write.c

linux 内核源代码

/* handling of writes to regular files and writing back to the server
 *
 * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
 * Written by David Howells ([e-mail])
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 */
#include <linux/backing-dev.h>
#include <linux/slab.h>
#include <linux/fs.h>
#include <linux/pagemap.h>
#include <linux/writeback.h>
#include <linux/pagevec.h>
#include "internal.h"

static int afs_write_back_from_locked_page(struct afs_writeback *wb,
					   struct page *page);

/*
 * mark a page as having been made dirty and thus needing writeback
 */
int afs_set_page_dirty(struct page *page)
{
	_enter("");
	return __set_page_dirty_nobuffers(page);
}

/*
 * unlink a writeback record because its usage has reached zero
 * - must be called with the wb->vnode->writeback_lock held
 */
static void afs_unlink_writeback(struct afs_writeback *wb)
{
	struct afs_writeback *front;
	struct afs_vnode *vnode = wb->vnode;

	list_del_init(&wb->link);
	if (!list_empty(&vnode->writebacks)) {
		/* if an fsync rises to the front of the queue then wake it
		 * up */
		front = list_entry(vnode->writebacks.next,
				   struct afs_writeback, link);
		if (front->state == AFS_WBACK_SYNCING) {
			_debug("wake up sync");
			front->state = AFS_WBACK_COMPLETE;
			wake_up(&front->waitq);
		}
	}
}

/*
 * free a writeback record
 */
static void afs_free_writeback(struct afs_writeback *wb)
{
	_enter("");
	key_put(wb->key);
	kfree(wb);
}

/*
 * dispose of a reference to a writeback record
 */
void afs_put_writeback(struct afs_writeback *wb)
{
	struct afs_vnode *vnode = wb->vnode;

	_enter("{%d}", wb->usage);

	spin_lock(&vnode->writeback_lock);
	if (--wb->usage == 0)
		afs_unlink_writeback(wb);
	else
		wb = NULL;
	spin_unlock(&vnode->writeback_lock);
	if (wb)
		afs_free_writeback(wb);
}

/*
 * partly or wholly fill a page that's under preparation for writing
 */
static int afs_fill_page(struct afs_vnode *vnode, struct key *key,
			 unsigned start, unsigned len, struct page *page)
{
	int ret;

	_enter(",,%u,%u", start, len);

	ASSERTCMP(start + len, <=, PAGE_SIZE);

	ret = afs_vnode_fetch_data(vnode, key, start, len, page);
	if (ret < 0) {
		if (ret == -ENOENT) {
			_debug("got NOENT from server"
			       " - marking file deleted and stale");
			set_bit(AFS_VNODE_DELETED, &vnode->flags);
			ret = -ESTALE;
		}
	}

	_leave(" = %d", ret);
	return ret;
}

/*
 * prepare a page for being written to
 */
static int afs_prepare_page(struct afs_vnode *vnode, struct page *page,
			    struct key *key, unsigned offset, unsigned to)
{
	unsigned eof, tail, start, stop, len;
	loff_t i_size, pos;
	void *p;
	int ret;

	_enter("");

	if (offset == 0 && to == PAGE_SIZE)
		return 0;

	p = kmap_atomic(page, KM_USER0);

	i_size = i_size_read(&vnode->vfs_inode);
	pos = (loff_t) page->index << PAGE_SHIFT;
	if (pos >= i_size) {
		/* partial write, page beyond EOF */
		_debug("beyond");
		if (offset > 0)
			memset(p, 0, offset);
		if (to < PAGE_SIZE)
			memset(p + to, 0, PAGE_SIZE - to);
		kunmap_atomic(p, KM_USER0);
		return 0;
	}

	if (i_size - pos >= PAGE_SIZE) {
		/* partial write, page entirely before EOF */
		_debug("before");
		tail = eof = PAGE_SIZE;
	} else {
		/* partial write, page overlaps EOF */
		eof = i_size - pos;
		_debug("overlap %u", eof);
		tail = max(eof, to);
		if (tail < PAGE_SIZE)
			memset(p + tail, 0, PAGE_SIZE - tail);
		if (offset > eof)
			memset(p + eof, 0, PAGE_SIZE - eof);
	}

	kunmap_atomic(p, KM_USER0);

	ret = 0;
	if (offset > 0 || eof > to) {
		/* need to fill one or two bits that aren't going to be written
		 * (cover both fillers in one read if there are two) */
		start = (offset > 0) ? 0 : to;
		stop = (eof > to) ? eof : offset;
		len = stop - start;
		_debug("wr=%u-%u av=0-%u rd=%u@%u",
		       offset, to, eof, start, len);
		ret = afs_fill_page(vnode, key, start, len, page);
	}

	_leave(" = %d", ret);
	return ret;
}

/*
 * prepare to perform part of a write to a page
 * - the caller holds the page locked, preventing it from being written out or
 *   modified by anyone else
 */
int afs_prepare_write(struct file *file, struct page *page,
		      unsigned offset, unsigned to)
{
	struct afs_writeback *candidate, *wb;
	struct afs_vnode *vnode = AFS_FS_I(file->f_dentry->d_inode);
	struct key *key = file->private_data;
	pgoff_t index;
	int ret;

	_enter("{%x:%u},{%lx},%u,%u",
	       vnode->fid.vid, vnode->fid.vnode, page->index, offset, to);

	candidate = kzalloc(sizeof(*candidate), GFP_KERNEL);
	if (!candidate)
		return -ENOMEM;
	candidate->vnode = vnode;
	candidate->first = candidate->last = page->index;
	candidate->offset_first = offset;
	candidate->to_last = to;
	candidate->usage = 1;
	candidate->state = AFS_WBACK_PENDING;
	init_waitqueue_head(&candidate->waitq);

	if (!PageUptodate(page)) {
		_debug("not up to date");
		ret = afs_prepare_page(vnode, page, key, offset, to);
		if (ret < 0) {
			kfree(candidate);
			_leave(" = %d [prep]", ret);
			return ret;
		}
	}

try_again:
	index = page->index;
	spin_lock(&vnode->writeback_lock);

	/* see if this page is already pending a writeback under a suitable key
	 * - if so we can just join onto that one */
	wb = (struct afs_writeback *) page_private(page);
	if (wb) {
		if (wb->key == key && wb->state == AFS_WBACK_PENDING)
			goto subsume_in_current_wb;
		goto flush_conflicting_wb;
	}

	if (index > 0) {
		/* see if we can find an already pending writeback that we can
		 * append this page to */
		list_for_each_entry(wb, &vnode->writebacks, link) {
			if (wb->last == index - 1 && wb->key == key &&
			    wb->state == AFS_WBACK_PENDING)
				goto append_to_previous_wb;
		}
	}

	list_add_tail(&candidate->link, &vnode->writebacks);
	candidate->key = key_get(key);
	spin_unlock(&vnode->writeback_lock);
	SetPagePrivate(page);
	set_page_private(page, (unsigned long) candidate);
	_leave(" = 0 [new]");
	return 0;

subsume_in_current_wb:
	_debug("subsume");
	ASSERTRANGE(wb->first, <=, index, <=, wb->last);
	if (index == wb->first && offset < wb->offset_first)
		wb->offset_first = offset;
	if (index == wb->last && to > wb->to_last)
		wb->to_last = to;
	spin_unlock(&vnode->writeback_lock);
	kfree(candidate);
	_leave(" = 0 [sub]");
	return 0;

append_to_previous_wb:
	_debug("append into %lx-%lx", wb->first, wb->last);
	wb->usage++;
	wb->last++;
	wb->to_last = to;
	spin_unlock(&vnode->writeback_lock);
	SetPagePrivate(page);
	set_page_private(page, (unsigned long) wb);
	kfree(candidate);
	_leave(" = 0 [app]");
	return 0;

	/* the page is currently bound to another context, so if it's dirty we
	 * need to flush it before we can use the new context */
flush_conflicting_wb:
	_debug("flush conflict");
	if (wb->state == AFS_WBACK_PENDING)
		wb->state = AFS_WBACK_CONFLICTING;
	spin_unlock(&vnode->writeback_lock);
	if (PageDirty(page)) {
		ret = afs_write_back_from_locked_page(wb, page);
		if (ret < 0) {
			afs_put_writeback(candidate);
			_leave(" = %d", ret);
			return ret;
		}
	}

	/* the page holds a ref on the writeback record */
	afs_put_writeback(wb);
	set_page_private(page, 0);
	ClearPagePrivate(page);
	goto try_again;
}

/*
 * finalise part of a write to a page
 */
int afs_commit_write(struct file *file, struct page *page,
		     unsigned offset, unsigned to)
{
	struct afs_vnode *vnode = AFS_FS_I(file->f_dentry->d_inode);
	loff_t i_size, maybe_i_size;

	_enter("{%x:%u},{%lx},%u,%u",
	       vnode->fid.vid, vnode->fid.vnode, page->index, offset, to);

	maybe_i_size = (loff_t) page->index << PAGE_SHIFT;
	maybe_i_size += to;

	i_size = i_size_read(&vnode->vfs_inode);
	if (maybe_i_size > i_size) {
		spin_lock(&vnode->writeback_lock);
		i_size = i_size_read(&vnode->vfs_inode);
		if (maybe_i_size > i_size)
			i_size_write(&vnode->vfs_inode, maybe_i_size);
		spin_unlock(&vnode->writeback_lock);
	}

	SetPageUptodate(page);
	set_page_dirty(page);
	if (PageDirty(page))
		_debug("dirtied");

	return 0;
}

/*
 * kill all the pages in the given range
 */
static void afs_kill_pages(struct afs_vnode *vnode, bool error,
			   pgoff_t first, pgoff_t last)
{
	struct pagevec pv;
	unsigned count, loop;

	_enter("{%x:%u},%lx-%lx",
	       vnode->fid.vid, vnode->fid.vnode, first, last);

	pagevec_init(&pv, 0);

	do {
		_debug("kill %lx-%lx", first, last);

		count = last - first + 1;
		if (count > PAGEVEC_SIZE)
			count = PAGEVEC_SIZE;
		pv.nr = find_get_pages_contig(vnode->vfs_inode.i_mapping,
					      first, count, pv.pages);
		ASSERTCMP(pv.nr, ==, count);

		for (loop = 0; loop < count; loop++) {
			ClearPageUptodate(pv.pages[loop]);
			if (error)
				SetPageError(pv.pages[loop]);
			end_page_writeback(pv.pages[loop]);
		}

		__pagevec_release(&pv);
	} while (first < last);

	_leave("");
}

/*
 * synchronously write back the locked page and any subsequent non-locked dirty
 * pages also covered by the same writeback record
 */
static int afs_write_back_from_locked_page(struct afs_writeback *wb,
					   struct page *primary_page)
{
	struct page *pages[8], *page;
	unsigned long count;
	unsigned n, offset, to;
	pgoff_t start, first, last;
	int loop, ret;

	_enter(",%lx", primary_page->index);

	count = 1;
	if (!clear_page_dirty_for_io(primary_page))
		BUG();
	if (test_set_page_writeback(primary_page))
		BUG();

	/* find all consecutive lockable dirty pages, stopping when we find a
	 * page that is not immediately lockable, is not dirty or is missing,
	 * or we reach the end of the range */
	start = primary_page->index;
	if (start >= wb->last)
		goto no_more;
	start++;
	do {
		_debug("more %lx [%lx]", start, count);
		n = wb->last - start + 1;
		if (n > ARRAY_SIZE(pages))
			n = ARRAY_SIZE(pages);
		n = find_get_pages_contig(wb->vnode->vfs_inode.i_mapping,
					  start, n, pages);
		_debug("fgpc %u", n);
		if (n == 0)
			goto no_more;
		if (pages[0]->index != start) {
			do {
				put_page(pages[--n]);
			} while (n > 0);
			goto no_more;
		}

		for (loop = 0; loop < n; loop++) {
			page = pages[loop];
			if (page->index > wb->last)
				break;
			if (TestSetPageLocked(page))
				break;
			if (!PageDirty(page) ||
			    page_private(page) != (unsigned long) wb) {
				unlock_page(page);
				break;
			}
			if (!clear_page_dirty_for_io(page))