1 files changed, 330 insertions, 145 deletions

diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c
index a7555238c41..3f9cafd7393 100644
--- a/fs/exofs/inode.c
+++ b/fs/exofs/inode.c
@@ -37,17 +37,20 @@
 
 #define EXOFS_DBGMSG2(M...) do {} while (0)
 
-enum { BIO_MAX_PAGES_KMALLOC =
-		(PAGE_SIZE - sizeof(struct bio)) / sizeof(struct bio_vec),
-	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,
+			       layout->max_io_length / PAGE_SIZE);
+
+	return pages;
+}
 
 struct page_collect {
 	struct exofs_sb_info *sbi;
 	struct inode *inode;
 	unsigned expected_pages;
-	struct exofs_io_state *ios;
+	struct ore_io_state *ios;
 
 	struct page **pages;
 	unsigned alloc_pages;
@@ -57,6 +60,7 @@ struct page_collect {
 	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,
@@ -75,6 +79,7 @@ static void _pcol_init(struct page_collect *pcol, unsigned expected_pages,
 	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)
@@ -87,29 +92,22 @@ static void _pcol_reset(struct page_collect *pcol)
 	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;
+		pcol->expected_pages =
+				exofs_max_io_pages(&pcol->sbi->layout, ~0);
 }
 
 static int pcol_try_alloc(struct page_collect *pcol)
 {
-	unsigned pages = min_t(unsigned, pcol->expected_pages,
-			  MAX_PAGES_KMALLOC);
-
-	if (!pcol->ios) { /* First time allocate io_state */
-		int ret = exofs_get_io_state(&pcol->sbi->layout, &pcol->ios);
-
-		if (ret)
-			return ret;
-	}
+	unsigned pages;
 
 	/* TODO: easily support bio chaining */
-	pages =  min_t(unsigned, pages,
-		       pcol->sbi->layout.group_width * BIO_MAX_PAGES_KMALLOC);
+	pages =  exofs_max_io_pages(&pcol->sbi->layout, pcol->expected_pages);
 
 	for (; pages; pages >>= 1) {
 		pcol->pages = kmalloc(pages * sizeof(struct page *),
@@ -131,7 +129,7 @@ static void pcol_free(struct page_collect *pcol)
 	pcol->pages = NULL;
 
 	if (pcol->ios) {
-		exofs_put_io_state(pcol->ios);
+		ore_put_io_state(pcol->ios);
 		pcol->ios = NULL;
 	}
 }
@@ -147,14 +145,17 @@ static int pcol_add_page(struct page_collect *pcol, struct page *page,
 	return 0;
 }
 
+enum {PAGE_WAS_NOT_IN_IO = 17};
 static int update_read_page(struct page *page, int ret)
 {
-	if (ret == 0) {
+	switch (ret) {
+	case 0:
 		/* Everything is OK */
 		SetPageUptodate(page);
 		if (PageError(page))
 			ClearPageError(page);
-	} else if (ret == -EFAULT) {
+		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
@@ -165,16 +166,22 @@ static int update_read_page(struct page *page, int ret)
 		SetPageUptodate(page);
 		if (PageError(page))
 			ClearPageError(page);
-		ret = 0; /* recovered error */
 		EXOFS_DBGMSG("recovered read error\n");
-	} else /* Error */
+		/* 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);
@@ -188,15 +195,16 @@ static void update_write_page(struct page *page, int ret)
 static int __readpages_done(struct page_collect *pcol)
 {
 	int i;
-	u64 resid;
 	u64 good_bytes;
 	u64 length = 0;
-	int ret = exofs_check_io(pcol->ios, &resid);
+	int ret = ore_check_io(pcol->ios, NULL);
 
-	if (likely(!ret))
+	if (likely(!ret)) {
 		good_bytes = pcol->length;
-	else
-		good_bytes = pcol->length - resid;
+		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",
@@ -232,7 +240,7 @@ static int __readpages_done(struct page_collect *pcol)
 }
 
 /* callback of async reads */
-static void readpages_done(struct exofs_io_state *ios, void *p)
+static void readpages_done(struct ore_io_state *ios, void *p)
 {
 	struct page_collect *pcol = p;
 
@@ -257,23 +265,70 @@ static void _unlock_pcol_pages(struct page_collect *pcol, int ret, int rw)
 	}
 }
 
+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 exofs_io_state *ios = pcol->ios;
+	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;
-	ios->nr_pages = pcol->nr_pages;
-	ios->length = pcol->length;
-	ios->offset = pcol->pg_first << PAGE_CACHE_SHIFT;
 
 	if (pcol->read_4_write) {
-		exofs_oi_read(oi, pcol->ios);
+		ore_read(pcol->ios);
 		return __readpages_done(pcol);
 	}
 
@@ -286,26 +341,32 @@ static int read_exec(struct page_collect *pcol)
 	*pcol_copy = *pcol;
 	ios->done = readpages_done;
 	ios->private = pcol_copy;
-	ret = exofs_oi_read(oi, ios);
+
+	/* 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;
 
-	atomic_inc(&pcol->sbi->s_curr_pending);
+	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;
 
-	EXOFS_DBGMSG2("read_exec obj=0x%llx start=0x%llx length=0x%lx\n",
-		  ios->obj.id, _LLU(ios->offset), pcol->length);
+	atomic_inc(&pcol->sbi->s_curr_pending);
 
-	/* pages ownership was passed to pcol_copy */
-	_pcol_reset(pcol);
 	return 0;
 
 err:
-	if (!pcol->read_4_write)
-		_unlock_pcol_pages(pcol, ret, READ);
-
-	pcol_free(pcol);
-
+	if (!pcol_copy) /* Failed before ownership transfer */
+		pcol_copy = pcol;
+	_unlock_pcol_pages(pcol_copy, ret, READ);
+	pcol_free(pcol_copy);
 	kfree(pcol_copy);
+
 	return ret;
 }
 
@@ -326,11 +387,15 @@ static int readpage_strip(void *data, struct page *page)
 	size_t len;
 	int ret;
 
+	BUG_ON(!PageLocked(page));
+
 	/* 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)
@@ -350,8 +415,10 @@ static int readpage_strip(void *data, struct page *page)
 
 		if (!pcol->read_4_write)
 			unlock_page(page);
-		EXOFS_DBGMSG("readpage_strip(0x%lx, 0x%lx) empty page,"
-			     " splitting\n", inode->i_ino, page->index);
+		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);
 	}
@@ -417,6 +484,10 @@ static int exofs_readpages(struct file *file, struct address_space *mapping,
 		return ret;
 	}
 
+	ret = read_exec(&pcol);
+	if (unlikely(ret))
+		return ret;
+
 	return read_exec(&pcol);
 }
 
@@ -446,21 +517,22 @@ static int exofs_readpage(struct file *file, struct page *page)
 }
 
 /* Callback for osd_write. All writes are asynchronous */
-static void writepages_done(struct exofs_io_state *ios, void *p)
+static void writepages_done(struct ore_io_state *ios, void *p)
 {
 	struct page_collect *pcol = p;
 	int i;
-	u64 resid;
 	u64  good_bytes;
 	u64  length = 0;
-	int ret = exofs_check_io(ios, &resid);
+	int ret = ore_check_io(ios, NULL);
 
 	atomic_dec(&pcol->sbi->s_curr_pending);
 
-	if (likely(!ret))
+	if (likely(!ret)) {
 		good_bytes = pcol->length;
-	else
-		good_bytes = pcol->length - resid;
+		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",
@@ -493,16 +565,82 @@ static void writepages_done(struct exofs_io_state *ios, void *p)
 	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;
+		loff_t i_size = i_size_read(pcol->inode);
+
+		if (offset >= i_size) {
+			*uptodate = true;
+			EXOFS_DBGMSG2("offset >= i_size index=0x%lx\n", index);
+			return ZERO_PAGE(0);
+		}
+
+		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_DBGMSG2("index=0x%lx uptodate=%d\n", index, *uptodate);
+		return page;
+	} else {
+		EXOFS_DBGMSG2("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) && (ZERO_PAGE(0) != page)) {
+		EXOFS_DBGMSG2("index=0x%lx\n", page->index);
+		page_cache_release(page);
+		return;
+	}
+	EXOFS_DBGMSG2("that_locked_page index=0x%lx\n",
+		     ZERO_PAGE(0) == page ? -1 : 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 exofs_io_state *ios = pcol->ios;
+	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");
@@ -512,30 +650,36 @@ static int write_exec(struct page_collect *pcol)
 
 	*pcol_copy = *pcol;
 
+	ios = pcol->ios;
 	ios->pages = pcol_copy->pages;
-	ios->nr_pages = pcol_copy->nr_pages;
-	ios->offset = pcol_copy->pg_first << PAGE_CACHE_SHIFT;
-	ios->length = pcol_copy->length;
 	ios->done = writepages_done;
+	ios->r4w = &_r4w_op;
 	ios->private = pcol_copy;
 
-	ret = exofs_oi_write(oi, ios);
+	/* 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: exofs_oi_write() Failed\n");
+		EXOFS_ERR("write_exec: ore_write() Failed\n");
 		goto err;
 	}
 
 	atomic_inc(&pcol->sbi->s_curr_pending);
-	EXOFS_DBGMSG2("write_exec(0x%lx, 0x%llx) start=0x%llx length=0x%lx\n",
-		  pcol->inode->i_ino, pcol->pg_first, _LLU(ios->offset),
-		  pcol->length);
-	/* pages ownership was passed to pcol_copy */
-	_pcol_reset(pcol);
 	return 0;
 
 err:
-	_unlock_pcol_pages(pcol, ret, WRITE);
-	pcol_free(pcol);
+	if (!pcol_copy) /* Failed before ownership transfer */
+		pcol_copy = pcol;
+	_unlock_pcol_pages(pcol_copy, ret, WRITE);
+	pcol_free(pcol_copy);
 	kfree(pcol_copy);
 
 	return ret;
@@ -670,14 +814,33 @@ static int exofs_writepages(struct address_space *mapping,
 	_pcol_init(&pcol, expected_pages, mapping->host);
 
 	ret = write_cache_pages(mapping, wbc, writepage_strip, &pcol);
-	if (ret) {
+	if (unlikely(ret)) {
 		EXOFS_ERR("write_cache_pages => %d\n", ret);
 		return ret;
 	}
 
-	return write_exec(&pcol);
+	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;
@@ -693,12 +856,12 @@ static int exofs_writepage(struct page *page, struct writeback_control *wbc)
 
 	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);
+		truncate_pagecache(inode, inode->i_size);
 }
 
 int exofs_write_begin(struct file *file, struct address_space *mapping,
@@ -722,11 +885,28 @@ int exofs_write_begin(struct file *file, struct address_space *mapping,
 
 	 /* 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_filler failed\n");
+			EXOFS_DBGMSG("__readpage failed\n");
 		}
 	}
 out:
@@ -773,16 +953,26 @@ static int exofs_releasepage(struct page *page, gfp_t gfp)
 	return 0;
 }
 
-static void exofs_invalidatepage(struct page *page, unsigned long offset)
+static void exofs_invalidatepage(struct page *page, unsigned int offset,
+				 unsigned int length)
 {
-	EXOFS_DBGMSG("page 0x%lx offset 0x%lx\n", page->index, offset);
+	EXOFS_DBGMSG("page 0x%lx offset 0x%x length 0x%x\n",
+		     page->index, offset, length);
 	WARN_ON(1);
 }
 
+
+ /* TODO: Should be easy enough to do proprly */
+static ssize_t exofs_direct_IO(int rw, struct kiocb *iocb,
+		struct iov_iter *iter, loff_t offset)
+{
+	return 0;
+}
+
 const struct address_space_operations exofs_aops = {
 	.readpage	= exofs_readpage,
 	.readpages	= exofs_readpages,
-	.writepage	= exofs_writepage,
+	.writepage	= NULL,
 	.writepages	= exofs_writepages,
 	.write_begin	= exofs_write_begin_export,
 	.write_end	= exofs_write_end,
@@ -792,10 +982,9 @@ const struct address_space_operations exofs_aops = {
 
 	/* Not implemented Yet */
 	.bmap		= NULL, /* TODO: use osd's OSD_ACT_READ_MAP */
-	.direct_IO	= NULL, /* TODO: Should be trivial to do */
+	.direct_IO	= exofs_direct_IO,
 
 	/* With these NULL has special meaning or default is not exported */
-	.sync_page	= NULL,
 	.get_xip_mem	= NULL,
 	.migratepage	= NULL,
 	.launder_page	= NULL,
@@ -817,21 +1006,19 @@ static inline int exofs_inode_is_fast_symlink(struct inode *inode)
 	return S_ISLNK(inode->i_mode) && (oi->i_data[0] != 0);
 }
 
-const struct osd_attr g_attr_logical_length = ATTR_DEF(
-	OSD_APAGE_OBJECT_INFORMATION, OSD_ATTR_OI_LOGICAL_LENGTH, 8);
-
 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 = exofs_oi_truncate(oi, (u64)newsize);
+	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",
+	EXOFS_DBGMSG2("(0x%lx) size=0x%llx ret=>%d\n",
 		     inode->i_ino, newsize, ret);
 	return ret;
 }
@@ -890,43 +1077,38 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi,
 		[1] = g_attr_inode_file_layout,
 		[2] = g_attr_inode_dir_layout,
 	};
-	struct exofs_io_state *ios;
+	struct ore_io_state *ios;
 	struct exofs_on_disk_inode_layout *layout;
 	int ret;
 
-	ret = exofs_get_io_state(&sbi->layout, &ios);
+	ret = ore_get_io_state(&sbi->layout, &oi->oc, &ios);
 	if (unlikely(ret)) {
-		EXOFS_ERR("%s: exofs_get_io_state failed.\n", __func__);
+		EXOFS_ERR("%s: ore_get_io_state failed.\n", __func__);
 		return ret;
 	}
 
-	ios->obj.id = exofs_oi_objno(oi);
-	exofs_make_credential(oi->i_cred, &ios->obj);
-	ios->cred = oi->i_cred;
-
-	attrs[1].len = exofs_on_disk_inode_layout_size(sbi->layout.s_numdevs);
-	attrs[2].len = exofs_on_disk_inode_layout_size(sbi->layout.s_numdevs);
+	attrs[1].len = exofs_on_disk_inode_layout_size(sbi->oc.numdevs);
+	attrs[2].len = exofs_on_disk_inode_layout_size(sbi->oc.numdevs);
 
 	ios->in_attr = attrs;
 	ios->in_attr_len = ARRAY_SIZE(attrs);
 
-	ret = exofs_sbi_read(ios);
+	ret = ore_read(ios);
 	if (unlikely(ret)) {
 		EXOFS_ERR("object(0x%llx) corrupted, return empty file=>%d\n",
-			  _LLU(ios->obj.id), ret);
+			  _LLU(oi->one_comp.obj.id), ret);
 		memset(inode, 0, sizeof(*inode));
 		inode->i_mode = 0040000 | (0777 & ~022);
 		/* If object is lost on target we might as well enable it's
 		 * delete.
 		 */
-		if ((ret == -ENOENT) || (ret == -EINVAL))
-			ret = 0;
+		ret = 0;
 		goto out;
 	}
 
 	ret = extract_attr_from_ios(ios, &attrs[0]);
 	if (ret) {
-		EXOFS_ERR("%s: extract_attr of inode_data failed\n", __func__);
+		EXOFS_ERR("%s: extract_attr 0 of inode failed\n", __func__);
 		goto out;
 	}
 	WARN_ON(attrs[0].len != EXOFS_INO_ATTR_SIZE);
@@ -934,7 +1116,7 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi,
 
 	ret = extract_attr_from_ios(ios, &attrs[1]);
 	if (ret) {
-		EXOFS_ERR("%s: extract_attr of inode_data failed\n", __func__);
+		EXOFS_ERR("%s: extract_attr 1 of inode failed\n", __func__);
 		goto out;
 	}
 	if (attrs[1].len) {
@@ -949,7 +1131,7 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi,
 
 	ret = extract_attr_from_ios(ios, &attrs[2]);
 	if (ret) {
-		EXOFS_ERR("%s: extract_attr of inode_data failed\n", __func__);
+		EXOFS_ERR("%s: extract_attr 2 of inode failed\n", __func__);
 		goto out;
 	}
 	if (attrs[2].len) {
@@ -963,7 +1145,7 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi,
 	}
 
 out:
-	exofs_put_io_state(ios);
+	ore_put_io_state(ios);
 	return ret;
 }
 
@@ -989,6 +1171,8 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino)
 		return inode;
 	oi = exofs_i(inode);
 	__oi_init(oi);
+	exofs_init_comps(&oi->oc, &oi->one_comp, sb->s_fs_info,
+			 exofs_oi_objno(oi));
 
 	/* read the inode from the osd */
 	ret = exofs_get_inode(sb, oi, &fcb);
@@ -999,9 +1183,9 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino)
 
 	/* copy stuff from on-disk struct to in-memory struct */
 	inode->i_mode = le16_to_cpu(fcb.i_mode);
-	inode->i_uid = le32_to_cpu(fcb.i_uid);
-	inode->i_gid = le32_to_cpu(fcb.i_gid);
-	inode->i_nlink = le16_to_cpu(fcb.i_links_count);
+	i_uid_write(inode, le32_to_cpu(fcb.i_uid));
+	i_gid_write(inode, le32_to_cpu(fcb.i_gid));
+	set_nlink(inode, le16_to_cpu(fcb.i_links_count));
 	inode->i_ctime.tv_sec = (signed)le32_to_cpu(fcb.i_ctime);
 	inode->i_atime.tv_sec = (signed)le32_to_cpu(fcb.i_atime);
 	inode->i_mtime.tv_sec = (signed)le32_to_cpu(fcb.i_mtime);
@@ -1030,6 +1214,7 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino)
 		memcpy(oi->i_data, fcb.i_data, sizeof(fcb.i_data));
 	}
 
+	inode->i_mapping->backing_dev_info = sb->s_bdi;
 	if (S_ISREG(inode->i_mode)) {
 		inode->i_op = &exofs_file_inode_operations;
 		inode->i_fop = &exofs_file_operations;
@@ -1073,26 +1258,28 @@ int __exofs_wait_obj_created(struct exofs_i_info *oi)
 	}
 	return unlikely(is_bad_inode(&oi->vfs_inode)) ? -EIO : 0;
 }
+
 /*
  * Callback function from exofs_new_inode().  The important thing is that we
  * set the obj_created flag so that other methods know that the object exists on
  * the OSD.
  */
-static void create_done(struct exofs_io_state *ios, void *p)
+static void create_done(struct ore_io_state *ios, void *p)
 {
 	struct inode *inode = p;
 	struct exofs_i_info *oi = exofs_i(inode);
 	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
 	int ret;
 
-	ret = exofs_check_io(ios, NULL);
-	exofs_put_io_state(ios);
+	ret = ore_check_io(ios, NULL);
+	ore_put_io_state(ios);
 
 	atomic_dec(&sbi->s_curr_pending);
 
 	if (unlikely(ret)) {
 		EXOFS_ERR("object=0x%llx creation failed in pid=0x%llx",
-			  _LLU(exofs_oi_objno(oi)), _LLU(sbi->layout.s_pid));
+			  _LLU(exofs_oi_objno(oi)),
+			  _LLU(oi->one_comp.obj.partition));
 		/*TODO: When FS is corrupted creation can fail, object already
 		 * exist. Get rid of this asynchronous creation, if exist
 		 * increment the obj counter and try the next object. Until we
@@ -1109,16 +1296,15 @@ static void create_done(struct exofs_io_state *ios, void *p)
 /*
  * Set up a new inode and create an object for it on the OSD
  */
-struct inode *exofs_new_inode(struct inode *dir, int mode)
+struct inode *exofs_new_inode(struct inode *dir, umode_t mode)
 {
-	struct super_block *sb;
+	struct super_block *sb = dir->i_sb;
+	struct exofs_sb_info *sbi = sb->s_fs_info;
 	struct inode *inode;
 	struct exofs_i_info *oi;
-	struct exofs_sb_info *sbi;
-	struct exofs_io_state *ios;
+	struct ore_io_state *ios;
 	int ret;
 
-	sb = dir->i_sb;
 	inode = new_inode(sb);
 	if (!inode)
 		return ERR_PTR(-ENOMEM);
@@ -1128,9 +1314,7 @@ struct inode *exofs_new_inode(struct inode *dir, int mode)
 
 	set_obj_2bcreated(oi);
 
-	sbi = sb->s_fs_info;
-
-	sb->s_dirt = 1;
+	inode->i_mapping->backing_dev_info = sb->s_bdi;
 	inode_init_owner(inode, dir, mode);
 	inode->i_ino = sbi->s_nextid++;
 	inode->i_blkbits = EXOFS_BLKSHIFT;
@@ -1141,23 +1325,24 @@ struct inode *exofs_new_inode(struct inode *dir, int mode)
 	spin_unlock(&sbi->s_next_gen_lock);
 	insert_inode_hash(inode);
 
+	exofs_init_comps(&oi->oc, &oi->one_comp, sb->s_fs_info,
+			 exofs_oi_objno(oi));
+	exofs_sbi_write_stats(sbi); /* Make sure new sbi->s_nextid is on disk */
+
 	mark_inode_dirty(inode);
 
-	ret = exofs_get_io_state(&sbi->layout, &ios);
+	ret = ore_get_io_state(&sbi->layout, &oi->oc, &ios);
 	if (unlikely(ret)) {
-		EXOFS_ERR("exofs_new_inode: exofs_get_io_state failed\n");
+		EXOFS_ERR("exofs_new_inode: ore_get_io_state failed\n");
 		return ERR_PTR(ret);
 	}
 
-	ios->obj.id = exofs_oi_objno(oi);
-	exofs_make_credential(oi->i_cred, &ios->obj);
-
 	ios->done = create_done;
 	ios->private = inode;
-	ios->cred = oi->i_cred;
-	ret = exofs_sbi_create(ios);
+
+	ret = ore_create(ios);
 	if (ret) {
-		exofs_put_io_state(ios);
+		ore_put_io_state(ios);
 		return ERR_PTR(ret);
 	}
 	atomic_inc(&sbi->s_curr_pending);
@@ -1176,11 +1361,11 @@ struct updatei_args {
 /*
  * Callback function from exofs_update_inode().
  */
-static void updatei_done(struct exofs_io_state *ios, void *p)
+static void updatei_done(struct ore_io_state *ios, void *p)
 {
 	struct updatei_args *args = p;
 
-	exofs_put_io_state(ios);
+	ore_put_io_state(ios);
 
 	atomic_dec(&args->sbi->s_curr_pending);
 
@@ -1196,7 +1381,7 @@ static int exofs_update_inode(struct inode *inode, int do_sync)
 	struct exofs_i_info *oi = exofs_i(inode);
 	struct super_block *sb = inode->i_sb;
 	struct exofs_sb_info *sbi = sb->s_fs_info;
-	struct exofs_io_state *ios;
+	struct ore_io_state *ios;
 	struct osd_attr attr;
 	struct exofs_fcb *fcb;
 	struct updatei_args *args;
@@ -1211,8 +1396,8 @@ static int exofs_update_inode(struct inode *inode, int do_sync)
 	fcb = &args->fcb;
 
 	fcb->i_mode = cpu_to_le16(inode->i_mode);
-	fcb->i_uid = cpu_to_le32(inode->i_uid);
-	fcb->i_gid = cpu_to_le32(inode->i_gid);
+	fcb->i_uid = cpu_to_le32(i_uid_read(inode));
+	fcb->i_gid = cpu_to_le32(i_gid_read(inode));
 	fcb->i_links_count = cpu_to_le16(inode->i_nlink);
 	fcb->i_ctime = cpu_to_le32(inode->i_ctime.tv_sec);
 	fcb->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
@@ -1235,9 +1420,9 @@ static int exofs_update_inode(struct inode *inode, int do_sync)
 	} else
 		memcpy(fcb->i_data, oi->i_data, sizeof(fcb->i_data));
 
-	ret = exofs_get_io_state(&sbi->layout, &ios);
+	ret = ore_get_io_state(&sbi->layout, &oi->oc, &ios);
 	if (unlikely(ret)) {
-		EXOFS_ERR("%s: exofs_get_io_state failed.\n", __func__);
+		EXOFS_ERR("%s: ore_get_io_state failed.\n", __func__);
 		goto free_args;
 	}
 
@@ -1254,13 +1439,13 @@ static int exofs_update_inode(struct inode *inode, int do_sync)
 		ios->private = args;
 	}
 
-	ret = exofs_oi_write(oi, ios);
+	ret = ore_write(ios);
 	if (!do_sync && !ret) {
 		atomic_inc(&sbi->s_curr_pending);
 		goto out; /* deallocation in updatei_done */
 	}
 
-	exofs_put_io_state(ios);
+	ore_put_io_state(ios);
 free_args:
 	kfree(args);
 out:
@@ -1271,18 +1456,19 @@ out:
 
 int exofs_write_inode(struct inode *inode, struct writeback_control *wbc)
 {
-	return exofs_update_inode(inode, wbc->sync_mode == WB_SYNC_ALL);
+	/* FIXME: fix fsync and use wbc->sync_mode == WB_SYNC_ALL */
+	return exofs_update_inode(inode, 1);
 }
 
 /*
  * Callback function from exofs_delete_inode() - don't have much cleaning up to
  * do.
  */
-static void delete_done(struct exofs_io_state *ios, void *p)
+static void delete_done(struct ore_io_state *ios, void *p)
 {
 	struct exofs_sb_info *sbi = p;
 
-	exofs_put_io_state(ios);
+	ore_put_io_state(ios);
 
 	atomic_dec(&sbi->s_curr_pending);
 }
@@ -1297,17 +1483,17 @@ void exofs_evict_inode(struct inode *inode)
 	struct exofs_i_info *oi = exofs_i(inode);
 	struct super_block *sb = inode->i_sb;
 	struct exofs_sb_info *sbi = sb->s_fs_info;
-	struct exofs_io_state *ios;
+	struct ore_io_state *ios;
 	int ret;
 
-	truncate_inode_pages(&inode->i_data, 0);
+	truncate_inode_pages_final(&inode->i_data);
 
 	/* TODO: should do better here */
 	if (inode->i_nlink || is_bad_inode(inode))
 		goto no_delete;
 
 	inode->i_size = 0;
-	end_writeback(inode);
+	clear_inode(inode);
 
 	/* if we are deleting an obj that hasn't been created yet, wait.
 	 * This also makes sure that create_done cannot be called with an
@@ -1317,20 +1503,19 @@ void exofs_evict_inode(struct inode *inode)
 	/* ignore the error, attempt a remove anyway */
 
 	/* Now Remove the OSD objects */
-	ret = exofs_get_io_state(&sbi->layout, &ios);
+	ret = ore_get_io_state(&sbi->layout, &oi->oc, &ios);
 	if (unlikely(ret)) {
-		EXOFS_ERR("%s: exofs_get_io_state failed\n", __func__);
+		EXOFS_ERR("%s: ore_get_io_state failed\n", __func__);
 		return;
 	}
 
-	ios->obj.id = exofs_oi_objno(oi);
 	ios->done = delete_done;
 	ios->private = sbi;
-	ios->cred = oi->i_cred;
-	ret = exofs_sbi_remove(ios);
+
+	ret = ore_remove(ios);
 	if (ret) {
-		EXOFS_ERR("%s: exofs_sbi_remove failed\n", __func__);
-		exofs_put_io_state(ios);
+		EXOFS_ERR("%s: ore_remove failed\n", __func__);
+		ore_put_io_state(ios);
 		return;
 	}
 	atomic_inc(&sbi->s_curr_pending);
@@ -1338,5 +1523,5 @@ void exofs_evict_inode(struct inode *inode)
 	return;
 
 no_delete:
-	end_writeback(inode);
+	clear_inode(inode);
 }