inode.c

/*
 * Copyright (C) 2005, 2006
 * Avishay Traeger (avishay@gmail.com)
 * Copyright (C) 2008, 2009
 * Boaz Harrosh <bharrosh@panasas.com>
 *
 * Copyrights for code taken from ext2:
 *     Copyright (C) 1992, 1993, 1994, 1995
 *     Remy Card (card@masi.ibp.fr)
 *     Laboratoire MASI - Institut Blaise Pascal
 *     Universite Pierre et Marie Curie (Paris VI)
 *     from
 *     linux/fs/minix/inode.c
 *     Copyright (C) 1991, 1992  Linus Torvalds
 *
 * This file is part of exofs.
 *
 * exofs 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.  Since it is based on ext2, and the only
 * valid version of GPL for the Linux kernel is version 2, the only valid
 * version of GPL for exofs is version 2.
 *
 * exofs is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with exofs; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */

#include <linux/slab.h>

#include "exofs.h"

#define EXOFS_DBGMSG2(M...) do {} while (0)

enum {MAX_PAGES_KMALLOC = PAGE_SIZE / sizeof(struct page *), };

unsigned exofs_max_io_pages(struct ore_layout *layout,
			    unsigned expected_pages)
{
	unsigned pages = min_t(unsigned, expected_pages, MAX_PAGES_KMALLOC);

	/* TODO: easily support bio chaining */
	pages =  min_t(unsigned, pages, layout->max_io_length / PAGE_SIZE);
	return pages;
}

struct page_collect {
	struct exofs_sb_info *sbi;
	struct inode *inode;
	unsigned expected_pages;
	struct ore_io_state *ios;

	struct page **pages;
	unsigned alloc_pages;
	unsigned nr_pages;
	unsigned long length;
	loff_t pg_first; /* keep 64bit also in 32-arches */
	bool read_4_write; /* This means two things: that the read is sync
			    * And the pages should not be unlocked.
			    */
	struct page *that_locked_page;
};

static void _pcol_init(struct page_collect *pcol, unsigned expected_pages,
		       struct inode *inode)
{
	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;

	pcol->sbi = sbi;
	pcol->inode = inode;
	pcol->expected_pages = expected_pages;

	pcol->ios = NULL;
	pcol->pages = NULL;
	pcol->alloc_pages = 0;
	pcol->nr_pages = 0;
	pcol->length = 0;
	pcol->pg_first = -1;
	pcol->read_4_write = false;
	pcol->that_locked_page = NULL;
}

static void _pcol_reset(struct page_collect *pcol)
{
	pcol->expected_pages -= min(pcol->nr_pages, pcol->expected_pages);

	pcol->pages = NULL;
	pcol->alloc_pages = 0;
	pcol->nr_pages = 0;
	pcol->length = 0;
	pcol->pg_first = -1;
	pcol->ios = NULL;
	pcol->that_locked_page = NULL;

	/* this is probably the end of the loop but in writes
	 * it might not end here. don't be left with nothing
	 */
	if (!pcol->expected_pages)
		pcol->expected_pages = MAX_PAGES_KMALLOC;
}

static int pcol_try_alloc(struct page_collect *pcol)
{
	unsigned pages;

	/* TODO: easily support bio chaining */
	pages =  exofs_max_io_pages(&pcol->sbi->layout, pcol->expected_pages);

	for (; pages; pages >>= 1) {
		pcol->pages = kmalloc(pages * sizeof(struct page *),
				      GFP_KERNEL);
		if (likely(pcol->pages)) {
			pcol->alloc_pages = pages;
			return 0;
		}
	}

	EXOFS_ERR("Failed to kmalloc expected_pages=%u\n",
		  pcol->expected_pages);
	return -ENOMEM;
}

static void pcol_free(struct page_collect *pcol)
{
	kfree(pcol->pages);
	pcol->pages = NULL;

	if (pcol->ios) {
		ore_put_io_state(pcol->ios);
		pcol->ios = NULL;
	}
}

static int pcol_add_page(struct page_collect *pcol, struct page *page,
			 unsigned len)
{
	if (unlikely(pcol->nr_pages >= pcol->alloc_pages))
		return -ENOMEM;

	pcol->pages[pcol->nr_pages++] = page;
	pcol->length += len;
	return 0;
}

enum {PAGE_WAS_NOT_IN_IO = 17};
static int update_read_page(struct page *page, int ret)
{
	switch (ret) {
	case 0:
		/* Everything is OK */
		SetPageUptodate(page);
		if (PageError(page))
			ClearPageError(page);
		break;
	case -EFAULT:
		/* In this case we were trying to read something that wasn't on
		 * disk yet - return a page full of zeroes.  This should be OK,
		 * because the object should be empty (if there was a write
		 * before this read, the read would be waiting with the page
		 * locked */
		clear_highpage(page);

		SetPageUptodate(page);
		if (PageError(page))
			ClearPageError(page);
		EXOFS_DBGMSG("recovered read error\n");
		/* fall through */
	case PAGE_WAS_NOT_IN_IO:
		ret = 0; /* recovered error */
		break;
	default:
		SetPageError(page);
	}
	return ret;
}

static void update_write_page(struct page *page, int ret)
{
	if (unlikely(ret == PAGE_WAS_NOT_IN_IO))
		return; /* don't pass start don't collect $200 */

	if (ret) {
		mapping_set_error(page->mapping, ret);
		SetPageError(page);
	}
	end_page_writeback(page);
}

/* Called at the end of reads, to optionally unlock pages and update their
 * status.
 */
static int __readpages_done(struct page_collect *pcol)
{
	int i;
	u64 good_bytes;
	u64 length = 0;
	int ret = ore_check_io(pcol->ios, NULL);

	if (likely(!ret)) {
		good_bytes = pcol->length;
		ret = PAGE_WAS_NOT_IN_IO;
	} else {
		good_bytes = 0;
	}

	EXOFS_DBGMSG2("readpages_done(0x%lx) good_bytes=0x%llx"
		     " length=0x%lx nr_pages=%u\n",
		     pcol->inode->i_ino, _LLU(good_bytes), pcol->length,
		     pcol->nr_pages);

	for (i = 0; i < pcol->nr_pages; i++) {
		struct page *page = pcol->pages[i];
		struct inode *inode = page->mapping->host;
		int page_stat;

		if (inode != pcol->inode)
			continue; /* osd might add more pages at end */

		if (likely(length < good_bytes))
			page_stat = 0;
		else
			page_stat = ret;

		EXOFS_DBGMSG2("    readpages_done(0x%lx, 0x%lx) %s\n",
			  inode->i_ino, page->index,
			  page_stat ? "bad_bytes" : "good_bytes");

		ret = update_read_page(page, page_stat);
		if (!pcol->read_4_write)
			unlock_page(page);
		length += PAGE_SIZE;
	}

	pcol_free(pcol);
	EXOFS_DBGMSG2("readpages_done END\n");
	return ret;
}

/* callback of async reads */
static void readpages_done(struct ore_io_state *ios, void *p)
{
	struct page_collect *pcol = p;

	__readpages_done(pcol);
	atomic_dec(&pcol->sbi->s_curr_pending);
	kfree(pcol);
}

static void _unlock_pcol_pages(struct page_collect *pcol, int ret, int rw)
{
	int i;

	for (i = 0; i < pcol->nr_pages; i++) {
		struct page *page = pcol->pages[i];

		if (rw == READ)
			update_read_page(page, ret);
		else
			update_write_page(page, ret);

		unlock_page(page);
	}
}

static int _maybe_not_all_in_one_io(struct ore_io_state *ios,
	struct page_collect *pcol_src, struct page_collect *pcol)
{
	/* length was wrong or offset was not page aligned */
	BUG_ON(pcol_src->nr_pages < ios->nr_pages);

	if (pcol_src->nr_pages > ios->nr_pages) {
		struct page **src_page;
		unsigned pages_less = pcol_src->nr_pages - ios->nr_pages;
		unsigned long len_less = pcol_src->length - ios->length;
		unsigned i;
		int ret;

		/* This IO was trimmed */
		pcol_src->nr_pages = ios->nr_pages;
		pcol_src->length = ios->length;

		/* Left over pages are passed to the next io */
		pcol->expected_pages += pages_less;
		pcol->nr_pages = pages_less;
		pcol->length = len_less;
		src_page = pcol_src->pages + pcol_src->nr_pages;
		pcol->pg_first = (*src_page)->index;

		ret = pcol_try_alloc(pcol);
		if (unlikely(ret))
			return ret;

		for (i = 0; i < pages_less; ++i)
			pcol->pages[i] = *src_page++;

		EXOFS_DBGMSG("Length was adjusted nr_pages=0x%x "
			"pages_less=0x%x expected_pages=0x%x "
			"next_offset=0x%llx next_len=0x%lx\n",
			pcol_src->nr_pages, pages_less, pcol->expected_pages,
			pcol->pg_first * PAGE_SIZE, pcol->length);
	}
	return 0;
}

static int read_exec(struct page_collect *pcol)
{
	struct exofs_i_info *oi = exofs_i(pcol->inode);
	struct ore_io_state *ios;
	struct page_collect *pcol_copy = NULL;
	int ret;

	if (!pcol->pages)
		return 0;

	if (!pcol->ios) {
		int ret = ore_get_rw_state(&pcol->sbi->layout, &oi->oc, true,
					     pcol->pg_first << PAGE_CACHE_SHIFT,
					     pcol->length, &pcol->ios);

		if (ret)
			return ret;
	}

	ios = pcol->ios;
	ios->pages = pcol->pages;

	if (pcol->read_4_write) {
		ore_read(pcol->ios);
		return __readpages_done(pcol);
	}

	pcol_copy = kmalloc(sizeof(*pcol_copy), GFP_KERNEL);
	if (!pcol_copy) {
		ret = -ENOMEM;
		goto err;
	}

	*pcol_copy = *pcol;
	ios->done = readpages_done;
	ios->private = pcol_copy;

	/* pages ownership was passed to pcol_copy */
	_pcol_reset(pcol);

	ret = _maybe_not_all_in_one_io(ios, pcol_copy, pcol);
	if (unlikely(ret))
		goto err;

	EXOFS_DBGMSG2("read_exec(0x%lx) offset=0x%llx length=0x%llx\n",
		pcol->inode->i_ino, _LLU(ios->offset), _LLU(ios->length));

	ret = ore_read(ios);
	if (unlikely(ret))
		goto err;

	atomic_inc(&pcol->sbi->s_curr_pending);

	return 0;

err:
	if (!pcol->read_4_write)
		_unlock_pcol_pages(pcol, ret, READ);

	pcol_free(pcol);

	kfree(pcol_copy);
	return ret;
}

/* readpage_strip is called either directly from readpage() or by the VFS from
 * within read_cache_pages(), to add one more page to be read. It will try to
 * collect as many contiguous pages as posible. If a discontinuity is
 * encountered, or it runs out of resources, it will submit the previous segment
 * and will start a new collection. Eventually caller must submit the last
 * segment if present.
 */
static int readpage_strip(void *data, struct page *page)
{
	struct page_collect *pcol = data;
	struct inode *inode = pcol->inode;
	struct exofs_i_info *oi = exofs_i(inode);
	loff_t i_size = i_size_read(inode);
	pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT;
	size_t len;
	int ret;

	/* FIXME: Just for debugging, will be removed */
	if (PageUptodate(page))
		EXOFS_ERR("PageUptodate(0x%lx, 0x%lx)\n", pcol->inode->i_ino,
			  page->index);

	pcol->that_locked_page = page;

	if (page->index < end_index)
		len = PAGE_CACHE_SIZE;
	else if (page->index == end_index)
		len = i_size & ~PAGE_CACHE_MASK;
	else
		len = 0;

	if (!len || !obj_created(oi)) {
		/* this will be out of bounds, or doesn't exist yet.
		 * Current page is cleared and the request is split
		 */
		clear_highpage(page);

		SetPageUptodate(page);
		if (PageError(page))
			ClearPageError(page);

		if (!pcol->read_4_write)
			unlock_page(page);
		EXOFS_DBGMSG("readpage_strip(0x%lx) empty page len=%zx "
			     "read_4_write=%d index=0x%lx end_index=0x%lx "
			     "splitting\n", inode->i_ino, len,
			     pcol->read_4_write, page->index, end_index);

		return read_exec(pcol);
	}

try_again:

	if (unlikely(pcol->pg_first == -1)) {
		pcol->pg_first = page->index;
	} else if (unlikely((pcol->pg_first + pcol->nr_pages) !=
		   page->index)) {
		/* Discontinuity detected, split the request */
		ret = read_exec(pcol);
		if (unlikely(ret))
			goto fail;
		goto try_again;
	}

	if (!pcol->pages) {
		ret = pcol_try_alloc(pcol);
		if (unlikely(ret))
			goto fail;
	}

	if (len != PAGE_CACHE_SIZE)
		zero_user(page, len, PAGE_CACHE_SIZE - len);

	EXOFS_DBGMSG2("    readpage_strip(0x%lx, 0x%lx) len=0x%zx\n",
		     inode->i_ino, page->index, len);

	ret = pcol_add_page(pcol, page, len);
	if (ret) {
		EXOFS_DBGMSG2("Failed pcol_add_page pages[i]=%p "
			  "this_len=0x%zx nr_pages=%u length=0x%lx\n",
			  page, len, pcol->nr_pages, pcol->length);

		/* split the request, and start again with current page */
		ret = read_exec(pcol);
		if (unlikely(ret))
			goto fail;

		goto try_again;
	}

	return 0;

fail:
	/* SetPageError(page); ??? */
	unlock_page(page);
	return ret;
}

static int exofs_readpages(struct file *file, struct address_space *mapping,
			   struct list_head *pages, unsigned nr_pages)
{
	struct page_collect pcol;
	int ret;

	_pcol_init(&pcol, nr_pages, mapping->host);

	ret = read_cache_pages(mapping, pages, readpage_strip, &pcol);
	if (ret) {
		EXOFS_ERR("read_cache_pages => %d\n", ret);
		return ret;
	}

	ret = read_exec(&pcol);
	if (unlikely(ret))
		return ret;

	return read_exec(&pcol);
}

static int _readpage(struct page *page, bool read_4_write)
{
	struct page_collect pcol;
	int ret;

	_pcol_init(&pcol, 1, page->mapping->host);

	pcol.read_4_write = read_4_write;
	ret = readpage_strip(&pcol, page);
	if (ret) {
		EXOFS_ERR("_readpage => %d\n", ret);
		return ret;
	}

	return read_exec(&pcol);
}

/*
 * We don't need the file
 */
static int exofs_readpage(struct file *file, struct page *page)
{
	return _readpage(page, false);
}

/* Callback for osd_write. All writes are asynchronous */
static void writepages_done(struct ore_io_state *ios, void *p)
{
	struct page_collect *pcol = p;
	int i;
	u64  good_bytes;
	u64  length = 0;
	int ret = ore_check_io(ios, NULL);

	atomic_dec(&pcol->sbi->s_curr_pending);

	if (likely(!ret)) {
		good_bytes = pcol->length;
		ret = PAGE_WAS_NOT_IN_IO;
	} else {
		good_bytes = 0;
	}

	EXOFS_DBGMSG2("writepages_done(0x%lx) good_bytes=0x%llx"
		     " length=0x%lx nr_pages=%u\n",
		     pcol->inode->i_ino, _LLU(good_bytes), pcol->length,
		     pcol->nr_pages);

	for (i = 0; i < pcol->nr_pages; i++) {
		struct page *page = pcol->pages[i];
		struct inode *inode = page->mapping->host;
		int page_stat;

		if (inode != pcol->inode)
			continue; /* osd might add more pages to a bio */

		if (likely(length < good_bytes))
			page_stat = 0;
		else
			page_stat = ret;

		update_write_page(page, page_stat);
		unlock_page(page);
		EXOFS_DBGMSG2("    writepages_done(0x%lx, 0x%lx) status=%d\n",
			     inode->i_ino, page->index, page_stat);

		length += PAGE_SIZE;
	}

	pcol_free(pcol);
	kfree(pcol);
	EXOFS_DBGMSG2("writepages_done END\n");
}

static struct page *__r4w_get_page(void *priv, u64 offset, bool *uptodate)
{
	struct page_collect *pcol = priv;
	pgoff_t index = offset / PAGE_SIZE;

	if (!pcol->that_locked_page ||
	    (pcol->that_locked_page->index != index)) {
		struct page *page = find_get_page(pcol->inode->i_mapping, index);

		if (!page) {
			page = find_or_create_page(pcol->inode->i_mapping,
						   index, GFP_NOFS);
			if (unlikely(!page)) {
				EXOFS_DBGMSG("grab_cache_page Failed "
					"index=0x%llx\n", _LLU(index));
				return NULL;
			}
			unlock_page(page);
		}
		if (PageDirty(page) || PageWriteback(page))
			*uptodate = true;
		else
			*uptodate = PageUptodate(page);
		EXOFS_DBGMSG("index=0x%lx uptodate=%d\n", index, *uptodate);
		return page;
	} else {
		EXOFS_DBGMSG("YES that_locked_page index=0x%lx\n",
			     pcol->that_locked_page->index);
		*uptodate = true;
		return pcol->that_locked_page;
	}
}

static void __r4w_put_page(void *priv, struct page *page)
{
	struct page_collect *pcol = priv;

	if (pcol->that_locked_page != page) {
		EXOFS_DBGMSG("index=0x%lx\n", page->index);
		page_cache_release(page);
		return;
	}
	EXOFS_DBGMSG("that_locked_page index=0x%lx\n", page->index);
}

static const struct _ore_r4w_op _r4w_op = {
	.get_page = &__r4w_get_page,
	.put_page = &__r4w_put_page,
};

static int write_exec(struct page_collect *pcol)
{
	struct exofs_i_info *oi = exofs_i(pcol->inode);
	struct ore_io_state *ios;
	struct page_collect *pcol_copy = NULL;
	int ret;

	if (!pcol->pages)
		return 0;

	BUG_ON(pcol->ios);
	ret = ore_get_rw_state(&pcol->sbi->layout, &oi->oc, false,
				 pcol->pg_first << PAGE_CACHE_SHIFT,
				 pcol->length, &pcol->ios);
	if (unlikely(ret))
		goto err;

	pcol_copy = kmalloc(sizeof(*pcol_copy), GFP_KERNEL);
	if (!pcol_copy) {
		EXOFS_ERR("write_exec: Failed to kmalloc(pcol)\n");
		ret = -ENOMEM;
		goto err;
	}

	*pcol_copy = *pcol;

	ios = pcol->ios;
	ios->pages = pcol_copy->pages;
	ios->done = writepages_done;
	ios->r4w = &_r4w_op;
	ios->private = pcol_copy;

	/* pages ownership was passed to pcol_copy */
	_pcol_reset(pcol);

	ret = _maybe_not_all_in_one_io(ios, pcol_copy, pcol);
	if (unlikely(ret))
		goto err;

	EXOFS_DBGMSG2("write_exec(0x%lx) offset=0x%llx length=0x%llx\n",
		pcol->inode->i_ino, _LLU(ios->offset), _LLU(ios->length));

	ret = ore_write(ios);
	if (unlikely(ret)) {
		EXOFS_ERR("write_exec: ore_write() Failed\n");
		goto err;
	}

	atomic_inc(&pcol->sbi->s_curr_pending);
	return 0;

err:
	_unlock_pcol_pages(pcol, ret, WRITE);
	pcol_free(pcol);
	kfree(pcol_copy);

	return ret;
}

/* writepage_strip is called either directly from writepage() or by the VFS from
 * within write_cache_pages(), to add one more page to be written to storage.
 * It will try to collect as many contiguous pages as possible. If a
 * discontinuity is encountered or it runs out of resources it will submit the
 * previous segment and will start a new collection.
 * Eventually caller must submit the last segment if present.
 */
static int writepage_strip(struct page *page,
			   struct writeback_control *wbc_unused, void *data)
{
	struct page_collect *pcol = data;
	struct inode *inode = pcol->inode;
	struct exofs_i_info *oi = exofs_i(inode);
	loff_t i_size = i_size_read(inode);
	pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT;
	size_t len;
	int ret;

	BUG_ON(!PageLocked(page));

	ret = wait_obj_created(oi);
	if (unlikely(ret))
		goto fail;

	if (page->index < end_index)
		/* in this case, the page is within the limits of the file */
		len = PAGE_CACHE_SIZE;
	else {
		len = i_size & ~PAGE_CACHE_MASK;

		if (page->index > end_index || !len) {
			/* in this case, the page is outside the limits
			 * (truncate in progress)
			 */
			ret = write_exec(pcol);
			if (unlikely(ret))
				goto fail;
			if (PageError(page))
				ClearPageError(page);
			unlock_page(page);
			EXOFS_DBGMSG("writepage_strip(0x%lx, 0x%lx) "
				     "outside the limits\n",
				     inode->i_ino, page->index);
			return 0;
		}
	}

try_again:

	if (unlikely(pcol->pg_first == -1)) {
		pcol->pg_first = page->index;
	} else if (unlikely((pcol->pg_first + pcol->nr_pages) !=
		   page->index)) {
		/* Discontinuity detected, split the request */
		ret = write_exec(pcol);
		if (unlikely(ret))
			goto fail;

		EXOFS_DBGMSG("writepage_strip(0x%lx, 0x%lx) Discontinuity\n",
			     inode->i_ino, page->index);
		goto try_again;
	}

	if (!pcol->pages) {
		ret = pcol_try_alloc(pcol);
		if (unlikely(ret))
			goto fail;
	}

	EXOFS_DBGMSG2("    writepage_strip(0x%lx, 0x%lx) len=0x%zx\n",
		     inode->i_ino, page->index, len);

	ret = pcol_add_page(pcol, page, len);
	if (unlikely(ret)) {
		EXOFS_DBGMSG2("Failed pcol_add_page "
			     "nr_pages=%u total_length=0x%lx\n",
			     pcol->nr_pages, pcol->length);

		/* split the request, next loop will start again */
		ret = write_exec(pcol);
		if (unlikely(ret)) {
			EXOFS_DBGMSG("write_exec failed => %d", ret);
			goto fail;
		}

		goto try_again;
	}

	BUG_ON(PageWriteback(page));
	set_page_writeback(page);

	return 0;

fail:
	EXOFS_DBGMSG("Error: writepage_strip(0x%lx, 0x%lx)=>%d\n",
		     inode->i_ino, page->index, ret);
	set_bit(AS_EIO, &page->mapping->flags);
	unlock_page(page);
	return ret;
}

static int exofs_writepages(struct address_space *mapping,
		       struct writeback_control *wbc)
{
	struct page_collect pcol;
	long start, end, expected_pages;
	int ret;

	start = wbc->range_start >> PAGE_CACHE_SHIFT;
	end = (wbc->range_end == LLONG_MAX) ?
			start + mapping->nrpages :
			wbc->range_end >> PAGE_CACHE_SHIFT;

	if (start || end)
		expected_pages = end - start + 1;
	else
		expected_pages = mapping->nrpages;

	if (expected_pages < 32L)
		expected_pages = 32L;

	EXOFS_DBGMSG2("inode(0x%lx) wbc->start=0x%llx wbc->end=0x%llx "
		     "nrpages=%lu start=0x%lx end=0x%lx expected_pages=%ld\n",
		     mapping->host->i_ino, wbc->range_start, wbc->range_end,
		     mapping->nrpages, start, end, expected_pages);

	_pcol_init(&pcol, expected_pages, mapping->host);

	ret = write_cache_pages(mapping, wbc, writepage_strip, &pcol);
	if (unlikely(ret)) {
		EXOFS_ERR("write_cache_pages => %d\n", ret);
		return ret;
	}

	ret = write_exec(&pcol);
	if (unlikely(ret))
		return ret;

	if (wbc->sync_mode == WB_SYNC_ALL) {
		return write_exec(&pcol); /* pump the last reminder */
	} else if (pcol.nr_pages) {
		/* not SYNC let the reminder join the next writeout */
		unsigned i;

		for (i = 0; i < pcol.nr_pages; i++) {
			struct page *page = pcol.pages[i];

			end_page_writeback(page);
			set_page_dirty(page);
			unlock_page(page);
		}
	}
	return 0;
}

/*
static int exofs_writepage(struct page *page, struct writeback_control *wbc)
{
	struct page_collect pcol;
	int ret;

	_pcol_init(&pcol, 1, page->mapping->host);

	ret = writepage_strip(page, NULL, &pcol);
	if (ret) {
		EXOFS_ERR("exofs_writepage => %d\n", ret);
		return ret;
	}

	return write_exec(&pcol);
}
*/
/* i_mutex held using inode->i_size directly */
static void _write_failed(struct inode *inode, loff_t to)
{
	if (to > inode->i_size)
		truncate_pagecache(inode, to, inode->i_size);
}

int exofs_write_begin(struct file *file, struct address_space *mapping,
		loff_t pos, unsigned len, unsigned flags,
		struct page **pagep, void **fsdata)
{
	int ret = 0;
	struct page *page;

	page = *pagep;
	if (page == NULL) {
		ret = simple_write_begin(file, mapping, pos, len, flags, pagep,
					 fsdata);
		if (ret) {
			EXOFS_DBGMSG("simple_write_begin failed\n");
			goto out;
		}

		page = *pagep;
	}

	 /* read modify write */
	if (!PageUptodate(page) && (len != PAGE_CACHE_SIZE)) {
		loff_t i_size = i_size_read(mapping->host);
		pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT;
		size_t rlen;

		if (page->index < end_index)
			rlen = PAGE_CACHE_SIZE;
		else if (page->index == end_index)
			rlen = i_size & ~PAGE_CACHE_MASK;
		else
			rlen = 0;

		if (!rlen) {
			clear_highpage(page);
			SetPageUptodate(page);
			goto out;
		}

		ret = _readpage(page, true);
		if (ret) {
			/*SetPageError was done by _readpage. Is it ok?*/
			unlock_page(page);
			EXOFS_DBGMSG("__readpage failed\n");
		}
	}
out:
	if (unlikely(ret))
		_write_failed(mapping->host, pos + len);

	return ret;
}

static int exofs_write_begin_export(struct file *file,
		struct address_space *mapping,
		loff_t pos, unsigned len, unsigned flags,
		struct page **pagep, void **fsdata)
{
	*pagep = NULL;

	return exofs_write_begin(file, mapping, pos, len, flags, pagep,
					fsdata);
}

static int exofs_write_end(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned copied,
			struct page *page, void *fsdata)
{
	struct inode *inode = mapping->host;
	/* According to comment in simple_write_end i_mutex is held */
	loff_t i_size = inode->i_size;
	int ret;

	ret = simple_write_end(file, mapping,pos, len, copied, page, fsdata);
	if (unlikely(ret))
		_write_failed(inode, pos + len);

	/* TODO: once simple_write_end marks inode dirty remove */
	if (i_size != inode->i_size)
		mark_inode_dirty(inode);
	return ret;
}

static int exofs_releasepage(struct page *page, gfp_t gfp)
{
	EXOFS_DBGMSG("page 0x%lx\n", page->index);
	WARN_ON(1);
	return 0;
}

static void exofs_invalidatepage(struct page *page, unsigned long offset)
{
	EXOFS_DBGMSG("page 0x%lx offset 0x%lx\n", page->index, offset);
	WARN_ON(1);
}

const struct address_space_operations exofs_aops = {
	.readpage	= exofs_readpage,
	.readpages	= exofs_readpages,
	.writepage	= NULL,
	.writepages	= exofs_writepages,
	.write_begin	= exofs_write_begin_export,
	.write_end	= exofs_write_end,
	.releasepage	= exofs_releasepage,
	.set_page_dirty	= __set_page_dirty_nobuffers,
	.invalidatepage = exofs_invalidatepage,

	/* Not implemented Yet */
	.bmap		= NULL, /* TODO: use osd's OSD_ACT_READ_MAP */
	.direct_IO	= NULL, /* TODO: Should be trivial to do */

	/* With these NULL has special meaning or default is not exported */
	.get_xip_mem	= NULL,
	.migratepage	= NULL,
	.launder_page	= NULL,
	.is_partially_uptodate = NULL,
	.error_remove_page = NULL,
};

/******************************************************************************
 * INODE OPERATIONS
 *****************************************************************************/

/*
 * Test whether an inode is a fast symlink.
 */
static inline int exofs_inode_is_fast_symlink(struct inode *inode)
{
	struct exofs_i_info *oi = exofs_i(inode);

	return S_ISLNK(inode->i_mode) && (oi->i_data[0] != 0);
}

static int _do_truncate(struct inode *inode, loff_t newsize)
{
	struct exofs_i_info *oi = exofs_i(inode);
	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
	int ret;

	inode->i_mtime = inode->i_ctime = CURRENT_TIME;

	ret = ore_truncate(&sbi->layout, &oi->oc, (u64)newsize);
	if (likely(!ret))
		truncate_setsize(inode, newsize);

	EXOFS_DBGMSG("(0x%lx) size=0x%llx ret=>%d\n",