->scsi.designator_len) {
-	case 8:
-		devname = kasprintf(GFP_KERNEL, "/dev/disk/by-id/wwn-0x%8phN",
-				v->scsi.designator);
-		break;
-	case 12:
-		devname = kasprintf(GFP_KERNEL, "/dev/disk/by-id/wwn-0x%12phN",
-				v->scsi.designator);
-		break;
-	case 16:
-		devname = kasprintf(GFP_KERNEL, "/dev/disk/by-id/wwn-0x%16phN",
-				v->scsi.designator);
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	d->bdev = blkdev_get_by_path(devname, FMODE_READ, NULL);
-	if (IS_ERR(d->bdev)) {
-		pr_warn("pNFS: failed to open device %s (%ld)\n",
-			devname, PTR_ERR(d->bdev));
-		kfree(devname);
-		return PTR_ERR(d->bdev);
-	}
-
-	kfree(devname);
+	bdev = bl_open_dm_mpath_udev_path(v);
+	if (IS_ERR(bdev))
+		bdev = bl_open_udev_path(v);
+	if (IS_ERR(bdev))
+		return PTR_ERR(bdev);
+	d->bdev = bdev;
 
 	d->len = i_size_read(d->bdev->bd_inode);
 	d->map = bl_map_simple;
@@ -352,7 +392,7 @@ bl_parse_scsi(struct nfs_server *server, struct pnfs_block_dev *d,
 	return 0;
 
 out_blkdev_put:
-	blkdev_put(d->bdev, FMODE_READ);
+	blkdev_put(d->bdev, FMODE_READ | FMODE_WRITE);
 	return error;
 }
 

fs/nfs/blocklayout/extent_tree.c

@@ -121,6 +121,16 @@ ext_try_to_merge_right(struct rb_root *root, struct pnfs_block_extent *be)
 	return be;
 }
 
+static void __ext_put_deviceids(struct list_head *head)
+{
+	struct pnfs_block_extent *be, *tmp;
+
+	list_for_each_entry_safe(be, tmp, head, be_list) {
+		nfs4_put_deviceid_node(be->be_device);
+		kfree(be);
+	}
+}
+
 static void
 __ext_tree_insert(struct rb_root *root,
 		struct pnfs_block_extent *new, bool merge_ok)
@@ -163,7 +173,8 @@ __ext_tree_insert(struct rb_root *root,
 }
 
 static int
-__ext_tree_remove(struct rb_root *root, sector_t start, sector_t end)
+__ext_tree_remove(struct rb_root *root,
+		sector_t start, sector_t end, struct list_head *tmp)
 {
 	struct pnfs_block_extent *be;
 	sector_t len1 = 0, len2 = 0;
@@ -223,8 +234,7 @@ __ext_tree_remove(struct rb_root *root, sector_t start, sector_t end)
 			struct pnfs_block_extent *next = ext_tree_next(be);
 
 			rb_erase(&be->be_node, root);
-			nfs4_put_deviceid_node(be->be_device);
-			kfree(be);
+			list_add_tail(&be->be_list, tmp);
 			be = next;
 		}
 
@@ -350,16 +360,18 @@ int ext_tree_remove(struct pnfs_block_layout *bl, bool rw,
 		sector_t start, sector_t end)
 {
 	int err, err2;
+	LIST_HEAD(tmp);
 
 	spin_lock(&bl->bl_ext_lock);
-	err = __ext_tree_remove(&bl->bl_ext_ro, start, end);
+	err = __ext_tree_remove(&bl->bl_ext_ro, start, end, &tmp);
 	if (rw) {
-		err2 = __ext_tree_remove(&bl->bl_ext_rw, start, end);
+		err2 = __ext_tree_remove(&bl->bl_ext_rw, start, end, &tmp);
 		if (!err)
 			err = err2;
 	}
 	spin_unlock(&bl->bl_ext_lock);
 
+	__ext_put_deviceids(&tmp);
 	return err;
 }
 
@@ -396,12 +408,13 @@ ext_tree_mark_written(struct pnfs_block_layout *bl, sector_t start,
 	sector_t end = start + len;
 	struct pnfs_block_extent *be;
 	int err = 0;
+	LIST_HEAD(tmp);
 
 	spin_lock(&bl->bl_ext_lock);
 	/*
 	 * First remove all COW extents or holes from written to range.
 	 */
-	err = __ext_tree_remove(&bl->bl_ext_ro, start, end);
+	err = __ext_tree_remove(&bl->bl_ext_ro, start, end, &tmp);
 	if (err)
 		goto out;
 
@@ -459,6 +472,8 @@ ext_tree_mark_written(struct pnfs_block_layout *bl, sector_t start,
 	}
 out:
 	spin_unlock(&bl->bl_ext_lock);
+
+	__ext_put_deviceids(&tmp);
 	return err;
 }