Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse

Pull fuse updates from Miklos Szeredi:
 "The biggest part of this pull request is a patch series from Maxim
  Patlasov to optimize scatter-gather direct IO.  There's also the
  addition of a "readdirplus" API, poll events and various fixes and
  cleanups.

  There's a one line change outside of fuse to mm/filemap.c which makes
  the argument of iov_iter_single_seg_count() const, required by Maxim's
  patches."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse: (22 commits)
  fuse: allow control of adaptive readdirplus use
  Synchronize fuse header with one used in library
  fuse: send poll events
  fuse: don't WARN when nlink is zero
  fuse: avoid out-of-scope stack access
  fuse: bump version for READDIRPLUS
  FUSE: Adapt readdirplus to application usage patterns
  Do not use RCU for current process credentials
  fuse: cleanup fuse_direct_io()
  fuse: optimize __fuse_direct_io()
  fuse: optimize fuse_get_user_pages()
  fuse: pass iov[] to fuse_get_user_pages()
  mm: minor cleanup of iov_iter_single_seg_count()
  fuse: use req->page_descs[] for argpages cases
  fuse: add per-page descriptor <offset, length> to fuse_req
  fuse: rework fuse_do_ioctl()
  fuse: rework fuse_perform_write()
  fuse: rework fuse_readpages()
  fuse: rework fuse_retrieve()
  fuse: categorize fuse_get_req()
  ...

fs/fuse/cuse.c

@@ -91,19 +91,22 @@ static ssize_t cuse_read(struct file *file, char __user *buf, size_t count,
 			 loff_t *ppos)
 {
 	loff_t pos = 0;
+	struct iovec iov = { .iov_base = buf, .iov_len = count };
 
-	return fuse_direct_io(file, buf, count, &pos, 0);
+	return fuse_direct_io(file, &iov, 1, count, &pos, 0);
 }
 
 static ssize_t cuse_write(struct file *file, const char __user *buf,
 			  size_t count, loff_t *ppos)
 {
 	loff_t pos = 0;
+	struct iovec iov = { .iov_base = (void __user *)buf, .iov_len = count };
+
 	/*
 	 * No locking or generic_write_checks(), the server is
 	 * responsible for locking and sanity checks.
 	 */
-	return fuse_direct_io(file, buf, count, &pos, 1);
+	return fuse_direct_io(file, &iov, 1, count, &pos, 1);
 }
 
 static int cuse_open(struct inode *inode, struct file *file)
@@ -419,7 +422,7 @@ static int cuse_send_init(struct cuse_conn *cc)
 
 	BUILD_BUG_ON(CUSE_INIT_INFO_MAX > PAGE_SIZE);
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req(fc, 1);
 	if (IS_ERR(req)) {
 		rc = PTR_ERR(req);
 		goto err;
@@ -449,6 +452,7 @@ static int cuse_send_init(struct cuse_conn *cc)
 	req->out.argvar = 1;
 	req->out.argpages = 1;
 	req->pages[0] = page;
+	req->page_descs[0].length = req->out.args[1].size;
 	req->num_pages = 1;
 	req->end = cuse_process_init_reply;
 	fuse_request_send_background(fc, req);

fs/fuse/dev.c

@@ -34,34 +34,67 @@ static struct fuse_conn *fuse_get_conn(struct file *file)
 	return file->private_data;
 }
 
-static void fuse_request_init(struct fuse_req *req)
+static void fuse_request_init(struct fuse_req *req, struct page **pages,
+			      struct fuse_page_desc *page_descs,
+			      unsigned npages)
 {
 	memset(req, 0, sizeof(*req));
+	memset(pages, 0, sizeof(*pages) * npages);
+	memset(page_descs, 0, sizeof(*page_descs) * npages);
 	INIT_LIST_HEAD(&req->list);
 	INIT_LIST_HEAD(&req->intr_entry);
 	init_waitqueue_head(&req->waitq);
 	atomic_set(&req->count, 1);
+	req->pages = pages;
+	req->page_descs = page_descs;
+	req->max_pages = npages;
 }
 
-struct fuse_req *fuse_request_alloc(void)
+static struct fuse_req *__fuse_request_alloc(unsigned npages, gfp_t flags)
 {
-	struct fuse_req *req = kmem_cache_alloc(fuse_req_cachep, GFP_KERNEL);
+	struct fuse_req *req = kmem_cache_alloc(fuse_req_cachep, flags);
-	if (req)
+	if (req) {
-		fuse_request_init(req);
+		struct page **pages;
+		struct fuse_page_desc *page_descs;
+
+		if (npages <= FUSE_REQ_INLINE_PAGES) {
+			pages = req->inline_pages;
+			page_descs = req->inline_page_descs;
+		} else {
+			pages = kmalloc(sizeof(struct page *) * npages, flags);
+			page_descs = kmalloc(sizeof(struct fuse_page_desc) *
+					     npages, flags);
+		}
+
+		if (!pages || !page_descs) {
+			kfree(pages);
+			kfree(page_descs);
+			kmem_cache_free(fuse_req_cachep, req);
+			return NULL;
+		}
+
+		fuse_request_init(req, pages, page_descs, npages);
+	}
 	return req;
 }
+
+struct fuse_req *fuse_request_alloc(unsigned npages)
+{
+	return __fuse_request_alloc(npages, GFP_KERNEL);
+}
 EXPORT_SYMBOL_GPL(fuse_request_alloc);
 
-struct fuse_req *fuse_request_alloc_nofs(void)
+struct fuse_req *fuse_request_alloc_nofs(unsigned npages)
 {
-	struct fuse_req *req = kmem_cache_alloc(fuse_req_cachep, GFP_NOFS);
+	return __fuse_request_alloc(npages, GFP_NOFS);
-	if (req)
-		fuse_request_init(req);
-	return req;
 }
 
 void fuse_request_free(struct fuse_req *req)
 {
+	if (req->pages != req->inline_pages) {
+		kfree(req->pages);
+		kfree(req->page_descs);
+	}
 	kmem_cache_free(fuse_req_cachep, req);
 }
 
@@ -97,7 +130,7 @@ static void fuse_req_init_context(struct fuse_req *req)
 	req->in.h.pid = current->pid;
 }
 
-struct fuse_req *fuse_get_req(struct fuse_conn *fc)
+struct fuse_req *fuse_get_req(struct fuse_conn *fc, unsigned npages)
 {
 	struct fuse_req *req;
 	sigset_t oldset;
@@ -116,7 +149,7 @@ struct fuse_req *fuse_get_req(struct fuse_conn *fc)
 	if (!fc->connected)
 		goto out;
 
-	req = fuse_request_alloc();
+	req = fuse_request_alloc(npages);
 	err = -ENOMEM;
 	if (!req)
 		goto out;
@@ -165,7 +198,7 @@ static void put_reserved_req(struct fuse_conn *fc, struct fuse_req *req)
 	struct fuse_file *ff = file->private_data;
 
 	spin_lock(&fc->lock);
-	fuse_request_init(req);
+	fuse_request_init(req, req->pages, req->page_descs, req->max_pages);
 	BUG_ON(ff->reserved_req);
 	ff->reserved_req = req;
 	wake_up_all(&fc->reserved_req_waitq);
@@ -186,13 +219,14 @@ static void put_reserved_req(struct fuse_conn *fc, struct fuse_req *req)
  * filesystem should not have it's own file open.  If deadlock is
  * intentional, it can still be broken by "aborting" the filesystem.
  */
-struct fuse_req *fuse_get_req_nofail(struct fuse_conn *fc, struct file *file)
+struct fuse_req *fuse_get_req_nofail_nopages(struct fuse_conn *fc,
+					     struct file *file)
 {
 	struct fuse_req *req;
 
 	atomic_inc(&fc->num_waiting);
 	wait_event(fc->blocked_waitq, !fc->blocked);
-	req = fuse_request_alloc();
+	req = fuse_request_alloc(0);
 	if (!req)
 		req = get_reserved_req(fc, file);
 
@@ -406,9 +440,8 @@ __acquires(fc->lock)
 	}
 }
 
-void fuse_request_send(struct fuse_conn *fc, struct fuse_req *req)
+static void __fuse_request_send(struct fuse_conn *fc, struct fuse_req *req)
 {
-	req->isreply = 1;
 	spin_lock(&fc->lock);
 	if (!fc->connected)
 		req->out.h.error = -ENOTCONN;
@@ -425,6 +458,12 @@ void fuse_request_send(struct fuse_conn *fc, struct fuse_req *req)
 	}
 	spin_unlock(&fc->lock);
 }
+
+void fuse_request_send(struct fuse_conn *fc, struct fuse_req *req)
+{
+	req->isreply = 1;
+	__fuse_request_send(fc, req);
+}
 EXPORT_SYMBOL_GPL(fuse_request_send);
 
 static void fuse_request_send_nowait_locked(struct fuse_conn *fc,
@@ -491,6 +530,27 @@ void fuse_request_send_background_locked(struct fuse_conn *fc,
 	fuse_request_send_nowait_locked(fc, req);
 }
 
+void fuse_force_forget(struct file *file, u64 nodeid)
+{
+	struct inode *inode = file->f_path.dentry->d_inode;
+	struct fuse_conn *fc = get_fuse_conn(inode);
+	struct fuse_req *req;
+	struct fuse_forget_in inarg;
+
+	memset(&inarg, 0, sizeof(inarg));
+	inarg.nlookup = 1;
+	req = fuse_get_req_nofail_nopages(fc, file);
+	req->in.h.opcode = FUSE_FORGET;
+	req->in.h.nodeid = nodeid;
+	req->in.numargs = 1;
+	req->in.args[0].size = sizeof(inarg);
+	req->in.args[0].value = &inarg;
+	req->isreply = 0;
+	__fuse_request_send(fc, req);
+	/* ignore errors */
+	fuse_put_request(fc, req);
+}
+
 /*
  * Lock the request.  Up to the next unlock_request() there mustn't be
  * anything that could cause a page-fault.  If the request was already
@@ -850,11 +910,11 @@ static int fuse_copy_pages(struct fuse_copy_state *cs, unsigned nbytes,
 {
 	unsigned i;
 	struct fuse_req *req = cs->req;
-	unsigned offset = req->page_offset;
-	unsigned count = min(nbytes, (unsigned) PAGE_SIZE - offset);
 
 	for (i = 0; i < req->num_pages && (nbytes || zeroing); i++) {
 		int err;
+		unsigned offset = req->page_descs[i].offset;
+		unsigned count = min(nbytes, req->page_descs[i].length);
 
 		err = fuse_copy_page(cs, &req->pages[i], offset, count,
 				     zeroing);
@@ -862,8 +922,6 @@ static int fuse_copy_pages(struct fuse_copy_state *cs, unsigned nbytes,
 			return err;
 
 		nbytes -= count;
-		count = min(nbytes, (unsigned) PAGE_SIZE);
-		offset = 0;
 	}
 	return 0;
 }
@@ -1536,29 +1594,34 @@ static int fuse_retrieve(struct fuse_conn *fc, struct inode *inode,
 	unsigned int num;
 	unsigned int offset;
 	size_t total_len = 0;
-
+	int num_pages;
-	req = fuse_get_req(fc);
-	if (IS_ERR(req))
-		return PTR_ERR(req);
 
 	offset = outarg->offset & ~PAGE_CACHE_MASK;
-
-	req->in.h.opcode = FUSE_NOTIFY_REPLY;
-	req->in.h.nodeid = outarg->nodeid;
-	req->in.numargs = 2;
-	req->in.argpages = 1;
-	req->page_offset = offset;
-	req->end = fuse_retrieve_end;
-
-	index = outarg->offset >> PAGE_CACHE_SHIFT;
 	file_size = i_size_read(inode);
+
 	num = outarg->size;
 	if (outarg->offset > file_size)
 		num = 0;
 	else if (outarg->offset + num > file_size)
 		num = file_size - outarg->offset;
 
-	while (num && req->num_pages < FUSE_MAX_PAGES_PER_REQ) {
+	num_pages = (num + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
+	num_pages = min(num_pages, FUSE_MAX_PAGES_PER_REQ);
+
+	req = fuse_get_req(fc, num_pages);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
+
+	req->in.h.opcode = FUSE_NOTIFY_REPLY;
+	req->in.h.nodeid = outarg->nodeid;
+	req->in.numargs = 2;
+	req->in.argpages = 1;
+	req->page_descs[0].offset = offset;
+	req->end = fuse_retrieve_end;
+
+	index = outarg->offset >> PAGE_CACHE_SHIFT;
+
+	while (num && req->num_pages < num_pages) {
 		struct page *page;
 		unsigned int this_num;
 
@@ -1568,6 +1631,7 @@ static int fuse_retrieve(struct fuse_conn *fc, struct inode *inode,
 
 		this_num = min_t(unsigned, num, PAGE_CACHE_SIZE - offset);
 		req->pages[req->num_pages] = page;
+		req->page_descs[req->num_pages].length = this_num;
 		req->num_pages++;
 
 		offset = 0;

fs/fuse/dir.c

@@ -14,6 +14,29 @@
 #include <linux/namei.h>
 #include <linux/slab.h>
 
+static bool fuse_use_readdirplus(struct inode *dir, struct file *filp)
+{
+	struct fuse_conn *fc = get_fuse_conn(dir);
+	struct fuse_inode *fi = get_fuse_inode(dir);
+
+	if (!fc->do_readdirplus)
+		return false;
+	if (!fc->readdirplus_auto)
+		return true;
+	if (test_and_clear_bit(FUSE_I_ADVISE_RDPLUS, &fi->state))
+		return true;
+	if (filp->f_pos == 0)
+		return true;
+	return false;
+}
+
+static void fuse_advise_use_readdirplus(struct inode *dir)
+{
+	struct fuse_inode *fi = get_fuse_inode(dir);
+
+	set_bit(FUSE_I_ADVISE_RDPLUS, &fi->state);
+}
+
 #if BITS_PER_LONG >= 64
 static inline void fuse_dentry_settime(struct dentry *entry, u64 time)
 {
@@ -178,7 +201,7 @@ static int fuse_dentry_revalidate(struct dentry *entry, unsigned int flags)
 			return -ECHILD;
 
 		fc = get_fuse_conn(inode);
-		req = fuse_get_req(fc);
+		req = fuse_get_req_nopages(fc);
 		if (IS_ERR(req))
 			return 0;
 
@@ -219,6 +242,7 @@ static int fuse_dentry_revalidate(struct dentry *entry, unsigned int flags)
 				       attr_version);
 		fuse_change_entry_timeout(entry, &outarg);
 	}
+	fuse_advise_use_readdirplus(inode);
 	return 1;
 }
 
@@ -271,7 +295,7 @@ int fuse_lookup_name(struct super_block *sb, u64 nodeid, struct qstr *name,
 	if (name->len > FUSE_NAME_MAX)
 		goto out;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	err = PTR_ERR(req);
 	if (IS_ERR(req))
 		goto out;
@@ -355,6 +379,7 @@ static struct dentry *fuse_lookup(struct inode *dir, struct dentry *entry,
 	else
 		fuse_invalidate_entry_cache(entry);
 
+	fuse_advise_use_readdirplus(dir);
 	return newent;
 
  out_iput:
@@ -391,7 +416,7 @@ static int fuse_create_open(struct inode *dir, struct dentry *entry,
 	if (!forget)
 		goto out_err;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	err = PTR_ERR(req);
 	if (IS_ERR(req))
 		goto out_put_forget_req;
@@ -592,7 +617,7 @@ static int fuse_mknod(struct inode *dir, struct dentry *entry, umode_t mode,
 {
 	struct fuse_mknod_in inarg;
 	struct fuse_conn *fc = get_fuse_conn(dir);
-	struct fuse_req *req = fuse_get_req(fc);
+	struct fuse_req *req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -623,7 +648,7 @@ static int fuse_mkdir(struct inode *dir, struct dentry *entry, umode_t mode)
 {
 	struct fuse_mkdir_in inarg;
 	struct fuse_conn *fc = get_fuse_conn(dir);
-	struct fuse_req *req = fuse_get_req(fc);
+	struct fuse_req *req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -647,7 +672,7 @@ static int fuse_symlink(struct inode *dir, struct dentry *entry,
 {
 	struct fuse_conn *fc = get_fuse_conn(dir);
 	unsigned len = strlen(link) + 1;
-	struct fuse_req *req = fuse_get_req(fc);
+	struct fuse_req *req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -664,7 +689,7 @@ static int fuse_unlink(struct inode *dir, struct dentry *entry)
 {
 	int err;
 	struct fuse_conn *fc = get_fuse_conn(dir);
-	struct fuse_req *req = fuse_get_req(fc);
+	struct fuse_req *req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -682,7 +707,14 @@ static int fuse_unlink(struct inode *dir, struct dentry *entry)
 
 		spin_lock(&fc->lock);
 		fi->attr_version = ++fc->attr_version;
-		drop_nlink(inode);
+		/*
+		 * If i_nlink == 0 then unlink doesn't make sense, yet this can
+		 * happen if userspace filesystem is careless.  It would be
+		 * difficult to enforce correct nlink usage so just ignore this
+		 * condition here
+		 */
+		if (inode->i_nlink > 0)
+			drop_nlink(inode);
 		spin_unlock(&fc->lock);
 		fuse_invalidate_attr(inode);
 		fuse_invalidate_attr(dir);
@@ -696,7 +728,7 @@ static int fuse_rmdir(struct inode *dir, struct dentry *entry)
 {
 	int err;
 	struct fuse_conn *fc = get_fuse_conn(dir);
-	struct fuse_req *req = fuse_get_req(fc);
+	struct fuse_req *req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -723,7 +755,7 @@ static int fuse_rename(struct inode *olddir, struct dentry *oldent,
 	int err;
 	struct fuse_rename_in inarg;
 	struct fuse_conn *fc = get_fuse_conn(olddir);
-	struct fuse_req *req = fuse_get_req(fc);
+	struct fuse_req *req = fuse_get_req_nopages(fc);
 
 	if (IS_ERR(req))
 		return PTR_ERR(req);
@@ -776,7 +808,7 @@ static int fuse_link(struct dentry *entry, struct inode *newdir,
 	struct fuse_link_in inarg;
 	struct inode *inode = entry->d_inode;
 	struct fuse_conn *fc = get_fuse_conn(inode);
-	struct fuse_req *req = fuse_get_req(fc);
+	struct fuse_req *req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -848,7 +880,7 @@ static int fuse_do_getattr(struct inode *inode, struct kstat *stat,
 	struct fuse_req *req;
 	u64 attr_version;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -985,7 +1017,7 @@ int fuse_reverse_inval_entry(struct super_block *sb, u64 parent_nodeid,
 
 /*
  * Calling into a user-controlled filesystem gives the filesystem
- * daemon ptrace-like capabilities over the requester process.  This
+ * daemon ptrace-like capabilities over the current process.  This
  * means, that the filesystem daemon is able to record the exact
  * filesystem operations performed, and can also control the behavior
  * of the requester process in otherwise impossible ways.  For example
@@ -996,27 +1028,23 @@ int fuse_reverse_inval_entry(struct super_block *sb, u64 parent_nodeid,
  * for which the owner of the mount has ptrace privilege.  This
  * excludes processes started by other users, suid or sgid processes.
  */
-int fuse_allow_task(struct fuse_conn *fc, struct task_struct *task)
+int fuse_allow_current_process(struct fuse_conn *fc)
 {
 	const struct cred *cred;
-	int ret;
 
 	if (fc->flags & FUSE_ALLOW_OTHER)
 		return 1;
 
-	rcu_read_lock();
+	cred = current_cred();
-	ret = 0;
-	cred = __task_cred(task);
 	if (uid_eq(cred->euid, fc->user_id) &&
 	    uid_eq(cred->suid, fc->user_id) &&
 	    uid_eq(cred->uid,  fc->user_id) &&
 	    gid_eq(cred->egid, fc->group_id) &&
 	    gid_eq(cred->sgid, fc->group_id) &&
 	    gid_eq(cred->gid,  fc->group_id))
-		ret = 1;
+		return 1;
-	rcu_read_unlock();
 
-	return ret;
+	return 0;
 }
 
 static int fuse_access(struct inode *inode, int mask)
@@ -1029,7 +1057,7 @@ static int fuse_access(struct inode *inode, int mask)
 	if (fc->no_access)
 		return 0;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -1077,7 +1105,7 @@ static int fuse_permission(struct inode *inode, int mask)
 	bool refreshed = false;
 	int err = 0;
 
-	if (!fuse_allow_task(fc, current))
+	if (!fuse_allow_current_process(fc))
 		return -EACCES;
 
 	/*
@@ -1155,19 +1183,157 @@ static int parse_dirfile(char *buf, size_t nbytes, struct file *file,
 	return 0;
 }
 
-static int fuse_readdir(struct file *file, void *dstbuf, filldir_t filldir)
+static int fuse_direntplus_link(struct file *file,
+				struct fuse_direntplus *direntplus,
+				u64 attr_version)
 {
 	int err;
+	struct fuse_entry_out *o = &direntplus->entry_out;
+	struct fuse_dirent *dirent = &direntplus->dirent;
+	struct dentry *parent = file->f_path.dentry;
+	struct qstr name = QSTR_INIT(dirent->name, dirent->namelen);
+	struct dentry *dentry;
+	struct dentry *alias;
+	struct inode *dir = parent->d_inode;
+	struct fuse_conn *fc;
+	struct inode *inode;
+
+	if (!o->nodeid) {
+		/*
+		 * Unlike in the case of fuse_lookup, zero nodeid does not mean
+		 * ENOENT. Instead, it only means the userspace filesystem did
+		 * not want to return attributes/handle for this entry.
+		 *
+		 * So do nothing.
+		 */
+		return 0;
+	}
+
+	if (name.name[0] == '.') {
+		/*
+		 * We could potentially refresh the attributes of the directory
+		 * and its parent?
+		 */
+		if (name.len == 1)
+			return 0;
+		if (name.name[1] == '.' && name.len == 2)
+			return 0;
+	}
+	fc = get_fuse_conn(dir);
+
+	name.hash = full_name_hash(name.name, name.len);
+	dentry = d_lookup(parent, &name);
+	if (dentry && dentry->d_inode) {
+		inode = dentry->d_inode;
+		if (get_node_id(inode) == o->nodeid) {
+			struct fuse_inode *fi;
+			fi = get_fuse_inode(inode);
+			spin_lock(&fc->lock);
+			fi->nlookup++;
+			spin_unlock(&fc->lock);
+
+			/*
+			 * The other branch to 'found' comes via fuse_iget()
+			 * which bumps nlookup inside
+			 */
+			goto found;
+		}
+		err = d_invalidate(dentry);
+		if (err)
+			goto out;
+		dput(dentry);
+		dentry = NULL;
+	}
+
+	dentry = d_alloc(parent, &name);
+	err = -ENOMEM;
+	if (!dentry)
+		goto out;
+
+	inode = fuse_iget(dir->i_sb, o->nodeid, o->generation,
+			  &o->attr, entry_attr_timeout(o), attr_version);
+	if (!inode)
+		goto out;
+
+	alias = d_materialise_unique(dentry, inode);
+	err = PTR_ERR(alias);
+	if (IS_ERR(alias))
+		goto out;
+	if (alias) {
+		dput(dentry);
+		dentry = alias;
+	}
+
+found:
+	fuse_change_attributes(inode, &o->attr, entry_attr_timeout(o),
+			       attr_version);
+
+	fuse_change_entry_timeout(dentry, o);
+
+	err = 0;
+out:
+	if (dentry)
+		dput(dentry);
+	return err;
+}
+
+static int parse_dirplusfile(char *buf, size_t nbytes, struct file *file,
+			     void *dstbuf, filldir_t filldir, u64 attr_version)
+{
+	struct fuse_direntplus *direntplus;
+	struct fuse_dirent *dirent;
+	size_t reclen;
+	int over = 0;
+	int ret;
+
+	while (nbytes >= FUSE_NAME_OFFSET_DIRENTPLUS) {
+		direntplus = (struct fuse_direntplus *) buf;
+		dirent = &direntplus->dirent;
+		reclen = FUSE_DIRENTPLUS_SIZE(direntplus);
+
+		if (!dirent->namelen || dirent->namelen > FUSE_NAME_MAX)
+			return -EIO;
+		if (reclen > nbytes)
+			break;
+
+		if (!over) {
+			/* We fill entries into dstbuf only as much as
+			   it can hold. But we still continue iterating
+			   over remaining entries to link them. If not,
+			   we need to send a FORGET for each of those
+			   which we did not link.
+			*/
+			over = filldir(dstbuf, dirent->name, dirent->namelen,
+				       file->f_pos, dirent->ino,
+				       dirent->type);
+			file->f_pos = dirent->off;
+		}
+
+		buf += reclen;
+		nbytes -= reclen;
+
+		ret = fuse_direntplus_link(file, direntplus, attr_version);
+		if (ret)
+			fuse_force_forget(file, direntplus->entry_out.nodeid);
+	}
+
+	return 0;
+}
+
+static int fuse_readdir(struct file *file, void *dstbuf, filldir_t filldir)
+{
+	int plus, err;
 	size_t nbytes;
 	struct page *page;
 	struct inode *inode = file->f_path.dentry->d_inode;
 	struct fuse_conn *fc = get_fuse_conn(inode);
 	struct fuse_req *req;
+	u64 attr_version = 0;
 
 	if (is_bad_inode(inode))
 		return -EIO;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req(fc, 1);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -1176,17 +1342,34 @@ static int fuse_readdir(struct file *file, void *dstbuf, filldir_t filldir)
 		fuse_put_request(fc, req);
 		return -ENOMEM;
 	}
+
+	plus = fuse_use_readdirplus(inode, file);
 	req->out.argpages = 1;
 	req->num_pages = 1;
 	req->pages[0] = page;
-	fuse_read_fill(req, file, file->f_pos, PAGE_SIZE, FUSE_READDIR);
+	req->page_descs[0].length = PAGE_SIZE;
+	if (plus) {
+		attr_version = fuse_get_attr_version(fc);
+		fuse_read_fill(req, file, file->f_pos, PAGE_SIZE,
+			       FUSE_READDIRPLUS);
+	} else {
+		fuse_read_fill(req, file, file->f_pos, PAGE_SIZE,
+			       FUSE_READDIR);
+	}
 	fuse_request_send(fc, req);
 	nbytes = req->out.args[0].size;
 	err = req->out.h.error;
 	fuse_put_request(fc, req);
-	if (!err)
+	if (!err) {
-		err = parse_dirfile(page_address(page), nbytes, file, dstbuf,
+		if (plus) {
-				    filldir);
+			err = parse_dirplusfile(page_address(page), nbytes,
+						file, dstbuf, filldir,
+						attr_version);
+		} else {
+			err = parse_dirfile(page_address(page), nbytes, file,
+					    dstbuf, filldir);
+		}
+	}
 
 	__free_page(page);
 	fuse_invalidate_attr(inode); /* atime changed */
@@ -1197,7 +1380,7 @@ static char *read_link(struct dentry *dentry)
 {
 	struct inode *inode = dentry->d_inode;
 	struct fuse_conn *fc = get_fuse_conn(inode);
-	struct fuse_req *req = fuse_get_req(fc);
+	struct fuse_req *req = fuse_get_req_nopages(fc);
 	char *link;
 
 	if (IS_ERR(req))
@@ -1391,7 +1574,7 @@ static int fuse_do_setattr(struct dentry *entry, struct iattr *attr,
 	loff_t oldsize;
 	int err;
 
-	if (!fuse_allow_task(fc, current))
+	if (!fuse_allow_current_process(fc))
 		return -EACCES;
 
 	if (!(fc->flags & FUSE_DEFAULT_PERMISSIONS))
@@ -1410,7 +1593,7 @@ static int fuse_do_setattr(struct dentry *entry, struct iattr *attr,
 	if (attr->ia_valid & ATTR_SIZE)
 		is_truncate = true;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -1500,7 +1683,7 @@ static int fuse_getattr(struct vfsmount *mnt, struct dentry *entry,
 	struct inode *inode = entry->d_inode;
 	struct fuse_conn *fc = get_fuse_conn(inode);
 
-	if (!fuse_allow_task(fc, current))
+	if (!fuse_allow_current_process(fc))
 		return -EACCES;
 
 	return fuse_update_attributes(inode, stat, NULL, NULL);
@@ -1518,7 +1701,7 @@ static int fuse_setxattr(struct dentry *entry, const char *name,
 	if (fc->no_setxattr)
 		return -EOPNOTSUPP;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -1557,7 +1740,7 @@ static ssize_t fuse_getxattr(struct dentry *entry, const char *name,
 	if (fc->no_getxattr)
 		return -EOPNOTSUPP;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -1603,13 +1786,13 @@ static ssize_t fuse_listxattr(struct dentry *entry, char *list, size_t size)
 	struct fuse_getxattr_out outarg;
 	ssize_t ret;
 
-	if (!fuse_allow_task(fc, current))
+	if (!fuse_allow_current_process(fc))
 		return -EACCES;
 
 	if (fc->no_listxattr)
 		return -EOPNOTSUPP;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -1654,7 +1837,7 @@ static int fuse_removexattr(struct dentry *entry, const char *name)
 	if (fc->no_removexattr)
 		return -EOPNOTSUPP;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 

fs/fuse/file.c

@@ -25,7 +25,7 @@ static int fuse_send_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
 	struct fuse_req *req;
 	int err;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -57,7 +57,7 @@ struct fuse_file *fuse_file_alloc(struct fuse_conn *fc)
 		return NULL;
 
 	ff->fc = fc;
-	ff->reserved_req = fuse_request_alloc();
+	ff->reserved_req = fuse_request_alloc(0);
 	if (unlikely(!ff->reserved_req)) {
 		kfree(ff);
 		return NULL;
@@ -368,7 +368,7 @@ static int fuse_flush(struct file *file, fl_owner_t id)
 	if (fc->no_flush)
 		return 0;
 
-	req = fuse_get_req_nofail(fc, file);
+	req = fuse_get_req_nofail_nopages(fc, file);
 	memset(&inarg, 0, sizeof(inarg));
 	inarg.fh = ff->fh;
 	inarg.lock_owner = fuse_lock_owner_id(fc, id);
@@ -436,7 +436,7 @@ int fuse_fsync_common(struct file *file, loff_t start, loff_t end,
 
 	fuse_sync_writes(inode);
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req)) {
 		err = PTR_ERR(req);
 		goto out;
@@ -544,7 +544,7 @@ static int fuse_readpage(struct file *file, struct page *page)
 	 */
 	fuse_wait_on_page_writeback(inode, page->index);
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req(fc, 1);
 	err = PTR_ERR(req);
 	if (IS_ERR(req))
 		goto out;
@@ -555,6 +555,7 @@ static int fuse_readpage(struct file *file, struct page *page)
 	req->out.argpages = 1;
 	req->num_pages = 1;
 	req->pages[0] = page;
+	req->page_descs[0].length = count;
 	num_read = fuse_send_read(req, file, pos, count, NULL);
 	err = req->out.h.error;
 	fuse_put_request(fc, req);
@@ -641,6 +642,7 @@ struct fuse_fill_data {
 	struct fuse_req *req;
 	struct file *file;
 	struct inode *inode;
+	unsigned nr_pages;
 };
 
 static int fuse_readpages_fill(void *_data, struct page *page)
@@ -656,16 +658,26 @@ static int fuse_readpages_fill(void *_data, struct page *page)
 	    (req->num_pages == FUSE_MAX_PAGES_PER_REQ ||
 	     (req->num_pages + 1) * PAGE_CACHE_SIZE > fc->max_read ||
 	     req->pages[req->num_pages - 1]->index + 1 != page->index)) {
+		int nr_alloc = min_t(unsigned, data->nr_pages,
+				     FUSE_MAX_PAGES_PER_REQ);
 		fuse_send_readpages(req, data->file);
-		data->req = req = fuse_get_req(fc);
+		data->req = req = fuse_get_req(fc, nr_alloc);
 		if (IS_ERR(req)) {
 			unlock_page(page);
 			return PTR_ERR(req);
 		}
 	}
+
+	if (WARN_ON(req->num_pages >= req->max_pages)) {
+		fuse_put_request(fc, req);
+		return -EIO;
+	}
+
 	page_cache_get(page);
 	req->pages[req->num_pages] = page;
+	req->page_descs[req->num_pages].length = PAGE_SIZE;
 	req->num_pages++;
+	data->nr_pages--;
 	return 0;
 }
 
@@ -676,6 +688,7 @@ static int fuse_readpages(struct file *file, struct address_space *mapping,
 	struct fuse_conn *fc = get_fuse_conn(inode);
 	struct fuse_fill_data data;
 	int err;
+	int nr_alloc = min_t(unsigned, nr_pages, FUSE_MAX_PAGES_PER_REQ);
 
 	err = -EIO;
 	if (is_bad_inode(inode))
@@ -683,7 +696,8 @@ static int fuse_readpages(struct file *file, struct address_space *mapping,
 
 	data.file = file;
 	data.inode = inode;
-	data.req = fuse_get_req(fc);
+	data.req = fuse_get_req(fc, nr_alloc);
+	data.nr_pages = nr_pages;
 	err = PTR_ERR(data.req);
 	if (IS_ERR(data.req))
 		goto out;
@@ -786,7 +800,7 @@ static size_t fuse_send_write_pages(struct fuse_req *req, struct file *file,
 
 	res = fuse_send_write(req, file, pos, count, NULL);
 
-	offset = req->page_offset;
+	offset = req->page_descs[0].offset;
 	count = res;
 	for (i = 0; i < req->num_pages; i++) {
 		struct page *page = req->pages[i];
@@ -817,7 +831,7 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
 	int err;
 
 	req->in.argpages = 1;
-	req->page_offset = offset;
+	req->page_descs[0].offset = offset;
 
 	do {
 		size_t tmp;
@@ -857,6 +871,7 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
 
 		err = 0;
 		req->pages[req->num_pages] = page;
+		req->page_descs[req->num_pages].length = tmp;
 		req->num_pages++;
 
 		iov_iter_advance(ii, tmp);
@@ -869,11 +884,19 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
 		if (!fc->big_writes)
 			break;
 	} while (iov_iter_count(ii) && count < fc->max_write &&
-		 req->num_pages < FUSE_MAX_PAGES_PER_REQ && offset == 0);
+		 req->num_pages < req->max_pages && offset == 0);
 
 	return count > 0 ? count : err;
 }
 
+static inline unsigned fuse_wr_pages(loff_t pos, size_t len)
+{
+	return min_t(unsigned,
+		     ((pos + len - 1) >> PAGE_CACHE_SHIFT) -
+		     (pos >> PAGE_CACHE_SHIFT) + 1,
+		     FUSE_MAX_PAGES_PER_REQ);
+}
+
 static ssize_t fuse_perform_write(struct file *file,
 				  struct address_space *mapping,
 				  struct iov_iter *ii, loff_t pos)
@@ -889,8 +912,9 @@ static ssize_t fuse_perform_write(struct file *file,
 	do {
 		struct fuse_req *req;
 		ssize_t count;
+		unsigned nr_pages = fuse_wr_pages(pos, iov_iter_count(ii));
 
-		req = fuse_get_req(fc);
+		req = fuse_get_req(fc, nr_pages);
 		if (IS_ERR(req)) {
 			err = PTR_ERR(req);
 			break;
@@ -1023,47 +1047,110 @@ static void fuse_release_user_pages(struct fuse_req *req, int write)
 	}
 }
 
-static int fuse_get_user_pages(struct fuse_req *req, const char __user *buf,
+static inline void fuse_page_descs_length_init(struct fuse_req *req,
+		unsigned index, unsigned nr_pages)
+{
+	int i;
+
+	for (i = index; i < index + nr_pages; i++)
+		req->page_descs[i].length = PAGE_SIZE -
+			req->page_descs[i].offset;
+}
+
+static inline unsigned long fuse_get_user_addr(const struct iov_iter *ii)
+{
+	return (unsigned long)ii->iov->iov_base + ii->iov_offset;
+}
+
+static inline size_t fuse_get_frag_size(const struct iov_iter *ii,
+					size_t max_size)
+{
+	return min(iov_iter_single_seg_count(ii), max_size);
+}
+
+static int fuse_get_user_pages(struct fuse_req *req, struct iov_iter *ii,
 			       size_t *nbytesp, int write)
 {
-	size_t nbytes = *nbytesp;
+	size_t nbytes = 0;  /* # bytes already packed in req */
-	unsigned long user_addr = (unsigned long) buf;
-	unsigned offset = user_addr & ~PAGE_MASK;
-	int npages;
 
 	/* Special case for kernel I/O: can copy directly into the buffer */
 	if (segment_eq(get_fs(), KERNEL_DS)) {
+		unsigned long user_addr = fuse_get_user_addr(ii);
+		size_t frag_size = fuse_get_frag_size(ii, *nbytesp);
+
 		if (write)
 			req->in.args[1].value = (void *) user_addr;
 		else
 			req->out.args[0].value = (void *) user_addr;
 
+		iov_iter_advance(ii, frag_size);
+		*nbytesp = frag_size;
 		return 0;
 	}
 
-	nbytes = min_t(size_t, nbytes, FUSE_MAX_PAGES_PER_REQ << PAGE_SHIFT);
+	while (nbytes < *nbytesp && req->num_pages < req->max_pages) {
-	npages = (nbytes + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
+		unsigned npages;
-	npages = clamp(npages, 1, FUSE_MAX_PAGES_PER_REQ);
+		unsigned long user_addr = fuse_get_user_addr(ii);
-	npages = get_user_pages_fast(user_addr, npages, !write, req->pages);
+		unsigned offset = user_addr & ~PAGE_MASK;
-	if (npages < 0)
+		size_t frag_size = fuse_get_frag_size(ii, *nbytesp - nbytes);
-		return npages;
+		int ret;
 
-	req->num_pages = npages;
+		unsigned n = req->max_pages - req->num_pages;
-	req->page_offset = offset;
+		frag_size = min_t(size_t, frag_size, n << PAGE_SHIFT);
+
+		npages = (frag_size + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
+		npages = clamp(npages, 1U, n);
+
+		ret = get_user_pages_fast(user_addr, npages, !write,
+					  &req->pages[req->num_pages]);
+		if (ret < 0)
+			return ret;
+
+		npages = ret;
+		frag_size = min_t(size_t, frag_size,
+				  (npages << PAGE_SHIFT) - offset);
+		iov_iter_advance(ii, frag_size);
+
+		req->page_descs[req->num_pages].offset = offset;
+		fuse_page_descs_length_init(req, req->num_pages, npages);
+
+		req->num_pages += npages;
+		req->page_descs[req->num_pages - 1].length -=
+			(npages << PAGE_SHIFT) - offset - frag_size;
+
+		nbytes += frag_size;
+	}
 
 	if (write)
 		req->in.argpages = 1;
 	else
 		req->out.argpages = 1;
 
-	nbytes = (req->num_pages << PAGE_SHIFT) - req->page_offset;
+	*nbytesp = nbytes;
-	*nbytesp = min(*nbytesp, nbytes);
 
 	return 0;
 }
 
-ssize_t fuse_direct_io(struct file *file, const char __user *buf,
+static inline int fuse_iter_npages(const struct iov_iter *ii_p)
-		       size_t count, loff_t *ppos, int write)
+{
+	struct iov_iter ii = *ii_p;
+	int npages = 0;
+
+	while (iov_iter_count(&ii) && npages < FUSE_MAX_PAGES_PER_REQ) {
+		unsigned long user_addr = fuse_get_user_addr(&ii);
+		unsigned offset = user_addr & ~PAGE_MASK;
+		size_t frag_size = iov_iter_single_seg_count(&ii);
+
+		npages += (frag_size + offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
+		iov_iter_advance(&ii, frag_size);
+	}
+
+	return min(npages, FUSE_MAX_PAGES_PER_REQ);
+}
+
+ssize_t fuse_direct_io(struct file *file, const struct iovec *iov,
+		       unsigned long nr_segs, size_t count, loff_t *ppos,
+		       int write)
 {
 	struct fuse_file *ff = file->private_data;
 	struct fuse_conn *fc = ff->fc;
@@ -1071,8 +1158,11 @@ ssize_t fuse_direct_io(struct file *file, const char __user *buf,
 	loff_t pos = *ppos;
 	ssize_t res = 0;
 	struct fuse_req *req;
+	struct iov_iter ii;
 
-	req = fuse_get_req(fc);
+	iov_iter_init(&ii, iov, nr_segs, count, 0);
+
+	req = fuse_get_req(fc, fuse_iter_npages(&ii));
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -1080,7 +1170,7 @@ ssize_t fuse_direct_io(struct file *file, const char __user *buf,
 		size_t nres;
 		fl_owner_t owner = current->files;
 		size_t nbytes = min(count, nmax);
-		int err = fuse_get_user_pages(req, buf, &nbytes, write);
+		int err = fuse_get_user_pages(req, &ii, &nbytes, write);
 		if (err) {
 			res = err;
 			break;
@@ -1103,12 +1193,11 @@ ssize_t fuse_direct_io(struct file *file, const char __user *buf,
 		count -= nres;
 		res += nres;
 		pos += nres;
-		buf += nres;
 		if (nres != nbytes)
 			break;
 		if (count) {
 			fuse_put_request(fc, req);
-			req = fuse_get_req(fc);
+			req = fuse_get_req(fc, fuse_iter_npages(&ii));
 			if (IS_ERR(req))
 				break;
 		}
@@ -1122,8 +1211,8 @@ ssize_t fuse_direct_io(struct file *file, const char __user *buf,
 }
 EXPORT_SYMBOL_GPL(fuse_direct_io);
 
-static ssize_t fuse_direct_read(struct file *file, char __user *buf,
+static ssize_t __fuse_direct_read(struct file *file, const struct iovec *iov,
-				     size_t count, loff_t *ppos)
+				  unsigned long nr_segs, loff_t *ppos)
 {
 	ssize_t res;
 	struct inode *inode = file->f_path.dentry->d_inode;
@@ -1131,22 +1220,31 @@ static ssize_t fuse_direct_read(struct file *file, char __user *buf,
 	if (is_bad_inode(inode))
 		return -EIO;
 
-	res = fuse_direct_io(file, buf, count, ppos, 0);
+	res = fuse_direct_io(file, iov, nr_segs, iov_length(iov, nr_segs),
+			     ppos, 0);
 
 	fuse_invalidate_attr(inode);
 
 	return res;
 }
 
-static ssize_t __fuse_direct_write(struct file *file, const char __user *buf,
+static ssize_t fuse_direct_read(struct file *file, char __user *buf,
-				   size_t count, loff_t *ppos)
+				     size_t count, loff_t *ppos)
+{
+	struct iovec iov = { .iov_base = buf, .iov_len = count };
+	return __fuse_direct_read(file, &iov, 1, ppos);
+}
+
+static ssize_t __fuse_direct_write(struct file *file, const struct iovec *iov,
+				   unsigned long nr_segs, loff_t *ppos)
 {
 	struct inode *inode = file->f_path.dentry->d_inode;
+	size_t count = iov_length(iov, nr_segs);
 	ssize_t res;
 
 	res = generic_write_checks(file, ppos, &count, 0);
 	if (!res) {
-		res = fuse_direct_io(file, buf, count, ppos, 1);
+		res = fuse_direct_io(file, iov, nr_segs, count, ppos, 1);
 		if (res > 0)
 			fuse_write_update_size(inode, *ppos);
 	}
@@ -1159,6 +1257,7 @@ static ssize_t __fuse_direct_write(struct file *file, const char __user *buf,
 static ssize_t fuse_direct_write(struct file *file, const char __user *buf,
 				 size_t count, loff_t *ppos)
 {
+	struct iovec iov = { .iov_base = (void __user *)buf, .iov_len = count };
 	struct inode *inode = file->f_path.dentry->d_inode;
 	ssize_t res;
 
@@ -1167,7 +1266,7 @@ static ssize_t fuse_direct_write(struct file *file, const char __user *buf,
 
 	/* Don't allow parallel writes to the same file */
 	mutex_lock(&inode->i_mutex);
-	res = __fuse_direct_write(file, buf, count, ppos);
+	res = __fuse_direct_write(file, &iov, 1, ppos);
 	mutex_unlock(&inode->i_mutex);
 
 	return res;
@@ -1272,7 +1371,7 @@ static int fuse_writepage_locked(struct page *page)
 
 	set_page_writeback(page);
 
-	req = fuse_request_alloc_nofs();
+	req = fuse_request_alloc_nofs(1);
 	if (!req)
 		goto err;
 
@@ -1293,7 +1392,8 @@ static int fuse_writepage_locked(struct page *page)
 	req->in.argpages = 1;
 	req->num_pages = 1;
 	req->pages[0] = tmp_page;
-	req->page_offset = 0;
+	req->page_descs[0].offset = 0;
+	req->page_descs[0].length = PAGE_SIZE;
 	req->end = fuse_writepage_end;
 	req->inode = inode;
 
@@ -1471,7 +1571,7 @@ static int fuse_getlk(struct file *file, struct file_lock *fl)
 	struct fuse_lk_out outarg;
 	int err;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -1506,7 +1606,7 @@ static int fuse_setlk(struct file *file, struct file_lock *fl, int flock)
 	if (fl->fl_flags & FL_CLOSE)
 		return 0;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -1575,7 +1675,7 @@ static sector_t fuse_bmap(struct address_space *mapping, sector_t block)
 	if (!inode->i_sb->s_bdev || fc->no_bmap)
 		return 0;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return 0;
 
@@ -1873,7 +1973,7 @@ long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
 		num_pages++;
 	}
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req(fc, num_pages);
 	if (IS_ERR(req)) {
 		err = PTR_ERR(req);
 		req = NULL;
@@ -1881,6 +1981,7 @@ long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
 	}
 	memcpy(req->pages, pages, sizeof(req->pages[0]) * num_pages);
 	req->num_pages = num_pages;
+	fuse_page_descs_length_init(req, 0, req->num_pages);
 
 	/* okay, let's send it to the client */
 	req->in.h.opcode = FUSE_IOCTL;
@@ -1981,7 +2082,7 @@ long fuse_ioctl_common(struct file *file, unsigned int cmd,
 	struct inode *inode = file->f_dentry->d_inode;
 	struct fuse_conn *fc = get_fuse_conn(inode);
 
-	if (!fuse_allow_task(fc, current))
+	if (!fuse_allow_current_process(fc))
 		return -EACCES;
 
 	if (is_bad_inode(inode))
@@ -2066,6 +2167,7 @@ unsigned fuse_file_poll(struct file *file, poll_table *wait)
 		return DEFAULT_POLLMASK;
 
 	poll_wait(file, &ff->poll_wait, wait);
+	inarg.events = (__u32)poll_requested_events(wait);
 
 	/*
 	 * Ask for notification iff there's someone waiting for it.
@@ -2076,7 +2178,7 @@ unsigned fuse_file_poll(struct file *file, poll_table *wait)
 		fuse_register_polled_file(fc, ff);
 	}
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return POLLERR;
 
@@ -2126,41 +2228,6 @@ int fuse_notify_poll_wakeup(struct fuse_conn *fc,
 	return 0;
 }
 
-static ssize_t fuse_loop_dio(struct file *filp, const struct iovec *iov,
-			     unsigned long nr_segs, loff_t *ppos, int rw)
-{
-	const struct iovec *vector = iov;
-	ssize_t ret = 0;
-
-	while (nr_segs > 0) {
-		void __user *base;
-		size_t len;
-		ssize_t nr;
-
-		base = vector->iov_base;
-		len = vector->iov_len;
-		vector++;
-		nr_segs--;
-
-		if (rw == WRITE)
-			nr = __fuse_direct_write(filp, base, len, ppos);
-		else
-			nr = fuse_direct_read(filp, base, len, ppos);
-
-		if (nr < 0) {
-			if (!ret)
-				ret = nr;
-			break;
-		}
-		ret += nr;
-		if (nr != len)
-			break;
-	}
-
-	return ret;
-}
-
-
 static ssize_t
 fuse_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov,
 			loff_t offset, unsigned long nr_segs)
@@ -2172,7 +2239,10 @@ fuse_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov,
 	file = iocb->ki_filp;
 	pos = offset;
 
-	ret = fuse_loop_dio(file, iov, nr_segs, &pos, rw);
+	if (rw == WRITE)
+		ret = __fuse_direct_write(file, iov, nr_segs, &pos);
+	else
+		ret = __fuse_direct_read(file, iov, nr_segs, &pos);
 
 	return ret;
 }
@@ -2194,7 +2264,7 @@ static long fuse_file_fallocate(struct file *file, int mode, loff_t offset,
 	if (fc->no_fallocate)
 		return -EOPNOTSUPP;
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 

fs/fuse/fuse_i.h

@@ -44,6 +44,9 @@
     doing the mount will be allowed to access the filesystem */
 #define FUSE_ALLOW_OTHER         (1 << 1)
 
+/** Number of page pointers embedded in fuse_req */
+#define FUSE_REQ_INLINE_PAGES 1
+
 /** List of active connections */
 extern struct list_head fuse_conn_list;
 
@@ -103,6 +106,15 @@ struct fuse_inode {
 
 	/** List of writepage requestst (pending or sent) */
 	struct list_head writepages;
+
+	/** Miscellaneous bits describing inode state */
+	unsigned long state;
+};
+
+/** FUSE inode state bits */
+enum {
+	/** Advise readdirplus  */
+	FUSE_I_ADVISE_RDPLUS,
 };
 
 struct fuse_conn;
@@ -200,6 +212,12 @@ struct fuse_out {
 	struct fuse_arg args[3];
 };
 
+/** FUSE page descriptor */
+struct fuse_page_desc {
+	unsigned int length;
+	unsigned int offset;
+};
+
 /** The request state */
 enum fuse_req_state {
 	FUSE_REQ_INIT = 0,
@@ -291,14 +309,23 @@ struct fuse_req {
 	} misc;
 
 	/** page vector */
-	struct page *pages[FUSE_MAX_PAGES_PER_REQ];
+	struct page **pages;
+
+	/** page-descriptor vector */
+	struct fuse_page_desc *page_descs;
+
+	/** size of the 'pages' array */
+	unsigned max_pages;
+
+	/** inline page vector */
+	struct page *inline_pages[FUSE_REQ_INLINE_PAGES];
+
+	/** inline page-descriptor vector */
+	struct fuse_page_desc inline_page_descs[FUSE_REQ_INLINE_PAGES];
 
 	/** number of pages in vector */
 	unsigned num_pages;
 
-	/** offset of data on first page */
-	unsigned page_offset;
-
 	/** File used in the request (or NULL) */
 	struct fuse_file *ff;
 
@@ -487,6 +514,12 @@ struct fuse_conn {
 	/** Use enhanced/automatic page cache invalidation. */
 	unsigned auto_inval_data:1;
 
+	/** Does the filesystem support readdirplus? */
+	unsigned do_readdirplus:1;
+
+	/** Does the filesystem want adaptive readdirplus? */
+	unsigned readdirplus_auto:1;
+
 	/** The number of requests waiting for completion */
 	atomic_t num_waiting;
 
@@ -578,6 +611,9 @@ void fuse_queue_forget(struct fuse_conn *fc, struct fuse_forget_link *forget,
 
 struct fuse_forget_link *fuse_alloc_forget(void);
 
+/* Used by READDIRPLUS */
+void fuse_force_forget(struct file *file, u64 nodeid);
+
 /**
  * Initialize READ or READDIR request
  */
@@ -658,9 +694,9 @@ void fuse_ctl_cleanup(void);
 /**
  * Allocate a request
  */
-struct fuse_req *fuse_request_alloc(void);
+struct fuse_req *fuse_request_alloc(unsigned npages);
 
-struct fuse_req *fuse_request_alloc_nofs(void);
+struct fuse_req *fuse_request_alloc_nofs(unsigned npages);
 
 /**
  * Free a request
@@ -668,14 +704,25 @@ struct fuse_req *fuse_request_alloc_nofs(void);
 void fuse_request_free(struct fuse_req *req);
 
 /**
- * Get a request, may fail with -ENOMEM
+ * Get a request, may fail with -ENOMEM,
+ * caller should specify # elements in req->pages[] explicitly
  */
-struct fuse_req *fuse_get_req(struct fuse_conn *fc);
+struct fuse_req *fuse_get_req(struct fuse_conn *fc, unsigned npages);
+
+/**
+ * Get a request, may fail with -ENOMEM,
+ * useful for callers who doesn't use req->pages[]
+ */
+static inline struct fuse_req *fuse_get_req_nopages(struct fuse_conn *fc)
+{
+	return fuse_get_req(fc, 0);
+}
 
 /**
  * Gets a requests for a file operation, always succeeds
  */
-struct fuse_req *fuse_get_req_nofail(struct fuse_conn *fc, struct file *file);
+struct fuse_req *fuse_get_req_nofail_nopages(struct fuse_conn *fc,
+					     struct file *file);
 
 /**
  * Decrement reference count of a request.  If count goes to zero free
@@ -739,9 +786,9 @@ void fuse_ctl_remove_conn(struct fuse_conn *fc);
 int fuse_valid_type(int m);
 
 /**
- * Is task allowed to perform filesystem operation?
+ * Is current process allowed to perform filesystem operation?
  */
-int fuse_allow_task(struct fuse_conn *fc, struct task_struct *task);
+int fuse_allow_current_process(struct fuse_conn *fc);
 
 u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id);
 
@@ -776,8 +823,9 @@ int fuse_reverse_inval_entry(struct super_block *sb, u64 parent_nodeid,
 
 int fuse_do_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
 		 bool isdir);
-ssize_t fuse_direct_io(struct file *file, const char __user *buf,
+ssize_t fuse_direct_io(struct file *file, const struct iovec *iov,
-		       size_t count, loff_t *ppos, int write);
+		       unsigned long nr_segs, size_t count, loff_t *ppos,
+		       int write);
 long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
 		   unsigned int flags);
 long fuse_ioctl_common(struct file *file, unsigned int cmd,

fs/fuse/inode.c

@@ -92,6 +92,7 @@ static struct inode *fuse_alloc_inode(struct super_block *sb)
 	fi->attr_version = 0;
 	fi->writectr = 0;
 	fi->orig_ino = 0;
+	fi->state = 0;
 	INIT_LIST_HEAD(&fi->write_files);
 	INIT_LIST_HEAD(&fi->queued_writes);
 	INIT_LIST_HEAD(&fi->writepages);
@@ -408,12 +409,12 @@ static int fuse_statfs(struct dentry *dentry, struct kstatfs *buf)
 	struct fuse_statfs_out outarg;
 	int err;
 
-	if (!fuse_allow_task(fc, current)) {
+	if (!fuse_allow_current_process(fc)) {
 		buf->f_type = FUSE_SUPER_MAGIC;
 		return 0;
 	}
 
-	req = fuse_get_req(fc);
+	req = fuse_get_req_nopages(fc);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
@@ -863,6 +864,10 @@ static void process_init_reply(struct fuse_conn *fc, struct fuse_req *req)
 				fc->dont_mask = 1;
 			if (arg->flags & FUSE_AUTO_INVAL_DATA)
 				fc->auto_inval_data = 1;
+			if (arg->flags & FUSE_DO_READDIRPLUS)
+				fc->do_readdirplus = 1;
+			if (arg->flags & FUSE_READDIRPLUS_AUTO)
+				fc->readdirplus_auto = 1;
 		} else {
 			ra_pages = fc->max_read / PAGE_CACHE_SIZE;
 			fc->no_lock = 1;
@@ -889,7 +894,8 @@ static void fuse_send_init(struct fuse_conn *fc, struct fuse_req *req)
 	arg->flags |= FUSE_ASYNC_READ | FUSE_POSIX_LOCKS | FUSE_ATOMIC_O_TRUNC |
 		FUSE_EXPORT_SUPPORT | FUSE_BIG_WRITES | FUSE_DONT_MASK |
 		FUSE_SPLICE_WRITE | FUSE_SPLICE_MOVE | FUSE_SPLICE_READ |
-		FUSE_FLOCK_LOCKS | FUSE_IOCTL_DIR | FUSE_AUTO_INVAL_DATA;
+		FUSE_FLOCK_LOCKS | FUSE_IOCTL_DIR | FUSE_AUTO_INVAL_DATA |
+		FUSE_DO_READDIRPLUS | FUSE_READDIRPLUS_AUTO;
 	req->in.h.opcode = FUSE_INIT;
 	req->in.numargs = 1;
 	req->in.args[0].size = sizeof(*arg);
@@ -1034,12 +1040,12 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 	/* only now - we want root dentry with NULL ->d_op */
 	sb->s_d_op = &fuse_dentry_operations;
 
-	init_req = fuse_request_alloc();
+	init_req = fuse_request_alloc(0);
 	if (!init_req)
 		goto err_put_root;
 
 	if (is_bdev) {
-		fc->destroy_req = fuse_request_alloc();
+		fc->destroy_req = fuse_request_alloc(0);
 		if (!fc->destroy_req)
 			goto err_free_init_req;
 	}

include/linux/fs.h

@@ -301,7 +301,7 @@ size_t iov_iter_copy_from_user(struct page *page,
 		struct iov_iter *i, unsigned long offset, size_t bytes);
 void iov_iter_advance(struct iov_iter *i, size_t bytes);
 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
-size_t iov_iter_single_seg_count(struct iov_iter *i);
+size_t iov_iter_single_seg_count(const struct iov_iter *i);
 
 static inline void iov_iter_init(struct iov_iter *i,
 			const struct iovec *iov, unsigned long nr_segs,

include/uapi/linux/fuse.h

@@ -1,9 +1,35 @@
 /*
-    FUSE: Filesystem in Userspace
+    This file defines the kernel interface of FUSE
     Copyright (C) 2001-2008  Miklos Szeredi <miklos@szeredi.hu>
 
     This program can be distributed under the terms of the GNU GPL.
     See the file COPYING.
+
+    This -- and only this -- header file may also be distributed under
+    the terms of the BSD Licence as follows:
+
+    Copyright (C) 2001-2007 Miklos Szeredi. All rights reserved.
+
+    Redistribution and use in source and binary forms, with or without
+    modification, are permitted provided that the following conditions
+    are met:
+    1. Redistributions of source code must retain the above copyright
+       notice, this list of conditions and the following disclaimer.
+    2. Redistributions in binary form must reproduce the above copyright
+       notice, this list of conditions and the following disclaimer in the
+       documentation and/or other materials provided with the distribution.
+
+    THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+    ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+    IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+    ARE DISCLAIMED.  IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE
+    FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+    DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+    OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+    HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+    LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+    OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+    SUCH DAMAGE.
 */
 
 /*
@@ -60,12 +86,25 @@
  *
  * 7.20
  *  - add FUSE_AUTO_INVAL_DATA
+ *
+ * 7.21
+ *  - add FUSE_READDIRPLUS
+ *  - send the requested events in POLL request
  */
 
 #ifndef _LINUX_FUSE_H
 #define _LINUX_FUSE_H
 
+#ifdef __linux__
 #include <linux/types.h>
+#else
+#include <stdint.h>
+#define __u64 uint64_t
+#define __s64 int64_t
+#define __u32 uint32_t
+#define __s32 int32_t
+#define __u16 uint16_t
+#endif
 
 /*
  * Version negotiation:
@@ -91,7 +130,7 @@
 #define FUSE_KERNEL_VERSION 7
 
 /** Minor version number of this interface */
-#define FUSE_KERNEL_MINOR_VERSION 20
+#define FUSE_KERNEL_MINOR_VERSION 21
 
 /** The node ID of the root inode */
 #define FUSE_ROOT_ID 1
@@ -179,6 +218,8 @@ struct fuse_file_lock {
  * FUSE_FLOCK_LOCKS: remote locking for BSD style file locks
  * FUSE_HAS_IOCTL_DIR: kernel supports ioctl on directories
  * FUSE_AUTO_INVAL_DATA: automatically invalidate cached pages
+ * FUSE_DO_READDIRPLUS: do READDIRPLUS (READDIR+LOOKUP in one)
+ * FUSE_READDIRPLUS_AUTO: adaptive readdirplus
  */
 #define FUSE_ASYNC_READ		(1 << 0)
 #define FUSE_POSIX_LOCKS	(1 << 1)
@@ -193,6 +234,8 @@ struct fuse_file_lock {
 #define FUSE_FLOCK_LOCKS	(1 << 10)
 #define FUSE_HAS_IOCTL_DIR	(1 << 11)
 #define FUSE_AUTO_INVAL_DATA	(1 << 12)
+#define FUSE_DO_READDIRPLUS	(1 << 13)
+#define FUSE_READDIRPLUS_AUTO	(1 << 14)
 
 /**
  * CUSE INIT request/reply flags
@@ -299,6 +342,7 @@ enum fuse_opcode {
 	FUSE_NOTIFY_REPLY  = 41,
 	FUSE_BATCH_FORGET  = 42,
 	FUSE_FALLOCATE     = 43,
+	FUSE_READDIRPLUS   = 44,
 
 	/* CUSE specific operations */
 	CUSE_INIT          = 4096,
@@ -580,7 +624,7 @@ struct fuse_poll_in {
 	__u64	fh;
 	__u64	kh;
 	__u32	flags;
-	__u32   padding;
+	__u32   events;
 };
 
 struct fuse_poll_out {
@@ -630,6 +674,16 @@ struct fuse_dirent {
 #define FUSE_DIRENT_SIZE(d) \
 	FUSE_DIRENT_ALIGN(FUSE_NAME_OFFSET + (d)->namelen)
 
+struct fuse_direntplus {
+	struct fuse_entry_out entry_out;
+	struct fuse_dirent dirent;
+};
+
+#define FUSE_NAME_OFFSET_DIRENTPLUS \
+	offsetof(struct fuse_direntplus, dirent.name)
+#define FUSE_DIRENTPLUS_SIZE(d) \
+	FUSE_DIRENT_ALIGN(FUSE_NAME_OFFSET_DIRENTPLUS + (d)->dirent.namelen)
+
 struct fuse_notify_inval_inode_out {
 	__u64	ino;
 	__s64	off;

mm/filemap.c

@@ -2056,7 +2056,7 @@ EXPORT_SYMBOL(iov_iter_fault_in_readable);
 /*
  * Return the count of just the current iov_iter segment.
  */
-size_t iov_iter_single_seg_count(struct iov_iter *i)
+size_t iov_iter_single_seg_count(const struct iov_iter *i)
 {
 	const struct iovec *iov = i->iov;
 	if (i->nr_segs == 1)