write.c

Linux Kernel 2.6.9 for OMAP1710

#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
/*
 * Add a request to the inode's commit list.
 */
static void
nfs_mark_request_commit(struct nfs_page *req)
{
	struct inode *inode = req->wb_context->dentry->d_inode;
	struct nfs_inode *nfsi = NFS_I(inode);

	spin_lock(&nfsi->req_lock);
	nfs_list_add_request(req, &nfsi->commit);
	nfsi->ncommit++;
	spin_unlock(&nfsi->req_lock);
	inc_page_state(nr_unstable);
	mark_inode_dirty(inode);
}
#endif

/*
 * Wait for a request to complete.
 *
 * Interruptible by signals only if mounted with intr flag.
 */
static int
nfs_wait_on_requests(struct inode *inode, unsigned long idx_start, unsigned int npages)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_page *req;
	unsigned long		idx_end, next;
	unsigned int		res = 0;
	int			error;

	if (npages == 0)
		idx_end = ~0;
	else
		idx_end = idx_start + npages - 1;

	spin_lock(&nfsi->req_lock);
	next = idx_start;
	while (radix_tree_gang_lookup(&nfsi->nfs_page_tree, (void **)&req, next, 1)) {
		if (req->wb_index > idx_end)
			break;

		next = req->wb_index + 1;
		if (!NFS_WBACK_BUSY(req))
			continue;

		atomic_inc(&req->wb_count);
		spin_unlock(&nfsi->req_lock);
		error = nfs_wait_on_request(req);
		nfs_release_request(req);
		if (error < 0)
			return error;
		spin_lock(&nfsi->req_lock);
		res++;
	}
	spin_unlock(&nfsi->req_lock);
	return res;
}

/*
 * nfs_scan_dirty - Scan an inode for dirty requests
 * @inode: NFS inode to scan
 * @dst: destination list
 * @idx_start: lower bound of page->index to scan.
 * @npages: idx_start + npages sets the upper bound to scan.
 *
 * Moves requests from the inode's dirty page list.
 * The requests are *not* checked to ensure that they form a contiguous set.
 */
static int
nfs_scan_dirty(struct inode *inode, struct list_head *dst, unsigned long idx_start, unsigned int npages)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	int	res;
	res = nfs_scan_list(&nfsi->dirty, dst, idx_start, npages);
	nfsi->ndirty -= res;
	sub_page_state(nr_dirty,res);
	if ((nfsi->ndirty == 0) != list_empty(&nfsi->dirty))
		printk(KERN_ERR "NFS: desynchronized value of nfs_i.ndirty.\n");
	return res;
}

#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
/*
 * nfs_scan_commit - Scan an inode for commit requests
 * @inode: NFS inode to scan
 * @dst: destination list
 * @idx_start: lower bound of page->index to scan.
 * @npages: idx_start + npages sets the upper bound to scan.
 *
 * Moves requests from the inode's 'commit' request list.
 * The requests are *not* checked to ensure that they form a contiguous set.
 */
static int
nfs_scan_commit(struct inode *inode, struct list_head *dst, unsigned long idx_start, unsigned int npages)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	int	res;
	res = nfs_scan_list(&nfsi->commit, dst, idx_start, npages);
	nfsi->ncommit -= res;
	if ((nfsi->ncommit == 0) != list_empty(&nfsi->commit))
		printk(KERN_ERR "NFS: desynchronized value of nfs_i.ncommit.\n");
	return res;
}
#endif

static int nfs_wait_on_write_congestion(struct address_space *mapping, int intr)
{
	struct backing_dev_info *bdi = mapping->backing_dev_info;
	DEFINE_WAIT(wait);
	int ret = 0;

	might_sleep();

	if (!bdi_write_congested(bdi))
		return 0;
	if (intr) {
		struct rpc_clnt *clnt = NFS_CLIENT(mapping->host);
		sigset_t oldset;

		rpc_clnt_sigmask(clnt, &oldset);
		prepare_to_wait(&nfs_write_congestion, &wait, TASK_INTERRUPTIBLE);
		if (bdi_write_congested(bdi)) {
			if (signalled())
				ret = -ERESTARTSYS;
			else
				schedule();
		}
		rpc_clnt_sigunmask(clnt, &oldset);
	} else {
		prepare_to_wait(&nfs_write_congestion, &wait, TASK_UNINTERRUPTIBLE);
		if (bdi_write_congested(bdi))
			schedule();
	}
	finish_wait(&nfs_write_congestion, &wait);
	return ret;
}


/*
 * Try to update any existing write request, or create one if there is none.
 * In order to match, the request's credentials must match those of
 * the calling process.
 *
 * Note: Should always be called with the Page Lock held!
 */
static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx,
		struct inode *inode, struct page *page,
		unsigned int offset, unsigned int bytes)
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_page		*req, *new = NULL;
	unsigned long		rqend, end;

	end = offset + bytes;

	if (nfs_wait_on_write_congestion(page->mapping, server->flags & NFS_MOUNT_INTR))
		return ERR_PTR(-ERESTARTSYS);
	for (;;) {
		/* Loop over all inode entries and see if we find
		 * A request for the page we wish to update
		 */
		spin_lock(&nfsi->req_lock);
		req = _nfs_find_request(inode, page->index);
		if (req) {
			if (!nfs_lock_request_dontget(req)) {
				int error;
				spin_unlock(&nfsi->req_lock);
				error = nfs_wait_on_request(req);
				nfs_release_request(req);
				if (error < 0)
					return ERR_PTR(error);
				continue;
			}
			spin_unlock(&nfsi->req_lock);
			if (new)
				nfs_release_request(new);
			break;
		}

		if (new) {
			int error;
			nfs_lock_request_dontget(new);
			error = nfs_inode_add_request(inode, new);
			if (error) {
				spin_unlock(&nfsi->req_lock);
				nfs_unlock_request(new);
				return ERR_PTR(error);
			}
			spin_unlock(&nfsi->req_lock);
			nfs_mark_request_dirty(new);
			return new;
		}
		spin_unlock(&nfsi->req_lock);

		new = nfs_create_request(ctx, inode, page, offset, bytes);
		if (IS_ERR(new))
			return new;
	}

	/* We have a request for our page.
	 * If the creds don't match, or the
	 * page addresses don't match,
	 * tell the caller to wait on the conflicting
	 * request.
	 */
	rqend = req->wb_offset + req->wb_bytes;
	if (req->wb_context != ctx
	    || req->wb_page != page
	    || !nfs_dirty_request(req)
	    || offset > rqend || end < req->wb_offset) {
		nfs_unlock_request(req);
		return ERR_PTR(-EBUSY);
	}

	/* Okay, the request matches. Update the region */
	if (offset < req->wb_offset) {
		req->wb_offset = offset;
		req->wb_pgbase = offset;
		req->wb_bytes = rqend - req->wb_offset;
	}

	if (end > rqend)
		req->wb_bytes = end - req->wb_offset;

	return req;
}

int nfs_flush_incompatible(struct file *file, struct page *page)
{
	struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
	struct inode	*inode = page->mapping->host;
	struct nfs_page	*req;
	int		status = 0;
	/*
	 * Look for a request corresponding to this page. If there
	 * is one, and it belongs to another file, we flush it out
	 * before we try to copy anything into the page. Do this
	 * due to the lack of an ACCESS-type call in NFSv2.
	 * Also do the same if we find a request from an existing
	 * dropped page.
	 */
	req = nfs_find_request(inode, page->index);
	if (req) {
		if (req->wb_page != page || ctx != req->wb_context)
			status = nfs_wb_page(inode, page);
		nfs_release_request(req);
	}
	return (status < 0) ? status : 0;
}

/*
 * Update and possibly write a cached page of an NFS file.
 *
 * XXX: Keep an eye on generic_file_read to make sure it doesn't do bad
 * things with a page scheduled for an RPC call (e.g. invalidate it).
 */
int nfs_updatepage(struct file *file, struct page *page,
		unsigned int offset, unsigned int count)
{
	struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
	struct dentry	*dentry = file->f_dentry;
	struct inode	*inode = page->mapping->host;
	struct nfs_page	*req;
	int		status = 0;

	dprintk("NFS:      nfs_updatepage(%s/%s %d@%Ld)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name,
		count, (long long)(page_offset(page) +offset));

	if (IS_SYNC(inode)) {
		status = nfs_writepage_sync(ctx, inode, page, offset, count, 0);
		if (status > 0) {
			if (offset == 0 && status == PAGE_CACHE_SIZE)
				SetPageUptodate(page);
			return 0;
		}
		return status;
	}

	/* If we're not using byte range locks, and we know the page
	 * is entirely in cache, it may be more efficient to avoid
	 * fragmenting write requests.
	 */
	if (PageUptodate(page) && inode->i_flock == NULL) {
		loff_t end_offs = i_size_read(inode) - 1;
		unsigned long end_index = end_offs >> PAGE_CACHE_SHIFT;

		count += offset;
		offset = 0;
		if (unlikely(end_offs < 0)) {
			/* Do nothing */
		} else if (page->index == end_index) {
			unsigned int pglen;
			pglen = (unsigned int)(end_offs & (PAGE_CACHE_SIZE-1)) + 1;
			if (count < pglen)
				count = pglen;
		} else if (page->index < end_index)
			count = PAGE_CACHE_SIZE;
	}

	/*
	 * Try to find an NFS request corresponding to this page
	 * and update it.
	 * If the existing request cannot be updated, we must flush
	 * it out now.
	 */
	do {
		req = nfs_update_request(ctx, inode, page, offset, count);
		status = (IS_ERR(req)) ? PTR_ERR(req) : 0;
		if (status != -EBUSY)
			break;
		/* Request could not be updated. Flush it out and try again */
		status = nfs_wb_page(inode, page);
	} while (status >= 0);
	if (status < 0)
		goto done;

	status = 0;

	/* Update file length */
	nfs_grow_file(page, offset, count);
	/* Set the PG_uptodate flag? */
	nfs_mark_uptodate(page, req->wb_pgbase, req->wb_bytes);
	nfs_unlock_request(req);
done:
        dprintk("NFS:      nfs_updatepage returns %d (isize %Ld)\n",
			status, (long long)i_size_read(inode));
	if (status < 0)
		ClearPageUptodate(page);
	return status;
}

static void nfs_writepage_release(struct nfs_page *req)
{
	end_page_writeback(req->wb_page);

#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
	if (!PageError(req->wb_page)) {
		if (NFS_NEED_RESCHED(req)) {
			nfs_mark_request_dirty(req);
			goto out;
		} else if (NFS_NEED_COMMIT(req)) {
			nfs_mark_request_commit(req);
			goto out;
		}
	}
	nfs_inode_remove_request(req);

out:
	nfs_clear_commit(req);
	nfs_clear_reschedule(req);
#else
	nfs_inode_remove_request(req);
#endif
	nfs_unlock_request(req);
}

static inline int flush_task_priority(int how)
{
	switch (how & (FLUSH_HIGHPRI|FLUSH_LOWPRI)) {
		case FLUSH_HIGHPRI:
			return RPC_PRIORITY_HIGH;
		case FLUSH_LOWPRI:
			return RPC_PRIORITY_LOW;
	}
	return RPC_PRIORITY_NORMAL;
}

/*
 * Set up the argument/result storage required for the RPC call.
 */
static void nfs_write_rpcsetup(struct nfs_page *req,
		struct nfs_write_data *data,
		unsigned int count, unsigned int offset,
		int how)
{
	struct rpc_task		*task = &data->task;
	struct inode		*inode;

	/* Set up the RPC argument and reply structs
	 * NB: take care not to mess about with data->commit et al. */

	data->req = req;
	data->inode = inode = req->wb_context->dentry->d_inode;
	data->cred = req->wb_context->cred;

	data->args.fh     = NFS_FH(inode);
	data->args.offset = req_offset(req) + offset;
	data->args.pgbase = req->wb_pgbase + offset;
	data->args.pages  = data->pagevec;
	data->args.count  = count;
	data->args.context = req->wb_context;

	data->res.fattr   = &data->fattr;
	data->res.count   = count;
	data->res.verf    = &data->verf;

	NFS_PROTO(inode)->write_setup(data, how);

	data->task.tk_priority = flush_task_priority(how);
	data->task.tk_cookie = (unsigned long)inode;
	data->task.tk_calldata = data;
	/* Release requests */
	data->task.tk_release = nfs_writedata_release;

	dprintk("NFS: %4d initiated write call (req %s/%Ld, %u bytes @ offset %Lu)\n",
		task->tk_pid,
		inode->i_sb->s_id,
		(long long)NFS_FILEID(inode),
		count,
		(unsigned long long)data->args.offset);
}

static void nfs_execute_write(struct nfs_write_data *data)
{
	struct rpc_clnt *clnt = NFS_CLIENT(data->inode);
	sigset_t oldset;

	rpc_clnt_sigmask(clnt, &oldset);
	lock_kernel();
	rpc_execute(&data->task);
	unlock_kernel();
	rpc_clnt_sigunmask(clnt, &oldset);
}

/*
 * Generate multiple small requests to write out a single
 * contiguous dirty area on one page.
 */
static int nfs_flush_multi(struct list_head *head, struct inode *inode, int how)
{
	struct nfs_page *req = nfs_list_entry(head->next);
	struct page *page = req->wb_page;
	struct nfs_write_data *data;
	unsigned int wsize = NFS_SERVER(inode)->wsize;
	unsigned int nbytes, offset;
	int requests = 0;
	LIST_HEAD(list);

	nfs_list_remove_request(req);

	nbytes = req->wb_bytes;
	for (;;) {
		data = nfs_writedata_alloc();
		if (!data)
			goto out_bad;
		list_add(&data->pages, &list);
		requests++;
		if (nbytes <= wsize)
			break;
		nbytes -= wsize;
	}
	atomic_set(&req->wb_complete, requests);

	ClearPageError(page);
	SetPageWriteback(page);
	offset = 0;
	nbytes = req->wb_bytes;
	do {
		data = list_entry(list.next, struct nfs_write_data, pages);
		list_del_init(&data->pages);

		data->pagevec[0] = page;
		data->complete = nfs_writeback_done_partial;

		if (nbytes > wsize) {
			nfs_write_rpcsetup(req, data, wsize, offset, how);
			offset += wsize;
			nbytes -= wsize;
		} else {
			nfs_write_rpcsetup(req, data, nbytes, offset, how);
			nbytes = 0;
		}
		nfs_execute_write(data);
	} while (nbytes != 0);

	return 0;

out_bad: