dev.c

UNIX/LINUX下面的用户文件系统

/*
  FUSE: Filesystem in Userspace
  Copyright (C) 2001-2006  Miklos Szeredi <[e-mail]>

  This program can be distributed under the terms of the GNU GPL.
  See the file COPYING.
*/

#include "fuse_i.h"

#include <linux/init.h>
#include <linux/module.h>
#include <linux/poll.h>
#include <linux/uio.h>
#include <linux/miscdevice.h>
#include <linux/pagemap.h>
#include <linux/file.h>
#include <linux/slab.h>

#ifdef MODULE_ALIAS_MISCDEV
MODULE_ALIAS_MISCDEV(FUSE_MINOR);
#endif

static kmem_cache_t *fuse_req_cachep;

static inline struct fuse_conn *fuse_get_conn(struct file *file)
{
	struct fuse_conn *fc;
	spin_lock(&fuse_lock);
	fc = file->private_data;
	if (fc && !fc->mounted)
		fc = NULL;
	spin_unlock(&fuse_lock);
	return fc;
}

static inline void fuse_request_init(struct fuse_req *req)
{
	memset(req, 0, sizeof(*req));
	INIT_LIST_HEAD(&req->list);
	init_waitqueue_head(&req->waitq);
	atomic_set(&req->count, 1);
}

struct fuse_req *fuse_request_alloc(void)
{
	struct fuse_req *req = kmem_cache_alloc(fuse_req_cachep, SLAB_KERNEL);
	if (req)
		fuse_request_init(req);
	return req;
}

void fuse_request_free(struct fuse_req *req)
{
	kmem_cache_free(fuse_req_cachep, req);
}

#ifdef KERNEL_2_6
static inline void block_sigs(sigset_t *oldset)
{
	sigset_t mask;

	siginitsetinv(&mask, sigmask(SIGKILL));
	sigprocmask(SIG_BLOCK, &mask, oldset);
}

static inline void restore_sigs(sigset_t *oldset)
{
	sigprocmask(SIG_SETMASK, oldset, NULL);
}
#else
#ifdef HAVE_RECALC_SIGPENDING_TSK
static inline void block_sigs(sigset_t *oldset)
{
	spin_lock_irq(&current->sighand->siglock);
	*oldset = current->blocked;
	siginitsetinv(&current->blocked, sigmask(SIGKILL) & ~oldset->sig[0]);
	recalc_sigpending();
	spin_unlock_irq(&current->sighand->siglock);
}

static inline void restore_sigs(sigset_t *oldset)
{
	spin_lock_irq(&current->sighand->siglock);
	current->blocked = *oldset;
	recalc_sigpending();
	spin_unlock_irq(&current->sighand->siglock);
}
#else
static inline void block_sigs(sigset_t *oldset)
{
	spin_lock_irq(&current->sigmask_lock);
	*oldset = current->blocked;
	siginitsetinv(&current->blocked, sigmask(SIGKILL) & ~oldset->sig[0]);
	recalc_sigpending(current);
	spin_unlock_irq(&current->sigmask_lock);
}

static inline void restore_sigs(sigset_t *oldset)
{
	spin_lock_irq(&current->sigmask_lock);
	current->blocked = *oldset;
	recalc_sigpending(current);
	spin_unlock_irq(&current->sigmask_lock);
}
#endif
#endif

void fuse_reset_request(struct fuse_req *req)
{
	int preallocated = req->preallocated;
	BUG_ON(atomic_read(&req->count) != 1);
	fuse_request_init(req);
	req->preallocated = preallocated;
}

static void __fuse_get_request(struct fuse_req *req)
{
	atomic_inc(&req->count);
}

/* Must be called with > 1 refcount */
static void __fuse_put_request(struct fuse_req *req)
{
	BUG_ON(atomic_read(&req->count) < 2);
	atomic_dec(&req->count);
}

static struct fuse_req *do_get_request(struct fuse_conn *fc)
{
	struct fuse_req *req;

	spin_lock(&fuse_lock);
	BUG_ON(list_empty(&fc->unused_list));
	req = list_entry(fc->unused_list.next, struct fuse_req, list);
	list_del_init(&req->list);
	spin_unlock(&fuse_lock);
	fuse_request_init(req);
	req->preallocated = 1;
	req->in.h.uid = current->fsuid;
	req->in.h.gid = current->fsgid;
	req->in.h.pid = current->pid;
	return req;
}

/* This can return NULL, but only in case it's interrupted by a SIGKILL */
struct fuse_req *fuse_get_request(struct fuse_conn *fc)
{
	int intr;
	sigset_t oldset;

	block_sigs(&oldset);
	intr = down_interruptible(&fc->outstanding_sem);
	restore_sigs(&oldset);
	return intr ? NULL : do_get_request(fc);
}

static void fuse_putback_request(struct fuse_conn *fc, struct fuse_req *req)
{
	if (req->preallocated)
		list_add(&req->list, &fc->unused_list);
	else
		fuse_request_free(req);

	/* If we are in debt decrease that first */
	if (fc->outstanding_debt)
		fc->outstanding_debt--;
	else
		up(&fc->outstanding_sem);
}

void fuse_put_request(struct fuse_conn *fc, struct fuse_req *req)
{
	if (atomic_dec_and_test(&req->count)) {
		spin_lock(&fuse_lock);
		fuse_putback_request(fc, req);
		spin_unlock(&fuse_lock);
	}
}

static void fuse_put_request_locked(struct fuse_conn *fc, struct fuse_req *req)
{
	if (atomic_dec_and_test(&req->count))
		fuse_putback_request(fc, req);
}

void fuse_release_background(struct fuse_req *req)
{
	iput(req->inode);
	iput(req->inode2);
	if (req->file)
		fput(req->file);
	spin_lock(&fuse_lock);
	list_del(&req->bg_entry);
	spin_unlock(&fuse_lock);
}

static void process_init_reply(struct fuse_conn *fc, struct fuse_req *req)
{
	int i;
	struct fuse_init_out *arg = &req->misc.init_out;

	if (arg->major != FUSE_KERNEL_VERSION)
		fc->conn_error = 1;
	else {
		fc->minor = arg->minor;
		fc->max_write = arg->minor < 5 ? 4096 : arg->max_write;
	}

	/* After INIT reply is received other requests can go
	   out.  So do (FUSE_MAX_OUTSTANDING - 1) number of
	   up()s on outstanding_sem.  The last up() is done in
	   fuse_putback_request() */
	for (i = 1; i < FUSE_MAX_OUTSTANDING; i++)
		up(&fc->outstanding_sem);
}

/*
 * This function is called when a request is finished.  Either a reply
 * has arrived or it was interrupted (and not yet sent) or some error
 * occurred during communication with userspace, or the device file was
 * closed.  It decreases the reference count for the request.  In case
 * of a background request the reference to the stored objects are
 * released.  The requester thread is woken up (if still waiting), and
 * finally the request is either freed or put on the unused_list
 *
 * Called with fuse_lock, unlocks it
 */
static void request_end(struct fuse_conn *fc, struct fuse_req *req)
{
	req->finished = 1;
	if (!req->background) {
		wake_up(&req->waitq);
		fuse_put_request_locked(fc, req);
		spin_unlock(&fuse_lock);
	} else {
		spin_unlock(&fuse_lock);
		down_read(&fc->sbput_sem);
		if (fc->mounted)
			fuse_release_background(req);
		up_read(&fc->sbput_sem);
		if (req->in.h.opcode == FUSE_INIT)
			process_init_reply(fc, req);
		else if (req->in.h.opcode == FUSE_RELEASE &&
			 req->inode == NULL) {
			/* Special case for failed iget in CREATE */
			u64 nodeid = req->in.h.nodeid;
			fuse_reset_request(req);
			fuse_send_forget(fc, req, nodeid, 1);
			return;
		}
		fuse_put_request(fc, req);
	}
}

/*
 * Unfortunately request interruption not just solves the deadlock
 * problem, it causes problems too.  These stem from the fact, that an
 * interrupted request is continued to be processed in userspace,
 * while all the locks and object references (inode and file) held
 * during the operation are released.
 *
 * To release the locks is exactly why there's a need to interrupt the
 * request, so there's not a lot that can be done about this, except
 * introduce additional locking in userspace.
 *
 * More important is to keep inode and file references until userspace
 * has replied, otherwise FORGET and RELEASE could be sent while the
 * inode/file is still used by the filesystem.
 *
 * For this reason the concept of "background" request is introduced.
 * An interrupted request is backgrounded if it has been already sent
 * to userspace.  Backgrounding involves getting an extra reference to
 * inode(s) or file used in the request, and adding the request to
 * fc->background list.  When a reply is received for a background
 * request, the object references are released, and the request is
 * removed from the list.  If the filesystem is unmounted while there
 * are still background requests, the list is walked and references
 * are released as if a reply was received.
 *
 * There's one more use for a background request.  The RELEASE message is
 * always sent as background, since it doesn't return an error or
 * data.
 */
static void background_request(struct fuse_conn *fc, struct fuse_req *req)
{
	req->background = 1;
	list_add(&req->bg_entry, &fc->background);
	if (req->inode)
		req->inode = igrab(req->inode);
	if (req->inode2)
		req->inode2 = igrab(req->inode2);
	if (req->file)
		get_file(req->file);
}

/* Called with fuse_lock held.  Releases, and then reacquires it. */
static void request_wait_answer(struct fuse_conn *fc, struct fuse_req *req)
{
	sigset_t oldset;

	spin_unlock(&fuse_lock);
	block_sigs(&oldset);
	wait_event_interruptible(req->waitq, req->finished);
	restore_sigs(&oldset);
	spin_lock(&fuse_lock);
	if (req->finished)
		return;

	req->out.h.error = -EINTR;
	req->interrupted = 1;
	if (req->locked) {
		/* This is uninterruptible sleep, because data is
		   being copied to/from the buffers of req.  During
		   locked state, there mustn't be any filesystem
		   operation (e.g. page fault), since that could lead
		   to deadlock */
		spin_unlock(&fuse_lock);
		wait_event(req->waitq, !req->locked);
		spin_lock(&fuse_lock);
	}
	if (!req->sent && !list_empty(&req->list)) {
		list_del(&req->list);
		__fuse_put_request(req);
	} else if (!req->finished && req->sent)
		background_request(fc, req);
}

static unsigned len_args(unsigned numargs, struct fuse_arg *args)
{
	unsigned nbytes = 0;
	unsigned i;

	for (i = 0; i < numargs; i++)
		nbytes += args[i].size;

	return nbytes;
}

static void queue_request(struct fuse_conn *fc, struct fuse_req *req)
{
	fc->reqctr++;
	/* zero is special */
	if (fc->reqctr == 0)
		fc->reqctr = 1;
	req->in.h.unique = fc->reqctr;
	req->in.h.len = sizeof(struct fuse_in_header) +
		len_args(req->in.numargs, (struct fuse_arg *) req->in.args);
	if (!req->preallocated) {
		/* If request is not preallocated (either FORGET or
		   RELEASE), then still decrease outstanding_sem, so
		   user can't open infinite number of files while not
		   processing the RELEASE requests.  However for
		   efficiency do it without blocking, so if down()
		   would block, just increase the debt instead */
		if (down_trylock(&fc->outstanding_sem))
			fc->outstanding_debt++;
	}
	list_add_tail(&req->list, &fc->pending);
	wake_up(&fc->waitq);
}

/*
 * This can only be interrupted by a SIGKILL
 */
void request_send(struct fuse_conn *fc, struct fuse_req *req)
{
	req->isreply = 1;
	spin_lock(&fuse_lock);
	if (!fc->connected)
		req->out.h.error = -ENOTCONN;
	else if (fc->conn_error)
		req->out.h.error = -ECONNREFUSED;
	else {
		queue_request(fc, req);
		/* acquire extra reference, since request is still needed
		   after request_end() */
		__fuse_get_request(req);

		request_wait_answer(fc, req);
	}
	spin_unlock(&fuse_lock);
}

static void request_send_nowait(struct fuse_conn *fc, struct fuse_req *req)
{
	spin_lock(&fuse_lock);
	if (fc->connected) {
		queue_request(fc, req);
		spin_unlock(&fuse_lock);
	} else {
		req->out.h.error = -ENOTCONN;
		request_end(fc, req);
	}
}

void request_send_noreply(struct fuse_conn *fc, struct fuse_req *req)
{
	req->isreply = 0;
	request_send_nowait(fc, req);
}

void request_send_background(struct fuse_conn *fc, struct fuse_req *req)
{
	req->isreply = 1;
	spin_lock(&fuse_lock);
	background_request(fc, req);
	spin_unlock(&fuse_lock);
	request_send_nowait(fc, req);
}

void fuse_send_init(struct fuse_conn *fc)
{
	/* This is called from fuse_read_super() so there's guaranteed
	   to be a request available */
	struct fuse_req *req = do_get_request(fc);
	struct fuse_init_in *arg = &req->misc.init_in;
	arg->major = FUSE_KERNEL_VERSION;
	arg->minor = FUSE_KERNEL_MINOR_VERSION;
	req->in.h.opcode = FUSE_INIT;
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(*arg);
	req->in.args[0].value = arg;
	req->out.numargs = 1;
	/* Variable length arguement used for backward compatibility
	   with interface version < 7.5.  Rest of init_out is zeroed
	   by do_get_request(), so a short reply is not a problem */
	req->out.argvar = 1;
	req->out.args[0].size = sizeof(struct fuse_init_out);
	req->out.args[0].value = &req->misc.init_out;
	request_send_background(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
 * interrupted bail out.
 */
static inline int lock_request(struct fuse_req *req)
{
	int err = 0;
	if (req) {
		spin_lock(&fuse_lock);
		if (req->interrupted)
			err = -ENOENT;
		else
			req->locked = 1;
		spin_unlock(&fuse_lock);
	}
	return err;
}

/*
 * Unlock request.  If it was interrupted during being locked, the
 * requester thread is currently waiting for it to be unlocked, so
 * wake it up.
 */
static inline void unlock_request(struct fuse_req *req)
{
	if (req) {
		spin_lock(&fuse_lock);
		req->locked = 0;
		if (req->interrupted)
			wake_up(&req->waitq);
		spin_unlock(&fuse_lock);
	}
}

struct fuse_copy_state {
	int write;
	struct fuse_req *req;
	const struct iovec *iov;
	unsigned long nr_segs;
	unsigned long seglen;
	unsigned long addr;
	struct page *pg;
	void *mapaddr;
	void *buf;
	unsigned len;
};

static void fuse_copy_init(struct fuse_copy_state *cs, int write,
			   struct fuse_req *req, const struct iovec *iov,