FUSE: extend FUSE to support more operations, take #2

index 350fe97..7caa473 100644
@@ -40,9 +40,6 @@
 /** The major number of the fuse character device */
 #define FUSE_MAJOR MISC_MAJOR
 
-/** The minor number of the fuse character device */
-#define FUSE_MINOR 229
-
 /* Make sure all structures are padded to 64bit boundary, so 32bit
    userspace works under 64bit kernels */
 

54 lines not on this page · raw message

index 34930a9..71fde8b 100644
@@ -1470,6 +1470,275 @@ static loff_t fuse_file_llseek(struct file *file, loff_t offset, int origin)
 	return retval;
 }
 
+static int fuse_ioctl_copy_user(struct page **pages, struct iovec *iov,
+			unsigned int nr_segs, size_t bytes, bool to_user)
+{
+	struct iov_iter ii;
+	int page_idx = 0;
+
+	if (!bytes)
+		return 0;
+
+	iov_iter_init(&ii, iov, nr_segs, bytes, 0);
+
+	while (iov_iter_count(&ii)) {
+		struct page *page = pages[page_idx++];
+		size_t todo = min_t(size_t, PAGE_SIZE, iov_iter_count(&ii));
+		void *kaddr, *map;
+
+		kaddr = map = kmap(page);
+
+		while (todo) {
+			char __user *uaddr = ii.iov->iov_base + ii.iov_offset;
+			size_t iov_len = ii.iov->iov_len - ii.iov_offset;
+			size_t copy = min(todo, iov_len);
+			size_t left;
+
+			if (!to_user)
+				left = copy_from_user(kaddr, uaddr, copy);
+			else
+				left = copy_to_user(uaddr, kaddr, copy);
+
+			if (unlikely(left))
+				return -EFAULT;
+
+			iov_iter_advance(&ii, copy);
+			todo -= copy;
+			kaddr += copy;
+		}
+
+		kunmap(map);
+	}
+
+	return 0;
+}
+
+/*
+ * For ioctls, there is no generic way to determine how much memory
+ * needs to be read and/or written.  Furthermore, ioctls are allowed
+ * to dereference the passed pointer, so the parameter requires deep
+ * copying but FUSE has no idea whatsoever about what to copy in or
+ * out.
+ *
+ * This is solved by allowing FUSE server to retry ioctl with
+ * necessary in/out iovecs.  Let's assume the ioctl implementation
+ * needs to read in the following structure.
+ *
+ * struct a {
+ *	char	*buf;
+ *	size_t	buflen;
+ * }
+ *
+ * On the first callout to FUSE server, inarg->in_size and
+ * inarg->out_size will be NULL; then, the server completes the ioctl
+ * with FUSE_IOCTL_RETRY set in out->flags, out->in_iovs set to 1 and
+ * the actual iov array to
+ *
+ * { { .iov_base = inarg.arg,	.iov_len = sizeof(struct a) } }
+ *
+ * which tells FUSE to copy in the requested area and retry the ioctl.
+ * On the second round, the server has access to the structure and
+ * from that it can tell what to look for next, so on the invocation,
+ * it sets FUSE_IOCTL_RETRY, out->in_iovs to 2 and iov array to
+ *
+ * { { .iov_base = inarg.arg,	.iov_len = sizeof(struct a)	},
+ *   { .iov_base = a.buf,	.iov_len = a.buflen		} }
+ *
+ * FUSE will copy both struct a and the pointed buffer from the
+ * process doing the ioctl and retry ioctl with both struct a and the
+ * buffer.
+ *
+ * This time, FUSE server has everything it needs and completes ioctl
+ * without FUSE_IOCTL_RETRY which finishes the ioctl call.
+ *
+ * Copying data out works the same way.
+ *
+ * Note that if FUSE_IOCTL_UNRESTRICTED is clear, the kernel
+ * automatically initializes in and out iovs by decoding @cmd with
+ * _IOC_* macros and the server is not allowed to request RETRY.  This
+ * limits ioctl data transfers to well-formed ioctls and is the forced
+ * behavior for all FUSE servers.
+ */
+static long fuse_file_do_ioctl(struct file *file, unsigned int cmd,
+			       unsigned long arg, unsigned int flags)
+{
+	struct inode *inode = file->f_dentry->d_inode;
+	struct fuse_file *ff = file->private_data;
+	struct fuse_conn *fc = get_fuse_conn(inode);
+	struct fuse_ioctl_in inarg = { .fh = ff->fh, .cmd = cmd, .arg = arg,
+				       .flags = flags };
+	struct fuse_ioctl_out outarg;
+	struct fuse_req *req = NULL;
+	struct page **pages = NULL;
+	struct page *iov_page = NULL;
+	struct iovec *in_iov = NULL, *out_iov = NULL;
+	unsigned int in_iovs = 0, out_iovs = 0, num_pages = 0, max_pages;
+	size_t in_size, out_size, transferred;
+	int err;
+
+	/* assume all the iovs returned by client always fits in a page */
+	BUILD_BUG_ON(sizeof(struct iovec) * FUSE_IOCTL_MAX_IOV > PAGE_SIZE);
+
+	err = -EIO;
+	if (is_bad_inode(inode))
+		goto out;
+
+	err = -ENOMEM;
+	pages = kzalloc(sizeof(pages[0]) * FUSE_MAX_PAGES_PER_REQ, GFP_KERNEL);
+	iov_page = alloc_page(GFP_KERNEL);
+	if (!pages || !iov_page)
+		goto out;
+
+	/*
+	 * If restricted, initialize IO parameters as encoded in @cmd.
+	 * RETRY from server is not allowed.
+	 */
+	if (!(flags & FUSE_IOCTL_UNRESTRICTED)) {
+		struct iovec *iov = page_address(iov_page);
+
+		iov->iov_base = (void *)arg;
+		iov->iov_len = _IOC_SIZE(cmd);
+
+		if (_IOC_DIR(cmd) & _IOC_READ) {
+			in_iov = iov;
+			in_iovs = 1;
+		}
+
+		if (_IOC_DIR(cmd) & _IOC_WRITE) {
+			out_iov = iov;
+			out_iovs = 1;
+		}
+	}
+
+ retry:
+	inarg.in_size = in_size = iov_length(in_iov, in_iovs);
+	inarg.out_size = out_size = iov_length(out_iov, out_iovs);
+
+	/*
+	 * Out data can be used either for actual out data or iovs,
+	 * make sure there always is at least one page.
+	 */
+	out_size = max_t(size_t, out_size, PAGE_SIZE);
+	max_pages = DIV_ROUND_UP(max(in_size, out_size), PAGE_SIZE);
+
+	/* make sure there are enough buffer pages and init request with them */
+	err = -ENOMEM;
+	if (max_pages > FUSE_MAX_PAGES_PER_REQ)
+		goto out;
+	while (num_pages < max_pages) {
+		pages[num_pages] = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
+		if (!pages[num_pages])
+			goto out;
+		num_pages++;
+	}
+
+	req = fuse_get_req(fc);
+	if (IS_ERR(req)) {
+		req = NULL;
+		err = PTR_ERR(req);
+		goto out;
+	}
+	memcpy(req->pages, pages, sizeof(req->pages[0]) * num_pages);
+	req->num_pages = num_pages;
+
+	/* okay, let's send it to the client */
+	req->in.h.opcode = FUSE_IOCTL;
+	req->in.h.nodeid = get_node_id(inode);
+	req->in.numargs = 1;
+	req->in.args[0].size = sizeof(inarg);
+	req->in.args[0].value = &inarg;
+	if (in_size) {
+		req->in.numargs++;
+		req->in.args[1].size = in_size;
+		req->in.argpages = 1;
+
+		err = fuse_ioctl_copy_user(pages, in_iov, in_iovs, in_size,
+					   false);
+		if (err)
+			goto out;
+	}
+
+	req->out.numargs = 2;
+	req->out.args[0].size = sizeof(outarg);
+	req->out.args[0].value = &outarg;
+	req->out.args[1].size = out_size;
+	req->out.argpages = 1;
+	req->out.argvar = 1;
+
+	request_send(fc, req);
+	err = req->out.h.error;
+	transferred = req->out.args[1].size;
+	fuse_put_request(fc, req);
+	req = NULL;
+	if (err)
+		goto out;
+
+	/* did it ask for retry? */
+	if (outarg.flags & FUSE_IOCTL_RETRY) {
+		char *vaddr;
+
+		/* no retry if in restricted mode */
+		err = -EBADE;
+		if (!(flags & FUSE_IOCTL_UNRESTRICTED))
+			goto out;
+
+		in_iovs = outarg.in_iovs;
+		out_iovs = outarg.out_iovs;
+
+		/*
+		 * Make sure things are in boundary, separate checks
+		 * are to protect against overflow.
+		 */
+		err = -ENOMEM;
+		if (in_iovs > FUSE_IOCTL_MAX_IOV ||
+		    out_iovs > FUSE_IOCTL_MAX_IOV ||
+		    in_iovs + out_iovs > FUSE_IOCTL_MAX_IOV)
+			goto out;
+
+		err = -EBADE;
+		if ((in_iovs + out_iovs) * sizeof(struct iovec) != transferred)
+			goto out;
+
+		/* okay, copy in iovs and retry */
+		vaddr = kmap_atomic(pages[0], KM_USER0);
+		memcpy(page_address(iov_page), vaddr, transferred);
+		kunmap_atomic(vaddr, KM_USER0);
+
+		in_iov = page_address(iov_page);
+		out_iov = in_iov + in_iovs;
+
+		goto retry;
+	}
+
+	err = -EBADE;
+	if (transferred > inarg.out_size)
+		goto out;
+
+	err = fuse_ioctl_copy_user(pages, out_iov, out_iovs, transferred, true);
+ out:
+	if (req)
+		fuse_put_request(fc, req);
+	if (iov_page)
+		__free_page(iov_page);
+	while (num_pages)
+		__free_page(pages[--num_pages]);
+	kfree(pages);
+
+	return err ? err : outarg.result;
+}
+
+static long fuse_file_ioctl(struct file *file, unsigned int cmd,
+			    unsigned long arg)
+{
+	return fuse_file_do_ioctl(file, cmd, arg, 0);
+}
+
+static long fuse_file_compat_ioctl(struct file *file, unsigned int cmd,
+				   unsigned long arg)
+{
+	return fuse_file_do_ioctl(file, cmd, arg, FUSE_IOCTL_COMPAT);
+}
+
 static const struct file_operations fuse_file_operations = {
 	.llseek		= fuse_file_llseek,
 	.read		= do_sync_read,

@@ -1484,6 +1753,8 @@ static const struct file_operations fuse_file_operations = {
 	.lock		= fuse_file_lock,
 	.flock		= fuse_file_flock,
 	.splice_read	= generic_file_splice_read,
+	.unlocked_ioctl	= fuse_file_ioctl,
+	.compat_ioctl	= fuse_file_compat_ioctl,
 };
 
 static const struct file_operations fuse_direct_io_file_operations = {

@@ -1496,6 +1767,8 @@ static const struct file_operations fuse_direct_io_file_operations = {
 	.fsync		= fuse_fsync,
 	.lock		= fuse_file_lock,
 	.flock		= fuse_file_flock,
+	.unlocked_ioctl	= fuse_file_ioctl,
+	.compat_ioctl	= fuse_file_compat_ioctl,
 	/* no mmap and splice_read */
 };
 

index 7caa473..608e300 100644
@@ -148,6 +148,21 @@ struct fuse_file_lock {
  */
 #define FUSE_READ_LOCKOWNER	(1 << 1)
 
+/**
+ * Ioctl flags
+ *
+ * FUSE_IOCTL_COMPAT: 32bit compat ioctl on 64bit machine
+ * FUSE_IOCTL_UNRESTRICTED: not restricted to well-formed ioctls, retry allowed
+ * FUSE_IOCTL_RETRY: retry with new iovecs
+ *
+ * FUSE_IOCTL_MAX_IOV: maximum of in_iovecs + out_iovecs
+ */
+#define FUSE_IOCTL_COMPAT	(1 << 0)
+#define FUSE_IOCTL_UNRESTRICTED	(1 << 1)
+#define FUSE_IOCTL_RETRY	(1 << 2)
+
+#define FUSE_IOCTL_MAX_IOV	256
+
 enum fuse_opcode {
 	FUSE_LOOKUP	   = 1,
 	FUSE_FORGET	   = 2,  /* no reply */

@@ -185,6 +200,7 @@ enum fuse_opcode {
 	FUSE_INTERRUPT     = 36,
 	FUSE_BMAP          = 37,
 	FUSE_DESTROY       = 38,
+	FUSE_IOCTL         = 39,
 };
 
 /* The read buffer is required to be at least 8k, but may be much larger */

@@ -385,6 +401,22 @@ struct fuse_bmap_out {
 	__u64	block;
 };
 
+struct fuse_ioctl_in {
+	__u64	fh;
+	__u32	flags;
+	__u32	cmd;
+	__u64	arg;
+	__u32	in_size;
+	__u32	out_size;
+};
+
+struct fuse_ioctl_out {
+	__s32	result;
+	__u32	flags;
+	__u32	in_iovs;
+	__u32	out_iovs;
+};
+
 struct fuse_in_header {
 	__u32	len;
 	__u32	opcode;

index fd03330..7c6821e 100644
@@ -408,7 +408,7 @@ static int fuse_create_open(struct inode *dir, struct dentry *entry, int mode,
 		goto out_put_forget_req;
 
 	err = -ENOMEM;
-	ff = fuse_file_alloc();
+	ff = fuse_file_alloc(fc);
 	if (!ff)
 		goto out_put_request;
 

index 71fde8b..1f662f1 100644
@@ -46,7 +46,7 @@ static int fuse_send_open(struct inode *inode, struct file *file, int isdir,
 	return err;
 }
 
-struct fuse_file *fuse_file_alloc(void)
+struct fuse_file *fuse_file_alloc(struct fuse_conn *fc)
 {
 	struct fuse_file *ff;
 	ff = kmalloc(sizeof(struct fuse_file), GFP_KERNEL);

@@ -58,6 +58,9 @@ struct fuse_file *fuse_file_alloc(void)
 		} else {
 			INIT_LIST_HEAD(&ff->write_entry);
 			atomic_set(&ff->count, 0);
+			spin_lock(&fc->lock);
+			ff->kh = ++fc->khctr;
+			spin_unlock(&fc->lock);
 		}
 	}
 	return ff;

@@ -109,6 +112,7 @@ void fuse_finish_open(struct inode *inode, struct file *file,
 
 int fuse_open_common(struct inode *inode, struct file *file, int isdir)
 {
+	struct fuse_conn *fc = get_fuse_conn(inode);
 	struct fuse_open_out outarg;
 	struct fuse_file *ff;
 	int err;

@@ -121,7 +125,7 @@ int fuse_open_common(struct inode *inode, struct file *file, int isdir)
 	if (err)
 		return err;
 
-	ff = fuse_file_alloc();
+	ff = fuse_file_alloc(fc);
 	if (!ff)
 		return -ENOMEM;
 

index 35accfd..fdfdfb0 100644
@@ -100,6 +100,9 @@ struct fuse_file {
 	/** Request reserved for flush and release */
 	struct fuse_req *reserved_req;
 
+	/** Kernel file handle guaranteed to be unique */
+	u64 kh;
+
 	/** File handle used by userspace */
 	u64 fh;
 

@@ -322,6 +325,9 @@ struct fuse_conn {
 	/** The list of requests under I/O */
 	struct list_head io;
 
+	/** The next unique kernel file handle */
+	u64 khctr;
+
 	/** Number of requests currently in the background */
 	unsigned num_background;
 

@@ -499,7 +505,7 @@ void fuse_read_fill(struct fuse_req *req, struct file *file,
  */
 int fuse_open_common(struct inode *inode, struct file *file, int isdir);
 
-struct fuse_file *fuse_file_alloc(void);
+struct fuse_file *fuse_file_alloc(struct fuse_conn *fc);
 void fuse_file_free(struct fuse_file *ff);
 void fuse_finish_open(struct inode *inode, struct file *file,
 		      struct fuse_file *ff, struct fuse_open_out *outarg);

index 2e99f34..9f27441 100644
@@ -485,6 +485,7 @@ static struct fuse_conn *new_conn(struct super_block *sb)
 		fc->bdi.unplug_io_fn = default_unplug_io_fn;
 		/* fuse does it's own writeback accounting */
 		fc->bdi.capabilities = BDI_CAP_NO_ACCT_WB;
+		fc->khctr = 0;
 		fc->dev = sb->s_dev;
 		err = bdi_init(&fc->bdi);
 		if (err)

index b723614..1408b18 100644
@@ -813,6 +813,21 @@ static ssize_t fuse_dev_read(struct kiocb *iocb, const struct iovec *iov,
 	return err;
 }
 
+static int fuse_handle_notify(struct fuse_conn *fc, enum fuse_notify_code code,
+			      unsigned int size, struct fuse_copy_state *cs)
+{
+	int err;
+
+	switch (code) {
+	default:
+		err = -EINVAL;
+		break;
+	}
+
+	fuse_copy_finish(cs);
+	return err;
+}
+
 /* Look up request on processing list by unique ID */
 static struct fuse_req *request_find(struct fuse_conn *fc, u64 unique)
 {

@@ -876,9 +891,24 @@ static ssize_t fuse_dev_write(struct kiocb *iocb, const struct iovec *iov,
 	err = fuse_copy_one(&cs, &oh, sizeof(oh));
 	if (err)
 		goto err_finish;
+
+	if (oh.len != nbytes)
+		goto err_finish;
+
+	/*
+	 * Zero oh.unique indicates unsolicited notification message
+	 * and error contains notification code.
+	 */
+	if (!oh.unique) {
+		err = fuse_handle_notify(fc, oh.error, nbytes - sizeof(oh),
+					 &cs);
+		if (err)
+			return err;
+		return nbytes;
+	}
+
 	err = -EINVAL;
-	if (!oh.unique || oh.error <= -1000 || oh.error > 0 ||
-	    oh.len != nbytes)
+	if (oh.error <= -1000 || oh.error > 0)
 		goto err_finish;
 
 	spin_lock(&fc->lock);

index 608e300..abde994 100644
@@ -203,6 +203,10 @@ enum fuse_opcode {
 	FUSE_IOCTL         = 39,
 };
 
+enum fuse_notify_code {
+	FUSE_NOTIFY_CODE_MAX,
+};
+
 /* The read buffer is required to be at least 8k, but may be much larger */
 #define FUSE_MIN_READ_BUFFER 8192
 

index 1408b18..b71cc89 100644
@@ -819,11 +819,26 @@ static int fuse_handle_notify(struct fuse_conn *fc, enum fuse_notify_code code,
 	int err;
 
 	switch (code) {
+	case FUSE_NOTIFY_POLL: {
+		struct fuse_notify_poll_wakeup_out outarg;
+
+		err = -EINVAL;
+		if (size != sizeof(outarg))
+			goto out;
+
+		err = fuse_copy_one(cs, &outarg, sizeof(outarg));
+		if (err)
+			goto out;
+
+		err = fuse_notify_poll_wakeup(fc, &outarg);
+		break;
+	}
 	default:
 		err = -EINVAL;
 		break;
 	}
 
+out:
 	fuse_copy_finish(cs);
 	return err;
 }

166 lines not on this page · raw message

index fdfdfb0..66f2400 100644
@@ -19,6 +19,8 @@
 #include <linux/backing-dev.h>
 #include <linux/mutex.h>
 #include <linux/rwsem.h>
+#include <linux/rbtree.h>
+#include <linux/poll.h>
 
 /** Max number of pages that can be used in a single read request */
 #define FUSE_MAX_PAGES_PER_REQ 32

@@ -111,6 +113,12 @@ struct fuse_file {
 
 	/** Entry on inode's write_files list */
 	struct list_head write_entry;
+
+	/** RB node to be linked on fuse_conn->polled_files */
+	struct rb_node polled_node;
+
+	/** Wait queue head for poll */
+	wait_queue_head_t poll_wait;
 };
 
 /** One input argument of a request */

@@ -328,6 +336,9 @@ struct fuse_conn {
 	/** The next unique kernel file handle */
 	u64 khctr;
 
+	/** rbtree of fuse_files waiting for poll events indexed by ph */
+	struct rb_root polled_files;
+
 	/** Number of requests currently in the background */
 	unsigned num_background;
 

@@ -416,6 +427,9 @@ struct fuse_conn {
 	/** Is bmap not implemented by fs? */
 	unsigned no_bmap : 1;
 
+	/** Is poll not implemented by fs? */
+	unsigned no_poll : 1;
+
 	/** Do multi-page cached writes */
 	unsigned big_writes : 1;
 

@@ -525,6 +539,12 @@ int fuse_fsync_common(struct file *file, struct dentry *de, int datasync,
 		      int isdir);
 
 /**
+ * Notify poll wakeup
+ */
+int fuse_notify_poll_wakeup(struct fuse_conn *fc,
+			    struct fuse_notify_poll_wakeup_out *outarg);
+
+/**
  * Initialize file operations on a regular file
  */
 void fuse_init_file_inode(struct inode *inode);

index 9f27441..77fcedb 100644
@@ -486,6 +486,7 @@ static struct fuse_conn *new_conn(struct super_block *sb)
 		/* fuse does it's own writeback accounting */
 		fc->bdi.capabilities = BDI_CAP_NO_ACCT_WB;
 		fc->khctr = 0;
+		fc->polled_files = RB_ROOT;
 		fc->dev = sb->s_dev;
 		err = bdi_init(&fc->bdi);
 		if (err)

index abde994..5650cf0 100644
@@ -163,6 +163,13 @@ struct fuse_file_lock {
 
 #define FUSE_IOCTL_MAX_IOV	256
 
+/**
+ * Poll flags
+ *
+ * FUSE_POLL_SCHEDULE_NOTIFY: request poll notify
+ */
+#define FUSE_POLL_SCHEDULE_NOTIFY (1 << 0)
+
 enum fuse_opcode {
 	FUSE_LOOKUP	   = 1,
 	FUSE_FORGET	   = 2,  /* no reply */

@@ -201,9 +208,11 @@ enum fuse_opcode {
 	FUSE_BMAP          = 37,
 	FUSE_DESTROY       = 38,
 	FUSE_IOCTL         = 39,
+	FUSE_POLL          = 40,
 };
 
 enum fuse_notify_code {
+	FUSE_NOTIFY_POLL   = 1,
 	FUSE_NOTIFY_CODE_MAX,
 };
 

@@ -421,6 +430,22 @@ struct fuse_ioctl_out {
 	__u32	out_iovs;
 };
 
+struct fuse_poll_in {
+	__u64	fh;
+	__u64	kh;
+	__u32	flags;
+	__u32   padding;
+};
+
+struct fuse_poll_out {
+	__u32	revents;
+	__u32	padding;
+};
+
+struct fuse_notify_poll_wakeup_out {
+	__u64	kh;
+};
+
 struct fuse_in_header {
 	__u32	len;
 	__u32	opcode;