write.c

linux 内核源代码

		if (req->wb_index > idx_end)
			break;

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

		kref_get(&req->wb_kref);
		spin_unlock(&inode->i_lock);
		error = nfs_wait_on_request(req);
		nfs_release_request(req);
		spin_lock(&inode->i_lock);
		if (error < 0)
			return error;
		res++;
	}
	return res;
}

static void nfs_cancel_commit_list(struct list_head *head)
{
	struct nfs_page *req;

	while(!list_empty(head)) {
		req = nfs_list_entry(head->next);
		dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
		dec_bdi_stat(req->wb_page->mapping->backing_dev_info,
				BDI_RECLAIMABLE);
		nfs_list_remove_request(req);
		clear_bit(PG_NEED_COMMIT, &(req)->wb_flags);
		nfs_inode_remove_request(req);
		nfs_unlock_request(req);
	}
}

#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, pgoff_t idx_start, unsigned int npages)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	int res = 0;

	if (nfsi->ncommit != 0) {
		res = nfs_scan_list(nfsi, dst, idx_start, npages,
				NFS_PAGE_TAG_COMMIT);
		nfsi->ncommit -= res;
	}
	return res;
}
#else
static inline int nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, unsigned int npages)
{
	return 0;
}
#endif

/*
 * 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 page *page, unsigned int offset, unsigned int bytes)
{
	struct address_space *mapping = page->mapping;
	struct inode *inode = mapping->host;
	struct nfs_page		*req, *new = NULL;
	pgoff_t		rqend, end;

	end = offset + bytes;

	for (;;) {
		/* Loop over all inode entries and see if we find
		 * A request for the page we wish to update
		 */
		spin_lock(&inode->i_lock);
		req = nfs_page_find_request_locked(page);
		if (req) {
			if (!nfs_lock_request_dontget(req)) {
				int error;

				spin_unlock(&inode->i_lock);
				error = nfs_wait_on_request(req);
				nfs_release_request(req);
				if (error < 0) {
					if (new)
						nfs_release_request(new);
					return ERR_PTR(error);
				}
				continue;
			}
			spin_unlock(&inode->i_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(&inode->i_lock);
				nfs_unlock_request(new);
				return ERR_PTR(error);
			}
			spin_unlock(&inode->i_lock);
			req = new;
			goto zero_page;
		}
		spin_unlock(&inode->i_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 = max(end, rqend) - req->wb_offset;
		goto zero_page;
	}

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

	return req;
zero_page:
	/* If this page might potentially be marked as up to date,
	 * then we need to zero any uninitalised data. */
	if (req->wb_pgbase == 0 && req->wb_bytes != PAGE_CACHE_SIZE
			&& !PageUptodate(req->wb_page))
		zero_user_page(req->wb_page, req->wb_bytes,
				PAGE_CACHE_SIZE - req->wb_bytes,
				KM_USER0);
	return req;
}

int nfs_flush_incompatible(struct file *file, struct page *page)
{
	struct nfs_open_context *ctx = nfs_file_open_context(file);
	struct nfs_page	*req;
	int do_flush, status;
	/*
	 * 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.
	 */
	do {
		req = nfs_page_find_request(page);
		if (req == NULL)
			return 0;
		do_flush = req->wb_page != page || req->wb_context != ctx
			|| !nfs_dirty_request(req);
		nfs_release_request(req);
		if (!do_flush)
			return 0;
		status = nfs_wb_page(page->mapping->host, page);
	} while (status == 0);
	return status;
}

/*
 * 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 = nfs_file_open_context(file);
	struct inode	*inode = page->mapping->host;
	int		status = 0;

	nfs_inc_stats(inode, NFSIOS_VFSUPDATEPAGE);

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

	/* 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 && !(file->f_mode & O_SYNC)) {
		count = max(count + offset, nfs_page_length(page));
		offset = 0;
	}

	status = nfs_writepage_setup(ctx, page, offset, count);
	__set_page_dirty_nobuffers(page);

        dprintk("NFS:      nfs_updatepage returns %d (isize %Ld)\n",
			status, (long long)i_size_read(inode));
	if (status < 0)
		nfs_set_pageerror(page);
	return status;
}

static void nfs_writepage_release(struct nfs_page *req)
{

	if (PageError(req->wb_page)) {
		nfs_end_page_writeback(req->wb_page);
		nfs_inode_remove_request(req);
	} else if (!nfs_reschedule_unstable_write(req)) {
		/* Set the PG_uptodate flag */
		nfs_mark_uptodate(req->wb_page, req->wb_pgbase, req->wb_bytes);
		nfs_end_page_writeback(req->wb_page);
		nfs_inode_remove_request(req);
	} else
		nfs_end_page_writeback(req->wb_page);
	nfs_clear_page_tag_locked(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,
		const struct rpc_call_ops *call_ops,
		unsigned int count, unsigned int offset,
		int how)
{
	struct inode		*inode;
	int flags;

	/* 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->path.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_fattr_init(&data->fattr);

	/* Set up the initial task struct.  */
	flags = (how & FLUSH_SYNC) ? 0 : RPC_TASK_ASYNC;
	rpc_init_task(&data->task, NFS_CLIENT(inode), flags, call_ops, data);
	NFS_PROTO(inode)->write_setup(data, how);

	data->task.tk_priority = flush_task_priority(how);
	data->task.tk_cookie = (unsigned long)inode;

	dprintk("NFS: %5u initiated write call "
		"(req %s/%Ld, %u bytes @ offset %Lu)\n",
		data->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);
	rpc_execute(&data->task);
	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 inode *inode, struct list_head *head, unsigned int npages, size_t count, int how)
{
	struct nfs_page *req = nfs_list_entry(head->next);
	struct page *page = req->wb_page;
	struct nfs_write_data *data;
	size_t wsize = NFS_SERVER(inode)->wsize, nbytes;
	unsigned int offset;
	int requests = 0;
	LIST_HEAD(list);

	nfs_list_remove_request(req);

	nbytes = count;
	do {
		size_t len = min(nbytes, wsize);

		data = nfs_writedata_alloc(1);
		if (!data)
			goto out_bad;
		list_add(&data->pages, &list);
		requests++;
		nbytes -= len;
	} while (nbytes != 0);
	atomic_set(&req->wb_complete, requests);

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

		data->pagevec[0] = page;

		if (nbytes < wsize)
			wsize = nbytes;
		nfs_write_rpcsetup(req, data, &nfs_write_partial_ops,
				   wsize, offset, how);
		offset += wsize;
		nbytes -= wsize;
		nfs_execute_write(data);
	} while (nbytes != 0);

	return 0;

out_bad:
	while (!list_empty(&list)) {
		data = list_entry(list.next, struct nfs_write_data, pages);
		list_del(&data->pages);
		nfs_writedata_release(data);
	}
	nfs_redirty_request(req);
	nfs_end_page_writeback(req->wb_page);
	nfs_clear_page_tag_locked(req);
	return -ENOMEM;
}

/*
 * Create an RPC task for the given write request and kick it.
 * The page must have been locked by the caller.
 *
 * It may happen that the page we're passed is not marked dirty.
 * This is the case if nfs_updatepage detects a conflicting request
 * that has been written but not committed.
 */
static int nfs_flush_one(struct inode *inode, struct list_head *head, unsigned int npages, size_t count, int how)
{
	struct nfs_page		*req;
	struct page		**pages;
	struct nfs_write_data	*data;

	data = nfs_writedata_alloc(npages);
	if (!data)
		goto out_bad;

	pages = data->pagevec;
	while (!list_empty(head)) {
		req = nfs_list_entry(head->next);
		nfs_list_remove_request(req);
		nfs_list_add_request(req, &data->pages);
		ClearPageError(req->wb_page);
		*pages++ = req->wb_page;
	}
	req = nfs_list_entry(data->pages.next);

	/* Set up the argument struct */
	nfs_write_rpcsetup(req, data, &nfs_write_full_ops, count, 0, how);

	nfs_execute_write(data);
	return 0;
 out_bad:
	while (!list_empty(head)) {
		req = nfs_list_entry(head->next);
		nfs_list_remove_request(req);
		nfs_redirty_request(req);
		nfs_end_page_writeback(req->wb_page);
		nfs_clear_page_tag_locked(req);
	}
	return -ENOMEM;
}

static void nfs_pageio_init_write(struct nfs_pageio_descriptor *pgio,
				  struct inode *inode, int ioflags)
{
	int wsize = NFS_SERVER(inode)->wsize;

	if (wsize < PAGE_CACHE_SIZE)
		nfs_pageio_init(pgio, inode, nfs_flush_multi, wsize, ioflags);
	else
		nfs_pageio_init(pgio, inode, nfs_flush_one, wsize, ioflags);
}

/*
 * Handle a write reply that flushed part of a page.
 */
static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata)
{
	struct nfs_write_data	*data = calldata;
	struct nfs_page		*req = data->req;
	struct page		*page = req->wb_page;

	dprintk("NFS: write (%s/%Ld %d@%Ld)",
		req->wb_context->path.dentry->d_inode->i_sb->s_id,
		(long long)NFS_FILEID(req->wb_context->path.dentry->d_inode),
		req->wb_bytes,
		(long long)req_offset(req));

	if (nfs_writeback_done(task, data) != 0)
		return;

	if (task->tk_status < 0) {
		nfs_set_pageerror(page);
		nfs_context_set_write_error(req->wb_context, task->tk_status);
		dprintk(", error = %d\n", task->tk_status);
		goto out;
	}

	if (nfs_write_need_commit(data)) {
		struct inode *inode = page->mapping->host;

		spin_lock(&inode->i_lock);
		if (test_bit(PG_NEED_RESCHED, &req->wb_flags)) {
			/* Do nothing we need to resend the writes */
		} else if (!test_and_set_bit(PG_NEED_COMMIT, &req->wb_flags)) {
			memcpy(&req->wb_verf, &data->verf, sizeof(req->wb_verf));
			dprintk(" defer commit\n");
		} else if (memcmp(&req->wb_verf, &data->verf, sizeof(req->wb_verf))) {
			set_bit(PG_NEED_RESCHED, &req->wb_flags);
			clear_bit(PG_NEED_COMMIT, &req->wb_flags);
			dprintk(" server reboot detected\n");
		}
		spin_unlock(&inode->i_lock);
	} else
		dprintk(" OK\n");

out:
	if (atomic_dec_and_test(&req->wb_complete))
		nfs_writepage_release(req);
}

static const struct rpc_call_ops nfs_write_partial_ops = {
	.rpc_call_done = nfs_writeback_done_partial,
	.rpc_release = nfs_writedata_release,
};

/*
 * Handle a write reply that flushes a whole page.
 *
 * FIXME: There is an inherent race with invalidate_inode_pages and
 *	  writebacks since the page->count is kept > 1 for as long
 *	  as the page has a write request pending.
 */
static void nfs_writeback_done_full(struct rpc_task *task, void *calldata)
{
	struct nfs_write_data	*data = calldata;
	struct nfs_page		*req;
	struct page		*page;

	if (nfs_writeback_done(task, data) != 0)
		return;

	/* Update attributes as result of writeback. */
	while (!list_empty(&data->pages)) {
		req = nfs_list_entry(data->pages.next);
		nfs_list_remove_request(req);
		page = req->wb_page;

		dprintk("NFS: write (%s/%Ld %d@%Ld)",