Resguardo inicial

linux-7.2.3/fs/nilfs2/Kconfig

@@ -0,0 +1,27 @@
+# SPDX-License-Identifier: GPL-2.0-only
+config NILFS2_FS
+	tristate "NILFS2 file system support"
+	select BUFFER_HEAD
+	select CRC32
+	select LEGACY_DIRECT_IO
+	help
+	  NILFS2 is a log-structured file system (LFS) supporting continuous
+	  snapshotting.  In addition to versioning capability of the entire
+	  file system, users can even restore files mistakenly overwritten or
+	  destroyed just a few seconds ago.  Since this file system can keep
+	  consistency like conventional LFS, it achieves quick recovery after
+	  system crashes.
+
+	  NILFS2 creates a number of checkpoints every few seconds or per
+	  synchronous write basis (unless there is no change).  Users can
+	  select significant versions among continuously created checkpoints,
+	  and can change them into snapshots which will be preserved for long
+	  periods until they are changed back to checkpoints.  Each
+	  snapshot is mountable as a read-only file system concurrently with
+	  its writable mount, and this feature is convenient for online backup.
+
+	  Some features including atime, extended attributes, and POSIX ACLs,
+	  are not supported yet.
+
+	  To compile this file system support as a module, choose M here: the
+	  module will be called nilfs2.  If unsure, say N.

linux-7.2.3/fs/nilfs2/Makefile

@@ -0,0 +1,6 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-$(CONFIG_NILFS2_FS) += nilfs2.o
+nilfs2-y := inode.o file.o dir.o super.o namei.o page.o mdt.o \
+	btnode.o bmap.o btree.o direct.o dat.o recovery.o \
+	the_nilfs.o segbuf.o segment.o cpfile.o sufile.o \
+	ifile.o alloc.o gcinode.o ioctl.o sysfs.o

linux-7.2.3/fs/nilfs2/alloc.c

@@ -0,0 +1,958 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS dat/inode allocator
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Originally written by Koji Sato.
+ * Two allocators were unified by Ryusuke Konishi and Amagai Yoshiji.
+ */
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include <linux/fs.h>
+#include <linux/bitops.h>
+#include <linux/slab.h>
+#include "mdt.h"
+#include "alloc.h"
+
+
+/**
+ * nilfs_palloc_groups_per_desc_block - get the number of groups that a group
+ *					descriptor block can maintain
+ * @inode: inode of metadata file using this allocator
+ *
+ * Return: Number of groups that a group descriptor block can maintain.
+ */
+static inline unsigned long
+nilfs_palloc_groups_per_desc_block(const struct inode *inode)
+{
+	return i_blocksize(inode) /
+		sizeof(struct nilfs_palloc_group_desc);
+}
+
+/**
+ * nilfs_palloc_groups_count - get maximum number of groups
+ * @inode: inode of metadata file using this allocator
+ *
+ * Return: Maximum number of groups.
+ */
+static inline unsigned long
+nilfs_palloc_groups_count(const struct inode *inode)
+{
+	return 1UL << (BITS_PER_LONG - (inode->i_blkbits + 3 /* log2(8) */));
+}
+
+/**
+ * nilfs_palloc_init_blockgroup - initialize private variables for allocator
+ * @inode: inode of metadata file using this allocator
+ * @entry_size: size of the persistent object
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int nilfs_palloc_init_blockgroup(struct inode *inode, unsigned int entry_size)
+{
+	struct nilfs_mdt_info *mi = NILFS_MDT(inode);
+
+	mi->mi_bgl = kmalloc_obj(*mi->mi_bgl, GFP_NOFS);
+	if (!mi->mi_bgl)
+		return -ENOMEM;
+
+	bgl_lock_init(mi->mi_bgl);
+
+	nilfs_mdt_set_entry_size(inode, entry_size, 0);
+
+	mi->mi_blocks_per_group =
+		DIV_ROUND_UP(nilfs_palloc_entries_per_group(inode),
+			     mi->mi_entries_per_block) + 1;
+		/*
+		 * Number of blocks in a group including entry blocks
+		 * and a bitmap block
+		 */
+	mi->mi_blocks_per_desc_block =
+		nilfs_palloc_groups_per_desc_block(inode) *
+		mi->mi_blocks_per_group + 1;
+		/*
+		 * Number of blocks per descriptor including the
+		 * descriptor block
+		 */
+	return 0;
+}
+
+/**
+ * nilfs_palloc_group - get group number and offset from an entry number
+ * @inode: inode of metadata file using this allocator
+ * @nr: serial number of the entry (e.g. inode number)
+ * @offset: pointer to store offset number in the group
+ *
+ * Return: Number of the group that contains the entry with the index
+ * specified by @nr.
+ */
+static unsigned long nilfs_palloc_group(const struct inode *inode, __u64 nr,
+					unsigned long *offset)
+{
+	__u64 group = nr;
+
+	*offset = do_div(group, nilfs_palloc_entries_per_group(inode));
+	return group;
+}
+
+/**
+ * nilfs_palloc_desc_blkoff - get block offset of a group descriptor block
+ * @inode: inode of metadata file using this allocator
+ * @group: group number
+ *
+ * Return: Index number in the metadata file of the descriptor block of
+ * the group specified by @group.
+ */
+static unsigned long
+nilfs_palloc_desc_blkoff(const struct inode *inode, unsigned long group)
+{
+	unsigned long desc_block =
+		group / nilfs_palloc_groups_per_desc_block(inode);
+	return desc_block * NILFS_MDT(inode)->mi_blocks_per_desc_block;
+}
+
+/**
+ * nilfs_palloc_bitmap_blkoff - get block offset of a bitmap block
+ * @inode: inode of metadata file using this allocator
+ * @group: group number
+ *
+ * nilfs_palloc_bitmap_blkoff() returns block offset of the bitmap
+ * block used to allocate/deallocate entries in the specified group.
+ *
+ * Return: Index number in the metadata file of the bitmap block of
+ * the group specified by @group.
+ */
+static unsigned long
+nilfs_palloc_bitmap_blkoff(const struct inode *inode, unsigned long group)
+{
+	unsigned long desc_offset =
+		group % nilfs_palloc_groups_per_desc_block(inode);
+	return nilfs_palloc_desc_blkoff(inode, group) + 1 +
+		desc_offset * NILFS_MDT(inode)->mi_blocks_per_group;
+}
+
+/**
+ * nilfs_palloc_group_desc_nfrees - get the number of free entries in a group
+ * @desc: pointer to descriptor structure for the group
+ * @lock: spin lock protecting @desc
+ *
+ * Return: Number of free entries written in the group descriptor @desc.
+ */
+static unsigned long
+nilfs_palloc_group_desc_nfrees(const struct nilfs_palloc_group_desc *desc,
+			       spinlock_t *lock)
+{
+	unsigned long nfree;
+
+	spin_lock(lock);
+	nfree = le32_to_cpu(desc->pg_nfrees);
+	spin_unlock(lock);
+	return nfree;
+}
+
+/**
+ * nilfs_palloc_group_desc_add_entries - adjust count of free entries
+ * @desc: pointer to descriptor structure for the group
+ * @lock: spin lock protecting @desc
+ * @n: delta to be added
+ *
+ * Return: Number of free entries after adjusting the group descriptor
+ * @desc.
+ */
+static u32
+nilfs_palloc_group_desc_add_entries(struct nilfs_palloc_group_desc *desc,
+				    spinlock_t *lock, u32 n)
+{
+	u32 nfree;
+
+	spin_lock(lock);
+	le32_add_cpu(&desc->pg_nfrees, n);
+	nfree = le32_to_cpu(desc->pg_nfrees);
+	spin_unlock(lock);
+	return nfree;
+}
+
+/**
+ * nilfs_palloc_entry_blkoff - get block offset of an entry block
+ * @inode: inode of metadata file using this allocator
+ * @nr: serial number of the entry (e.g. inode number)
+ *
+ * Return: Index number in the metadata file of the block containing
+ * the entry specified by @nr.
+ */
+static unsigned long
+nilfs_palloc_entry_blkoff(const struct inode *inode, __u64 nr)
+{
+	unsigned long group, group_offset;
+
+	group = nilfs_palloc_group(inode, nr, &group_offset);
+
+	return nilfs_palloc_bitmap_blkoff(inode, group) + 1 +
+		group_offset / NILFS_MDT(inode)->mi_entries_per_block;
+}
+
+/**
+ * nilfs_palloc_desc_block_init - initialize buffer of a group descriptor block
+ * @inode: inode of metadata file
+ * @bh: buffer head of the buffer to be initialized
+ * @from: kernel address mapped for a chunk of the block
+ *
+ * This function does not yet support the case where block size > PAGE_SIZE.
+ */
+static void nilfs_palloc_desc_block_init(struct inode *inode,
+					 struct buffer_head *bh, void *from)
+{
+	struct nilfs_palloc_group_desc *desc = from;
+	unsigned long n = nilfs_palloc_groups_per_desc_block(inode);
+	__le32 nfrees;
+
+	nfrees = cpu_to_le32(nilfs_palloc_entries_per_group(inode));
+	while (n-- > 0) {
+		desc->pg_nfrees = nfrees;
+		desc++;
+	}
+}
+
+static int nilfs_palloc_get_block(struct inode *inode, unsigned long blkoff,
+				  int create,
+				  void (*init_block)(struct inode *,
+						     struct buffer_head *,
+						     void *),
+				  struct buffer_head **bhp,
+				  struct nilfs_bh_assoc *prev,
+				  spinlock_t *lock)
+{
+	int ret;
+
+	spin_lock(lock);
+	if (prev->bh && blkoff == prev->blkoff &&
+	    likely(buffer_uptodate(prev->bh))) {
+		get_bh(prev->bh);
+		*bhp = prev->bh;
+		spin_unlock(lock);
+		return 0;
+	}
+	spin_unlock(lock);
+
+	ret = nilfs_mdt_get_block(inode, blkoff, create, init_block, bhp);
+	if (!ret) {
+		spin_lock(lock);
+		/*
+		 * The following code must be safe for change of the
+		 * cache contents during the get block call.
+		 */
+		brelse(prev->bh);
+		get_bh(*bhp);
+		prev->bh = *bhp;
+		prev->blkoff = blkoff;
+		spin_unlock(lock);
+	}
+	return ret;
+}
+
+/**
+ * nilfs_palloc_delete_block - delete a block on the persistent allocator file
+ * @inode: inode of metadata file using this allocator
+ * @blkoff: block offset
+ * @prev: nilfs_bh_assoc struct of the last used buffer
+ * @lock: spin lock protecting @prev
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- Non-existent block.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+static int nilfs_palloc_delete_block(struct inode *inode, unsigned long blkoff,
+				     struct nilfs_bh_assoc *prev,
+				     spinlock_t *lock)
+{
+	spin_lock(lock);
+	if (prev->bh && blkoff == prev->blkoff) {
+		brelse(prev->bh);
+		prev->bh = NULL;
+	}
+	spin_unlock(lock);
+	return nilfs_mdt_delete_block(inode, blkoff);
+}
+
+/**
+ * nilfs_palloc_get_desc_block - get buffer head of a group descriptor block
+ * @inode: inode of metadata file using this allocator
+ * @group: group number
+ * @create: create flag
+ * @bhp: pointer to store the resultant buffer head
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static int nilfs_palloc_get_desc_block(struct inode *inode,
+				       unsigned long group,
+				       int create, struct buffer_head **bhp)
+{
+	struct nilfs_palloc_cache *cache = NILFS_MDT(inode)->mi_palloc_cache;
+
+	return nilfs_palloc_get_block(inode,
+				      nilfs_palloc_desc_blkoff(inode, group),
+				      create, nilfs_palloc_desc_block_init,
+				      bhp, &cache->prev_desc, &cache->lock);
+}
+
+/**
+ * nilfs_palloc_get_bitmap_block - get buffer head of a bitmap block
+ * @inode: inode of metadata file using this allocator
+ * @group: group number
+ * @create: create flag
+ * @bhp: pointer to store the resultant buffer head
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static int nilfs_palloc_get_bitmap_block(struct inode *inode,
+					 unsigned long group,
+					 int create, struct buffer_head **bhp)
+{
+	struct nilfs_palloc_cache *cache = NILFS_MDT(inode)->mi_palloc_cache;
+
+	return nilfs_palloc_get_block(inode,
+				      nilfs_palloc_bitmap_blkoff(inode, group),
+				      create, NULL, bhp,
+				      &cache->prev_bitmap, &cache->lock);
+}
+
+/**
+ * nilfs_palloc_delete_bitmap_block - delete a bitmap block
+ * @inode: inode of metadata file using this allocator
+ * @group: group number
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static int nilfs_palloc_delete_bitmap_block(struct inode *inode,
+					    unsigned long group)
+{
+	struct nilfs_palloc_cache *cache = NILFS_MDT(inode)->mi_palloc_cache;
+
+	return nilfs_palloc_delete_block(inode,
+					 nilfs_palloc_bitmap_blkoff(inode,
+								    group),
+					 &cache->prev_bitmap, &cache->lock);
+}
+
+/**
+ * nilfs_palloc_get_entry_block - get buffer head of an entry block
+ * @inode: inode of metadata file using this allocator
+ * @nr: serial number of the entry (e.g. inode number)
+ * @create: create flag
+ * @bhp: pointer to store the resultant buffer head
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int nilfs_palloc_get_entry_block(struct inode *inode, __u64 nr,
+				 int create, struct buffer_head **bhp)
+{
+	struct nilfs_palloc_cache *cache = NILFS_MDT(inode)->mi_palloc_cache;
+
+	return nilfs_palloc_get_block(inode,
+				      nilfs_palloc_entry_blkoff(inode, nr),
+				      create, NULL, bhp,
+				      &cache->prev_entry, &cache->lock);
+}
+
+/**
+ * nilfs_palloc_delete_entry_block - delete an entry block
+ * @inode: inode of metadata file using this allocator
+ * @nr: serial number of the entry
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static int nilfs_palloc_delete_entry_block(struct inode *inode, __u64 nr)
+{
+	struct nilfs_palloc_cache *cache = NILFS_MDT(inode)->mi_palloc_cache;
+
+	return nilfs_palloc_delete_block(inode,
+					 nilfs_palloc_entry_blkoff(inode, nr),
+					 &cache->prev_entry, &cache->lock);
+}
+
+/**
+ * nilfs_palloc_group_desc_offset - calculate the byte offset of a group
+ *                                  descriptor in the folio containing it
+ * @inode: inode of metadata file using this allocator
+ * @group: group number
+ * @bh:    buffer head of the group descriptor block
+ *
+ * Return: Byte offset in the folio of the group descriptor for @group.
+ */
+static size_t nilfs_palloc_group_desc_offset(const struct inode *inode,
+					     unsigned long group,
+					     const struct buffer_head *bh)
+{
+	return offset_in_folio(bh->b_folio, bh->b_data) +
+		sizeof(struct nilfs_palloc_group_desc) *
+		(group % nilfs_palloc_groups_per_desc_block(inode));
+}
+
+/**
+ * nilfs_palloc_bitmap_offset - calculate the byte offset of a bitmap block
+ *                              in the folio containing it
+ * @bh: buffer head of the bitmap block
+ *
+ * Return: Byte offset in the folio of the bitmap block for @bh.
+ */
+static size_t nilfs_palloc_bitmap_offset(const struct buffer_head *bh)
+{
+	return offset_in_folio(bh->b_folio, bh->b_data);
+}
+
+/**
+ * nilfs_palloc_entry_offset - calculate the byte offset of an entry in the
+ *                             folio containing it
+ * @inode: inode of metadata file using this allocator
+ * @nr:    serial number of the entry (e.g. inode number)
+ * @bh:    buffer head of the entry block
+ *
+ * Return: Byte offset in the folio of the entry @nr.
+ */
+size_t nilfs_palloc_entry_offset(const struct inode *inode, __u64 nr,
+				 const struct buffer_head *bh)
+{
+	unsigned long entry_index_in_group, entry_index_in_block;
+
+	nilfs_palloc_group(inode, nr, &entry_index_in_group);
+	entry_index_in_block = entry_index_in_group %
+		NILFS_MDT(inode)->mi_entries_per_block;
+
+	return offset_in_folio(bh->b_folio, bh->b_data) +
+		entry_index_in_block * NILFS_MDT(inode)->mi_entry_size;
+}
+
+/**
+ * nilfs_palloc_find_available_slot - find available slot in a group
+ * @bitmap: bitmap of the group
+ * @target: offset number of an entry in the group (start point)
+ * @bsize: size in bits
+ * @lock: spin lock protecting @bitmap
+ * @wrap: whether to wrap around
+ *
+ * Return: Offset number within the group of the found free entry, or
+ * %-ENOSPC if not found.
+ */
+static int nilfs_palloc_find_available_slot(unsigned char *bitmap,
+					    unsigned long target,
+					    unsigned int bsize,
+					    spinlock_t *lock, bool wrap)
+{
+	int pos, end = bsize;
+
+	if (likely(target < bsize)) {
+		pos = target;
+		do {
+			pos = nilfs_find_next_zero_bit(bitmap, end, pos);
+			if (pos >= end)
+				break;
+			if (!nilfs_set_bit_atomic(lock, pos, bitmap))
+				return pos;
+		} while (++pos < end);
+
+		end = target;
+	}
+	if (!wrap)
+		return -ENOSPC;
+
+	/* wrap around */
+	for (pos = 0; pos < end; pos++) {
+		pos = nilfs_find_next_zero_bit(bitmap, end, pos);
+		if (pos >= end)
+			break;
+		if (!nilfs_set_bit_atomic(lock, pos, bitmap))
+			return pos;
+	}
+
+	return -ENOSPC;
+}
+
+/**
+ * nilfs_palloc_rest_groups_in_desc_block - get the remaining number of groups
+ *					    in a group descriptor block
+ * @inode: inode of metadata file using this allocator
+ * @curr: current group number
+ * @max: maximum number of groups
+ *
+ * Return: Number of remaining descriptors (= groups) managed by the descriptor
+ * block.
+ */
+static unsigned long
+nilfs_palloc_rest_groups_in_desc_block(const struct inode *inode,
+				       unsigned long curr, unsigned long max)
+{
+	return min_t(unsigned long,
+		     nilfs_palloc_groups_per_desc_block(inode) -
+		     curr % nilfs_palloc_groups_per_desc_block(inode),
+		     max - curr + 1);
+}
+
+/**
+ * nilfs_palloc_count_desc_blocks - count descriptor blocks number
+ * @inode: inode of metadata file using this allocator
+ * @desc_blocks: descriptor blocks number [out]
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static int nilfs_palloc_count_desc_blocks(struct inode *inode,
+					    unsigned long *desc_blocks)
+{
+	__u64 blknum;
+	int ret;
+
+	ret = nilfs_bmap_last_key(NILFS_I(inode)->i_bmap, &blknum);
+	if (likely(!ret))
+		*desc_blocks = DIV_ROUND_UP(
+			(unsigned long)blknum,
+			NILFS_MDT(inode)->mi_blocks_per_desc_block);
+	return ret;
+}
+
+/**
+ * nilfs_palloc_mdt_file_can_grow - check potential opportunity for
+ *					MDT file growing
+ * @inode: inode of metadata file using this allocator
+ * @desc_blocks: known current descriptor blocks count
+ *
+ * Return: true if a group can be added in the metadata file, false if not.
+ */
+static inline bool nilfs_palloc_mdt_file_can_grow(struct inode *inode,
+						    unsigned long desc_blocks)
+{
+	return (nilfs_palloc_groups_per_desc_block(inode) * desc_blocks) <
+			nilfs_palloc_groups_count(inode);
+}
+
+/**
+ * nilfs_palloc_count_max_entries - count max number of entries that can be
+ *					described by descriptor blocks count
+ * @inode: inode of metadata file using this allocator
+ * @nused: current number of used entries
+ * @nmaxp: max number of entries [out]
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ * * %-ERANGE	- Number of entries in use is out of range.
+ */
+int nilfs_palloc_count_max_entries(struct inode *inode, u64 nused, u64 *nmaxp)
+{
+	unsigned long desc_blocks = 0;
+	u64 entries_per_desc_block, nmax;
+	int err;
+
+	err = nilfs_palloc_count_desc_blocks(inode, &desc_blocks);
+	if (unlikely(err))
+		return err;
+
+	entries_per_desc_block = (u64)nilfs_palloc_entries_per_group(inode) *
+				nilfs_palloc_groups_per_desc_block(inode);
+	nmax = entries_per_desc_block * desc_blocks;
+
+	if (nused == nmax &&
+			nilfs_palloc_mdt_file_can_grow(inode, desc_blocks))
+		nmax += entries_per_desc_block;
+
+	if (nused > nmax)
+		return -ERANGE;
+
+	*nmaxp = nmax;
+	return 0;
+}
+
+/**
+ * nilfs_palloc_prepare_alloc_entry - prepare to allocate a persistent object
+ * @inode: inode of metadata file using this allocator
+ * @req: nilfs_palloc_req structure exchanged for the allocation
+ * @wrap: whether to wrap around
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ * * %-ENOSPC	- Entries exhausted (No entries available for allocation).
+ * * %-EROFS	- Read only filesystem
+ */
+int nilfs_palloc_prepare_alloc_entry(struct inode *inode,
+				     struct nilfs_palloc_req *req, bool wrap)
+{
+	struct buffer_head *desc_bh, *bitmap_bh;
+	struct nilfs_palloc_group_desc *desc;
+	unsigned char *bitmap;
+	size_t doff, boff;
+	unsigned long group, maxgroup, ngroups;
+	unsigned long group_offset, maxgroup_offset;
+	unsigned long n, entries_per_group;
+	unsigned long i, j;
+	spinlock_t *lock;
+	int pos, ret;
+
+	ngroups = nilfs_palloc_groups_count(inode);
+	maxgroup = ngroups - 1;
+	group = nilfs_palloc_group(inode, req->pr_entry_nr, &group_offset);
+	entries_per_group = nilfs_palloc_entries_per_group(inode);
+
+	for (i = 0; i < ngroups; i += n) {
+		if (group >= ngroups && wrap) {
+			/* wrap around */
+			group = 0;
+			maxgroup = nilfs_palloc_group(inode, req->pr_entry_nr,
+						      &maxgroup_offset) - 1;
+		}
+		ret = nilfs_palloc_get_desc_block(inode, group, 1, &desc_bh);
+		if (ret < 0)
+			return ret;
+
+		doff = nilfs_palloc_group_desc_offset(inode, group, desc_bh);
+		desc = kmap_local_folio(desc_bh->b_folio, doff);
+		n = nilfs_palloc_rest_groups_in_desc_block(inode, group,
+							   maxgroup);
+		for (j = 0; j < n; j++, group++, group_offset = 0) {
+			lock = nilfs_mdt_bgl_lock(inode, group);
+			if (nilfs_palloc_group_desc_nfrees(&desc[j], lock) == 0)
+				continue;
+
+			kunmap_local(desc);
+			ret = nilfs_palloc_get_bitmap_block(inode, group, 1,
+							    &bitmap_bh);
+			if (unlikely(ret < 0)) {
+				brelse(desc_bh);
+				return ret;
+			}
+
+			/*
+			 * Re-kmap the folio containing the first (and
+			 * subsequent) group descriptors.
+			 */
+			desc = kmap_local_folio(desc_bh->b_folio, doff);
+
+			boff = nilfs_palloc_bitmap_offset(bitmap_bh);
+			bitmap = kmap_local_folio(bitmap_bh->b_folio, boff);
+			pos = nilfs_palloc_find_available_slot(
+				bitmap, group_offset, entries_per_group, lock,
+				wrap);
+			/*
+			 * Since the search for a free slot in the second and
+			 * subsequent bitmap blocks always starts from the
+			 * beginning, the wrap flag only has an effect on the
+			 * first search.
+			 */
+			kunmap_local(bitmap);
+			if (pos >= 0)
+				goto found;
+
+			brelse(bitmap_bh);
+		}
+
+		kunmap_local(desc);
+		brelse(desc_bh);
+	}
+
+	/* no entries left */
+	return -ENOSPC;
+
+found:
+	/* found a free entry */
+	nilfs_palloc_group_desc_add_entries(&desc[j], lock, -1);
+	req->pr_entry_nr = entries_per_group * group + pos;
+	kunmap_local(desc);
+
+	req->pr_desc_bh = desc_bh;
+	req->pr_bitmap_bh = bitmap_bh;
+	return 0;
+}
+
+/**
+ * nilfs_palloc_commit_alloc_entry - finish allocation of a persistent object
+ * @inode: inode of metadata file using this allocator
+ * @req: nilfs_palloc_req structure exchanged for the allocation
+ */
+void nilfs_palloc_commit_alloc_entry(struct inode *inode,
+				     struct nilfs_palloc_req *req)
+{
+	mark_buffer_dirty(req->pr_bitmap_bh);
+	mark_buffer_dirty(req->pr_desc_bh);
+	nilfs_mdt_mark_dirty(inode);
+
+	brelse(req->pr_bitmap_bh);
+	brelse(req->pr_desc_bh);
+}
+
+/**
+ * nilfs_palloc_commit_free_entry - finish deallocating a persistent object
+ * @inode: inode of metadata file using this allocator
+ * @req: nilfs_palloc_req structure exchanged for the removal
+ */
+void nilfs_palloc_commit_free_entry(struct inode *inode,
+				    struct nilfs_palloc_req *req)
+{
+	unsigned long group, group_offset;
+	size_t doff, boff;
+	struct nilfs_palloc_group_desc *desc;
+	unsigned char *bitmap;
+	spinlock_t *lock;
+
+	group = nilfs_palloc_group(inode, req->pr_entry_nr, &group_offset);
+	doff = nilfs_palloc_group_desc_offset(inode, group, req->pr_desc_bh);
+	desc = kmap_local_folio(req->pr_desc_bh->b_folio, doff);
+
+	boff = nilfs_palloc_bitmap_offset(req->pr_bitmap_bh);
+	bitmap = kmap_local_folio(req->pr_bitmap_bh->b_folio, boff);
+	lock = nilfs_mdt_bgl_lock(inode, group);
+
+	if (!nilfs_clear_bit_atomic(lock, group_offset, bitmap))
+		nilfs_warn(inode->i_sb,
+			   "%s (ino=%llu): entry number %llu already freed",
+			   __func__, inode->i_ino,
+			   (unsigned long long)req->pr_entry_nr);
+	else
+		nilfs_palloc_group_desc_add_entries(desc, lock, 1);
+
+	kunmap_local(bitmap);
+	kunmap_local(desc);
+
+	mark_buffer_dirty(req->pr_desc_bh);
+	mark_buffer_dirty(req->pr_bitmap_bh);
+	nilfs_mdt_mark_dirty(inode);
+
+	brelse(req->pr_bitmap_bh);
+	brelse(req->pr_desc_bh);
+}
+
+/**
+ * nilfs_palloc_abort_alloc_entry - cancel allocation of a persistent object
+ * @inode: inode of metadata file using this allocator
+ * @req: nilfs_palloc_req structure exchanged for the allocation
+ */
+void nilfs_palloc_abort_alloc_entry(struct inode *inode,
+				    struct nilfs_palloc_req *req)
+{
+	struct nilfs_palloc_group_desc *desc;
+	size_t doff, boff;
+	unsigned char *bitmap;
+	unsigned long group, group_offset;
+	spinlock_t *lock;
+
+	group = nilfs_palloc_group(inode, req->pr_entry_nr, &group_offset);
+	doff = nilfs_palloc_group_desc_offset(inode, group, req->pr_desc_bh);
+	desc = kmap_local_folio(req->pr_desc_bh->b_folio, doff);
+
+	boff = nilfs_palloc_bitmap_offset(req->pr_bitmap_bh);
+	bitmap = kmap_local_folio(req->pr_bitmap_bh->b_folio, boff);
+	lock = nilfs_mdt_bgl_lock(inode, group);
+
+	if (!nilfs_clear_bit_atomic(lock, group_offset, bitmap))
+		nilfs_warn(inode->i_sb,
+			   "%s (ino=%llu): entry number %llu already freed",
+			   __func__, inode->i_ino,
+			   (unsigned long long)req->pr_entry_nr);
+	else
+		nilfs_palloc_group_desc_add_entries(desc, lock, 1);
+
+	kunmap_local(bitmap);
+	kunmap_local(desc);
+
+	brelse(req->pr_bitmap_bh);
+	brelse(req->pr_desc_bh);
+
+	req->pr_entry_nr = 0;
+	req->pr_bitmap_bh = NULL;
+	req->pr_desc_bh = NULL;
+}
+
+/**
+ * nilfs_palloc_prepare_free_entry - prepare to deallocate a persistent object
+ * @inode: inode of metadata file using this allocator
+ * @req: nilfs_palloc_req structure exchanged for the removal
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int nilfs_palloc_prepare_free_entry(struct inode *inode,
+				    struct nilfs_palloc_req *req)
+{
+	struct buffer_head *desc_bh, *bitmap_bh;
+	unsigned long group, group_offset;
+	int ret;
+
+	group = nilfs_palloc_group(inode, req->pr_entry_nr, &group_offset);
+	ret = nilfs_palloc_get_desc_block(inode, group, 1, &desc_bh);
+	if (ret < 0)
+		return ret;
+	ret = nilfs_palloc_get_bitmap_block(inode, group, 1, &bitmap_bh);
+	if (ret < 0) {
+		brelse(desc_bh);
+		return ret;
+	}
+
+	req->pr_desc_bh = desc_bh;
+	req->pr_bitmap_bh = bitmap_bh;
+	return 0;
+}
+
+/**
+ * nilfs_palloc_abort_free_entry - cancel deallocating a persistent object
+ * @inode: inode of metadata file using this allocator
+ * @req: nilfs_palloc_req structure exchanged for the removal
+ */
+void nilfs_palloc_abort_free_entry(struct inode *inode,
+				   struct nilfs_palloc_req *req)
+{
+	brelse(req->pr_bitmap_bh);
+	brelse(req->pr_desc_bh);
+
+	req->pr_entry_nr = 0;
+	req->pr_bitmap_bh = NULL;
+	req->pr_desc_bh = NULL;
+}
+
+/**
+ * nilfs_palloc_freev - deallocate a set of persistent objects
+ * @inode: inode of metadata file using this allocator
+ * @entry_nrs: array of entry numbers to be deallocated
+ * @nitems: number of entries stored in @entry_nrs
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int nilfs_palloc_freev(struct inode *inode, __u64 *entry_nrs, size_t nitems)
+{
+	struct buffer_head *desc_bh, *bitmap_bh;
+	struct nilfs_palloc_group_desc *desc;
+	unsigned char *bitmap;
+	size_t doff, boff;
+	unsigned long group, group_offset;
+	__u64 group_min_nr, last_nrs[8];
+	const unsigned long epg = nilfs_palloc_entries_per_group(inode);
+	const unsigned int epb = NILFS_MDT(inode)->mi_entries_per_block;
+	unsigned int entry_start, end, pos;
+	spinlock_t *lock;
+	int i, j, k, ret;
+	u32 nfree;
+
+	for (i = 0; i < nitems; i = j) {
+		int change_group = false;
+		int nempties = 0, n = 0;
+
+		group = nilfs_palloc_group(inode, entry_nrs[i], &group_offset);
+		ret = nilfs_palloc_get_desc_block(inode, group, 0, &desc_bh);
+		if (ret < 0)
+			return ret;
+		ret = nilfs_palloc_get_bitmap_block(inode, group, 0,
+						    &bitmap_bh);
+		if (ret < 0) {
+			brelse(desc_bh);
+			return ret;
+		}
+
+		/* Get the first entry number of the group */
+		group_min_nr = (__u64)group * epg;
+
+		boff = nilfs_palloc_bitmap_offset(bitmap_bh);
+		bitmap = kmap_local_folio(bitmap_bh->b_folio, boff);
+		lock = nilfs_mdt_bgl_lock(inode, group);
+
+		j = i;
+		entry_start = rounddown(group_offset, epb);
+		do {
+			if (!nilfs_clear_bit_atomic(lock, group_offset,
+						    bitmap)) {
+				nilfs_warn(inode->i_sb,
+					   "%s (ino=%llu): entry number %llu already freed",
+					   __func__, inode->i_ino,
+					   (unsigned long long)entry_nrs[j]);
+			} else {
+				n++;
+			}
+
+			j++;
+			if (j >= nitems || entry_nrs[j] < group_min_nr ||
+			    entry_nrs[j] >= group_min_nr + epg) {
+				change_group = true;
+			} else {
+				group_offset = entry_nrs[j] - group_min_nr;
+				if (group_offset >= entry_start &&
+				    group_offset < entry_start + epb) {
+					/* This entry is in the same block */
+					continue;
+				}
+			}
+
+			/* Test if the entry block is empty or not */
+			end = entry_start + epb;
+			pos = nilfs_find_next_bit(bitmap, end, entry_start);
+			if (pos >= end) {
+				last_nrs[nempties++] = entry_nrs[j - 1];
+				if (nempties >= ARRAY_SIZE(last_nrs))
+					break;
+			}
+
+			if (change_group)
+				break;
+
+			/* Go on to the next entry block */
+			entry_start = rounddown(group_offset, epb);
+		} while (true);
+
+		kunmap_local(bitmap);
+		mark_buffer_dirty(bitmap_bh);
+		brelse(bitmap_bh);
+
+		for (k = 0; k < nempties; k++) {
+			ret = nilfs_palloc_delete_entry_block(inode,
+							      last_nrs[k]);
+			if (ret && ret != -ENOENT)
+				nilfs_warn(inode->i_sb,
+					   "error %d deleting block that object (entry=%llu, ino=%llu) belongs to",
+					   ret, (unsigned long long)last_nrs[k],
+					   inode->i_ino);
+		}
+
+		doff = nilfs_palloc_group_desc_offset(inode, group, desc_bh);
+		desc = kmap_local_folio(desc_bh->b_folio, doff);
+		nfree = nilfs_palloc_group_desc_add_entries(desc, lock, n);
+		kunmap_local(desc);
+		mark_buffer_dirty(desc_bh);
+		nilfs_mdt_mark_dirty(inode);
+		brelse(desc_bh);
+
+		if (nfree == nilfs_palloc_entries_per_group(inode)) {
+			ret = nilfs_palloc_delete_bitmap_block(inode, group);
+			if (ret && ret != -ENOENT)
+				nilfs_warn(inode->i_sb,
+					   "error %d deleting bitmap block of group=%lu, ino=%llu",
+					   ret, group, inode->i_ino);
+		}
+	}
+	return 0;
+}
+
+void nilfs_palloc_setup_cache(struct inode *inode,
+			      struct nilfs_palloc_cache *cache)
+{
+	NILFS_MDT(inode)->mi_palloc_cache = cache;
+	spin_lock_init(&cache->lock);
+}
+
+void nilfs_palloc_clear_cache(struct inode *inode)
+{
+	struct nilfs_palloc_cache *cache = NILFS_MDT(inode)->mi_palloc_cache;
+
+	spin_lock(&cache->lock);
+	brelse(cache->prev_desc.bh);
+	brelse(cache->prev_bitmap.bh);
+	brelse(cache->prev_entry.bh);
+	cache->prev_desc.bh = NULL;
+	cache->prev_bitmap.bh = NULL;
+	cache->prev_entry.bh = NULL;
+	spin_unlock(&cache->lock);
+}
+
+void nilfs_palloc_destroy_cache(struct inode *inode)
+{
+	nilfs_palloc_clear_cache(inode);
+	NILFS_MDT(inode)->mi_palloc_cache = NULL;
+}

linux-7.2.3/fs/nilfs2/alloc.h

@@ -0,0 +1,99 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Persistent object (dat entry/disk inode) allocator/deallocator
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Originally written by Koji Sato.
+ * Two allocators were unified by Ryusuke Konishi and Amagai Yoshiji.
+ */
+
+#ifndef _NILFS_ALLOC_H
+#define _NILFS_ALLOC_H
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include <linux/fs.h>
+
+/**
+ * nilfs_palloc_entries_per_group - get the number of entries per group
+ * @inode: inode of metadata file using this allocator
+ *
+ * The number of entries per group is defined by the number of bits
+ * that a bitmap block can maintain.
+ *
+ * Return: Number of entries per group.
+ */
+static inline unsigned long
+nilfs_palloc_entries_per_group(const struct inode *inode)
+{
+	return 1UL << (inode->i_blkbits + 3 /* log2(8 = CHAR_BITS) */);
+}
+
+int nilfs_palloc_init_blockgroup(struct inode *, unsigned int);
+int nilfs_palloc_get_entry_block(struct inode *, __u64, int,
+				 struct buffer_head **);
+size_t nilfs_palloc_entry_offset(const struct inode *inode, __u64 nr,
+				 const struct buffer_head *bh);
+
+int nilfs_palloc_count_max_entries(struct inode *, u64, u64 *);
+
+/**
+ * struct nilfs_palloc_req - persistent allocator request and reply
+ * @pr_entry_nr: entry number (vblocknr or inode number)
+ * @pr_desc_bh: buffer head of the buffer containing block group descriptors
+ * @pr_bitmap_bh: buffer head of the buffer containing a block group bitmap
+ * @pr_entry_bh: buffer head of the buffer containing translation entries
+ */
+struct nilfs_palloc_req {
+	__u64 pr_entry_nr;
+	struct buffer_head *pr_desc_bh;
+	struct buffer_head *pr_bitmap_bh;
+	struct buffer_head *pr_entry_bh;
+};
+
+int nilfs_palloc_prepare_alloc_entry(struct inode *inode,
+				     struct nilfs_palloc_req *req, bool wrap);
+void nilfs_palloc_commit_alloc_entry(struct inode *,
+				     struct nilfs_palloc_req *);
+void nilfs_palloc_abort_alloc_entry(struct inode *, struct nilfs_palloc_req *);
+void nilfs_palloc_commit_free_entry(struct inode *, struct nilfs_palloc_req *);
+int nilfs_palloc_prepare_free_entry(struct inode *, struct nilfs_palloc_req *);
+void nilfs_palloc_abort_free_entry(struct inode *, struct nilfs_palloc_req *);
+int nilfs_palloc_freev(struct inode *, __u64 *, size_t);
+
+#define nilfs_set_bit_atomic		ext2_set_bit_atomic
+#define nilfs_clear_bit_atomic		ext2_clear_bit_atomic
+#define nilfs_find_next_zero_bit	find_next_zero_bit_le
+#define nilfs_find_next_bit		find_next_bit_le
+
+/**
+ * struct nilfs_bh_assoc - block offset and buffer head association
+ * @blkoff: block offset
+ * @bh: buffer head
+ */
+struct nilfs_bh_assoc {
+	unsigned long blkoff;
+	struct buffer_head *bh;
+};
+
+/**
+ * struct nilfs_palloc_cache - persistent object allocator cache
+ * @lock: cache protecting lock
+ * @prev_desc: blockgroup descriptors cache
+ * @prev_bitmap: blockgroup bitmap cache
+ * @prev_entry: translation entries cache
+ */
+struct nilfs_palloc_cache {
+	spinlock_t lock;
+	struct nilfs_bh_assoc prev_desc;
+	struct nilfs_bh_assoc prev_bitmap;
+	struct nilfs_bh_assoc prev_entry;
+};
+
+void nilfs_palloc_setup_cache(struct inode *inode,
+			      struct nilfs_palloc_cache *cache);
+void nilfs_palloc_clear_cache(struct inode *inode);
+void nilfs_palloc_destroy_cache(struct inode *inode);
+
+#endif	/* _NILFS_ALLOC_H */

linux-7.2.3/fs/nilfs2/bmap.c

@@ -0,0 +1,568 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS block mapping.
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato.
+ */
+
+#include <linux/fs.h>
+#include <linux/string.h>
+#include <linux/errno.h>
+#include "nilfs.h"
+#include "bmap.h"
+#include "btree.h"
+#include "direct.h"
+#include "btnode.h"
+#include "mdt.h"
+#include "dat.h"
+#include "alloc.h"
+
+struct inode *nilfs_bmap_get_dat(const struct nilfs_bmap *bmap)
+{
+	struct the_nilfs *nilfs = bmap->b_inode->i_sb->s_fs_info;
+
+	return nilfs->ns_dat;
+}
+
+static int nilfs_bmap_convert_error(struct nilfs_bmap *bmap,
+				     const char *fname, int err)
+{
+	struct inode *inode = bmap->b_inode;
+
+	if (err == -EINVAL) {
+		__nilfs_error(inode->i_sb, fname,
+			      "broken bmap (inode number=%llu)", inode->i_ino);
+		err = -EIO;
+	}
+	return err;
+}
+
+/**
+ * nilfs_bmap_lookup_at_level - find a data block or node block
+ * @bmap: bmap
+ * @key: key
+ * @level: level
+ * @ptrp: place to store the value associated to @key
+ *
+ * Description: nilfs_bmap_lookup_at_level() finds a record whose key
+ * matches @key in the block at @level of the bmap.  The record associated
+ * with @key is stored in the place pointed to by @ptrp.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- A record associated with @key does not exist.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_bmap_lookup_at_level(struct nilfs_bmap *bmap, __u64 key, int level,
+			       __u64 *ptrp)
+{
+	sector_t blocknr;
+	int ret;
+
+	down_read(&bmap->b_sem);
+	ret = bmap->b_ops->bop_lookup(bmap, key, level, ptrp);
+	if (ret < 0)
+		goto out;
+
+	if (NILFS_BMAP_USE_VBN(bmap)) {
+		ret = nilfs_dat_translate(nilfs_bmap_get_dat(bmap), *ptrp,
+					  &blocknr);
+		if (!ret)
+			*ptrp = blocknr;
+		else if (ret == -ENOENT) {
+			/*
+			 * If there was no valid entry in DAT for the block
+			 * address obtained by b_ops->bop_lookup, then pass
+			 * internal code -EINVAL to nilfs_bmap_convert_error
+			 * to treat it as metadata corruption.
+			 */
+			ret = -EINVAL;
+		}
+	}
+
+ out:
+	up_read(&bmap->b_sem);
+	return nilfs_bmap_convert_error(bmap, __func__, ret);
+}
+
+int nilfs_bmap_lookup_contig(struct nilfs_bmap *bmap, __u64 key, __u64 *ptrp,
+			     unsigned int maxblocks)
+{
+	int ret;
+
+	down_read(&bmap->b_sem);
+	ret = bmap->b_ops->bop_lookup_contig(bmap, key, ptrp, maxblocks);
+	up_read(&bmap->b_sem);
+
+	return nilfs_bmap_convert_error(bmap, __func__, ret);
+}
+
+static int nilfs_bmap_do_insert(struct nilfs_bmap *bmap, __u64 key, __u64 ptr)
+{
+	__u64 keys[NILFS_BMAP_SMALL_HIGH + 1];
+	__u64 ptrs[NILFS_BMAP_SMALL_HIGH + 1];
+	int ret, n;
+
+	if (bmap->b_ops->bop_check_insert != NULL) {
+		ret = bmap->b_ops->bop_check_insert(bmap, key);
+		if (ret > 0) {
+			n = bmap->b_ops->bop_gather_data(
+				bmap, keys, ptrs, NILFS_BMAP_SMALL_HIGH + 1);
+			if (n < 0)
+				return n;
+			ret = nilfs_btree_convert_and_insert(
+				bmap, key, ptr, keys, ptrs, n);
+			if (ret == 0)
+				bmap->b_u.u_flags |= NILFS_BMAP_LARGE;
+
+			return ret;
+		} else if (ret < 0)
+			return ret;
+	}
+
+	return bmap->b_ops->bop_insert(bmap, key, ptr);
+}
+
+/**
+ * nilfs_bmap_insert - insert a new key-record pair into a bmap
+ * @bmap: bmap
+ * @key: key
+ * @rec: record
+ *
+ * Description: nilfs_bmap_insert() inserts the new key-record pair specified
+ * by @key and @rec into @bmap.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EEXIST	- A record associated with @key already exists.
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_bmap_insert(struct nilfs_bmap *bmap, __u64 key, unsigned long rec)
+{
+	int ret;
+
+	down_write(&bmap->b_sem);
+	ret = nilfs_bmap_do_insert(bmap, key, rec);
+	up_write(&bmap->b_sem);
+
+	return nilfs_bmap_convert_error(bmap, __func__, ret);
+}
+
+static int nilfs_bmap_do_delete(struct nilfs_bmap *bmap, __u64 key)
+{
+	__u64 keys[NILFS_BMAP_LARGE_LOW + 1];
+	__u64 ptrs[NILFS_BMAP_LARGE_LOW + 1];
+	int ret, n;
+
+	if (bmap->b_ops->bop_check_delete != NULL) {
+		ret = bmap->b_ops->bop_check_delete(bmap, key);
+		if (ret > 0) {
+			n = bmap->b_ops->bop_gather_data(
+				bmap, keys, ptrs, NILFS_BMAP_LARGE_LOW + 1);
+			if (n < 0)
+				return n;
+			ret = nilfs_direct_delete_and_convert(
+				bmap, key, keys, ptrs, n);
+			if (ret == 0)
+				bmap->b_u.u_flags &= ~NILFS_BMAP_LARGE;
+
+			return ret;
+		} else if (ret < 0)
+			return ret;
+	}
+
+	return bmap->b_ops->bop_delete(bmap, key);
+}
+
+/**
+ * nilfs_bmap_seek_key - seek a valid entry and return its key
+ * @bmap: bmap struct
+ * @start: start key number
+ * @keyp: place to store valid key
+ *
+ * Description: nilfs_bmap_seek_key() seeks a valid key on @bmap
+ * starting from @start, and stores it to @keyp if found.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- No valid entry was found.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_bmap_seek_key(struct nilfs_bmap *bmap, __u64 start, __u64 *keyp)
+{
+	int ret;
+
+	down_read(&bmap->b_sem);
+	ret = bmap->b_ops->bop_seek_key(bmap, start, keyp);
+	up_read(&bmap->b_sem);
+
+	if (ret < 0)
+		ret = nilfs_bmap_convert_error(bmap, __func__, ret);
+	return ret;
+}
+
+int nilfs_bmap_last_key(struct nilfs_bmap *bmap, __u64 *keyp)
+{
+	int ret;
+
+	down_read(&bmap->b_sem);
+	ret = bmap->b_ops->bop_last_key(bmap, keyp);
+	up_read(&bmap->b_sem);
+
+	if (ret < 0)
+		ret = nilfs_bmap_convert_error(bmap, __func__, ret);
+	return ret;
+}
+
+/**
+ * nilfs_bmap_delete - delete a key-record pair from a bmap
+ * @bmap: bmap
+ * @key: key
+ *
+ * Description: nilfs_bmap_delete() deletes the key-record pair specified by
+ * @key from @bmap.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- A record associated with @key does not exist.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_bmap_delete(struct nilfs_bmap *bmap, __u64 key)
+{
+	int ret;
+
+	down_write(&bmap->b_sem);
+	ret = nilfs_bmap_do_delete(bmap, key);
+	up_write(&bmap->b_sem);
+
+	return nilfs_bmap_convert_error(bmap, __func__, ret);
+}
+
+static int nilfs_bmap_do_truncate(struct nilfs_bmap *bmap, __u64 key)
+{
+	__u64 lastkey;
+	int ret;
+
+	ret = bmap->b_ops->bop_last_key(bmap, &lastkey);
+	if (ret < 0) {
+		if (ret == -ENOENT)
+			ret = 0;
+		return ret;
+	}
+
+	while (key <= lastkey) {
+		ret = nilfs_bmap_do_delete(bmap, lastkey);
+		if (ret < 0)
+			return ret;
+		ret = bmap->b_ops->bop_last_key(bmap, &lastkey);
+		if (ret < 0) {
+			if (ret == -ENOENT)
+				ret = 0;
+			return ret;
+		}
+	}
+	return 0;
+}
+
+/**
+ * nilfs_bmap_truncate - truncate a bmap to a specified key
+ * @bmap: bmap
+ * @key: key
+ *
+ * Description: nilfs_bmap_truncate() removes key-record pairs whose keys are
+ * greater than or equal to @key from @bmap.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_bmap_truncate(struct nilfs_bmap *bmap, __u64 key)
+{
+	int ret;
+
+	down_write(&bmap->b_sem);
+	ret = nilfs_bmap_do_truncate(bmap, key);
+	up_write(&bmap->b_sem);
+
+	return nilfs_bmap_convert_error(bmap, __func__, ret);
+}
+
+/**
+ * nilfs_bmap_clear - free resources a bmap holds
+ * @bmap: bmap
+ *
+ * Description: nilfs_bmap_clear() frees resources associated with @bmap.
+ */
+void nilfs_bmap_clear(struct nilfs_bmap *bmap)
+{
+	down_write(&bmap->b_sem);
+	if (bmap->b_ops->bop_clear != NULL)
+		bmap->b_ops->bop_clear(bmap);
+	up_write(&bmap->b_sem);
+}
+
+/**
+ * nilfs_bmap_propagate - propagate dirty state
+ * @bmap: bmap
+ * @bh: buffer head
+ *
+ * Description: nilfs_bmap_propagate() marks the buffers that directly or
+ * indirectly refer to the block specified by @bh dirty.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_bmap_propagate(struct nilfs_bmap *bmap, struct buffer_head *bh)
+{
+	int ret;
+
+	down_write(&bmap->b_sem);
+	ret = bmap->b_ops->bop_propagate(bmap, bh);
+	up_write(&bmap->b_sem);
+
+	return nilfs_bmap_convert_error(bmap, __func__, ret);
+}
+
+/**
+ * nilfs_bmap_lookup_dirty_buffers - collect dirty block buffers
+ * @bmap: bmap
+ * @listp: pointer to buffer head list
+ */
+void nilfs_bmap_lookup_dirty_buffers(struct nilfs_bmap *bmap,
+				     struct list_head *listp)
+{
+	if (bmap->b_ops->bop_lookup_dirty_buffers != NULL)
+		bmap->b_ops->bop_lookup_dirty_buffers(bmap, listp);
+}
+
+/**
+ * nilfs_bmap_assign - assign a new block number to a block
+ * @bmap:    bmap
+ * @bh:      place to store a pointer to the buffer head to which a block
+ *           address is assigned (in/out)
+ * @blocknr: block number
+ * @binfo:   block information
+ *
+ * Description: nilfs_bmap_assign() assigns the block number @blocknr to the
+ * buffer specified by @bh.  The block information is stored in the memory
+ * pointed to by @binfo, and the buffer head may be replaced as a block
+ * address is assigned, in which case a pointer to the new buffer head is
+ * stored in the memory pointed to by @bh.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_bmap_assign(struct nilfs_bmap *bmap,
+		      struct buffer_head **bh,
+		      unsigned long blocknr,
+		      union nilfs_binfo *binfo)
+{
+	int ret;
+
+	down_write(&bmap->b_sem);
+	ret = bmap->b_ops->bop_assign(bmap, bh, blocknr, binfo);
+	up_write(&bmap->b_sem);
+
+	return nilfs_bmap_convert_error(bmap, __func__, ret);
+}
+
+/**
+ * nilfs_bmap_mark - mark block dirty
+ * @bmap: bmap
+ * @key: key
+ * @level: level
+ *
+ * Description: nilfs_bmap_mark() marks the block specified by @key and @level
+ * as dirty.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_bmap_mark(struct nilfs_bmap *bmap, __u64 key, int level)
+{
+	int ret;
+
+	if (bmap->b_ops->bop_mark == NULL)
+		return 0;
+
+	down_write(&bmap->b_sem);
+	ret = bmap->b_ops->bop_mark(bmap, key, level);
+	up_write(&bmap->b_sem);
+
+	return nilfs_bmap_convert_error(bmap, __func__, ret);
+}
+
+/**
+ * nilfs_bmap_test_and_clear_dirty - test and clear a bmap dirty state
+ * @bmap: bmap
+ *
+ * Description: nilfs_test_and_clear() is the atomic operation to test and
+ * clear the dirty state of @bmap.
+ *
+ * Return: 1 if @bmap is dirty, or 0 if clear.
+ */
+int nilfs_bmap_test_and_clear_dirty(struct nilfs_bmap *bmap)
+{
+	int ret;
+
+	down_write(&bmap->b_sem);
+	ret = nilfs_bmap_dirty(bmap);
+	nilfs_bmap_clear_dirty(bmap);
+	up_write(&bmap->b_sem);
+	return ret;
+}
+
+
+/*
+ * Internal use only
+ */
+__u64 nilfs_bmap_data_get_key(const struct nilfs_bmap *bmap,
+			      const struct buffer_head *bh)
+{
+	loff_t pos = folio_pos(bh->b_folio) + bh_offset(bh);
+
+	return pos >> bmap->b_inode->i_blkbits;
+}
+
+__u64 nilfs_bmap_find_target_seq(const struct nilfs_bmap *bmap, __u64 key)
+{
+	__s64 diff;
+
+	diff = key - bmap->b_last_allocated_key;
+	if ((nilfs_bmap_keydiff_abs(diff) < NILFS_INODE_BMAP_SIZE) &&
+	    (bmap->b_last_allocated_ptr != NILFS_BMAP_INVALID_PTR) &&
+	    (bmap->b_last_allocated_ptr + diff > 0))
+		return bmap->b_last_allocated_ptr + diff;
+	else
+		return NILFS_BMAP_INVALID_PTR;
+}
+
+#define NILFS_BMAP_GROUP_DIV	8	/* must be power of 2 */
+
+__u64 nilfs_bmap_find_target_in_group(const struct nilfs_bmap *bmap)
+{
+	struct inode *dat = nilfs_bmap_get_dat(bmap);
+	unsigned long entries_per_group = nilfs_palloc_entries_per_group(dat);
+	unsigned long group;
+	u32 index;
+
+	BUILD_BUG_ON_NOT_POWER_OF_2(NILFS_BMAP_GROUP_DIV);
+
+	group = div_u64(bmap->b_inode->i_ino, entries_per_group);
+	index = bmap->b_inode->i_ino & (NILFS_BMAP_GROUP_DIV - 1);
+
+	return group * entries_per_group +
+	       index * (entries_per_group / NILFS_BMAP_GROUP_DIV);
+}
+
+
+static struct lock_class_key nilfs_bmap_dat_lock_key;
+static struct lock_class_key nilfs_bmap_mdt_lock_key;
+
+/**
+ * nilfs_bmap_read - read a bmap from an inode
+ * @bmap: bmap
+ * @raw_inode: on-disk inode
+ *
+ * Description: nilfs_bmap_read() initializes the bmap @bmap.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (corrupted bmap).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_bmap_read(struct nilfs_bmap *bmap, struct nilfs_inode *raw_inode)
+{
+	if (raw_inode == NULL)
+		memset(bmap->b_u.u_data, 0, NILFS_BMAP_SIZE);
+	else
+		memcpy(bmap->b_u.u_data, raw_inode->i_bmap, NILFS_BMAP_SIZE);
+
+	init_rwsem(&bmap->b_sem);
+	bmap->b_state = 0;
+	bmap->b_inode = &NILFS_BMAP_I(bmap)->vfs_inode;
+	switch (bmap->b_inode->i_ino) {
+	case NILFS_DAT_INO:
+		bmap->b_ptr_type = NILFS_BMAP_PTR_P;
+		bmap->b_last_allocated_key = 0;
+		bmap->b_last_allocated_ptr = NILFS_BMAP_NEW_PTR_INIT;
+		lockdep_set_class(&bmap->b_sem, &nilfs_bmap_dat_lock_key);
+		break;
+	case NILFS_CPFILE_INO:
+	case NILFS_SUFILE_INO:
+		bmap->b_ptr_type = NILFS_BMAP_PTR_VS;
+		bmap->b_last_allocated_key = 0;
+		bmap->b_last_allocated_ptr = NILFS_BMAP_INVALID_PTR;
+		lockdep_set_class(&bmap->b_sem, &nilfs_bmap_mdt_lock_key);
+		break;
+	case NILFS_IFILE_INO:
+		lockdep_set_class(&bmap->b_sem, &nilfs_bmap_mdt_lock_key);
+		fallthrough;
+	default:
+		bmap->b_ptr_type = NILFS_BMAP_PTR_VM;
+		bmap->b_last_allocated_key = 0;
+		bmap->b_last_allocated_ptr = NILFS_BMAP_INVALID_PTR;
+		break;
+	}
+
+	return (bmap->b_u.u_flags & NILFS_BMAP_LARGE) ?
+		nilfs_btree_init(bmap) : nilfs_direct_init(bmap);
+}
+
+/**
+ * nilfs_bmap_write - write back a bmap to an inode
+ * @bmap: bmap
+ * @raw_inode: on-disk inode
+ *
+ * Description: nilfs_bmap_write() stores @bmap in @raw_inode.
+ */
+void nilfs_bmap_write(struct nilfs_bmap *bmap, struct nilfs_inode *raw_inode)
+{
+	memcpy(raw_inode->i_bmap, bmap->b_u.u_data,
+	       NILFS_INODE_BMAP_SIZE * sizeof(__le64));
+	if (bmap->b_inode->i_ino == NILFS_DAT_INO)
+		bmap->b_last_allocated_ptr = NILFS_BMAP_NEW_PTR_INIT;
+}
+
+void nilfs_bmap_init_gc(struct nilfs_bmap *bmap)
+{
+	memset(&bmap->b_u, 0, NILFS_BMAP_SIZE);
+	init_rwsem(&bmap->b_sem);
+	bmap->b_inode = &NILFS_BMAP_I(bmap)->vfs_inode;
+	bmap->b_ptr_type = NILFS_BMAP_PTR_U;
+	bmap->b_last_allocated_key = 0;
+	bmap->b_last_allocated_ptr = NILFS_BMAP_INVALID_PTR;
+	bmap->b_state = 0;
+	nilfs_btree_init_gc(bmap);
+}
+
+void nilfs_bmap_save(const struct nilfs_bmap *bmap,
+		     struct nilfs_bmap_store *store)
+{
+	memcpy(store->data, bmap->b_u.u_data, sizeof(store->data));
+	store->last_allocated_key = bmap->b_last_allocated_key;
+	store->last_allocated_ptr = bmap->b_last_allocated_ptr;
+	store->state = bmap->b_state;
+}
+
+void nilfs_bmap_restore(struct nilfs_bmap *bmap,
+			const struct nilfs_bmap_store *store)
+{
+	memcpy(bmap->b_u.u_data, store->data, sizeof(store->data));
+	bmap->b_last_allocated_key = store->last_allocated_key;
+	bmap->b_last_allocated_ptr = store->last_allocated_ptr;
+	bmap->b_state = store->state;
+}

linux-7.2.3/fs/nilfs2/bmap.h

@@ -0,0 +1,283 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS block mapping.
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato.
+ */
+
+#ifndef _NILFS_BMAP_H
+#define _NILFS_BMAP_H
+
+#include <linux/types.h>
+#include <linux/fs.h>
+#include <linux/buffer_head.h>
+#include <linux/nilfs2_ondisk.h>	/* nilfs_binfo, nilfs_inode, etc */
+#include "alloc.h"
+#include "dat.h"
+
+#define NILFS_BMAP_INVALID_PTR	0
+
+#define nilfs_bmap_keydiff_abs(diff)	((diff) < 0 ? -(diff) : (diff))
+
+
+struct nilfs_bmap;
+
+/**
+ * union nilfs_bmap_ptr_req - request for bmap ptr
+ * @bpr_ptr: bmap pointer
+ * @bpr_req: request for persistent allocator
+ */
+union nilfs_bmap_ptr_req {
+	__u64 bpr_ptr;
+	struct nilfs_palloc_req bpr_req;
+};
+
+/**
+ * struct nilfs_bmap_stats - bmap statistics
+ * @bs_nblocks: number of blocks created or deleted
+ */
+struct nilfs_bmap_stats {
+	unsigned int bs_nblocks;
+};
+
+/**
+ * struct nilfs_bmap_operations - bmap operation table
+ * @bop_lookup:               single block search operation
+ * @bop_lookup_contig:        consecutive block search operation
+ * @bop_insert:               block insertion operation
+ * @bop_delete:               block delete operation
+ * @bop_clear:                block mapping resource release operation
+ * @bop_propagate:            operation to propagate dirty state towards the
+ *                            mapping root
+ * @bop_lookup_dirty_buffers: operation to collect dirty block buffers
+ * @bop_assign:               disk block address assignment operation
+ * @bop_mark:                 operation to mark in-use blocks as dirty for
+ *                            relocation by GC
+ * @bop_seek_key:             find valid block key operation
+ * @bop_last_key:             find last valid block key operation
+ */
+struct nilfs_bmap_operations {
+	int (*bop_lookup)(const struct nilfs_bmap *, __u64, int, __u64 *);
+	int (*bop_lookup_contig)(const struct nilfs_bmap *, __u64, __u64 *,
+				 unsigned int);
+	int (*bop_insert)(struct nilfs_bmap *, __u64, __u64);
+	int (*bop_delete)(struct nilfs_bmap *, __u64);
+	void (*bop_clear)(struct nilfs_bmap *);
+
+	int (*bop_propagate)(struct nilfs_bmap *, struct buffer_head *);
+	void (*bop_lookup_dirty_buffers)(struct nilfs_bmap *,
+					 struct list_head *);
+
+	int (*bop_assign)(struct nilfs_bmap *,
+			  struct buffer_head **,
+			  sector_t,
+			  union nilfs_binfo *);
+	int (*bop_mark)(struct nilfs_bmap *, __u64, int);
+
+	int (*bop_seek_key)(const struct nilfs_bmap *, __u64, __u64 *);
+	int (*bop_last_key)(const struct nilfs_bmap *, __u64 *);
+
+	/* private: internal use only */
+	int (*bop_check_insert)(const struct nilfs_bmap *, __u64);
+	int (*bop_check_delete)(struct nilfs_bmap *, __u64);
+	int (*bop_gather_data)(struct nilfs_bmap *, __u64 *, __u64 *, int);
+};
+
+
+#define NILFS_BMAP_SIZE		(NILFS_INODE_BMAP_SIZE * sizeof(__le64))
+#define NILFS_BMAP_KEY_BIT	BITS_PER_LONG
+#define NILFS_BMAP_NEW_PTR_INIT	(1UL << (BITS_PER_LONG - 1))
+
+static inline int nilfs_bmap_is_new_ptr(unsigned long ptr)
+{
+	return !!(ptr & NILFS_BMAP_NEW_PTR_INIT);
+}
+
+
+/**
+ * struct nilfs_bmap - bmap structure
+ * @b_u: raw data
+ * @b_sem: semaphore
+ * @b_inode: owner of bmap
+ * @b_ops: bmap operation table
+ * @b_last_allocated_key: last allocated key for data block
+ * @b_last_allocated_ptr: last allocated ptr for data block
+ * @b_ptr_type: pointer type
+ * @b_state: state
+ * @b_nchildren_per_block: maximum number of child nodes for non-root nodes
+ */
+struct nilfs_bmap {
+	union {
+		__u8 u_flags;
+		__le64 u_data[NILFS_BMAP_SIZE / sizeof(__le64)];
+	} b_u;
+	struct rw_semaphore b_sem;
+	struct inode *b_inode;
+	const struct nilfs_bmap_operations *b_ops;
+	__u64 b_last_allocated_key;
+	__u64 b_last_allocated_ptr;
+	int b_ptr_type;
+	int b_state;
+	__u16 b_nchildren_per_block;
+};
+
+/* pointer type */
+#define NILFS_BMAP_PTR_P	0	/* physical block number (i.e. LBN) */
+#define NILFS_BMAP_PTR_VS	1	/*
+					 * virtual block number (single
+					 * version)
+					 */
+#define NILFS_BMAP_PTR_VM	2	/*
+					 * virtual block number (has multiple
+					 * versions)
+					 */
+#define NILFS_BMAP_PTR_U	(-1)	/* never perform pointer operations */
+
+#define NILFS_BMAP_USE_VBN(bmap)	((bmap)->b_ptr_type > 0)
+
+/* state */
+#define NILFS_BMAP_DIRTY	0x00000001
+
+/**
+ * struct nilfs_bmap_store - shadow copy of bmap state
+ * @data: cached raw block mapping of on-disk inode
+ * @last_allocated_key: cached value of last allocated key for data block
+ * @last_allocated_ptr: cached value of last allocated ptr for data block
+ * @state: cached value of state field of bmap structure
+ */
+struct nilfs_bmap_store {
+	__le64 data[NILFS_BMAP_SIZE / sizeof(__le64)];
+	__u64 last_allocated_key;
+	__u64 last_allocated_ptr;
+	int state;
+};
+
+int nilfs_bmap_test_and_clear_dirty(struct nilfs_bmap *);
+int nilfs_bmap_read(struct nilfs_bmap *, struct nilfs_inode *);
+void nilfs_bmap_write(struct nilfs_bmap *, struct nilfs_inode *);
+int nilfs_bmap_lookup_contig(struct nilfs_bmap *, __u64, __u64 *, unsigned int);
+int nilfs_bmap_insert(struct nilfs_bmap *bmap, __u64 key, unsigned long rec);
+int nilfs_bmap_delete(struct nilfs_bmap *bmap, __u64 key);
+int nilfs_bmap_seek_key(struct nilfs_bmap *bmap, __u64 start, __u64 *keyp);
+int nilfs_bmap_last_key(struct nilfs_bmap *bmap, __u64 *keyp);
+int nilfs_bmap_truncate(struct nilfs_bmap *bmap, __u64 key);
+void nilfs_bmap_clear(struct nilfs_bmap *);
+int nilfs_bmap_propagate(struct nilfs_bmap *, struct buffer_head *);
+void nilfs_bmap_lookup_dirty_buffers(struct nilfs_bmap *, struct list_head *);
+int nilfs_bmap_assign(struct nilfs_bmap *, struct buffer_head **,
+		      unsigned long, union nilfs_binfo *);
+int nilfs_bmap_lookup_at_level(struct nilfs_bmap *, __u64, int, __u64 *);
+int nilfs_bmap_mark(struct nilfs_bmap *, __u64, int);
+
+void nilfs_bmap_init_gc(struct nilfs_bmap *);
+
+void nilfs_bmap_save(const struct nilfs_bmap *, struct nilfs_bmap_store *);
+void nilfs_bmap_restore(struct nilfs_bmap *, const struct nilfs_bmap_store *);
+
+static inline int nilfs_bmap_lookup(struct nilfs_bmap *bmap, __u64 key,
+				    __u64 *ptr)
+{
+	return nilfs_bmap_lookup_at_level(bmap, key, 1, ptr);
+}
+
+/*
+ * Internal use only
+ */
+struct inode *nilfs_bmap_get_dat(const struct nilfs_bmap *);
+
+static inline int nilfs_bmap_prepare_alloc_ptr(struct nilfs_bmap *bmap,
+					       union nilfs_bmap_ptr_req *req,
+					       struct inode *dat)
+{
+	if (dat)
+		return nilfs_dat_prepare_alloc(dat, &req->bpr_req);
+	/* ignore target ptr */
+	req->bpr_ptr = bmap->b_last_allocated_ptr++;
+	return 0;
+}
+
+static inline void nilfs_bmap_commit_alloc_ptr(struct nilfs_bmap *bmap,
+					       union nilfs_bmap_ptr_req *req,
+					       struct inode *dat)
+{
+	if (dat)
+		nilfs_dat_commit_alloc(dat, &req->bpr_req);
+}
+
+static inline void nilfs_bmap_abort_alloc_ptr(struct nilfs_bmap *bmap,
+					      union nilfs_bmap_ptr_req *req,
+					      struct inode *dat)
+{
+	if (dat)
+		nilfs_dat_abort_alloc(dat, &req->bpr_req);
+	else
+		bmap->b_last_allocated_ptr--;
+}
+
+static inline int nilfs_bmap_prepare_end_ptr(struct nilfs_bmap *bmap,
+					     union nilfs_bmap_ptr_req *req,
+					     struct inode *dat)
+{
+	return dat ? nilfs_dat_prepare_end(dat, &req->bpr_req) : 0;
+}
+
+static inline void nilfs_bmap_commit_end_ptr(struct nilfs_bmap *bmap,
+					     union nilfs_bmap_ptr_req *req,
+					     struct inode *dat)
+{
+	if (dat)
+		nilfs_dat_commit_end(dat, &req->bpr_req,
+				     bmap->b_ptr_type == NILFS_BMAP_PTR_VS);
+}
+
+static inline void nilfs_bmap_abort_end_ptr(struct nilfs_bmap *bmap,
+					    union nilfs_bmap_ptr_req *req,
+					    struct inode *dat)
+{
+	if (dat)
+		nilfs_dat_abort_end(dat, &req->bpr_req);
+}
+
+static inline void nilfs_bmap_set_target_v(struct nilfs_bmap *bmap, __u64 key,
+					   __u64 ptr)
+{
+	bmap->b_last_allocated_key = key;
+	bmap->b_last_allocated_ptr = ptr;
+}
+
+__u64 nilfs_bmap_data_get_key(const struct nilfs_bmap *,
+			      const struct buffer_head *);
+
+__u64 nilfs_bmap_find_target_seq(const struct nilfs_bmap *, __u64);
+__u64 nilfs_bmap_find_target_in_group(const struct nilfs_bmap *);
+
+
+/* Assume that bmap semaphore is locked. */
+static inline int nilfs_bmap_dirty(const struct nilfs_bmap *bmap)
+{
+	return !!(bmap->b_state & NILFS_BMAP_DIRTY);
+}
+
+/* Assume that bmap semaphore is locked. */
+static inline void nilfs_bmap_set_dirty(struct nilfs_bmap *bmap)
+{
+	bmap->b_state |= NILFS_BMAP_DIRTY;
+}
+
+/* Assume that bmap semaphore is locked. */
+static inline void nilfs_bmap_clear_dirty(struct nilfs_bmap *bmap)
+{
+	bmap->b_state &= ~NILFS_BMAP_DIRTY;
+}
+
+
+#define NILFS_BMAP_LARGE	0x1
+
+#define NILFS_BMAP_SMALL_LOW	NILFS_DIRECT_KEY_MIN
+#define NILFS_BMAP_SMALL_HIGH	NILFS_DIRECT_KEY_MAX
+#define NILFS_BMAP_LARGE_LOW	NILFS_BTREE_ROOT_NCHILDREN_MAX
+#define NILFS_BMAP_LARGE_HIGH	NILFS_BTREE_KEY_MAX
+
+#endif	/* _NILFS_BMAP_H */

linux-7.2.3/fs/nilfs2/btnode.c

@@ -0,0 +1,356 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS B-tree node cache
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Originally written by Seiji Kihara.
+ * Fully revised by Ryusuke Konishi for stabilization and simplification.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include <linux/mm.h>
+#include <linux/backing-dev.h>
+#include <linux/gfp.h>
+#include "nilfs.h"
+#include "mdt.h"
+#include "dat.h"
+#include "page.h"
+#include "btnode.h"
+
+
+/**
+ * nilfs_init_btnc_inode - initialize B-tree node cache inode
+ * @btnc_inode: inode to be initialized
+ *
+ * nilfs_init_btnc_inode() sets up an inode for B-tree node cache.
+ */
+void nilfs_init_btnc_inode(struct inode *btnc_inode)
+{
+	struct nilfs_inode_info *ii = NILFS_I(btnc_inode);
+
+	btnc_inode->i_mode = S_IFREG;
+	ii->i_flags = 0;
+	memset(&ii->i_bmap_data, 0, sizeof(struct nilfs_bmap));
+	mapping_set_gfp_mask(btnc_inode->i_mapping, GFP_NOFS);
+	btnc_inode->i_mapping->a_ops = &nilfs_buffer_cache_aops;
+}
+
+void nilfs_btnode_cache_clear(struct address_space *btnc)
+{
+	invalidate_mapping_pages(btnc, 0, -1);
+	truncate_inode_pages(btnc, 0);
+}
+
+struct buffer_head *
+nilfs_btnode_create_block(struct address_space *btnc, __u64 blocknr)
+{
+	struct inode *inode = btnc->host;
+	struct buffer_head *bh;
+
+	bh = nilfs_grab_buffer(inode, btnc, blocknr, BIT(BH_NILFS_Node));
+	if (unlikely(!bh))
+		return ERR_PTR(-ENOMEM);
+
+	if (unlikely(buffer_mapped(bh) || buffer_uptodate(bh) ||
+		     buffer_dirty(bh))) {
+		/*
+		 * The block buffer at the specified new address was already
+		 * in use.  This can happen if it is a virtual block number
+		 * and has been reallocated due to corruption of the bitmap
+		 * used to manage its allocation state (if not, the buffer
+		 * clearing of an abandoned b-tree node is missing somewhere).
+		 */
+		nilfs_error(inode->i_sb,
+			    "state inconsistency probably due to duplicate use of b-tree node block address %llu (ino=%llu)",
+			    (unsigned long long)blocknr, inode->i_ino);
+		goto failed;
+	}
+	memset(bh->b_data, 0, i_blocksize(inode));
+	bh->b_blocknr = blocknr;
+	set_buffer_mapped(bh);
+	set_buffer_uptodate(bh);
+
+	folio_unlock(bh->b_folio);
+	folio_put(bh->b_folio);
+	return bh;
+
+failed:
+	folio_unlock(bh->b_folio);
+	folio_put(bh->b_folio);
+	brelse(bh);
+	return ERR_PTR(-EIO);
+}
+
+int nilfs_btnode_submit_block(struct address_space *btnc, __u64 blocknr,
+			      sector_t pblocknr, blk_opf_t opf,
+			      struct buffer_head **pbh, sector_t *submit_ptr)
+{
+	struct buffer_head *bh;
+	struct inode *inode = btnc->host;
+	struct folio *folio;
+	int err;
+
+	bh = nilfs_grab_buffer(inode, btnc, blocknr, BIT(BH_NILFS_Node));
+	if (unlikely(!bh))
+		return -ENOMEM;
+
+	err = -EEXIST; /* internal code */
+	folio = bh->b_folio;
+
+	if (buffer_uptodate(bh) || buffer_dirty(bh))
+		goto found;
+
+	if (pblocknr == 0) {
+		pblocknr = blocknr;
+		if (inode->i_ino != NILFS_DAT_INO) {
+			struct the_nilfs *nilfs = inode->i_sb->s_fs_info;
+
+			/* blocknr is a virtual block number */
+			err = nilfs_dat_translate(nilfs->ns_dat, blocknr,
+						  &pblocknr);
+			if (unlikely(err)) {
+				brelse(bh);
+				goto out_locked;
+			}
+		}
+	}
+
+	if (opf & REQ_RAHEAD) {
+		if (pblocknr != *submit_ptr + 1 || !trylock_buffer(bh)) {
+			err = -EBUSY; /* internal code */
+			brelse(bh);
+			goto out_locked;
+		}
+	} else { /* opf == REQ_OP_READ */
+		lock_buffer(bh);
+	}
+	if (buffer_uptodate(bh)) {
+		unlock_buffer(bh);
+		err = -EEXIST; /* internal code */
+		goto found;
+	}
+	set_buffer_mapped(bh);
+	bh->b_blocknr = pblocknr; /* set block address for read */
+	bh_submit(bh, opf, bh_end_read);
+	bh->b_blocknr = blocknr; /* set back to the given block address */
+	*submit_ptr = pblocknr;
+	err = 0;
+found:
+	*pbh = bh;
+
+out_locked:
+	folio_unlock(folio);
+	folio_put(folio);
+	return err;
+}
+
+/**
+ * nilfs_btnode_delete - delete B-tree node buffer
+ * @bh: buffer to be deleted
+ *
+ * nilfs_btnode_delete() invalidates the specified buffer and delete the page
+ * including the buffer if the page gets unbusy.
+ */
+void nilfs_btnode_delete(struct buffer_head *bh)
+{
+	struct address_space *mapping;
+	struct folio *folio = bh->b_folio;
+	pgoff_t index = folio->index;
+	int still_dirty;
+
+	folio_get(folio);
+	folio_lock(folio);
+	folio_wait_writeback(folio);
+
+	nilfs_forget_buffer(bh);
+	still_dirty = folio_test_dirty(folio);
+	mapping = folio->mapping;
+	folio_unlock(folio);
+	folio_put(folio);
+
+	if (!still_dirty && mapping)
+		invalidate_inode_pages2_range(mapping, index, index);
+}
+
+/**
+ * nilfs_btnode_prepare_change_key - prepare to change the search key of a
+ *                                   b-tree node block
+ * @btnc: page cache in which the b-tree node block is buffered
+ * @ctxt: structure for exchanging context information for key change
+ *
+ * nilfs_btnode_prepare_change_key() prepares to move the contents of the
+ * b-tree node block of the old key given in the "oldkey" member of @ctxt to
+ * the position of the new key given in the "newkey" member of @ctxt in the
+ * page cache @btnc.  Here, the key of the block is an index in units of
+ * blocks, and if the page and block sizes match, it matches the page index
+ * in the page cache.
+ *
+ * If the page size and block size match, this function attempts to move the
+ * entire folio, and in preparation for this, inserts the original folio into
+ * the new index of the cache.  If this insertion fails or if the page size
+ * and block size are different, it falls back to a copy preparation using
+ * nilfs_btnode_create_block(), inserts a new block at the position
+ * corresponding to "newkey", and stores the buffer head pointer in the
+ * "newbh" member of @ctxt.
+ *
+ * Note that the current implementation does not support folio sizes larger
+ * than the page size.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_btnode_prepare_change_key(struct address_space *btnc,
+				    struct nilfs_btnode_chkey_ctxt *ctxt)
+{
+	struct buffer_head *obh, *nbh;
+	struct inode *inode = btnc->host;
+	__u64 oldkey = ctxt->oldkey, newkey = ctxt->newkey;
+	int err;
+
+	if (oldkey == newkey)
+		return 0;
+
+	obh = ctxt->bh;
+	ctxt->newbh = NULL;
+
+	if (inode->i_blkbits == PAGE_SHIFT) {
+		struct folio *ofolio = obh->b_folio;
+		folio_lock(ofolio);
+retry:
+		/* BUG_ON(oldkey != obh->b_folio->index); */
+		if (unlikely(oldkey != ofolio->index))
+			NILFS_FOLIO_BUG(ofolio,
+				       "invalid oldkey %lld (newkey=%lld)",
+				       (unsigned long long)oldkey,
+				       (unsigned long long)newkey);
+
+		xa_lock_irq(&btnc->i_pages);
+		err = __xa_insert(&btnc->i_pages, newkey, ofolio, GFP_NOFS);
+		xa_unlock_irq(&btnc->i_pages);
+		/*
+		 * Note: folio->index will not change to newkey until
+		 * nilfs_btnode_commit_change_key() will be called.
+		 * To protect the folio in intermediate state, the folio lock
+		 * is held.
+		 */
+		if (!err)
+			return 0;
+		else if (err != -EBUSY)
+			goto failed_unlock;
+
+		err = invalidate_inode_pages2_range(btnc, newkey, newkey);
+		if (!err)
+			goto retry;
+		/* fallback to copy mode */
+		folio_unlock(ofolio);
+	}
+
+	nbh = nilfs_btnode_create_block(btnc, newkey);
+	if (IS_ERR(nbh))
+		return PTR_ERR(nbh);
+
+	BUG_ON(nbh == obh);
+	ctxt->newbh = nbh;
+	return 0;
+
+ failed_unlock:
+	folio_unlock(obh->b_folio);
+	return err;
+}
+
+/**
+ * nilfs_btnode_commit_change_key - commit the change of the search key of
+ *                                  a b-tree node block
+ * @btnc: page cache in which the b-tree node block is buffered
+ * @ctxt: structure for exchanging context information for key change
+ *
+ * nilfs_btnode_commit_change_key() executes the key change based on the
+ * context @ctxt prepared by nilfs_btnode_prepare_change_key().  If no valid
+ * block buffer is prepared in "newbh" of @ctxt (i.e., a full folio move),
+ * this function removes the folio from the old index and completes the move.
+ * Otherwise, it copies the block data and inherited flag states of "oldbh"
+ * to "newbh" and clears the "oldbh" from the cache.  In either case, the
+ * relocated buffer is marked as dirty.
+ *
+ * As with nilfs_btnode_prepare_change_key(), the current implementation does
+ * not support folio sizes larger than the page size.
+ */
+void nilfs_btnode_commit_change_key(struct address_space *btnc,
+				    struct nilfs_btnode_chkey_ctxt *ctxt)
+{
+	struct buffer_head *obh = ctxt->bh, *nbh = ctxt->newbh;
+	__u64 oldkey = ctxt->oldkey, newkey = ctxt->newkey;
+	struct folio *ofolio;
+
+	if (oldkey == newkey)
+		return;
+
+	if (nbh == NULL) {	/* blocksize == pagesize */
+		ofolio = obh->b_folio;
+		if (unlikely(oldkey != ofolio->index))
+			NILFS_FOLIO_BUG(ofolio,
+				       "invalid oldkey %lld (newkey=%lld)",
+				       (unsigned long long)oldkey,
+				       (unsigned long long)newkey);
+		mark_buffer_dirty(obh);
+
+		xa_lock_irq(&btnc->i_pages);
+		__xa_erase(&btnc->i_pages, oldkey);
+		__xa_set_mark(&btnc->i_pages, newkey, PAGECACHE_TAG_DIRTY);
+		xa_unlock_irq(&btnc->i_pages);
+
+		ofolio->index = obh->b_blocknr = newkey;
+		folio_unlock(ofolio);
+	} else {
+		nilfs_copy_buffer(nbh, obh);
+		mark_buffer_dirty(nbh);
+
+		nbh->b_blocknr = newkey;
+		ctxt->bh = nbh;
+		nilfs_btnode_delete(obh); /* will decrement bh->b_count */
+	}
+}
+
+/**
+ * nilfs_btnode_abort_change_key - abort the change of the search key of a
+ *                                 b-tree node block
+ * @btnc: page cache in which the b-tree node block is buffered
+ * @ctxt: structure for exchanging context information for key change
+ *
+ * nilfs_btnode_abort_change_key() cancels the key change associated with the
+ * context @ctxt prepared via nilfs_btnode_prepare_change_key() and performs
+ * any necessary cleanup.  If no valid block buffer is prepared in "newbh" of
+ * @ctxt, this function removes the folio from the destination index and aborts
+ * the move.  Otherwise, it clears "newbh" from the cache.
+ *
+ * As with nilfs_btnode_prepare_change_key(), the current implementation does
+ * not support folio sizes larger than the page size.
+ */
+void nilfs_btnode_abort_change_key(struct address_space *btnc,
+				   struct nilfs_btnode_chkey_ctxt *ctxt)
+{
+	struct buffer_head *nbh = ctxt->newbh;
+	__u64 oldkey = ctxt->oldkey, newkey = ctxt->newkey;
+
+	if (oldkey == newkey)
+		return;
+
+	if (nbh == NULL) {	/* blocksize == pagesize */
+		xa_erase_irq(&btnc->i_pages, newkey);
+		folio_unlock(ctxt->bh->b_folio);
+	} else {
+		/*
+		 * When canceling a buffer that a prepare operation has
+		 * allocated to copy a node block to another location, use
+		 * nilfs_btnode_delete() to initialize and release the buffer
+		 * so that the buffer flags will not be in an inconsistent
+		 * state when it is reallocated.
+		 */
+		nilfs_btnode_delete(nbh);
+	}
+}

linux-7.2.3/fs/nilfs2/btnode.h

@@ -0,0 +1,47 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS B-tree node cache
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Seiji Kihara.
+ * Revised by Ryusuke Konishi.
+ */
+
+#ifndef _NILFS_BTNODE_H
+#define _NILFS_BTNODE_H
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include <linux/fs.h>
+#include <linux/backing-dev.h>
+
+/**
+ * struct nilfs_btnode_chkey_ctxt - change key context
+ * @oldkey: old key of block's moving content
+ * @newkey: new key for block's content
+ * @bh: buffer head of old buffer
+ * @newbh: buffer head of new buffer
+ */
+struct nilfs_btnode_chkey_ctxt {
+	__u64 oldkey;
+	__u64 newkey;
+	struct buffer_head *bh;
+	struct buffer_head *newbh;
+};
+
+void nilfs_init_btnc_inode(struct inode *btnc_inode);
+void nilfs_btnode_cache_clear(struct address_space *);
+struct buffer_head *nilfs_btnode_create_block(struct address_space *btnc,
+					      __u64 blocknr);
+int nilfs_btnode_submit_block(struct address_space *, __u64, sector_t,
+			      blk_opf_t, struct buffer_head **, sector_t *);
+void nilfs_btnode_delete(struct buffer_head *);
+int nilfs_btnode_prepare_change_key(struct address_space *,
+				    struct nilfs_btnode_chkey_ctxt *);
+void nilfs_btnode_commit_change_key(struct address_space *,
+				    struct nilfs_btnode_chkey_ctxt *);
+void nilfs_btnode_abort_change_key(struct address_space *,
+				   struct nilfs_btnode_chkey_ctxt *);
+
+#endif	/* _NILFS_BTNODE_H */

linux-7.2.3/fs/nilfs2/btree.h

@@ -0,0 +1,65 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS B-tree.
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato.
+ */
+
+#ifndef _NILFS_BTREE_H
+#define _NILFS_BTREE_H
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include <linux/list.h>
+#include <linux/nilfs2_ondisk.h>	/* nilfs_btree_node */
+#include "btnode.h"
+#include "bmap.h"
+
+/**
+ * struct nilfs_btree_path - A path on which B-tree operations are executed
+ * @bp_bh: buffer head of node block
+ * @bp_sib_bh: buffer head of sibling node block
+ * @bp_index: index of child node
+ * @bp_oldreq: ptr end request for old ptr
+ * @bp_newreq: ptr alloc request for new ptr
+ * @bp_ctxt: context information for changing the key of a b-tree node block
+ * @bp_op: rebalance operation
+ */
+struct nilfs_btree_path {
+	struct buffer_head *bp_bh;
+	struct buffer_head *bp_sib_bh;
+	int bp_index;
+	union nilfs_bmap_ptr_req bp_oldreq;
+	union nilfs_bmap_ptr_req bp_newreq;
+	struct nilfs_btnode_chkey_ctxt bp_ctxt;
+	void (*bp_op)(struct nilfs_bmap *, struct nilfs_btree_path *,
+		      int, __u64 *, __u64 *);
+};
+
+#define NILFS_BTREE_ROOT_SIZE		NILFS_BMAP_SIZE
+#define NILFS_BTREE_ROOT_NCHILDREN_MAX					\
+	((NILFS_BTREE_ROOT_SIZE - sizeof(struct nilfs_btree_node)) /	\
+	 (sizeof(__le64 /* dkey */) + sizeof(__le64 /* dptr */)))
+#define NILFS_BTREE_ROOT_NCHILDREN_MIN	0
+#define NILFS_BTREE_NODE_EXTRA_PAD_SIZE	(sizeof(__le64))
+#define NILFS_BTREE_NODE_NCHILDREN_MAX(nodesize)			\
+	(((nodesize) - sizeof(struct nilfs_btree_node) -		\
+		NILFS_BTREE_NODE_EXTRA_PAD_SIZE) /			\
+	 (sizeof(__le64 /* dkey */) + sizeof(__le64 /* dptr */)))
+#define NILFS_BTREE_NODE_NCHILDREN_MIN(nodesize)			\
+	((NILFS_BTREE_NODE_NCHILDREN_MAX(nodesize) - 1) / 2 + 1)
+#define NILFS_BTREE_KEY_MIN	((__u64)0)
+#define NILFS_BTREE_KEY_MAX	(~(__u64)0)
+
+extern struct kmem_cache *nilfs_btree_path_cache;
+
+int nilfs_btree_init(struct nilfs_bmap *);
+int nilfs_btree_convert_and_insert(struct nilfs_bmap *, __u64, __u64,
+				   const __u64 *, const __u64 *, int);
+void nilfs_btree_init_gc(struct nilfs_bmap *);
+
+int nilfs_btree_broken_node_block(struct buffer_head *bh);
+
+#endif	/* _NILFS_BTREE_H */

linux-7.2.3/fs/nilfs2/cpfile.h

@@ -0,0 +1,36 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS checkpoint file.
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato.
+ */
+
+#ifndef _NILFS_CPFILE_H
+#define _NILFS_CPFILE_H
+
+#include <linux/fs.h>
+#include <linux/buffer_head.h>
+#include <linux/nilfs2_api.h>		/* nilfs_cpstat */
+#include <linux/nilfs2_ondisk.h>	/* nilfs_inode, nilfs_checkpoint */
+
+
+int nilfs_cpfile_read_checkpoint(struct inode *cpfile, __u64 cno,
+				 struct nilfs_root *root, struct inode *ifile);
+int nilfs_cpfile_create_checkpoint(struct inode *cpfile, __u64 cno);
+int nilfs_cpfile_finalize_checkpoint(struct inode *cpfile, __u64 cno,
+				     struct nilfs_root *root, __u64 blkinc,
+				     time64_t ctime, bool minor);
+int nilfs_cpfile_delete_checkpoints(struct inode *, __u64, __u64);
+int nilfs_cpfile_delete_checkpoint(struct inode *, __u64);
+int nilfs_cpfile_change_cpmode(struct inode *, __u64, int);
+int nilfs_cpfile_is_snapshot(struct inode *, __u64);
+int nilfs_cpfile_get_stat(struct inode *, struct nilfs_cpstat *);
+ssize_t nilfs_cpfile_get_cpinfo(struct inode *, __u64 *, int, void *,
+				unsigned int, size_t);
+
+int nilfs_cpfile_read(struct super_block *sb, size_t cpsize,
+		      struct nilfs_inode *raw_inode, struct inode **inodep);
+
+#endif	/* _NILFS_CPFILE_H */

linux-7.2.3/fs/nilfs2/dat.c

@@ -0,0 +1,541 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS disk address translation.
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato.
+ */
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include <linux/string.h>
+#include <linux/errno.h>
+#include "nilfs.h"
+#include "mdt.h"
+#include "alloc.h"
+#include "dat.h"
+
+
+#define NILFS_CNO_MIN	((__u64)1)
+#define NILFS_CNO_MAX	(~(__u64)0)
+
+/**
+ * struct nilfs_dat_info - on-memory private data of DAT file
+ * @mi: on-memory private data of metadata file
+ * @palloc_cache: persistent object allocator cache of DAT file
+ * @shadow: shadow map of DAT file
+ */
+struct nilfs_dat_info {
+	struct nilfs_mdt_info mi;
+	struct nilfs_palloc_cache palloc_cache;
+	struct nilfs_shadow_map shadow;
+};
+
+static inline struct nilfs_dat_info *NILFS_DAT_I(struct inode *dat)
+{
+	return (struct nilfs_dat_info *)NILFS_MDT(dat);
+}
+
+static int nilfs_dat_prepare_entry(struct inode *dat,
+				   struct nilfs_palloc_req *req, int create)
+{
+	int ret;
+
+	ret = nilfs_palloc_get_entry_block(dat, req->pr_entry_nr,
+					   create, &req->pr_entry_bh);
+	if (unlikely(ret == -ENOENT)) {
+		nilfs_err(dat->i_sb,
+			  "DAT doesn't have a block to manage vblocknr = %llu",
+			  (unsigned long long)req->pr_entry_nr);
+		/*
+		 * Return internal code -EINVAL to notify bmap layer of
+		 * metadata corruption.
+		 */
+		ret = -EINVAL;
+	}
+	return ret;
+}
+
+static void nilfs_dat_commit_entry(struct inode *dat,
+				   struct nilfs_palloc_req *req)
+{
+	mark_buffer_dirty(req->pr_entry_bh);
+	nilfs_mdt_mark_dirty(dat);
+	brelse(req->pr_entry_bh);
+}
+
+static void nilfs_dat_abort_entry(struct inode *dat,
+				  struct nilfs_palloc_req *req)
+{
+	brelse(req->pr_entry_bh);
+}
+
+int nilfs_dat_prepare_alloc(struct inode *dat, struct nilfs_palloc_req *req)
+{
+	int ret;
+
+	ret = nilfs_palloc_prepare_alloc_entry(dat, req, true);
+	if (ret < 0)
+		return ret;
+
+	ret = nilfs_dat_prepare_entry(dat, req, 1);
+	if (ret < 0)
+		nilfs_palloc_abort_alloc_entry(dat, req);
+
+	return ret;
+}
+
+void nilfs_dat_commit_alloc(struct inode *dat, struct nilfs_palloc_req *req)
+{
+	struct nilfs_dat_entry *entry;
+	size_t offset;
+
+	offset = nilfs_palloc_entry_offset(dat, req->pr_entry_nr,
+					   req->pr_entry_bh);
+	entry = kmap_local_folio(req->pr_entry_bh->b_folio, offset);
+	entry->de_start = cpu_to_le64(NILFS_CNO_MIN);
+	entry->de_end = cpu_to_le64(NILFS_CNO_MAX);
+	entry->de_blocknr = cpu_to_le64(0);
+	kunmap_local(entry);
+
+	nilfs_palloc_commit_alloc_entry(dat, req);
+	nilfs_dat_commit_entry(dat, req);
+}
+
+void nilfs_dat_abort_alloc(struct inode *dat, struct nilfs_palloc_req *req)
+{
+	nilfs_dat_abort_entry(dat, req);
+	nilfs_palloc_abort_alloc_entry(dat, req);
+}
+
+static void nilfs_dat_commit_free(struct inode *dat,
+				  struct nilfs_palloc_req *req)
+{
+	struct nilfs_dat_entry *entry;
+	size_t offset;
+
+	offset = nilfs_palloc_entry_offset(dat, req->pr_entry_nr,
+					   req->pr_entry_bh);
+	entry = kmap_local_folio(req->pr_entry_bh->b_folio, offset);
+	entry->de_start = cpu_to_le64(NILFS_CNO_MIN);
+	entry->de_end = cpu_to_le64(NILFS_CNO_MIN);
+	entry->de_blocknr = cpu_to_le64(0);
+	kunmap_local(entry);
+
+	nilfs_dat_commit_entry(dat, req);
+
+	if (unlikely(req->pr_desc_bh == NULL || req->pr_bitmap_bh == NULL)) {
+		nilfs_error(dat->i_sb,
+			    "state inconsistency probably due to duplicate use of vblocknr = %llu",
+			    (unsigned long long)req->pr_entry_nr);
+		return;
+	}
+	nilfs_palloc_commit_free_entry(dat, req);
+}
+
+int nilfs_dat_prepare_start(struct inode *dat, struct nilfs_palloc_req *req)
+{
+	return nilfs_dat_prepare_entry(dat, req, 0);
+}
+
+void nilfs_dat_commit_start(struct inode *dat, struct nilfs_palloc_req *req,
+			    sector_t blocknr)
+{
+	struct nilfs_dat_entry *entry;
+	size_t offset;
+
+	offset = nilfs_palloc_entry_offset(dat, req->pr_entry_nr,
+					   req->pr_entry_bh);
+	entry = kmap_local_folio(req->pr_entry_bh->b_folio, offset);
+	entry->de_start = cpu_to_le64(nilfs_mdt_cno(dat));
+	entry->de_blocknr = cpu_to_le64(blocknr);
+	kunmap_local(entry);
+
+	nilfs_dat_commit_entry(dat, req);
+}
+
+int nilfs_dat_prepare_end(struct inode *dat, struct nilfs_palloc_req *req)
+{
+	struct nilfs_dat_entry *entry;
+	__u64 start;
+	sector_t blocknr;
+	size_t offset;
+	int ret;
+
+	ret = nilfs_dat_prepare_entry(dat, req, 0);
+	if (ret < 0)
+		return ret;
+
+	offset = nilfs_palloc_entry_offset(dat, req->pr_entry_nr,
+					   req->pr_entry_bh);
+	entry = kmap_local_folio(req->pr_entry_bh->b_folio, offset);
+	start = le64_to_cpu(entry->de_start);
+	blocknr = le64_to_cpu(entry->de_blocknr);
+	kunmap_local(entry);
+
+	if (blocknr == 0) {
+		ret = nilfs_palloc_prepare_free_entry(dat, req);
+		if (ret < 0) {
+			nilfs_dat_abort_entry(dat, req);
+			return ret;
+		}
+	}
+	if (unlikely(start > nilfs_mdt_cno(dat))) {
+		nilfs_err(dat->i_sb,
+			  "vblocknr = %llu has abnormal lifetime: start cno (= %llu) > current cno (= %llu)",
+			  (unsigned long long)req->pr_entry_nr,
+			  (unsigned long long)start,
+			  (unsigned long long)nilfs_mdt_cno(dat));
+		nilfs_dat_abort_entry(dat, req);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+void nilfs_dat_commit_end(struct inode *dat, struct nilfs_palloc_req *req,
+			  int dead)
+{
+	struct nilfs_dat_entry *entry;
+	__u64 start, end;
+	sector_t blocknr;
+	size_t offset;
+
+	offset = nilfs_palloc_entry_offset(dat, req->pr_entry_nr,
+					   req->pr_entry_bh);
+	entry = kmap_local_folio(req->pr_entry_bh->b_folio, offset);
+	end = start = le64_to_cpu(entry->de_start);
+	if (!dead) {
+		end = nilfs_mdt_cno(dat);
+		WARN_ON(start > end);
+	}
+	entry->de_end = cpu_to_le64(end);
+	blocknr = le64_to_cpu(entry->de_blocknr);
+	kunmap_local(entry);
+
+	if (blocknr == 0)
+		nilfs_dat_commit_free(dat, req);
+	else
+		nilfs_dat_commit_entry(dat, req);
+}
+
+void nilfs_dat_abort_end(struct inode *dat, struct nilfs_palloc_req *req)
+{
+	struct nilfs_dat_entry *entry;
+	__u64 start;
+	sector_t blocknr;
+	size_t offset;
+
+	offset = nilfs_palloc_entry_offset(dat, req->pr_entry_nr,
+					   req->pr_entry_bh);
+	entry = kmap_local_folio(req->pr_entry_bh->b_folio, offset);
+	start = le64_to_cpu(entry->de_start);
+	blocknr = le64_to_cpu(entry->de_blocknr);
+	kunmap_local(entry);
+
+	if (start == nilfs_mdt_cno(dat) && blocknr == 0)
+		nilfs_palloc_abort_free_entry(dat, req);
+	nilfs_dat_abort_entry(dat, req);
+}
+
+int nilfs_dat_prepare_update(struct inode *dat,
+			     struct nilfs_palloc_req *oldreq,
+			     struct nilfs_palloc_req *newreq)
+{
+	int ret;
+
+	ret = nilfs_dat_prepare_end(dat, oldreq);
+	if (!ret) {
+		ret = nilfs_dat_prepare_alloc(dat, newreq);
+		if (ret < 0)
+			nilfs_dat_abort_end(dat, oldreq);
+	}
+	return ret;
+}
+
+void nilfs_dat_commit_update(struct inode *dat,
+			     struct nilfs_palloc_req *oldreq,
+			     struct nilfs_palloc_req *newreq, int dead)
+{
+	nilfs_dat_commit_end(dat, oldreq, dead);
+	nilfs_dat_commit_alloc(dat, newreq);
+}
+
+void nilfs_dat_abort_update(struct inode *dat,
+			    struct nilfs_palloc_req *oldreq,
+			    struct nilfs_palloc_req *newreq)
+{
+	nilfs_dat_abort_end(dat, oldreq);
+	nilfs_dat_abort_alloc(dat, newreq);
+}
+
+/**
+ * nilfs_dat_mark_dirty - mark the DAT block buffer containing the specified
+ *                        virtual block address entry as dirty
+ * @dat:      DAT file inode
+ * @vblocknr: virtual block number
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EINVAL	- Invalid DAT entry (internal code).
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_dat_mark_dirty(struct inode *dat, __u64 vblocknr)
+{
+	struct nilfs_palloc_req req;
+	int ret;
+
+	req.pr_entry_nr = vblocknr;
+	ret = nilfs_dat_prepare_entry(dat, &req, 0);
+	if (ret == 0)
+		nilfs_dat_commit_entry(dat, &req);
+	return ret;
+}
+
+/**
+ * nilfs_dat_freev - free virtual block numbers
+ * @dat: DAT file inode
+ * @vblocknrs: array of virtual block numbers
+ * @nitems: number of virtual block numbers
+ *
+ * Description: nilfs_dat_freev() frees the virtual block numbers specified by
+ * @vblocknrs and @nitems.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- The virtual block number have not been allocated.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_dat_freev(struct inode *dat, __u64 *vblocknrs, size_t nitems)
+{
+	return nilfs_palloc_freev(dat, vblocknrs, nitems);
+}
+
+/**
+ * nilfs_dat_move - change a block number
+ * @dat: DAT file inode
+ * @vblocknr: virtual block number
+ * @blocknr: block number
+ *
+ * Description: nilfs_dat_move() changes the block number associated with
+ * @vblocknr to @blocknr.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_dat_move(struct inode *dat, __u64 vblocknr, sector_t blocknr)
+{
+	struct buffer_head *entry_bh;
+	struct nilfs_dat_entry *entry;
+	size_t offset;
+	int ret;
+
+	ret = nilfs_palloc_get_entry_block(dat, vblocknr, 0, &entry_bh);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * The given disk block number (blocknr) is not yet written to
+	 * the device at this point.
+	 *
+	 * To prevent nilfs_dat_translate() from returning the
+	 * uncommitted block number, this makes a copy of the entry
+	 * buffer and redirects nilfs_dat_translate() to the copy.
+	 */
+	if (!buffer_nilfs_redirected(entry_bh)) {
+		ret = nilfs_mdt_freeze_buffer(dat, entry_bh);
+		if (ret) {
+			brelse(entry_bh);
+			return ret;
+		}
+	}
+
+	offset = nilfs_palloc_entry_offset(dat, vblocknr, entry_bh);
+	entry = kmap_local_folio(entry_bh->b_folio, offset);
+	if (unlikely(entry->de_blocknr == cpu_to_le64(0))) {
+		nilfs_crit(dat->i_sb,
+			   "%s: invalid vblocknr = %llu, [%llu, %llu)",
+			   __func__, (unsigned long long)vblocknr,
+			   (unsigned long long)le64_to_cpu(entry->de_start),
+			   (unsigned long long)le64_to_cpu(entry->de_end));
+		kunmap_local(entry);
+		brelse(entry_bh);
+		return -EINVAL;
+	}
+	WARN_ON(blocknr == 0);
+	entry->de_blocknr = cpu_to_le64(blocknr);
+	kunmap_local(entry);
+
+	mark_buffer_dirty(entry_bh);
+	nilfs_mdt_mark_dirty(dat);
+
+	brelse(entry_bh);
+
+	return 0;
+}
+
+/**
+ * nilfs_dat_translate - translate a virtual block number to a block number
+ * @dat: DAT file inode
+ * @vblocknr: virtual block number
+ * @blocknrp: pointer to a block number
+ *
+ * Description: nilfs_dat_translate() maps the virtual block number @vblocknr
+ * to the corresponding block number.  The block number associated with
+ * @vblocknr is stored in the place pointed to by @blocknrp.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- A block number associated with @vblocknr does not exist.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_dat_translate(struct inode *dat, __u64 vblocknr, sector_t *blocknrp)
+{
+	struct buffer_head *entry_bh, *bh;
+	struct nilfs_dat_entry *entry;
+	sector_t blocknr;
+	size_t offset;
+	int ret;
+
+	ret = nilfs_palloc_get_entry_block(dat, vblocknr, 0, &entry_bh);
+	if (ret < 0)
+		return ret;
+
+	if (!nilfs_doing_gc() && buffer_nilfs_redirected(entry_bh)) {
+		bh = nilfs_mdt_get_frozen_buffer(dat, entry_bh);
+		if (bh) {
+			WARN_ON(!buffer_uptodate(bh));
+			brelse(entry_bh);
+			entry_bh = bh;
+		}
+	}
+
+	offset = nilfs_palloc_entry_offset(dat, vblocknr, entry_bh);
+	entry = kmap_local_folio(entry_bh->b_folio, offset);
+	blocknr = le64_to_cpu(entry->de_blocknr);
+	if (blocknr == 0) {
+		ret = -ENOENT;
+		goto out;
+	}
+	*blocknrp = blocknr;
+
+ out:
+	kunmap_local(entry);
+	brelse(entry_bh);
+	return ret;
+}
+
+ssize_t nilfs_dat_get_vinfo(struct inode *dat, void *buf, unsigned int visz,
+			    size_t nvi)
+{
+	struct buffer_head *entry_bh;
+	struct nilfs_dat_entry *entry, *first_entry;
+	struct nilfs_vinfo *vinfo = buf;
+	__u64 first, last;
+	size_t offset;
+	unsigned long entries_per_block = NILFS_MDT(dat)->mi_entries_per_block;
+	unsigned int entry_size = NILFS_MDT(dat)->mi_entry_size;
+	int i, j, n, ret;
+
+	for (i = 0; i < nvi; i += n) {
+		ret = nilfs_palloc_get_entry_block(dat, vinfo->vi_vblocknr,
+						   0, &entry_bh);
+		if (ret < 0)
+			return ret;
+
+		first = vinfo->vi_vblocknr;
+		first = div64_ul(first, entries_per_block);
+		first *= entries_per_block;
+		/* first virtual block number in this block */
+
+		last = first + entries_per_block - 1;
+		/* last virtual block number in this block */
+
+		offset = nilfs_palloc_entry_offset(dat, first, entry_bh);
+		first_entry = kmap_local_folio(entry_bh->b_folio, offset);
+		for (j = i, n = 0;
+		     j < nvi && vinfo->vi_vblocknr >= first &&
+			     vinfo->vi_vblocknr <= last;
+		     j++, n++, vinfo = (void *)vinfo + visz) {
+			entry = (void *)first_entry +
+				(vinfo->vi_vblocknr - first) * entry_size;
+			vinfo->vi_start = le64_to_cpu(entry->de_start);
+			vinfo->vi_end = le64_to_cpu(entry->de_end);
+			vinfo->vi_blocknr = le64_to_cpu(entry->de_blocknr);
+		}
+		kunmap_local(first_entry);
+		brelse(entry_bh);
+	}
+
+	return nvi;
+}
+
+/**
+ * nilfs_dat_read - read or get dat inode
+ * @sb: super block instance
+ * @entry_size: size of a dat entry
+ * @raw_inode: on-disk dat inode
+ * @inodep: buffer to store the inode
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int nilfs_dat_read(struct super_block *sb, size_t entry_size,
+		   struct nilfs_inode *raw_inode, struct inode **inodep)
+{
+	static struct lock_class_key dat_lock_key;
+	struct inode *dat;
+	struct nilfs_dat_info *di;
+	int err;
+
+	if (entry_size > sb->s_blocksize) {
+		nilfs_err(sb, "too large DAT entry size: %zu bytes",
+			  entry_size);
+		return -EINVAL;
+	} else if (entry_size < NILFS_MIN_DAT_ENTRY_SIZE) {
+		nilfs_err(sb, "too small DAT entry size: %zu bytes",
+			  entry_size);
+		return -EINVAL;
+	}
+
+	dat = nilfs_iget_locked(sb, NULL, NILFS_DAT_INO);
+	if (unlikely(!dat))
+		return -ENOMEM;
+	if (!(inode_state_read_once(dat) & I_NEW))
+		goto out;
+
+	err = nilfs_mdt_init(dat, NILFS_MDT_GFP, sizeof(*di));
+	if (err)
+		goto failed;
+
+	err = nilfs_palloc_init_blockgroup(dat, entry_size);
+	if (err)
+		goto failed;
+
+	di = NILFS_DAT_I(dat);
+	lockdep_set_class(&di->mi.mi_sem, &dat_lock_key);
+	nilfs_palloc_setup_cache(dat, &di->palloc_cache);
+	err = nilfs_mdt_setup_shadow_map(dat, &di->shadow);
+	if (err)
+		goto failed;
+
+	err = nilfs_attach_btree_node_cache(dat);
+	if (err)
+		goto failed;
+	err = nilfs_read_inode_common(dat, raw_inode);
+	if (err)
+		goto failed;
+
+	unlock_new_inode(dat);
+ out:
+	*inodep = dat;
+	return 0;
+ failed:
+	iget_failed(dat);
+	return err;
+}

linux-7.2.3/fs/nilfs2/dat.h

@@ -0,0 +1,47 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS disk address translation.
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato.
+ */
+
+#ifndef _NILFS_DAT_H
+#define _NILFS_DAT_H
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include <linux/fs.h>
+#include <linux/nilfs2_ondisk.h>	/* nilfs_inode, nilfs_checkpoint */
+
+
+struct nilfs_palloc_req;
+
+int nilfs_dat_translate(struct inode *, __u64, sector_t *);
+
+int nilfs_dat_prepare_alloc(struct inode *, struct nilfs_palloc_req *);
+void nilfs_dat_commit_alloc(struct inode *, struct nilfs_palloc_req *);
+void nilfs_dat_abort_alloc(struct inode *, struct nilfs_palloc_req *);
+int nilfs_dat_prepare_start(struct inode *, struct nilfs_palloc_req *);
+void nilfs_dat_commit_start(struct inode *, struct nilfs_palloc_req *,
+			    sector_t);
+int nilfs_dat_prepare_end(struct inode *, struct nilfs_palloc_req *);
+void nilfs_dat_commit_end(struct inode *, struct nilfs_palloc_req *, int);
+void nilfs_dat_abort_end(struct inode *, struct nilfs_palloc_req *);
+int nilfs_dat_prepare_update(struct inode *, struct nilfs_palloc_req *,
+			     struct nilfs_palloc_req *);
+void nilfs_dat_commit_update(struct inode *, struct nilfs_palloc_req *,
+			     struct nilfs_palloc_req *, int);
+void nilfs_dat_abort_update(struct inode *, struct nilfs_palloc_req *,
+			    struct nilfs_palloc_req *);
+
+int nilfs_dat_mark_dirty(struct inode *, __u64);
+int nilfs_dat_freev(struct inode *, __u64 *, size_t);
+int nilfs_dat_move(struct inode *, __u64, sector_t);
+ssize_t nilfs_dat_get_vinfo(struct inode *, void *, unsigned int, size_t);
+
+int nilfs_dat_read(struct super_block *sb, size_t entry_size,
+		   struct nilfs_inode *raw_inode, struct inode **inodep);
+
+#endif	/* _NILFS_DAT_H */

linux-7.2.3/fs/nilfs2/dir.c

@@ -0,0 +1,666 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS directory entry operations
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Modified for NILFS by Amagai Yoshiji.
+ */
+/*
+ *  linux/fs/ext2/dir.c
+ *
+ * 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/dir.c
+ *
+ *  Copyright (C) 1991, 1992  Linus Torvalds
+ *
+ *  ext2 directory handling functions
+ *
+ *  Big-endian to little-endian byte-swapping/bitmaps by
+ *        David S. Miller (davem@caip.rutgers.edu), 1995
+ *
+ * All code that works with directory layout had been switched to pagecache
+ * and moved here. AV
+ */
+
+#include <linux/pagemap.h>
+#include <linux/filelock.h>
+#include "nilfs.h"
+#include "page.h"
+
+static inline unsigned int nilfs_rec_len_from_disk(__le16 dlen)
+{
+	unsigned int len = le16_to_cpu(dlen);
+
+#if (PAGE_SIZE >= 65536)
+	if (len == NILFS_MAX_REC_LEN)
+		return 1 << 16;
+#endif
+	return len;
+}
+
+static inline __le16 nilfs_rec_len_to_disk(unsigned int len)
+{
+#if (PAGE_SIZE >= 65536)
+	if (len == (1 << 16))
+		return cpu_to_le16(NILFS_MAX_REC_LEN);
+
+	BUG_ON(len > (1 << 16));
+#endif
+	return cpu_to_le16(len);
+}
+
+/*
+ * nilfs uses block-sized chunks. Arguably, sector-sized ones would be
+ * more robust, but we have what we have
+ */
+static inline unsigned int nilfs_chunk_size(struct inode *inode)
+{
+	return inode->i_sb->s_blocksize;
+}
+
+/*
+ * Return the offset into page `page_nr' of the last valid
+ * byte in that page, plus one.
+ */
+static unsigned int nilfs_last_byte(struct inode *inode, unsigned long page_nr)
+{
+	u64 last_byte = inode->i_size;
+
+	last_byte -= page_nr << PAGE_SHIFT;
+	if (last_byte > PAGE_SIZE)
+		last_byte = PAGE_SIZE;
+	return last_byte;
+}
+
+static int nilfs_prepare_chunk(struct folio *folio, unsigned int from,
+			       unsigned int to)
+{
+	loff_t pos = folio_pos(folio) + from;
+
+	return __block_write_begin(folio, pos, to - from, nilfs_get_block);
+}
+
+static void nilfs_commit_chunk(struct folio *folio,
+		struct address_space *mapping, size_t from, size_t to)
+{
+	struct inode *dir = mapping->host;
+	loff_t pos = folio_pos(folio) + from;
+	size_t copied, len = to - from;
+	unsigned int nr_dirty;
+	int err;
+
+	nr_dirty = nilfs_page_count_clean_buffers(folio, from, to);
+	copied = block_write_end(pos, len, len, folio);
+	if (pos + copied > dir->i_size)
+		i_size_write(dir, pos + copied);
+	if (IS_DIRSYNC(dir))
+		nilfs_set_transaction_flag(NILFS_TI_SYNC);
+	err = nilfs_set_file_dirty(dir, nr_dirty);
+	WARN_ON(err); /* do not happen */
+	folio_unlock(folio);
+}
+
+static bool nilfs_check_folio(struct folio *folio, char *kaddr)
+{
+	struct inode *dir = folio->mapping->host;
+	struct super_block *sb = dir->i_sb;
+	unsigned int chunk_size = nilfs_chunk_size(dir);
+	size_t offs, rec_len;
+	size_t limit = folio_size(folio);
+	struct nilfs_dir_entry *p;
+	char *error;
+
+	if (dir->i_size < folio_pos(folio) + limit) {
+		limit = dir->i_size - folio_pos(folio);
+		if (limit & (chunk_size - 1))
+			goto Ebadsize;
+		if (!limit)
+			goto out;
+	}
+	for (offs = 0; offs <= limit - NILFS_DIR_REC_LEN(1); offs += rec_len) {
+		p = (struct nilfs_dir_entry *)(kaddr + offs);
+		rec_len = nilfs_rec_len_from_disk(p->rec_len);
+
+		if (rec_len < NILFS_DIR_REC_LEN(1))
+			goto Eshort;
+		if (rec_len & 3)
+			goto Ealign;
+		if (rec_len < NILFS_DIR_REC_LEN(p->name_len))
+			goto Enamelen;
+		if (((offs + rec_len - 1) ^ offs) & ~(chunk_size-1))
+			goto Espan;
+		if (unlikely(p->inode &&
+			     NILFS_PRIVATE_INODE(le64_to_cpu(p->inode))))
+			goto Einumber;
+	}
+	if (offs != limit)
+		goto Eend;
+out:
+	folio_set_checked(folio);
+	return true;
+
+	/* Too bad, we had an error */
+
+Ebadsize:
+	nilfs_error(sb,
+		    "size of directory #%llu is not a multiple of chunk size",
+		    dir->i_ino);
+	goto fail;
+Eshort:
+	error = "rec_len is smaller than minimal";
+	goto bad_entry;
+Ealign:
+	error = "unaligned directory entry";
+	goto bad_entry;
+Enamelen:
+	error = "rec_len is too small for name_len";
+	goto bad_entry;
+Espan:
+	error = "directory entry across blocks";
+	goto bad_entry;
+Einumber:
+	error = "disallowed inode number";
+bad_entry:
+	nilfs_error(sb,
+		    "bad entry in directory #%llu: %s - offset=%lu, inode=%lu, rec_len=%zd, name_len=%d",
+		    dir->i_ino, error, (folio->index << PAGE_SHIFT) + offs,
+		    (unsigned long)le64_to_cpu(p->inode),
+		    rec_len, p->name_len);
+	goto fail;
+Eend:
+	p = (struct nilfs_dir_entry *)(kaddr + offs);
+	nilfs_error(sb,
+		    "entry in directory #%llu spans the page boundary offset=%lu, inode=%lu",
+		    dir->i_ino, (folio->index << PAGE_SHIFT) + offs,
+		    (unsigned long)le64_to_cpu(p->inode));
+fail:
+	return false;
+}
+
+static void *nilfs_get_folio(struct inode *dir, unsigned long n,
+		struct folio **foliop)
+{
+	struct address_space *mapping = dir->i_mapping;
+	struct folio *folio = read_mapping_folio(mapping, n, NULL);
+	void *kaddr;
+
+	if (IS_ERR(folio))
+		return folio;
+
+	kaddr = kmap_local_folio(folio, 0);
+	if (unlikely(!folio_test_checked(folio))) {
+		if (!nilfs_check_folio(folio, kaddr))
+			goto fail;
+	}
+
+	*foliop = folio;
+	return kaddr;
+
+fail:
+	folio_release_kmap(folio, kaddr);
+	return ERR_PTR(-EIO);
+}
+
+/*
+ * NOTE! unlike strncmp, nilfs_match returns 1 for success, 0 for failure.
+ *
+ * len <= NILFS_NAME_LEN and de != NULL are guaranteed by caller.
+ */
+static int
+nilfs_match(int len, const unsigned char *name, struct nilfs_dir_entry *de)
+{
+	if (len != de->name_len)
+		return 0;
+	if (!de->inode)
+		return 0;
+	return !memcmp(name, de->name, len);
+}
+
+/*
+ * p is at least 6 bytes before the end of page
+ */
+static struct nilfs_dir_entry *nilfs_next_entry(struct nilfs_dir_entry *p)
+{
+	return (struct nilfs_dir_entry *)((char *)p +
+					  nilfs_rec_len_from_disk(p->rec_len));
+}
+
+static int nilfs_readdir(struct file *file, struct dir_context *ctx)
+{
+	loff_t pos = ctx->pos;
+	struct inode *inode = file_inode(file);
+	struct super_block *sb = inode->i_sb;
+	unsigned int offset = pos & ~PAGE_MASK;
+	unsigned long n = pos >> PAGE_SHIFT;
+	unsigned long npages = dir_pages(inode);
+
+	if (pos > inode->i_size - NILFS_DIR_REC_LEN(1))
+		return 0;
+
+	for ( ; n < npages; n++, offset = 0) {
+		char *kaddr, *limit;
+		struct nilfs_dir_entry *de;
+		struct folio *folio;
+
+		kaddr = nilfs_get_folio(inode, n, &folio);
+		if (IS_ERR(kaddr)) {
+			nilfs_error(sb, "bad page in #%llu", inode->i_ino);
+			ctx->pos += PAGE_SIZE - offset;
+			return -EIO;
+		}
+		de = (struct nilfs_dir_entry *)(kaddr + offset);
+		limit = kaddr + nilfs_last_byte(inode, n) -
+			NILFS_DIR_REC_LEN(1);
+		for ( ; (char *)de <= limit; de = nilfs_next_entry(de)) {
+			if (de->rec_len == 0) {
+				nilfs_error(sb, "zero-length directory entry");
+				folio_release_kmap(folio, kaddr);
+				return -EIO;
+			}
+			if (de->inode) {
+				unsigned char t;
+
+				t = fs_ftype_to_dtype(de->file_type);
+
+				if (!dir_emit(ctx, de->name, de->name_len,
+						le64_to_cpu(de->inode), t)) {
+					folio_release_kmap(folio, kaddr);
+					return 0;
+				}
+			}
+			ctx->pos += nilfs_rec_len_from_disk(de->rec_len);
+		}
+		folio_release_kmap(folio, kaddr);
+	}
+	return 0;
+}
+
+/*
+ * nilfs_find_entry()
+ *
+ * Finds an entry in the specified directory with the wanted name. It
+ * returns the folio in which the entry was found, and the entry itself.
+ * The folio is mapped and unlocked.  When the caller is finished with
+ * the entry, it should call folio_release_kmap().
+ *
+ * On failure, returns an error pointer and the caller should ignore foliop.
+ */
+struct nilfs_dir_entry *nilfs_find_entry(struct inode *dir,
+		const struct qstr *qstr, struct folio **foliop)
+{
+	const unsigned char *name = qstr->name;
+	int namelen = qstr->len;
+	unsigned int reclen = NILFS_DIR_REC_LEN(namelen);
+	unsigned long start, n;
+	unsigned long npages = dir_pages(dir);
+	struct nilfs_inode_info *ei = NILFS_I(dir);
+	struct nilfs_dir_entry *de;
+
+	if (npages == 0)
+		goto out;
+
+	start = ei->i_dir_start_lookup;
+	if (start >= npages)
+		start = 0;
+	n = start;
+	do {
+		char *kaddr = nilfs_get_folio(dir, n, foliop);
+
+		if (IS_ERR(kaddr))
+			return ERR_CAST(kaddr);
+
+		de = (struct nilfs_dir_entry *)kaddr;
+		kaddr += nilfs_last_byte(dir, n) - reclen;
+		while ((char *)de <= kaddr) {
+			if (de->rec_len == 0) {
+				nilfs_error(dir->i_sb,
+					    "zero-length directory entry");
+				folio_release_kmap(*foliop, kaddr);
+				goto out;
+			}
+			if (nilfs_match(namelen, name, de))
+				goto found;
+			de = nilfs_next_entry(de);
+		}
+		folio_release_kmap(*foliop, kaddr);
+
+		if (++n >= npages)
+			n = 0;
+		/* next folio is past the blocks we've got */
+		if (unlikely(n > (dir->i_blocks >> (PAGE_SHIFT - 9)))) {
+			nilfs_error(dir->i_sb,
+			       "dir %llu size %lld exceeds block count %llu",
+			       dir->i_ino, dir->i_size,
+			       (unsigned long long)dir->i_blocks);
+			goto out;
+		}
+	} while (n != start);
+out:
+	return ERR_PTR(-ENOENT);
+
+found:
+	ei->i_dir_start_lookup = n;
+	return de;
+}
+
+struct nilfs_dir_entry *nilfs_dotdot(struct inode *dir, struct folio **foliop)
+{
+	struct folio *folio;
+	struct nilfs_dir_entry *de, *next_de;
+	size_t limit;
+	char *msg;
+
+	de = nilfs_get_folio(dir, 0, &folio);
+	if (IS_ERR(de))
+		return NULL;
+
+	limit = nilfs_last_byte(dir, 0);  /* is a multiple of chunk size */
+	if (unlikely(!limit || le64_to_cpu(de->inode) != dir->i_ino ||
+		     !nilfs_match(1, ".", de))) {
+		msg = "missing '.'";
+		goto fail;
+	}
+
+	next_de = nilfs_next_entry(de);
+	/*
+	 * If "next_de" has not reached the end of the chunk, there is
+	 * at least one more record.  Check whether it matches "..".
+	 */
+	if (unlikely((char *)next_de == (char *)de + nilfs_chunk_size(dir) ||
+		     !nilfs_match(2, "..", next_de))) {
+		msg = "missing '..'";
+		goto fail;
+	}
+	*foliop = folio;
+	return next_de;
+
+fail:
+	nilfs_error(dir->i_sb, "directory #%llu %s", dir->i_ino, msg);
+	folio_release_kmap(folio, de);
+	return NULL;
+}
+
+int nilfs_inode_by_name(struct inode *dir, const struct qstr *qstr, ino_t *ino)
+{
+	struct nilfs_dir_entry *de;
+	struct folio *folio;
+
+	de = nilfs_find_entry(dir, qstr, &folio);
+	if (IS_ERR(de))
+		return PTR_ERR(de);
+
+	*ino = le64_to_cpu(de->inode);
+	folio_release_kmap(folio, de);
+	return 0;
+}
+
+int nilfs_set_link(struct inode *dir, struct nilfs_dir_entry *de,
+		    struct folio *folio, struct inode *inode)
+{
+	size_t from = offset_in_folio(folio, de);
+	size_t to = from + nilfs_rec_len_from_disk(de->rec_len);
+	struct address_space *mapping = folio->mapping;
+	int err;
+
+	folio_lock(folio);
+	err = nilfs_prepare_chunk(folio, from, to);
+	if (unlikely(err)) {
+		folio_unlock(folio);
+		return err;
+	}
+	de->inode = cpu_to_le64(inode->i_ino);
+	de->file_type = fs_umode_to_ftype(inode->i_mode);
+	nilfs_commit_chunk(folio, mapping, from, to);
+	inode_set_mtime_to_ts(dir, inode_set_ctime_current(dir));
+	return 0;
+}
+
+/*
+ *	Parent is locked.
+ */
+int nilfs_add_link(struct dentry *dentry, struct inode *inode)
+{
+	struct inode *dir = d_inode(dentry->d_parent);
+	const unsigned char *name = dentry->d_name.name;
+	int namelen = dentry->d_name.len;
+	unsigned int chunk_size = nilfs_chunk_size(dir);
+	unsigned int reclen = NILFS_DIR_REC_LEN(namelen);
+	unsigned short rec_len, name_len;
+	struct folio *folio = NULL;
+	struct nilfs_dir_entry *de;
+	unsigned long npages = dir_pages(dir);
+	unsigned long n;
+	size_t from, to;
+	int err;
+
+	/*
+	 * We take care of directory expansion in the same loop.
+	 * This code plays outside i_size, so it locks the folio
+	 * to protect that region.
+	 */
+	for (n = 0; n <= npages; n++) {
+		char *kaddr = nilfs_get_folio(dir, n, &folio);
+		char *dir_end;
+
+		if (IS_ERR(kaddr))
+			return PTR_ERR(kaddr);
+		folio_lock(folio);
+		dir_end = kaddr + nilfs_last_byte(dir, n);
+		de = (struct nilfs_dir_entry *)kaddr;
+		kaddr += folio_size(folio) - reclen;
+		while ((char *)de <= kaddr) {
+			if ((char *)de == dir_end) {
+				/* We hit i_size */
+				name_len = 0;
+				rec_len = chunk_size;
+				de->rec_len = nilfs_rec_len_to_disk(chunk_size);
+				de->inode = 0;
+				goto got_it;
+			}
+			if (de->rec_len == 0) {
+				nilfs_error(dir->i_sb,
+					    "zero-length directory entry");
+				err = -EIO;
+				goto out_unlock;
+			}
+			err = -EEXIST;
+			if (nilfs_match(namelen, name, de))
+				goto out_unlock;
+			name_len = NILFS_DIR_REC_LEN(de->name_len);
+			rec_len = nilfs_rec_len_from_disk(de->rec_len);
+			if (!de->inode && rec_len >= reclen)
+				goto got_it;
+			if (rec_len >= name_len + reclen)
+				goto got_it;
+			de = (struct nilfs_dir_entry *)((char *)de + rec_len);
+		}
+		folio_unlock(folio);
+		folio_release_kmap(folio, kaddr);
+	}
+	BUG();
+	return -EINVAL;
+
+got_it:
+	from = offset_in_folio(folio, de);
+	to = from + rec_len;
+	err = nilfs_prepare_chunk(folio, from, to);
+	if (err)
+		goto out_unlock;
+	if (de->inode) {
+		struct nilfs_dir_entry *de1;
+
+		de1 = (struct nilfs_dir_entry *)((char *)de + name_len);
+		de1->rec_len = nilfs_rec_len_to_disk(rec_len - name_len);
+		de->rec_len = nilfs_rec_len_to_disk(name_len);
+		de = de1;
+	}
+	de->name_len = namelen;
+	memcpy(de->name, name, namelen);
+	de->inode = cpu_to_le64(inode->i_ino);
+	de->file_type = fs_umode_to_ftype(inode->i_mode);
+	nilfs_commit_chunk(folio, folio->mapping, from, to);
+	inode_set_mtime_to_ts(dir, inode_set_ctime_current(dir));
+	nilfs_mark_inode_dirty(dir);
+	/* OFFSET_CACHE */
+out_put:
+	folio_release_kmap(folio, de);
+	return err;
+out_unlock:
+	folio_unlock(folio);
+	goto out_put;
+}
+
+/*
+ * nilfs_delete_entry deletes a directory entry by merging it with the
+ * previous entry. Folio is up-to-date.
+ */
+int nilfs_delete_entry(struct nilfs_dir_entry *dir, struct folio *folio)
+{
+	struct address_space *mapping = folio->mapping;
+	struct inode *inode = mapping->host;
+	char *kaddr = (char *)((unsigned long)dir & ~(folio_size(folio) - 1));
+	size_t from, to;
+	struct nilfs_dir_entry *de, *pde = NULL;
+	int err;
+
+	from = ((char *)dir - kaddr) & ~(nilfs_chunk_size(inode) - 1);
+	to = ((char *)dir - kaddr) + nilfs_rec_len_from_disk(dir->rec_len);
+	de = (struct nilfs_dir_entry *)(kaddr + from);
+
+	while ((char *)de < (char *)dir) {
+		if (de->rec_len == 0) {
+			nilfs_error(inode->i_sb,
+				    "zero-length directory entry");
+			err = -EIO;
+			goto out;
+		}
+		pde = de;
+		de = nilfs_next_entry(de);
+	}
+	if (pde)
+		from = (char *)pde - kaddr;
+	folio_lock(folio);
+	err = nilfs_prepare_chunk(folio, from, to);
+	if (unlikely(err)) {
+		folio_unlock(folio);
+		goto out;
+	}
+	if (pde)
+		pde->rec_len = nilfs_rec_len_to_disk(to - from);
+	dir->inode = 0;
+	nilfs_commit_chunk(folio, mapping, from, to);
+	inode_set_mtime_to_ts(inode, inode_set_ctime_current(inode));
+out:
+	return err;
+}
+
+/*
+ * Set the first fragment of directory.
+ */
+int nilfs_make_empty(struct inode *inode, struct inode *parent)
+{
+	struct address_space *mapping = inode->i_mapping;
+	struct folio *folio = filemap_grab_folio(mapping, 0);
+	unsigned int chunk_size = nilfs_chunk_size(inode);
+	struct nilfs_dir_entry *de;
+	int err;
+	void *kaddr;
+
+	if (IS_ERR(folio))
+		return PTR_ERR(folio);
+
+	err = nilfs_prepare_chunk(folio, 0, chunk_size);
+	if (unlikely(err)) {
+		folio_unlock(folio);
+		goto fail;
+	}
+	kaddr = kmap_local_folio(folio, 0);
+	memset(kaddr, 0, chunk_size);
+	de = (struct nilfs_dir_entry *)kaddr;
+	de->name_len = 1;
+	de->rec_len = nilfs_rec_len_to_disk(NILFS_DIR_REC_LEN(1));
+	memcpy(de->name, ".\0\0", 4);
+	de->inode = cpu_to_le64(inode->i_ino);
+	de->file_type = fs_umode_to_ftype(inode->i_mode);
+
+	de = (struct nilfs_dir_entry *)(kaddr + NILFS_DIR_REC_LEN(1));
+	de->name_len = 2;
+	de->rec_len = nilfs_rec_len_to_disk(chunk_size - NILFS_DIR_REC_LEN(1));
+	de->inode = cpu_to_le64(parent->i_ino);
+	memcpy(de->name, "..\0", 4);
+	de->file_type = fs_umode_to_ftype(inode->i_mode);
+	kunmap_local(kaddr);
+	nilfs_commit_chunk(folio, mapping, 0, chunk_size);
+fail:
+	folio_put(folio);
+	return err;
+}
+
+/*
+ * routine to check that the specified directory is empty (for rmdir)
+ */
+int nilfs_empty_dir(struct inode *inode)
+{
+	struct folio *folio = NULL;
+	char *kaddr;
+	unsigned long i, npages = dir_pages(inode);
+
+	for (i = 0; i < npages; i++) {
+		struct nilfs_dir_entry *de;
+
+		kaddr = nilfs_get_folio(inode, i, &folio);
+		if (IS_ERR(kaddr))
+			return 0;
+
+		de = (struct nilfs_dir_entry *)kaddr;
+		kaddr += nilfs_last_byte(inode, i) - NILFS_DIR_REC_LEN(1);
+
+		while ((char *)de <= kaddr) {
+			if (de->rec_len == 0) {
+				nilfs_error(inode->i_sb,
+					    "zero-length directory entry (kaddr=%p, de=%p)",
+					    kaddr, de);
+				goto not_empty;
+			}
+			if (de->inode != 0) {
+				/* check for . and .. */
+				if (de->name[0] != '.')
+					goto not_empty;
+				if (de->name_len > 2)
+					goto not_empty;
+				if (de->name_len < 2) {
+					if (de->inode !=
+					    cpu_to_le64(inode->i_ino))
+						goto not_empty;
+				} else if (de->name[1] != '.')
+					goto not_empty;
+			}
+			de = nilfs_next_entry(de);
+		}
+		folio_release_kmap(folio, kaddr);
+	}
+	return 1;
+
+not_empty:
+	folio_release_kmap(folio, kaddr);
+	return 0;
+}
+
+const struct file_operations nilfs_dir_operations = {
+	.llseek		= generic_file_llseek,
+	.read		= generic_read_dir,
+	.iterate_shared	= nilfs_readdir,
+	.unlocked_ioctl	= nilfs_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl	= nilfs_compat_ioctl,
+#endif	/* CONFIG_COMPAT */
+	.fsync		= nilfs_sync_file,
+	.setlease	= generic_setlease,
+};

linux-7.2.3/fs/nilfs2/direct.c

@@ -0,0 +1,387 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS direct block pointer.
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato.
+ */
+
+#include <linux/errno.h>
+#include "nilfs.h"
+#include "page.h"
+#include "direct.h"
+#include "alloc.h"
+#include "dat.h"
+
+static inline __le64 *nilfs_direct_dptrs(const struct nilfs_bmap *direct)
+{
+	return (__le64 *)
+		((struct nilfs_direct_node *)direct->b_u.u_data + 1);
+}
+
+static inline __u64
+nilfs_direct_get_ptr(const struct nilfs_bmap *direct, __u64 key)
+{
+	return le64_to_cpu(*(nilfs_direct_dptrs(direct) + key));
+}
+
+static inline void nilfs_direct_set_ptr(struct nilfs_bmap *direct,
+					__u64 key, __u64 ptr)
+{
+	*(nilfs_direct_dptrs(direct) + key) = cpu_to_le64(ptr);
+}
+
+static int nilfs_direct_lookup(const struct nilfs_bmap *direct,
+			       __u64 key, int level, __u64 *ptrp)
+{
+	__u64 ptr;
+
+	if (key > NILFS_DIRECT_KEY_MAX || level != 1)
+		return -ENOENT;
+	ptr = nilfs_direct_get_ptr(direct, key);
+	if (ptr == NILFS_BMAP_INVALID_PTR)
+		return -ENOENT;
+
+	*ptrp = ptr;
+	return 0;
+}
+
+static int nilfs_direct_lookup_contig(const struct nilfs_bmap *direct,
+				      __u64 key, __u64 *ptrp,
+				      unsigned int maxblocks)
+{
+	struct inode *dat = NULL;
+	__u64 ptr, ptr2;
+	sector_t blocknr;
+	int ret, cnt;
+
+	if (key > NILFS_DIRECT_KEY_MAX)
+		return -ENOENT;
+	ptr = nilfs_direct_get_ptr(direct, key);
+	if (ptr == NILFS_BMAP_INVALID_PTR)
+		return -ENOENT;
+
+	if (NILFS_BMAP_USE_VBN(direct)) {
+		dat = nilfs_bmap_get_dat(direct);
+		ret = nilfs_dat_translate(dat, ptr, &blocknr);
+		if (ret < 0)
+			goto dat_error;
+		ptr = blocknr;
+	}
+
+	maxblocks = min_t(unsigned int, maxblocks,
+			  NILFS_DIRECT_KEY_MAX - key + 1);
+	for (cnt = 1; cnt < maxblocks &&
+		     (ptr2 = nilfs_direct_get_ptr(direct, key + cnt)) !=
+		     NILFS_BMAP_INVALID_PTR;
+	     cnt++) {
+		if (dat) {
+			ret = nilfs_dat_translate(dat, ptr2, &blocknr);
+			if (ret < 0)
+				goto dat_error;
+			ptr2 = blocknr;
+		}
+		if (ptr2 != ptr + cnt)
+			break;
+	}
+	*ptrp = ptr;
+	return cnt;
+
+ dat_error:
+	if (ret == -ENOENT)
+		ret = -EINVAL;  /* Notify bmap layer of metadata corruption */
+	return ret;
+}
+
+static __u64
+nilfs_direct_find_target_v(const struct nilfs_bmap *direct, __u64 key)
+{
+	__u64 ptr;
+
+	ptr = nilfs_bmap_find_target_seq(direct, key);
+	if (ptr != NILFS_BMAP_INVALID_PTR)
+		/* sequential access */
+		return ptr;
+
+	/* block group */
+	return nilfs_bmap_find_target_in_group(direct);
+}
+
+static int nilfs_direct_insert(struct nilfs_bmap *bmap, __u64 key, __u64 ptr)
+{
+	union nilfs_bmap_ptr_req req;
+	struct inode *dat = NULL;
+	struct buffer_head *bh;
+	int ret;
+
+	if (key > NILFS_DIRECT_KEY_MAX)
+		return -ENOENT;
+	if (nilfs_direct_get_ptr(bmap, key) != NILFS_BMAP_INVALID_PTR)
+		return -EEXIST;
+
+	if (NILFS_BMAP_USE_VBN(bmap)) {
+		req.bpr_ptr = nilfs_direct_find_target_v(bmap, key);
+		dat = nilfs_bmap_get_dat(bmap);
+	}
+	ret = nilfs_bmap_prepare_alloc_ptr(bmap, &req, dat);
+	if (!ret) {
+		/* ptr must be a pointer to a buffer head. */
+		bh = (struct buffer_head *)((unsigned long)ptr);
+		set_buffer_nilfs_volatile(bh);
+
+		nilfs_bmap_commit_alloc_ptr(bmap, &req, dat);
+		nilfs_direct_set_ptr(bmap, key, req.bpr_ptr);
+
+		if (!nilfs_bmap_dirty(bmap))
+			nilfs_bmap_set_dirty(bmap);
+
+		if (NILFS_BMAP_USE_VBN(bmap))
+			nilfs_bmap_set_target_v(bmap, key, req.bpr_ptr);
+
+		nilfs_inode_add_blocks(bmap->b_inode, 1);
+	}
+	return ret;
+}
+
+static int nilfs_direct_delete(struct nilfs_bmap *bmap, __u64 key)
+{
+	union nilfs_bmap_ptr_req req;
+	struct inode *dat;
+	int ret;
+
+	if (key > NILFS_DIRECT_KEY_MAX ||
+	    nilfs_direct_get_ptr(bmap, key) == NILFS_BMAP_INVALID_PTR)
+		return -ENOENT;
+
+	dat = NILFS_BMAP_USE_VBN(bmap) ? nilfs_bmap_get_dat(bmap) : NULL;
+	req.bpr_ptr = nilfs_direct_get_ptr(bmap, key);
+
+	ret = nilfs_bmap_prepare_end_ptr(bmap, &req, dat);
+	if (!ret) {
+		nilfs_bmap_commit_end_ptr(bmap, &req, dat);
+		nilfs_direct_set_ptr(bmap, key, NILFS_BMAP_INVALID_PTR);
+		nilfs_inode_sub_blocks(bmap->b_inode, 1);
+	}
+	return ret;
+}
+
+static int nilfs_direct_seek_key(const struct nilfs_bmap *direct, __u64 start,
+				 __u64 *keyp)
+{
+	__u64 key;
+
+	for (key = start; key <= NILFS_DIRECT_KEY_MAX; key++) {
+		if (nilfs_direct_get_ptr(direct, key) !=
+		    NILFS_BMAP_INVALID_PTR) {
+			*keyp = key;
+			return 0;
+		}
+	}
+	return -ENOENT;
+}
+
+static int nilfs_direct_last_key(const struct nilfs_bmap *direct, __u64 *keyp)
+{
+	__u64 key, lastkey;
+
+	lastkey = NILFS_DIRECT_KEY_MAX + 1;
+	for (key = NILFS_DIRECT_KEY_MIN; key <= NILFS_DIRECT_KEY_MAX; key++)
+		if (nilfs_direct_get_ptr(direct, key) !=
+		    NILFS_BMAP_INVALID_PTR)
+			lastkey = key;
+
+	if (lastkey == NILFS_DIRECT_KEY_MAX + 1)
+		return -ENOENT;
+
+	*keyp = lastkey;
+
+	return 0;
+}
+
+static int nilfs_direct_check_insert(const struct nilfs_bmap *bmap, __u64 key)
+{
+	return key > NILFS_DIRECT_KEY_MAX;
+}
+
+static int nilfs_direct_gather_data(struct nilfs_bmap *direct,
+				    __u64 *keys, __u64 *ptrs, int nitems)
+{
+	__u64 key;
+	__u64 ptr;
+	int n;
+
+	if (nitems > NILFS_DIRECT_NBLOCKS)
+		nitems = NILFS_DIRECT_NBLOCKS;
+	n = 0;
+	for (key = 0; key < nitems; key++) {
+		ptr = nilfs_direct_get_ptr(direct, key);
+		if (ptr != NILFS_BMAP_INVALID_PTR) {
+			keys[n] = key;
+			ptrs[n] = ptr;
+			n++;
+		}
+	}
+	return n;
+}
+
+int nilfs_direct_delete_and_convert(struct nilfs_bmap *bmap,
+				    __u64 key, __u64 *keys, __u64 *ptrs, int n)
+{
+	__le64 *dptrs;
+	int ret, i, j;
+
+	/* no need to allocate any resource for conversion */
+
+	/* delete */
+	ret = bmap->b_ops->bop_delete(bmap, key);
+	if (ret < 0)
+		return ret;
+
+	/* free resources */
+	if (bmap->b_ops->bop_clear != NULL)
+		bmap->b_ops->bop_clear(bmap);
+
+	/* convert */
+	dptrs = nilfs_direct_dptrs(bmap);
+	for (i = 0, j = 0; i < NILFS_DIRECT_NBLOCKS; i++) {
+		if ((j < n) && (i == keys[j])) {
+			dptrs[i] = (i != key) ?
+				cpu_to_le64(ptrs[j]) :
+				NILFS_BMAP_INVALID_PTR;
+			j++;
+		} else
+			dptrs[i] = NILFS_BMAP_INVALID_PTR;
+	}
+
+	nilfs_direct_init(bmap);
+	return 0;
+}
+
+static int nilfs_direct_propagate(struct nilfs_bmap *bmap,
+				  struct buffer_head *bh)
+{
+	struct nilfs_palloc_req oldreq, newreq;
+	struct inode *dat;
+	__u64 key;
+	__u64 ptr;
+	int ret;
+
+	if (!NILFS_BMAP_USE_VBN(bmap))
+		return 0;
+
+	dat = nilfs_bmap_get_dat(bmap);
+	key = nilfs_bmap_data_get_key(bmap, bh);
+	ptr = nilfs_direct_get_ptr(bmap, key);
+	if (ptr == NILFS_BMAP_INVALID_PTR)
+		return -EINVAL;
+
+	if (!buffer_nilfs_volatile(bh)) {
+		oldreq.pr_entry_nr = ptr;
+		newreq.pr_entry_nr = ptr;
+		ret = nilfs_dat_prepare_update(dat, &oldreq, &newreq);
+		if (ret < 0)
+			return ret;
+		nilfs_dat_commit_update(dat, &oldreq, &newreq,
+					bmap->b_ptr_type == NILFS_BMAP_PTR_VS);
+		set_buffer_nilfs_volatile(bh);
+		nilfs_direct_set_ptr(bmap, key, newreq.pr_entry_nr);
+	} else
+		ret = nilfs_dat_mark_dirty(dat, ptr);
+
+	return ret;
+}
+
+static int nilfs_direct_assign_v(struct nilfs_bmap *direct,
+				 __u64 key, __u64 ptr,
+				 struct buffer_head **bh,
+				 sector_t blocknr,
+				 union nilfs_binfo *binfo)
+{
+	struct inode *dat = nilfs_bmap_get_dat(direct);
+	union nilfs_bmap_ptr_req req;
+	int ret;
+
+	req.bpr_ptr = ptr;
+	ret = nilfs_dat_prepare_start(dat, &req.bpr_req);
+	if (!ret) {
+		nilfs_dat_commit_start(dat, &req.bpr_req, blocknr);
+		binfo->bi_v.bi_vblocknr = cpu_to_le64(ptr);
+		binfo->bi_v.bi_blkoff = cpu_to_le64(key);
+	}
+	return ret;
+}
+
+static int nilfs_direct_assign_p(struct nilfs_bmap *direct,
+				 __u64 key, __u64 ptr,
+				 struct buffer_head **bh,
+				 sector_t blocknr,
+				 union nilfs_binfo *binfo)
+{
+	nilfs_direct_set_ptr(direct, key, blocknr);
+
+	binfo->bi_dat.bi_blkoff = cpu_to_le64(key);
+	binfo->bi_dat.bi_level = 0;
+	memset(binfo->bi_dat.bi_pad, 0, sizeof(binfo->bi_dat.bi_pad));
+
+	return 0;
+}
+
+static int nilfs_direct_assign(struct nilfs_bmap *bmap,
+			       struct buffer_head **bh,
+			       sector_t blocknr,
+			       union nilfs_binfo *binfo)
+{
+	__u64 key;
+	__u64 ptr;
+
+	key = nilfs_bmap_data_get_key(bmap, *bh);
+	if (unlikely(key > NILFS_DIRECT_KEY_MAX)) {
+		nilfs_crit(bmap->b_inode->i_sb,
+			   "%s (ino=%llu): invalid key: %llu",
+			   __func__,
+			   bmap->b_inode->i_ino, (unsigned long long)key);
+		return -EINVAL;
+	}
+	ptr = nilfs_direct_get_ptr(bmap, key);
+	if (unlikely(ptr == NILFS_BMAP_INVALID_PTR)) {
+		nilfs_crit(bmap->b_inode->i_sb,
+			   "%s (ino=%llu): invalid pointer: %llu",
+			   __func__,
+			   bmap->b_inode->i_ino, (unsigned long long)ptr);
+		return -EINVAL;
+	}
+
+	return NILFS_BMAP_USE_VBN(bmap) ?
+		nilfs_direct_assign_v(bmap, key, ptr, bh, blocknr, binfo) :
+		nilfs_direct_assign_p(bmap, key, ptr, bh, blocknr, binfo);
+}
+
+static const struct nilfs_bmap_operations nilfs_direct_ops = {
+	.bop_lookup		=	nilfs_direct_lookup,
+	.bop_lookup_contig	=	nilfs_direct_lookup_contig,
+	.bop_insert		=	nilfs_direct_insert,
+	.bop_delete		=	nilfs_direct_delete,
+	.bop_clear		=	NULL,
+
+	.bop_propagate		=	nilfs_direct_propagate,
+
+	.bop_lookup_dirty_buffers	=	NULL,
+
+	.bop_assign		=	nilfs_direct_assign,
+	.bop_mark		=	NULL,
+
+	.bop_seek_key		=	nilfs_direct_seek_key,
+	.bop_last_key		=	nilfs_direct_last_key,
+
+	.bop_check_insert	=	nilfs_direct_check_insert,
+	.bop_check_delete	=	NULL,
+	.bop_gather_data	=	nilfs_direct_gather_data,
+};
+
+
+int nilfs_direct_init(struct nilfs_bmap *bmap)
+{
+	bmap->b_ops = &nilfs_direct_ops;
+	return 0;
+}

linux-7.2.3/fs/nilfs2/direct.h

@@ -0,0 +1,28 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS direct block pointer.
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato.
+ */
+
+#ifndef _NILFS_DIRECT_H
+#define _NILFS_DIRECT_H
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include "bmap.h"
+
+
+#define NILFS_DIRECT_NBLOCKS	(NILFS_BMAP_SIZE / sizeof(__le64) - 1)
+#define NILFS_DIRECT_KEY_MIN	0
+#define NILFS_DIRECT_KEY_MAX	(NILFS_DIRECT_NBLOCKS - 1)
+
+
+int nilfs_direct_init(struct nilfs_bmap *);
+int nilfs_direct_delete_and_convert(struct nilfs_bmap *, __u64, __u64 *,
+				    __u64 *, int);
+
+
+#endif	/* _NILFS_DIRECT_H */

linux-7.2.3/fs/nilfs2/export.h

@@ -0,0 +1,26 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef NILFS_EXPORT_H
+#define NILFS_EXPORT_H
+
+#include <linux/exportfs.h>
+
+extern const struct export_operations nilfs_export_ops;
+
+/**
+ * struct nilfs_fid - NILFS file id type
+ * @cno: checkpoint number
+ * @ino: inode number
+ * @gen: file generation (version) for NFS
+ * @parent_gen: parent generation (version) for NFS
+ * @parent_ino: parent inode number
+ */
+struct nilfs_fid {
+	u64 cno;
+	u64 ino;
+	u32 gen;
+
+	u32 parent_gen;
+	u64 parent_ino;
+} __packed;
+
+#endif

linux-7.2.3/fs/nilfs2/file.c

@@ -0,0 +1,165 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS regular file handling primitives including fsync().
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Amagai Yoshiji and Ryusuke Konishi.
+ */
+
+#include <linux/fs.h>
+#include <linux/filelock.h>
+#include <linux/mm.h>
+#include <linux/writeback.h>
+#include "nilfs.h"
+#include "segment.h"
+
+int nilfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
+{
+	/*
+	 * Called from fsync() system call
+	 * This is the only entry point that can catch write and synch
+	 * timing for both data blocks and intermediate blocks.
+	 *
+	 * This function should be implemented when the writeback function
+	 * will be implemented.
+	 */
+	struct the_nilfs *nilfs;
+	struct inode *inode = file->f_mapping->host;
+	int err = 0;
+
+	if (nilfs_inode_dirty(inode)) {
+		if (datasync)
+			err = nilfs_construct_dsync_segment(inode->i_sb, inode,
+							    start, end);
+		else
+			err = nilfs_construct_segment(inode->i_sb);
+	}
+
+	nilfs = inode->i_sb->s_fs_info;
+	if (!err)
+		err = nilfs_flush_device(nilfs);
+
+	return err;
+}
+
+static vm_fault_t nilfs_page_mkwrite(struct vm_fault *vmf)
+{
+	struct vm_area_struct *vma = vmf->vma;
+	struct folio *folio = page_folio(vmf->page);
+	struct inode *inode = file_inode(vma->vm_file);
+	struct nilfs_transaction_info ti;
+	struct buffer_head *bh, *head;
+	int ret = 0;
+
+	if (unlikely(nilfs_near_disk_full(inode->i_sb->s_fs_info)))
+		return VM_FAULT_SIGBUS; /* -ENOSPC */
+
+	sb_start_pagefault(inode->i_sb);
+	folio_lock(folio);
+	if (folio->mapping != inode->i_mapping ||
+	    folio_pos(folio) >= i_size_read(inode) ||
+	    !folio_test_uptodate(folio)) {
+		folio_unlock(folio);
+		ret = -EFAULT;	/* make the VM retry the fault */
+		goto out;
+	}
+
+	/*
+	 * check to see if the folio is mapped already (no holes)
+	 */
+	if (folio_test_mappedtodisk(folio))
+		goto mapped;
+
+	head = folio_buffers(folio);
+	if (head) {
+		int fully_mapped = 1;
+
+		bh = head;
+		do {
+			if (!buffer_mapped(bh)) {
+				fully_mapped = 0;
+				break;
+			}
+		} while (bh = bh->b_this_page, bh != head);
+
+		if (fully_mapped) {
+			folio_set_mappedtodisk(folio);
+			goto mapped;
+		}
+	}
+	folio_unlock(folio);
+
+	/*
+	 * fill hole blocks
+	 */
+	ret = nilfs_transaction_begin(inode->i_sb, &ti, 1);
+	/* never returns -ENOMEM, but may return -ENOSPC */
+	if (unlikely(ret))
+		goto out;
+
+	file_update_time(vma->vm_file);
+	ret = block_page_mkwrite(vma, vmf, nilfs_get_block);
+	if (ret) {
+		nilfs_transaction_abort(inode->i_sb);
+		goto out;
+	}
+	nilfs_set_file_dirty(inode, 1 << (PAGE_SHIFT - inode->i_blkbits));
+	nilfs_transaction_commit(inode->i_sb);
+
+ mapped:
+	/*
+	 * Since checksumming including data blocks is performed to determine
+	 * the validity of the log to be written and used for recovery, it is
+	 * necessary to wait for writeback to finish here, regardless of the
+	 * stable write requirement of the backing device.
+	 */
+	folio_wait_writeback(folio);
+ out:
+	sb_end_pagefault(inode->i_sb);
+	return vmf_fs_error(ret);
+}
+
+static const struct vm_operations_struct nilfs_file_vm_ops = {
+	.fault		= filemap_fault,
+	.map_pages	= filemap_map_pages,
+	.page_mkwrite	= nilfs_page_mkwrite,
+};
+
+static int nilfs_file_mmap_prepare(struct vm_area_desc *desc)
+{
+	file_accessed(desc->file);
+	desc->vm_ops = &nilfs_file_vm_ops;
+	return 0;
+}
+
+/*
+ * We have mostly NULL's here: the current defaults are ok for
+ * the nilfs filesystem.
+ */
+const struct file_operations nilfs_file_operations = {
+	.llseek		= generic_file_llseek,
+	.read_iter	= generic_file_read_iter,
+	.write_iter	= generic_file_write_iter,
+	.unlocked_ioctl	= nilfs_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl	= nilfs_compat_ioctl,
+#endif	/* CONFIG_COMPAT */
+	.mmap_prepare	= nilfs_file_mmap_prepare,
+	.open		= generic_file_open,
+	/* .release	= nilfs_release_file, */
+	.fsync		= nilfs_sync_file,
+	.splice_read	= filemap_splice_read,
+	.splice_write   = iter_file_splice_write,
+	.setlease	= generic_setlease,
+};
+
+const struct inode_operations nilfs_file_inode_operations = {
+	.setattr	= nilfs_setattr,
+	.permission     = nilfs_permission,
+	.fiemap		= nilfs_fiemap,
+	.fileattr_get	= nilfs_fileattr_get,
+	.fileattr_set	= nilfs_fileattr_set,
+};
+
+/* end of file */

linux-7.2.3/fs/nilfs2/gcinode.c

@@ -0,0 +1,184 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Dummy inodes to buffer blocks for garbage collection
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Seiji Kihara, Amagai Yoshiji, and Ryusuke Konishi.
+ * Revised by Ryusuke Konishi.
+ *
+ */
+/*
+ * This file adds the cache of on-disk blocks to be moved in garbage
+ * collection.  The disk blocks are held with dummy inodes (called
+ * gcinodes), and this file provides lookup function of the dummy
+ * inodes and their buffer read function.
+ *
+ * Buffers and pages held by the dummy inodes will be released each
+ * time after they are copied to a new log.  Dirty blocks made on the
+ * current generation and the blocks to be moved by GC never overlap
+ * because the dirty blocks make a new generation; they rather must be
+ * written individually.
+ */
+
+#include <linux/buffer_head.h>
+#include <linux/mpage.h>
+#include <linux/hash.h>
+#include <linux/slab.h>
+#include <linux/swap.h>
+#include "nilfs.h"
+#include "btree.h"
+#include "btnode.h"
+#include "page.h"
+#include "mdt.h"
+#include "dat.h"
+#include "ifile.h"
+
+/*
+ * nilfs_gccache_submit_read_data() - add data buffer and submit read request
+ * @inode - gc inode
+ * @blkoff - dummy offset treated as the key for the page cache
+ * @pbn - physical block number of the block
+ * @vbn - virtual block number of the block, 0 for non-virtual block
+ * @out_bh - indirect pointer to a buffer_head struct to receive the results
+ *
+ * Description: nilfs_gccache_submit_read_data() registers the data buffer
+ * specified by @pbn to the GC pagecache with the key @blkoff.
+ * This function sets @vbn (@pbn if @vbn is zero) in b_blocknr of the buffer.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- The block specified with @pbn does not exist.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_gccache_submit_read_data(struct inode *inode, sector_t blkoff,
+				   sector_t pbn, __u64 vbn,
+				   struct buffer_head **out_bh)
+{
+	struct buffer_head *bh;
+	int err;
+
+	bh = nilfs_grab_buffer(inode, inode->i_mapping, blkoff, 0);
+	if (unlikely(!bh))
+		return -ENOMEM;
+
+	if (buffer_uptodate(bh))
+		goto out;
+
+	if (pbn == 0) {
+		struct the_nilfs *nilfs = inode->i_sb->s_fs_info;
+
+		err = nilfs_dat_translate(nilfs->ns_dat, vbn, &pbn);
+		if (unlikely(err)) /* -EIO, -ENOMEM, -ENOENT */
+			goto failed;
+	}
+
+	lock_buffer(bh);
+	if (buffer_uptodate(bh)) {
+		unlock_buffer(bh);
+		goto out;
+	}
+
+	if (!buffer_mapped(bh))
+		set_buffer_mapped(bh);
+	bh->b_blocknr = pbn;
+	bh_submit(bh, REQ_OP_READ, bh_end_read);
+	if (vbn)
+		bh->b_blocknr = vbn;
+ out:
+	err = 0;
+	*out_bh = bh;
+
+ failed:
+	folio_unlock(bh->b_folio);
+	folio_put(bh->b_folio);
+	if (unlikely(err))
+		brelse(bh);
+	return err;
+}
+
+/*
+ * nilfs_gccache_submit_read_node() - add node buffer and submit read request
+ * @inode - gc inode
+ * @pbn - physical block number for the block
+ * @vbn - virtual block number for the block
+ * @out_bh - indirect pointer to a buffer_head struct to receive the results
+ *
+ * Description: nilfs_gccache_submit_read_node() registers the node buffer
+ * specified by @vbn to the GC pagecache.  @pbn can be supplied by the
+ * caller to avoid translation of the disk block address.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- Invalid virtual block address.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_gccache_submit_read_node(struct inode *inode, sector_t pbn,
+				   __u64 vbn, struct buffer_head **out_bh)
+{
+	struct inode *btnc_inode = NILFS_I(inode)->i_assoc_inode;
+	int ret;
+
+	ret = nilfs_btnode_submit_block(btnc_inode->i_mapping, vbn ? : pbn, pbn,
+					REQ_OP_READ, out_bh, &pbn);
+	if (ret == -EEXIST) /* internal code (cache hit) */
+		ret = 0;
+	return ret;
+}
+
+int nilfs_gccache_wait_and_mark_dirty(struct buffer_head *bh)
+{
+	wait_on_buffer(bh);
+	if (!buffer_uptodate(bh)) {
+		struct inode *inode = bh->b_folio->mapping->host;
+
+		nilfs_err(inode->i_sb,
+			  "I/O error reading %s block for GC (ino=%llu, vblocknr=%llu)",
+			  buffer_nilfs_node(bh) ? "node" : "data",
+			  inode->i_ino, (unsigned long long)bh->b_blocknr);
+		return -EIO;
+	}
+	if (buffer_dirty(bh))
+		return -EEXIST;
+
+	if (buffer_nilfs_node(bh) && nilfs_btree_broken_node_block(bh)) {
+		clear_buffer_uptodate(bh);
+		return -EIO;
+	}
+	mark_buffer_dirty(bh);
+	return 0;
+}
+
+int nilfs_init_gcinode(struct inode *inode)
+{
+	struct nilfs_inode_info *ii = NILFS_I(inode);
+
+	inode->i_mode = S_IFREG;
+	mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
+	inode->i_mapping->a_ops = &nilfs_buffer_cache_aops;
+
+	ii->i_flags = 0;
+	nilfs_bmap_init_gc(ii->i_bmap);
+
+	return nilfs_attach_btree_node_cache(inode);
+}
+
+/**
+ * nilfs_remove_all_gcinodes() - remove all unprocessed gc inodes
+ * @nilfs: NILFS filesystem instance
+ */
+void nilfs_remove_all_gcinodes(struct the_nilfs *nilfs)
+{
+	struct list_head *head = &nilfs->ns_gc_inodes;
+	struct nilfs_inode_info *ii;
+
+	while (!list_empty(head)) {
+		ii = list_first_entry(head, struct nilfs_inode_info, i_dirty);
+		list_del_init(&ii->i_dirty);
+		truncate_inode_pages(&ii->vfs_inode.i_data, 0);
+		nilfs_btnode_cache_clear(ii->i_assoc_inode->i_mapping);
+		iput(&ii->vfs_inode);
+	}
+}

linux-7.2.3/fs/nilfs2/ifile.c

@@ -0,0 +1,216 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS inode file
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Amagai Yoshiji.
+ * Revised by Ryusuke Konishi.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include "nilfs.h"
+#include "mdt.h"
+#include "alloc.h"
+#include "ifile.h"
+#include "cpfile.h"
+
+/**
+ * struct nilfs_ifile_info - on-memory private data of ifile
+ * @mi: on-memory private data of metadata file
+ * @palloc_cache: persistent object allocator cache of ifile
+ */
+struct nilfs_ifile_info {
+	struct nilfs_mdt_info mi;
+	struct nilfs_palloc_cache palloc_cache;
+};
+
+static inline struct nilfs_ifile_info *NILFS_IFILE_I(struct inode *ifile)
+{
+	return (struct nilfs_ifile_info *)NILFS_MDT(ifile);
+}
+
+/**
+ * nilfs_ifile_create_inode - create a new disk inode
+ * @ifile: ifile inode
+ * @out_ino: pointer to a variable to store inode number
+ * @out_bh: buffer_head contains newly allocated disk inode
+ *
+ * nilfs_ifile_create_inode() allocates a new inode in the ifile metadata
+ * file and stores the inode number in the variable pointed to by @out_ino,
+ * as well as storing the ifile's buffer with the disk inode in the location
+ * pointed to by @out_bh.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ * * %-ENOSPC	- No inode left.
+ */
+int nilfs_ifile_create_inode(struct inode *ifile, ino_t *out_ino,
+			     struct buffer_head **out_bh)
+{
+	struct nilfs_palloc_req req;
+	int ret;
+
+	req.pr_entry_nr = NILFS_FIRST_INO(ifile->i_sb);
+	req.pr_entry_bh = NULL;
+
+	ret = nilfs_palloc_prepare_alloc_entry(ifile, &req, false);
+	if (!ret) {
+		ret = nilfs_palloc_get_entry_block(ifile, req.pr_entry_nr, 1,
+						   &req.pr_entry_bh);
+		if (ret < 0)
+			nilfs_palloc_abort_alloc_entry(ifile, &req);
+	}
+	if (ret < 0) {
+		brelse(req.pr_entry_bh);
+		return ret;
+	}
+	nilfs_palloc_commit_alloc_entry(ifile, &req);
+	mark_buffer_dirty(req.pr_entry_bh);
+	nilfs_mdt_mark_dirty(ifile);
+	*out_ino = (ino_t)req.pr_entry_nr;
+	*out_bh = req.pr_entry_bh;
+	return 0;
+}
+
+/**
+ * nilfs_ifile_delete_inode - delete a disk inode
+ * @ifile: ifile inode
+ * @ino: inode number
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- Inode number unallocated.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_ifile_delete_inode(struct inode *ifile, ino_t ino)
+{
+	struct nilfs_palloc_req req = {
+		.pr_entry_nr = ino, .pr_entry_bh = NULL
+	};
+	struct nilfs_inode *raw_inode;
+	size_t offset;
+	int ret;
+
+	ret = nilfs_palloc_prepare_free_entry(ifile, &req);
+	if (!ret) {
+		ret = nilfs_palloc_get_entry_block(ifile, req.pr_entry_nr, 0,
+						   &req.pr_entry_bh);
+		if (ret < 0)
+			nilfs_palloc_abort_free_entry(ifile, &req);
+	}
+	if (ret < 0) {
+		brelse(req.pr_entry_bh);
+		return ret;
+	}
+
+	offset = nilfs_palloc_entry_offset(ifile, req.pr_entry_nr,
+					   req.pr_entry_bh);
+	raw_inode = kmap_local_folio(req.pr_entry_bh->b_folio, offset);
+	raw_inode->i_flags = 0;
+	kunmap_local(raw_inode);
+
+	mark_buffer_dirty(req.pr_entry_bh);
+	brelse(req.pr_entry_bh);
+
+	nilfs_palloc_commit_free_entry(ifile, &req);
+
+	return 0;
+}
+
+int nilfs_ifile_get_inode_block(struct inode *ifile, ino_t ino,
+				struct buffer_head **out_bh)
+{
+	struct super_block *sb = ifile->i_sb;
+	int err;
+
+	if (unlikely(!NILFS_VALID_INODE(sb, ino))) {
+		nilfs_error(sb, "bad inode number: %lu", (unsigned long)ino);
+		return -EINVAL;
+	}
+
+	err = nilfs_palloc_get_entry_block(ifile, ino, 0, out_bh);
+	if (unlikely(err))
+		nilfs_warn(sb, "error %d reading inode: ino=%lu",
+			   err, (unsigned long)ino);
+	return err;
+}
+
+/**
+ * nilfs_ifile_count_free_inodes - calculate free inodes count
+ * @ifile: ifile inode
+ * @nmaxinodes: current maximum of available inodes count [out]
+ * @nfreeinodes: free inodes count [out]
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int nilfs_ifile_count_free_inodes(struct inode *ifile,
+				    u64 *nmaxinodes, u64 *nfreeinodes)
+{
+	u64 nused;
+	int err;
+
+	*nmaxinodes = 0;
+	*nfreeinodes = 0;
+
+	nused = atomic64_read(&NILFS_I(ifile)->i_root->inodes_count);
+	err = nilfs_palloc_count_max_entries(ifile, nused, nmaxinodes);
+	if (likely(!err))
+		*nfreeinodes = *nmaxinodes - nused;
+	return err;
+}
+
+/**
+ * nilfs_ifile_read - read or get ifile inode
+ * @sb: super block instance
+ * @root: root object
+ * @cno: number of checkpoint entry to read
+ * @inode_size: size of an inode
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EINVAL	- Invalid checkpoint.
+ * * %-ENOMEM	- Insufficient memory available.
+ * * %-EIO	- I/O error (including metadata corruption).
+ */
+int nilfs_ifile_read(struct super_block *sb, struct nilfs_root *root,
+		     __u64 cno, size_t inode_size)
+{
+	struct the_nilfs *nilfs;
+	struct inode *ifile;
+	int err;
+
+	ifile = nilfs_iget_locked(sb, root, NILFS_IFILE_INO);
+	if (unlikely(!ifile))
+		return -ENOMEM;
+	if (!(inode_state_read_once(ifile) & I_NEW))
+		goto out;
+
+	err = nilfs_mdt_init(ifile, NILFS_MDT_GFP,
+			     sizeof(struct nilfs_ifile_info));
+	if (err)
+		goto failed;
+
+	err = nilfs_palloc_init_blockgroup(ifile, inode_size);
+	if (err)
+		goto failed;
+
+	nilfs_palloc_setup_cache(ifile, &NILFS_IFILE_I(ifile)->palloc_cache);
+
+	nilfs = sb->s_fs_info;
+	err = nilfs_cpfile_read_checkpoint(nilfs->ns_cpfile, cno, root, ifile);
+	if (err)
+		goto failed;
+
+	unlock_new_inode(ifile);
+ out:
+	return 0;
+ failed:
+	iget_failed(ifile);
+	return err;
+}

linux-7.2.3/fs/nilfs2/ifile.h

@@ -0,0 +1,43 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS inode file
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Amagai Yoshiji.
+ * Revised by Ryusuke Konishi.
+ *
+ */
+
+#ifndef _NILFS_IFILE_H
+#define _NILFS_IFILE_H
+
+#include <linux/fs.h>
+#include <linux/buffer_head.h>
+#include "mdt.h"
+#include "alloc.h"
+
+
+static inline struct nilfs_inode *
+nilfs_ifile_map_inode(struct inode *ifile, ino_t ino, struct buffer_head *ibh)
+{
+	size_t __offset_in_folio = nilfs_palloc_entry_offset(ifile, ino, ibh);
+
+	return kmap_local_folio(ibh->b_folio, __offset_in_folio);
+}
+
+static inline void nilfs_ifile_unmap_inode(struct nilfs_inode *raw_inode)
+{
+	kunmap_local(raw_inode);
+}
+
+int nilfs_ifile_create_inode(struct inode *, ino_t *, struct buffer_head **);
+int nilfs_ifile_delete_inode(struct inode *, ino_t);
+int nilfs_ifile_get_inode_block(struct inode *, ino_t, struct buffer_head **);
+
+int nilfs_ifile_count_free_inodes(struct inode *, u64 *, u64 *);
+
+int nilfs_ifile_read(struct super_block *sb, struct nilfs_root *root,
+		     __u64 cno, size_t inode_size);
+
+#endif	/* _NILFS_IFILE_H */

linux-7.2.3/fs/nilfs2/mdt.c

@@ -0,0 +1,678 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Meta data file for NILFS
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi.
+ */
+
+#include <linux/buffer_head.h>
+#include <linux/mpage.h>
+#include <linux/mm.h>
+#include <linux/writeback.h>
+#include <linux/backing-dev.h>
+#include <linux/swap.h>
+#include <linux/slab.h>
+#include "nilfs.h"
+#include "btnode.h"
+#include "segment.h"
+#include "page.h"
+#include "mdt.h"
+#include "alloc.h"		/* nilfs_palloc_destroy_cache() */
+
+#include <trace/events/nilfs2.h>
+
+#define NILFS_MDT_MAX_RA_BLOCKS		(16 - 1)
+
+
+static int
+nilfs_mdt_insert_new_block(struct inode *inode, unsigned long block,
+			   struct buffer_head *bh,
+			   void (*init_block)(struct inode *,
+					      struct buffer_head *, void *))
+{
+	struct nilfs_inode_info *ii = NILFS_I(inode);
+	struct folio *folio = bh->b_folio;
+	void *from;
+	int ret;
+
+	/* Caller exclude read accesses using page lock */
+
+	/* set_buffer_new(bh); */
+	bh->b_blocknr = 0;
+
+	ret = nilfs_bmap_insert(ii->i_bmap, block, (unsigned long)bh);
+	if (unlikely(ret))
+		return ret;
+
+	set_buffer_mapped(bh);
+
+	/* Initialize block (block size > PAGE_SIZE not yet supported) */
+	from = kmap_local_folio(folio, offset_in_folio(folio, bh->b_data));
+	memset(from, 0, bh->b_size);
+	if (init_block)
+		init_block(inode, bh, from);
+	kunmap_local(from);
+
+	flush_dcache_folio(folio);
+
+	set_buffer_uptodate(bh);
+	mark_buffer_dirty(bh);
+	nilfs_mdt_mark_dirty(inode);
+
+	trace_nilfs2_mdt_insert_new_block(inode, inode->i_ino, block);
+
+	return 0;
+}
+
+static int nilfs_mdt_create_block(struct inode *inode, unsigned long block,
+				  struct buffer_head **out_bh,
+				  void (*init_block)(struct inode *,
+						     struct buffer_head *,
+						     void *))
+{
+	struct super_block *sb = inode->i_sb;
+	struct nilfs_transaction_info ti;
+	struct buffer_head *bh;
+	int err;
+
+	nilfs_transaction_begin(sb, &ti, 0);
+
+	err = -ENOMEM;
+	bh = nilfs_grab_buffer(inode, inode->i_mapping, block, 0);
+	if (unlikely(!bh))
+		goto failed_unlock;
+
+	err = -EEXIST;
+	if (buffer_uptodate(bh))
+		goto failed_bh;
+
+	wait_on_buffer(bh);
+	if (buffer_uptodate(bh))
+		goto failed_bh;
+
+	err = nilfs_mdt_insert_new_block(inode, block, bh, init_block);
+	if (likely(!err)) {
+		get_bh(bh);
+		*out_bh = bh;
+	}
+
+ failed_bh:
+	folio_unlock(bh->b_folio);
+	folio_put(bh->b_folio);
+	brelse(bh);
+
+ failed_unlock:
+	if (likely(!err))
+		err = nilfs_transaction_commit(sb);
+	else
+		nilfs_transaction_abort(sb);
+
+	return err;
+}
+
+static int
+nilfs_mdt_submit_block(struct inode *inode, unsigned long blkoff, blk_opf_t opf,
+		       struct buffer_head **out_bh)
+{
+	struct buffer_head *bh;
+	__u64 blknum = 0;
+	int ret = -ENOMEM;
+
+	bh = nilfs_grab_buffer(inode, inode->i_mapping, blkoff, 0);
+	if (unlikely(!bh))
+		goto failed;
+
+	ret = -EEXIST; /* internal code */
+	if (buffer_uptodate(bh))
+		goto out;
+
+	if (opf & REQ_RAHEAD) {
+		if (!trylock_buffer(bh)) {
+			ret = -EBUSY;
+			goto failed_bh;
+		}
+	} else /* opf == REQ_OP_READ */
+		lock_buffer(bh);
+
+	if (buffer_uptodate(bh)) {
+		unlock_buffer(bh);
+		goto out;
+	}
+
+	ret = nilfs_bmap_lookup(NILFS_I(inode)->i_bmap, blkoff, &blknum);
+	if (unlikely(ret)) {
+		unlock_buffer(bh);
+		goto failed_bh;
+	}
+	map_bh(bh, inode->i_sb, (sector_t)blknum);
+
+	bh_submit(bh, opf, bh_end_read);
+	ret = 0;
+
+	trace_nilfs2_mdt_submit_block(inode, inode->i_ino, blkoff,
+				      opf & REQ_OP_MASK);
+ out:
+	get_bh(bh);
+	*out_bh = bh;
+
+ failed_bh:
+	folio_unlock(bh->b_folio);
+	folio_put(bh->b_folio);
+	brelse(bh);
+ failed:
+	return ret;
+}
+
+static int nilfs_mdt_read_block(struct inode *inode, unsigned long block,
+				int readahead, struct buffer_head **out_bh)
+{
+	struct buffer_head *first_bh, *bh;
+	unsigned long blkoff;
+	int i, nr_ra_blocks = NILFS_MDT_MAX_RA_BLOCKS;
+	int err;
+
+	err = nilfs_mdt_submit_block(inode, block, REQ_OP_READ, &first_bh);
+	if (err == -EEXIST) /* internal code */
+		goto out;
+
+	if (unlikely(err))
+		goto failed;
+
+	if (readahead) {
+		blkoff = block + 1;
+		for (i = 0; i < nr_ra_blocks; i++, blkoff++) {
+			err = nilfs_mdt_submit_block(inode, blkoff,
+						REQ_OP_READ | REQ_RAHEAD, &bh);
+			if (likely(!err || err == -EEXIST))
+				brelse(bh);
+			else if (err != -EBUSY)
+				break;
+				/* abort readahead if bmap lookup failed */
+			if (!buffer_locked(first_bh))
+				goto out_no_wait;
+		}
+	}
+
+	wait_on_buffer(first_bh);
+
+ out_no_wait:
+	err = -EIO;
+	if (!buffer_uptodate(first_bh)) {
+		nilfs_err(inode->i_sb,
+			  "I/O error reading meta-data file (ino=%llu, block-offset=%lu)",
+			  inode->i_ino, block);
+		goto failed_bh;
+	}
+ out:
+	*out_bh = first_bh;
+	return 0;
+
+ failed_bh:
+	brelse(first_bh);
+ failed:
+	return err;
+}
+
+/**
+ * nilfs_mdt_get_block - read or create a buffer on meta data file.
+ * @inode: inode of the meta data file
+ * @blkoff: block offset
+ * @create: create flag
+ * @init_block: initializer used for newly allocated block
+ * @out_bh: output of a pointer to the buffer_head
+ *
+ * nilfs_mdt_get_block() looks up the specified buffer and tries to create
+ * a new buffer if @create is not zero.  If (and only if) this function
+ * succeeds, it stores a pointer to the retrieved buffer head in the location
+ * pointed to by @out_bh.
+ *
+ * The retrieved buffer may be either an existing one or a newly allocated one.
+ * For a newly created buffer, if the callback function argument @init_block
+ * is non-NULL, the callback will be called with the buffer locked to format
+ * the block.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- The specified block does not exist (hole block).
+ * * %-ENOMEM	- Insufficient memory available.
+ * * %-EROFS	- Read only filesystem (for create mode).
+ */
+int nilfs_mdt_get_block(struct inode *inode, unsigned long blkoff, int create,
+			void (*init_block)(struct inode *,
+					   struct buffer_head *, void *),
+			struct buffer_head **out_bh)
+{
+	int ret;
+
+	/* Should be rewritten with merging nilfs_mdt_read_block() */
+ retry:
+	ret = nilfs_mdt_read_block(inode, blkoff, !create, out_bh);
+	if (!create || ret != -ENOENT)
+		return ret;
+
+	ret = nilfs_mdt_create_block(inode, blkoff, out_bh, init_block);
+	if (unlikely(ret == -EEXIST)) {
+		/* create = 0; */  /* limit read-create loop retries */
+		goto retry;
+	}
+	return ret;
+}
+
+/**
+ * nilfs_mdt_find_block - find and get a buffer on meta data file.
+ * @inode: inode of the meta data file
+ * @start: start block offset (inclusive)
+ * @end: end block offset (inclusive)
+ * @blkoff: block offset
+ * @out_bh: place to store a pointer to buffer_head struct
+ *
+ * nilfs_mdt_find_block() looks up an existing block in range of
+ * [@start, @end] and stores pointer to a buffer head of the block to
+ * @out_bh, and block offset to @blkoff, respectively.  @out_bh and
+ * @blkoff are substituted only when zero is returned.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- No block was found in the range.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_mdt_find_block(struct inode *inode, unsigned long start,
+			 unsigned long end, unsigned long *blkoff,
+			 struct buffer_head **out_bh)
+{
+	__u64 next;
+	int ret;
+
+	if (unlikely(start > end))
+		return -ENOENT;
+
+	ret = nilfs_mdt_read_block(inode, start, true, out_bh);
+	if (!ret) {
+		*blkoff = start;
+		goto out;
+	}
+	if (unlikely(ret != -ENOENT || start == ULONG_MAX))
+		goto out;
+
+	ret = nilfs_bmap_seek_key(NILFS_I(inode)->i_bmap, start + 1, &next);
+	if (!ret) {
+		if (next <= end) {
+			ret = nilfs_mdt_read_block(inode, next, true, out_bh);
+			if (!ret)
+				*blkoff = next;
+		} else {
+			ret = -ENOENT;
+		}
+	}
+out:
+	return ret;
+}
+
+/**
+ * nilfs_mdt_delete_block - make a hole on the meta data file.
+ * @inode: inode of the meta data file
+ * @block: block offset
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- Non-existent block.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_mdt_delete_block(struct inode *inode, unsigned long block)
+{
+	struct nilfs_inode_info *ii = NILFS_I(inode);
+	int err;
+
+	err = nilfs_bmap_delete(ii->i_bmap, block);
+	if (!err || err == -ENOENT) {
+		nilfs_mdt_mark_dirty(inode);
+		nilfs_mdt_forget_block(inode, block);
+	}
+	return err;
+}
+
+/**
+ * nilfs_mdt_forget_block - discard dirty state and try to remove the page
+ * @inode: inode of the meta data file
+ * @block: block offset
+ *
+ * nilfs_mdt_forget_block() clears a dirty flag of the specified buffer, and
+ * tries to release the page including the buffer from a page cache.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EBUSY	- Page has an active buffer.
+ * * %-ENOENT	- Page cache has no page addressed by the offset.
+ */
+int nilfs_mdt_forget_block(struct inode *inode, unsigned long block)
+{
+	pgoff_t index = block >> (PAGE_SHIFT - inode->i_blkbits);
+	struct folio *folio;
+	struct buffer_head *bh;
+	int ret = 0;
+	int still_dirty;
+
+	folio = filemap_lock_folio(inode->i_mapping, index);
+	if (IS_ERR(folio))
+		return -ENOENT;
+
+	folio_wait_writeback(folio);
+
+	bh = folio_buffers(folio);
+	if (bh) {
+		unsigned long first_block = index <<
+				(PAGE_SHIFT - inode->i_blkbits);
+		bh = get_nth_bh(bh, block - first_block);
+		nilfs_forget_buffer(bh);
+	}
+	still_dirty = folio_test_dirty(folio);
+	folio_unlock(folio);
+	folio_put(folio);
+
+	if (still_dirty ||
+	    invalidate_inode_pages2_range(inode->i_mapping, index, index) != 0)
+		ret = -EBUSY;
+	return ret;
+}
+
+int nilfs_mdt_fetch_dirty(struct inode *inode)
+{
+	struct nilfs_inode_info *ii = NILFS_I(inode);
+
+	if (nilfs_bmap_test_and_clear_dirty(ii->i_bmap)) {
+		set_bit(NILFS_I_DIRTY, &ii->i_state);
+		return 1;
+	}
+	return test_bit(NILFS_I_DIRTY, &ii->i_state);
+}
+
+static int nilfs_mdt_write_folio(struct folio *folio,
+		struct writeback_control *wbc)
+{
+	struct inode *inode = folio->mapping->host;
+	struct super_block *sb;
+	int err = 0;
+
+	if (inode && sb_rdonly(inode->i_sb)) {
+		/*
+		 * It means that filesystem was remounted in read-only
+		 * mode because of error or metadata corruption. But we
+		 * have dirty folios that try to be flushed in background.
+		 * So, here we simply discard this dirty folio.
+		 */
+		nilfs_clear_folio_dirty(folio);
+		folio_unlock(folio);
+		return -EROFS;
+	}
+
+	folio_redirty_for_writepage(wbc, folio);
+	folio_unlock(folio);
+
+	if (!inode)
+		return 0;
+
+	sb = inode->i_sb;
+
+	if (wbc->sync_mode == WB_SYNC_ALL)
+		err = nilfs_construct_segment(sb);
+
+	return err;
+}
+
+static int nilfs_mdt_writeback(struct address_space *mapping,
+		struct writeback_control *wbc)
+{
+	struct folio *folio = NULL;
+	int error;
+
+	while ((folio = writeback_iter(mapping, wbc, folio, &error)))
+		error = nilfs_mdt_write_folio(folio, wbc);
+
+	return error;
+}
+
+static const struct address_space_operations def_mdt_aops = {
+	.dirty_folio		= block_dirty_folio,
+	.invalidate_folio	= block_invalidate_folio,
+	.writepages		= nilfs_mdt_writeback,
+	.migrate_folio		= buffer_migrate_folio_norefs,
+};
+
+static const struct inode_operations def_mdt_iops;
+static const struct file_operations def_mdt_fops;
+
+
+int nilfs_mdt_init(struct inode *inode, gfp_t gfp_mask, size_t objsz)
+{
+	struct nilfs_mdt_info *mi;
+
+	mi = kzalloc(max(sizeof(*mi), objsz), GFP_NOFS);
+	if (!mi)
+		return -ENOMEM;
+
+	init_rwsem(&mi->mi_sem);
+	inode->i_private = mi;
+
+	inode->i_mode = S_IFREG;
+	mapping_set_gfp_mask(inode->i_mapping, gfp_mask);
+
+	inode->i_op = &def_mdt_iops;
+	inode->i_fop = &def_mdt_fops;
+	inode->i_mapping->a_ops = &def_mdt_aops;
+
+	return 0;
+}
+
+/**
+ * nilfs_mdt_clear - do cleanup for the metadata file
+ * @inode: inode of the metadata file
+ */
+void nilfs_mdt_clear(struct inode *inode)
+{
+	struct nilfs_mdt_info *mdi = NILFS_MDT(inode);
+	struct nilfs_shadow_map *shadow = mdi->mi_shadow;
+
+	if (mdi->mi_palloc_cache)
+		nilfs_palloc_destroy_cache(inode);
+
+	if (shadow) {
+		struct inode *s_inode = shadow->inode;
+
+		shadow->inode = NULL;
+		iput(s_inode);
+		mdi->mi_shadow = NULL;
+	}
+}
+
+/**
+ * nilfs_mdt_destroy - release resources used by the metadata file
+ * @inode: inode of the metadata file
+ */
+void nilfs_mdt_destroy(struct inode *inode)
+{
+	struct nilfs_mdt_info *mdi = NILFS_MDT(inode);
+
+	kfree(mdi->mi_bgl); /* kfree(NULL) is safe */
+	kfree(mdi);
+}
+
+void nilfs_mdt_set_entry_size(struct inode *inode, unsigned int entry_size,
+			      unsigned int header_size)
+{
+	struct nilfs_mdt_info *mi = NILFS_MDT(inode);
+
+	mi->mi_entry_size = entry_size;
+	mi->mi_entries_per_block = i_blocksize(inode) / entry_size;
+	mi->mi_first_entry_offset = DIV_ROUND_UP(header_size, entry_size);
+}
+
+/**
+ * nilfs_mdt_setup_shadow_map - setup shadow map and bind it to metadata file
+ * @inode: inode of the metadata file
+ * @shadow: shadow mapping
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int nilfs_mdt_setup_shadow_map(struct inode *inode,
+			       struct nilfs_shadow_map *shadow)
+{
+	struct nilfs_mdt_info *mi = NILFS_MDT(inode);
+	struct inode *s_inode;
+
+	INIT_LIST_HEAD(&shadow->frozen_buffers);
+
+	s_inode = nilfs_iget_for_shadow(inode);
+	if (IS_ERR(s_inode))
+		return PTR_ERR(s_inode);
+
+	shadow->inode = s_inode;
+	mi->mi_shadow = shadow;
+	return 0;
+}
+
+/**
+ * nilfs_mdt_save_to_shadow_map - copy bmap and dirty pages to shadow map
+ * @inode: inode of the metadata file
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int nilfs_mdt_save_to_shadow_map(struct inode *inode)
+{
+	struct nilfs_mdt_info *mi = NILFS_MDT(inode);
+	struct nilfs_inode_info *ii = NILFS_I(inode);
+	struct nilfs_shadow_map *shadow = mi->mi_shadow;
+	struct inode *s_inode = shadow->inode;
+	int ret;
+
+	ret = nilfs_copy_dirty_pages(s_inode->i_mapping, inode->i_mapping);
+	if (ret)
+		goto out;
+
+	ret = nilfs_copy_dirty_pages(NILFS_I(s_inode)->i_assoc_inode->i_mapping,
+				     ii->i_assoc_inode->i_mapping);
+	if (ret)
+		goto out;
+
+	nilfs_bmap_save(ii->i_bmap, &shadow->bmap_store);
+ out:
+	return ret;
+}
+
+int nilfs_mdt_freeze_buffer(struct inode *inode, struct buffer_head *bh)
+{
+	struct nilfs_shadow_map *shadow = NILFS_MDT(inode)->mi_shadow;
+	struct buffer_head *bh_frozen;
+	struct folio *folio;
+	int blkbits = inode->i_blkbits;
+
+	folio = filemap_grab_folio(shadow->inode->i_mapping,
+			bh->b_folio->index);
+	if (IS_ERR(folio))
+		return PTR_ERR(folio);
+
+	bh_frozen = folio_buffers(folio);
+	if (!bh_frozen)
+		bh_frozen = create_empty_buffers(folio, 1 << blkbits, 0);
+
+	bh_frozen = get_nth_bh(bh_frozen,
+			       offset_in_folio(folio, bh->b_data) >> blkbits);
+
+	if (!buffer_uptodate(bh_frozen))
+		nilfs_copy_buffer(bh_frozen, bh);
+	if (list_empty(&bh_frozen->b_assoc_buffers)) {
+		list_add_tail(&bh_frozen->b_assoc_buffers,
+			      &shadow->frozen_buffers);
+		set_buffer_nilfs_redirected(bh);
+	} else {
+		brelse(bh_frozen); /* already frozen */
+	}
+
+	folio_unlock(folio);
+	folio_put(folio);
+	return 0;
+}
+
+struct buffer_head *
+nilfs_mdt_get_frozen_buffer(struct inode *inode, struct buffer_head *bh)
+{
+	struct nilfs_shadow_map *shadow = NILFS_MDT(inode)->mi_shadow;
+	struct buffer_head *bh_frozen = NULL;
+	struct folio *folio;
+	int n;
+
+	folio = filemap_lock_folio(shadow->inode->i_mapping,
+			bh->b_folio->index);
+	if (!IS_ERR(folio)) {
+		bh_frozen = folio_buffers(folio);
+		if (bh_frozen) {
+			n = offset_in_folio(folio, bh->b_data) >>
+				inode->i_blkbits;
+			bh_frozen = get_nth_bh(bh_frozen, n);
+		}
+		folio_unlock(folio);
+		folio_put(folio);
+	}
+	return bh_frozen;
+}
+
+static void nilfs_release_frozen_buffers(struct nilfs_shadow_map *shadow)
+{
+	struct list_head *head = &shadow->frozen_buffers;
+	struct buffer_head *bh;
+
+	while (!list_empty(head)) {
+		bh = list_first_entry(head, struct buffer_head,
+				      b_assoc_buffers);
+		list_del_init(&bh->b_assoc_buffers);
+		brelse(bh); /* drop ref-count to make it releasable */
+	}
+}
+
+/**
+ * nilfs_mdt_restore_from_shadow_map - restore dirty pages and bmap state
+ * @inode: inode of the metadata file
+ */
+void nilfs_mdt_restore_from_shadow_map(struct inode *inode)
+{
+	struct nilfs_mdt_info *mi = NILFS_MDT(inode);
+	struct nilfs_inode_info *ii = NILFS_I(inode);
+	struct nilfs_shadow_map *shadow = mi->mi_shadow;
+
+	down_write(&mi->mi_sem);
+
+	if (mi->mi_palloc_cache)
+		nilfs_palloc_clear_cache(inode);
+
+	nilfs_clear_dirty_pages(inode->i_mapping);
+	nilfs_copy_back_pages(inode->i_mapping, shadow->inode->i_mapping);
+
+	nilfs_clear_dirty_pages(ii->i_assoc_inode->i_mapping);
+	nilfs_copy_back_pages(ii->i_assoc_inode->i_mapping,
+			      NILFS_I(shadow->inode)->i_assoc_inode->i_mapping);
+
+	nilfs_bmap_restore(ii->i_bmap, &shadow->bmap_store);
+
+	up_write(&mi->mi_sem);
+}
+
+/**
+ * nilfs_mdt_clear_shadow_map - truncate pages in shadow map caches
+ * @inode: inode of the metadata file
+ */
+void nilfs_mdt_clear_shadow_map(struct inode *inode)
+{
+	struct nilfs_mdt_info *mi = NILFS_MDT(inode);
+	struct nilfs_shadow_map *shadow = mi->mi_shadow;
+	struct inode *shadow_btnc_inode = NILFS_I(shadow->inode)->i_assoc_inode;
+
+	down_write(&mi->mi_sem);
+	nilfs_release_frozen_buffers(shadow);
+	truncate_inode_pages(shadow->inode->i_mapping, 0);
+	truncate_inode_pages(shadow_btnc_inode->i_mapping, 0);
+	up_write(&mi->mi_sem);
+}

linux-7.2.3/fs/nilfs2/mdt.h

@@ -0,0 +1,115 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS meta data file prototype and definitions
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi.
+ */
+
+#ifndef _NILFS_MDT_H
+#define _NILFS_MDT_H
+
+#include <linux/buffer_head.h>
+#include <linux/blockgroup_lock.h>
+#include "nilfs.h"
+#include "page.h"
+
+/**
+ * struct nilfs_shadow_map - shadow mapping of meta data file
+ * @bmap_store: shadow copy of bmap state
+ * @inode: holder of page caches used in shadow mapping
+ * @frozen_buffers: list of frozen buffers
+ */
+struct nilfs_shadow_map {
+	struct nilfs_bmap_store bmap_store;
+	struct inode *inode;
+	struct list_head frozen_buffers;
+};
+
+/**
+ * struct nilfs_mdt_info - on-memory private data of meta data files
+ * @mi_sem: reader/writer semaphore for meta data operations
+ * @mi_bgl: per-blockgroup locking
+ * @mi_entry_size: size of an entry
+ * @mi_first_entry_offset: offset to the first entry
+ * @mi_entries_per_block: number of entries in a block
+ * @mi_palloc_cache: persistent object allocator cache
+ * @mi_shadow: shadow of bmap and page caches
+ * @mi_blocks_per_group: number of blocks in a group
+ * @mi_blocks_per_desc_block: number of blocks per descriptor block
+ */
+struct nilfs_mdt_info {
+	struct rw_semaphore	mi_sem;
+	struct blockgroup_lock *mi_bgl;
+	unsigned int		mi_entry_size;
+	unsigned int		mi_first_entry_offset;
+	unsigned long		mi_entries_per_block;
+	struct nilfs_palloc_cache *mi_palloc_cache;
+	struct nilfs_shadow_map *mi_shadow;
+	unsigned long		mi_blocks_per_group;
+	unsigned long		mi_blocks_per_desc_block;
+};
+
+static inline struct nilfs_mdt_info *NILFS_MDT(const struct inode *inode)
+{
+	return inode->i_private;
+}
+
+static inline int nilfs_is_metadata_file_inode(const struct inode *inode)
+{
+	return inode->i_private != NULL;
+}
+
+/* Default GFP flags using highmem */
+#define NILFS_MDT_GFP      (__GFP_RECLAIM | __GFP_IO | __GFP_HIGHMEM)
+
+int nilfs_mdt_get_block(struct inode *, unsigned long, int,
+			void (*init_block)(struct inode *,
+					   struct buffer_head *, void *),
+			struct buffer_head **);
+int nilfs_mdt_find_block(struct inode *inode, unsigned long start,
+			 unsigned long end, unsigned long *blkoff,
+			 struct buffer_head **out_bh);
+int nilfs_mdt_delete_block(struct inode *, unsigned long);
+int nilfs_mdt_forget_block(struct inode *, unsigned long);
+int nilfs_mdt_fetch_dirty(struct inode *);
+
+int nilfs_mdt_init(struct inode *inode, gfp_t gfp_mask, size_t objsz);
+void nilfs_mdt_clear(struct inode *inode);
+void nilfs_mdt_destroy(struct inode *inode);
+
+void nilfs_mdt_set_entry_size(struct inode *, unsigned int, unsigned int);
+
+int nilfs_mdt_setup_shadow_map(struct inode *inode,
+			       struct nilfs_shadow_map *shadow);
+int nilfs_mdt_save_to_shadow_map(struct inode *inode);
+void nilfs_mdt_restore_from_shadow_map(struct inode *inode);
+void nilfs_mdt_clear_shadow_map(struct inode *inode);
+int nilfs_mdt_freeze_buffer(struct inode *inode, struct buffer_head *bh);
+struct buffer_head *nilfs_mdt_get_frozen_buffer(struct inode *inode,
+						struct buffer_head *bh);
+
+static inline void nilfs_mdt_mark_dirty(struct inode *inode)
+{
+	if (!test_bit(NILFS_I_DIRTY, &NILFS_I(inode)->i_state))
+		set_bit(NILFS_I_DIRTY, &NILFS_I(inode)->i_state);
+}
+
+static inline void nilfs_mdt_clear_dirty(struct inode *inode)
+{
+	clear_bit(NILFS_I_DIRTY, &NILFS_I(inode)->i_state);
+}
+
+static inline __u64 nilfs_mdt_cno(struct inode *inode)
+{
+	return ((struct the_nilfs *)inode->i_sb->s_fs_info)->ns_cno;
+}
+
+static inline spinlock_t *
+nilfs_mdt_bgl_lock(struct inode *inode, unsigned int block_group)
+{
+	return bgl_lock_ptr(NILFS_MDT(inode)->mi_bgl, block_group);
+}
+
+#endif /* _NILFS_MDT_H */

linux-7.2.3/fs/nilfs2/namei.c

@@ -0,0 +1,594 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS pathname lookup operations.
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Modified for NILFS by Amagai Yoshiji and Ryusuke Konishi.
+ */
+/*
+ *  linux/fs/ext2/namei.c
+ *
+ * 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/namei.c
+ *
+ *  Copyright (C) 1991, 1992  Linus Torvalds
+ *
+ *  Big-endian to little-endian byte-swapping/bitmaps by
+ *        David S. Miller (davem@caip.rutgers.edu), 1995
+ */
+
+#include <linux/pagemap.h>
+#include "nilfs.h"
+#include "export.h"
+
+#define NILFS_FID_SIZE_NON_CONNECTABLE \
+	(offsetof(struct nilfs_fid, parent_gen) / 4)
+#define NILFS_FID_SIZE_CONNECTABLE	(sizeof(struct nilfs_fid) / 4)
+
+static inline int nilfs_add_nondir(struct dentry *dentry, struct inode *inode)
+{
+	int err = nilfs_add_link(dentry, inode);
+
+	if (!err) {
+		d_instantiate_new(dentry, inode);
+		return 0;
+	}
+	inode_dec_link_count(inode);
+	unlock_new_inode(inode);
+	iput(inode);
+	return err;
+}
+
+/*
+ * Methods themselves.
+ */
+
+static struct dentry *
+nilfs_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
+{
+	struct inode *inode;
+	ino_t ino;
+	int res;
+
+	if (dentry->d_name.len > NILFS_NAME_LEN)
+		return ERR_PTR(-ENAMETOOLONG);
+
+	res = nilfs_inode_by_name(dir, &dentry->d_name, &ino);
+	if (res) {
+		if (res != -ENOENT)
+			return ERR_PTR(res);
+		inode = NULL;
+	} else {
+		inode = nilfs_iget(dir->i_sb, NILFS_I(dir)->i_root, ino);
+		if (inode == ERR_PTR(-ESTALE)) {
+			nilfs_error(dir->i_sb,
+					"deleted inode referenced: %lu", ino);
+			return ERR_PTR(-EIO);
+		}
+	}
+
+	return d_splice_alias(inode, dentry);
+}
+
+/*
+ * By the time this is called, we already have created
+ * the directory cache entry for the new file, but it
+ * is so far negative - it has no inode.
+ *
+ * If the create succeeds, we fill in the inode information
+ * with d_instantiate().
+ */
+static int nilfs_create(struct mnt_idmap *idmap, struct inode *dir,
+			struct dentry *dentry, umode_t mode, bool excl)
+{
+	struct inode *inode;
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return err;
+	inode = nilfs_new_inode(dir, mode);
+	err = PTR_ERR(inode);
+	if (!IS_ERR(inode)) {
+		inode->i_op = &nilfs_file_inode_operations;
+		inode->i_fop = &nilfs_file_operations;
+		inode->i_mapping->a_ops = &nilfs_aops;
+		nilfs_mark_inode_dirty(inode);
+		err = nilfs_add_nondir(dentry, inode);
+	}
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+}
+
+static int
+nilfs_mknod(struct mnt_idmap *idmap, struct inode *dir,
+	    struct dentry *dentry, umode_t mode, dev_t rdev)
+{
+	struct inode *inode;
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return err;
+	inode = nilfs_new_inode(dir, mode);
+	err = PTR_ERR(inode);
+	if (!IS_ERR(inode)) {
+		init_special_inode(inode, inode->i_mode, rdev);
+		nilfs_mark_inode_dirty(inode);
+		err = nilfs_add_nondir(dentry, inode);
+	}
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+}
+
+static int nilfs_symlink(struct mnt_idmap *idmap, struct inode *dir,
+			 struct dentry *dentry, const char *symname)
+{
+	struct nilfs_transaction_info ti;
+	struct super_block *sb = dir->i_sb;
+	unsigned int l = strlen(symname) + 1;
+	struct inode *inode;
+	int err;
+
+	if (l > sb->s_blocksize)
+		return -ENAMETOOLONG;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return err;
+
+	inode = nilfs_new_inode(dir, S_IFLNK | 0777);
+	err = PTR_ERR(inode);
+	if (IS_ERR(inode))
+		goto out;
+
+	/* slow symlink */
+	inode->i_op = &nilfs_symlink_inode_operations;
+	inode_nohighmem(inode);
+	mapping_set_gfp_mask(inode->i_mapping,
+			     mapping_gfp_constraint(inode->i_mapping,
+						    ~__GFP_FS));
+	inode->i_mapping->a_ops = &nilfs_aops;
+	err = page_symlink(inode, symname, l);
+	if (err)
+		goto out_fail;
+
+	/* mark_inode_dirty(inode); */
+	/* page_symlink() do this */
+
+	err = nilfs_add_nondir(dentry, inode);
+out:
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+
+out_fail:
+	drop_nlink(inode);
+	nilfs_mark_inode_dirty(inode);
+	unlock_new_inode(inode);
+	iput(inode);
+	goto out;
+}
+
+static int nilfs_link(struct dentry *old_dentry, struct inode *dir,
+		      struct dentry *dentry)
+{
+	struct inode *inode = d_inode(old_dentry);
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return err;
+
+	inode_set_ctime_current(inode);
+	inode_inc_link_count(inode);
+	ihold(inode);
+
+	err = nilfs_add_link(dentry, inode);
+	if (!err) {
+		d_instantiate(dentry, inode);
+		err = nilfs_transaction_commit(dir->i_sb);
+	} else {
+		inode_dec_link_count(inode);
+		iput(inode);
+		nilfs_transaction_abort(dir->i_sb);
+	}
+
+	return err;
+}
+
+static struct dentry *nilfs_mkdir(struct mnt_idmap *idmap, struct inode *dir,
+				  struct dentry *dentry, umode_t mode)
+{
+	struct inode *inode;
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return ERR_PTR(err);
+
+	inc_nlink(dir);
+
+	inode = nilfs_new_inode(dir, S_IFDIR | mode);
+	err = PTR_ERR(inode);
+	if (IS_ERR(inode))
+		goto out_dir;
+
+	inode->i_op = &nilfs_dir_inode_operations;
+	inode->i_fop = &nilfs_dir_operations;
+	inode->i_mapping->a_ops = &nilfs_aops;
+
+	inc_nlink(inode);
+
+	err = nilfs_make_empty(inode, dir);
+	if (err)
+		goto out_fail;
+
+	err = nilfs_add_link(dentry, inode);
+	if (err)
+		goto out_fail;
+
+	nilfs_mark_inode_dirty(inode);
+	d_instantiate_new(dentry, inode);
+out:
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err ? ERR_PTR(err) : NULL;
+
+out_fail:
+	drop_nlink(inode);
+	drop_nlink(inode);
+	nilfs_mark_inode_dirty(inode);
+	unlock_new_inode(inode);
+	iput(inode);
+out_dir:
+	drop_nlink(dir);
+	nilfs_mark_inode_dirty(dir);
+	goto out;
+}
+
+static int nilfs_do_unlink(struct inode *dir, struct dentry *dentry)
+{
+	struct inode *inode;
+	struct nilfs_dir_entry *de;
+	struct folio *folio;
+	int err;
+
+	de = nilfs_find_entry(dir, &dentry->d_name, &folio);
+	if (IS_ERR(de)) {
+		err = PTR_ERR(de);
+		goto out;
+	}
+
+	inode = d_inode(dentry);
+	err = -EIO;
+	if (le64_to_cpu(de->inode) != inode->i_ino)
+		goto out;
+
+	if (!inode->i_nlink) {
+		nilfs_warn(inode->i_sb,
+			   "deleting nonexistent file (ino=%llu), %d",
+			   inode->i_ino, inode->i_nlink);
+		set_nlink(inode, 1);
+	}
+	err = nilfs_delete_entry(de, folio);
+	folio_release_kmap(folio, de);
+	if (err)
+		goto out;
+
+	inode_set_ctime_to_ts(inode, inode_get_ctime(dir));
+	drop_nlink(inode);
+	err = 0;
+out:
+	return err;
+}
+
+static int nilfs_unlink(struct inode *dir, struct dentry *dentry)
+{
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 0);
+	if (err)
+		return err;
+
+	err = nilfs_do_unlink(dir, dentry);
+
+	if (!err) {
+		nilfs_mark_inode_dirty(dir);
+		nilfs_mark_inode_dirty(d_inode(dentry));
+		err = nilfs_transaction_commit(dir->i_sb);
+	} else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+}
+
+static int nilfs_rmdir(struct inode *dir, struct dentry *dentry)
+{
+	struct inode *inode = d_inode(dentry);
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 0);
+	if (err)
+		return err;
+
+	err = -ENOTEMPTY;
+	if (nilfs_empty_dir(inode)) {
+		err = nilfs_do_unlink(dir, dentry);
+		if (!err) {
+			inode->i_size = 0;
+			drop_nlink(inode);
+			nilfs_mark_inode_dirty(inode);
+			drop_nlink(dir);
+			nilfs_mark_inode_dirty(dir);
+		}
+	}
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+}
+
+static int nilfs_rename(struct mnt_idmap *idmap,
+			struct inode *old_dir, struct dentry *old_dentry,
+			struct inode *new_dir, struct dentry *new_dentry,
+			unsigned int flags)
+{
+	struct inode *old_inode = d_inode(old_dentry);
+	struct inode *new_inode = d_inode(new_dentry);
+	struct folio *dir_folio = NULL;
+	struct nilfs_dir_entry *dir_de = NULL;
+	struct folio *old_folio;
+	struct nilfs_dir_entry *old_de;
+	struct nilfs_transaction_info ti;
+	bool old_is_dir = S_ISDIR(old_inode->i_mode);
+	int err;
+
+	if (flags & ~RENAME_NOREPLACE)
+		return -EINVAL;
+
+	err = nilfs_transaction_begin(old_dir->i_sb, &ti, 1);
+	if (unlikely(err))
+		return err;
+
+	old_de = nilfs_find_entry(old_dir, &old_dentry->d_name, &old_folio);
+	if (IS_ERR(old_de)) {
+		err = PTR_ERR(old_de);
+		goto out;
+	}
+
+	if (old_is_dir && old_dir != new_dir) {
+		err = -EIO;
+		dir_de = nilfs_dotdot(old_inode, &dir_folio);
+		if (!dir_de)
+			goto out_old;
+	}
+
+	if (new_inode) {
+		struct folio *new_folio;
+		struct nilfs_dir_entry *new_de;
+
+		err = -ENOTEMPTY;
+		if (old_is_dir && !nilfs_empty_dir(new_inode))
+			goto out_dir;
+
+		new_de = nilfs_find_entry(new_dir, &new_dentry->d_name,
+					  &new_folio);
+		if (IS_ERR(new_de)) {
+			err = PTR_ERR(new_de);
+			goto out_dir;
+		}
+		err = nilfs_set_link(new_dir, new_de, new_folio, old_inode);
+		folio_release_kmap(new_folio, new_de);
+		if (unlikely(err))
+			goto out_dir;
+		nilfs_mark_inode_dirty(new_dir);
+		inode_set_ctime_current(new_inode);
+		if (old_is_dir)
+			drop_nlink(new_inode);
+		drop_nlink(new_inode);
+		nilfs_mark_inode_dirty(new_inode);
+	} else {
+		err = nilfs_add_link(new_dentry, old_inode);
+		if (err)
+			goto out_dir;
+		if (old_is_dir) {
+			inc_nlink(new_dir);
+			nilfs_mark_inode_dirty(new_dir);
+		}
+	}
+
+	/*
+	 * Like most other Unix systems, set the ctime for inodes on a
+	 * rename.
+	 */
+	inode_set_ctime_current(old_inode);
+
+	err = nilfs_delete_entry(old_de, old_folio);
+	if (likely(!err)) {
+		if (old_is_dir) {
+			if (old_dir != new_dir)
+				err = nilfs_set_link(old_inode, dir_de,
+						     dir_folio, new_dir);
+			drop_nlink(old_dir);
+		}
+		nilfs_mark_inode_dirty(old_dir);
+	}
+	nilfs_mark_inode_dirty(old_inode);
+
+out_dir:
+	if (dir_de)
+		folio_release_kmap(dir_folio, dir_de);
+out_old:
+	folio_release_kmap(old_folio, old_de);
+out:
+	if (likely(!err))
+		err = nilfs_transaction_commit(old_dir->i_sb);
+	else
+		nilfs_transaction_abort(old_dir->i_sb);
+	return err;
+}
+
+/*
+ * Export operations
+ */
+static struct dentry *nilfs_get_parent(struct dentry *child)
+{
+	ino_t ino;
+	int res;
+	struct nilfs_root *root;
+
+	res = nilfs_inode_by_name(d_inode(child), &dotdot_name, &ino);
+	if (res)
+		return ERR_PTR(res);
+
+	root = NILFS_I(d_inode(child))->i_root;
+
+	return d_obtain_alias(nilfs_iget(child->d_sb, root, ino));
+}
+
+static struct dentry *nilfs_get_dentry(struct super_block *sb, u64 cno,
+				       u64 ino, u32 gen)
+{
+	struct nilfs_root *root;
+	struct inode *inode;
+
+	if (ino < NILFS_FIRST_INO(sb) && ino != NILFS_ROOT_INO)
+		return ERR_PTR(-ESTALE);
+
+	root = nilfs_lookup_root(sb->s_fs_info, cno);
+	if (!root)
+		return ERR_PTR(-ESTALE);
+
+	inode = nilfs_iget(sb, root, ino);
+	nilfs_put_root(root);
+
+	if (IS_ERR(inode))
+		return ERR_CAST(inode);
+	if (gen && inode->i_generation != gen) {
+		iput(inode);
+		return ERR_PTR(-ESTALE);
+	}
+	return d_obtain_alias(inode);
+}
+
+static struct dentry *nilfs_fh_to_dentry(struct super_block *sb, struct fid *fh,
+					 int fh_len, int fh_type)
+{
+	struct nilfs_fid *fid = (struct nilfs_fid *)fh;
+
+	if (fh_len < NILFS_FID_SIZE_NON_CONNECTABLE ||
+	    (fh_type != FILEID_NILFS_WITH_PARENT &&
+	     fh_type != FILEID_NILFS_WITHOUT_PARENT))
+		return NULL;
+
+	return nilfs_get_dentry(sb, fid->cno, fid->ino, fid->gen);
+}
+
+static struct dentry *nilfs_fh_to_parent(struct super_block *sb, struct fid *fh,
+					 int fh_len, int fh_type)
+{
+	struct nilfs_fid *fid = (struct nilfs_fid *)fh;
+
+	if (fh_len < NILFS_FID_SIZE_CONNECTABLE ||
+	    fh_type != FILEID_NILFS_WITH_PARENT)
+		return NULL;
+
+	return nilfs_get_dentry(sb, fid->cno, fid->parent_ino, fid->parent_gen);
+}
+
+static int nilfs_encode_fh(struct inode *inode, __u32 *fh, int *lenp,
+			   struct inode *parent)
+{
+	struct nilfs_fid *fid = (struct nilfs_fid *)fh;
+	struct nilfs_root *root = NILFS_I(inode)->i_root;
+	int type;
+
+	if (parent && *lenp < NILFS_FID_SIZE_CONNECTABLE) {
+		*lenp = NILFS_FID_SIZE_CONNECTABLE;
+		return FILEID_INVALID;
+	}
+	if (*lenp < NILFS_FID_SIZE_NON_CONNECTABLE) {
+		*lenp = NILFS_FID_SIZE_NON_CONNECTABLE;
+		return FILEID_INVALID;
+	}
+
+	fid->cno = root->cno;
+	fid->ino = inode->i_ino;
+	fid->gen = inode->i_generation;
+
+	if (parent) {
+		fid->parent_ino = parent->i_ino;
+		fid->parent_gen = parent->i_generation;
+		type = FILEID_NILFS_WITH_PARENT;
+		*lenp = NILFS_FID_SIZE_CONNECTABLE;
+	} else {
+		type = FILEID_NILFS_WITHOUT_PARENT;
+		*lenp = NILFS_FID_SIZE_NON_CONNECTABLE;
+	}
+
+	return type;
+}
+
+const struct inode_operations nilfs_dir_inode_operations = {
+	.create		= nilfs_create,
+	.lookup		= nilfs_lookup,
+	.link		= nilfs_link,
+	.unlink		= nilfs_unlink,
+	.symlink	= nilfs_symlink,
+	.mkdir		= nilfs_mkdir,
+	.rmdir		= nilfs_rmdir,
+	.mknod		= nilfs_mknod,
+	.rename		= nilfs_rename,
+	.setattr	= nilfs_setattr,
+	.permission	= nilfs_permission,
+	.fiemap		= nilfs_fiemap,
+	.fileattr_get	= nilfs_fileattr_get,
+	.fileattr_set	= nilfs_fileattr_set,
+};
+
+const struct inode_operations nilfs_special_inode_operations = {
+	.setattr	= nilfs_setattr,
+	.permission	= nilfs_permission,
+};
+
+const struct inode_operations nilfs_symlink_inode_operations = {
+	.get_link	= page_get_link,
+	.permission     = nilfs_permission,
+};
+
+const struct export_operations nilfs_export_ops = {
+	.encode_fh = nilfs_encode_fh,
+	.fh_to_dentry = nilfs_fh_to_dentry,
+	.fh_to_parent = nilfs_fh_to_parent,
+	.get_parent = nilfs_get_parent,
+};

linux-7.2.3/fs/nilfs2/nilfs.h

@@ -0,0 +1,416 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS local header file.
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato and Ryusuke Konishi.
+ */
+
+#ifndef _NILFS_H
+#define _NILFS_H
+
+#include <linux/kernel.h>
+#include <linux/buffer_head.h>
+#include <linux/spinlock.h>
+#include <linux/blkdev.h>
+#include <linux/fs_struct.h>
+#include <linux/nilfs2_api.h>
+#include <linux/nilfs2_ondisk.h>
+#include "the_nilfs.h"
+#include "bmap.h"
+
+/**
+ * struct nilfs_inode_info - nilfs inode data in memory
+ * @i_flags: inode flags
+ * @i_type: inode type (combination of flags that inidicate usage)
+ * @i_state: dynamic state flags
+ * @i_bmap: pointer on i_bmap_data
+ * @i_bmap_data: raw block mapping
+ * @i_xattr: <TODO>
+ * @i_dir_start_lookup: page index of last successful search
+ * @i_cno: checkpoint number for GC inode
+ * @i_assoc_inode: associated inode (B-tree node cache holder or back pointer)
+ * @i_dirty: list for connecting dirty files
+ * @xattr_sem: semaphore for extended attributes processing
+ * @i_bh: buffer contains disk inode
+ * @i_root: root object of the current filesystem tree
+ * @vfs_inode: VFS inode object
+ */
+struct nilfs_inode_info {
+	__u32 i_flags;
+	unsigned int i_type;
+	unsigned long  i_state;		/* Dynamic state flags */
+	struct nilfs_bmap *i_bmap;
+	struct nilfs_bmap i_bmap_data;
+	__u64 i_xattr;	/* sector_t ??? */
+	__u32 i_dir_start_lookup;
+	__u64 i_cno;		/* check point number for GC inode */
+	struct inode *i_assoc_inode;
+	struct list_head i_dirty;	/* List for connecting dirty files */
+
+#ifdef CONFIG_NILFS_XATTR
+	/*
+	 * Extended attributes can be read independently of the main file
+	 * data. Taking i_sem even when reading would cause contention
+	 * between readers of EAs and writers of regular file data, so
+	 * instead we synchronize on xattr_sem when reading or changing
+	 * EAs.
+	 */
+	struct rw_semaphore xattr_sem;
+#endif
+	struct buffer_head *i_bh;	/*
+					 * i_bh contains a new or dirty
+					 * disk inode.
+					 */
+	struct nilfs_root *i_root;
+	struct inode vfs_inode;
+};
+
+static inline struct nilfs_inode_info *NILFS_I(const struct inode *inode)
+{
+	return container_of(inode, struct nilfs_inode_info, vfs_inode);
+}
+
+static inline struct nilfs_inode_info *
+NILFS_BMAP_I(const struct nilfs_bmap *bmap)
+{
+	return container_of(bmap, struct nilfs_inode_info, i_bmap_data);
+}
+
+/*
+ * Dynamic state flags of NILFS on-memory inode (i_state)
+ */
+enum {
+	NILFS_I_NEW = 0,		/* Inode is newly created */
+	NILFS_I_DIRTY,			/* The file is dirty */
+	NILFS_I_QUEUED,			/* inode is in dirty_files list */
+	NILFS_I_BUSY,			/*
+					 * Inode is grabbed by a segment
+					 * constructor
+					 */
+	NILFS_I_COLLECTED,		/* All dirty blocks are collected */
+	NILFS_I_UPDATED,		/* The file has been written back */
+	NILFS_I_INODE_SYNC,		/* dsync is not allowed for inode */
+	NILFS_I_BMAP,			/* has bmap and btnode_cache */
+};
+
+/*
+ * Flags to identify the usage of on-memory inodes (i_type)
+ */
+enum {
+	NILFS_I_TYPE_NORMAL =	0,
+	NILFS_I_TYPE_GC =	0x0001,	/* For data caching during GC */
+	NILFS_I_TYPE_BTNC =	0x0002,	/* For btree node cache */
+	NILFS_I_TYPE_SHADOW =	0x0004,	/* For shadowed page cache */
+};
+
+/*
+ * commit flags for nilfs_commit_super and nilfs_sync_super
+ */
+enum {
+	NILFS_SB_COMMIT = 0,	/* Commit a super block alternately */
+	NILFS_SB_COMMIT_ALL	/* Commit both super blocks */
+};
+
+/**
+ * define NILFS_MAX_VOLUME_NAME - maximum number of characters (bytes) in a
+ *                                file system volume name
+ *
+ * Defined by the size of the volume name field in the on-disk superblocks.
+ * This volume name does not include the terminating NULL byte if the string
+ * length matches the field size, so use (NILFS_MAX_VOLUME_NAME + 1) for the
+ * size of the buffer that requires a NULL byte termination.
+ */
+#define NILFS_MAX_VOLUME_NAME  \
+	sizeof_field(struct nilfs_super_block, s_volume_name)
+
+/*
+ * Macros to check inode numbers
+ */
+#define NILFS_MDT_INO_BITS						\
+	(BIT(NILFS_DAT_INO) | BIT(NILFS_CPFILE_INO) |			\
+	 BIT(NILFS_SUFILE_INO) | BIT(NILFS_IFILE_INO) |			\
+	 BIT(NILFS_ATIME_INO) | BIT(NILFS_SKETCH_INO))
+
+#define NILFS_SYS_INO_BITS (BIT(NILFS_ROOT_INO) | NILFS_MDT_INO_BITS)
+
+#define NILFS_FIRST_INO(sb) (((struct the_nilfs *)sb->s_fs_info)->ns_first_ino)
+
+#define NILFS_MDT_INODE(sb, ino) \
+	((ino) < NILFS_USER_INO && (NILFS_MDT_INO_BITS & BIT(ino)))
+#define NILFS_VALID_INODE(sb, ino) \
+	((ino) >= NILFS_FIRST_INO(sb) ||				\
+	 ((ino) < NILFS_USER_INO && (NILFS_SYS_INO_BITS & BIT(ino))))
+
+#define NILFS_PRIVATE_INODE(ino) ({					\
+	ino_t __ino = (ino);						\
+	((__ino) < NILFS_USER_INO && (__ino) != NILFS_ROOT_INO &&	\
+	 (__ino) != NILFS_SKETCH_INO); })
+
+/**
+ * struct nilfs_transaction_info: context information for synchronization
+ * @ti_magic: Magic number
+ * @ti_save: Backup of journal_info field of task_struct
+ * @ti_flags: Flags
+ * @ti_count: Nest level
+ */
+struct nilfs_transaction_info {
+	u32			ti_magic;
+	void		       *ti_save;
+				/*
+				 * This should never be used.  If it happens,
+				 * one of other filesystems has a bug.
+				 */
+	unsigned short		ti_flags;
+	unsigned short		ti_count;
+};
+
+/* ti_magic */
+#define NILFS_TI_MAGIC		0xd9e392fb
+
+/* ti_flags */
+#define NILFS_TI_DYNAMIC_ALLOC	0x0001  /* Allocated from slab */
+#define NILFS_TI_SYNC		0x0002	/*
+					 * Force to construct segment at the
+					 * end of transaction.
+					 */
+#define NILFS_TI_GC		0x0004	/* GC context */
+#define NILFS_TI_COMMIT		0x0008	/* Change happened or not */
+#define NILFS_TI_WRITER		0x0010	/* Constructor context */
+
+
+int nilfs_transaction_begin(struct super_block *,
+			    struct nilfs_transaction_info *, int);
+int nilfs_transaction_commit(struct super_block *);
+void nilfs_transaction_abort(struct super_block *);
+
+static inline void nilfs_set_transaction_flag(unsigned int flag)
+{
+	struct nilfs_transaction_info *ti = current->journal_info;
+
+	ti->ti_flags |= flag;
+}
+
+static inline int nilfs_test_transaction_flag(unsigned int flag)
+{
+	struct nilfs_transaction_info *ti = current->journal_info;
+
+	if (ti == NULL || ti->ti_magic != NILFS_TI_MAGIC)
+		return 0;
+	return !!(ti->ti_flags & flag);
+}
+
+static inline int nilfs_doing_gc(void)
+{
+	return nilfs_test_transaction_flag(NILFS_TI_GC);
+}
+
+static inline int nilfs_doing_construction(void)
+{
+	return nilfs_test_transaction_flag(NILFS_TI_WRITER);
+}
+
+/*
+ * function prototype
+ */
+#ifdef CONFIG_NILFS_POSIX_ACL
+#error "NILFS: not yet supported POSIX ACL"
+extern int nilfs_acl_chmod(struct inode *);
+extern int nilfs_init_acl(struct inode *, struct inode *);
+#else
+static inline int nilfs_acl_chmod(struct inode *inode)
+{
+	return 0;
+}
+
+static inline int nilfs_init_acl(struct inode *inode, struct inode *dir)
+{
+	if (S_ISLNK(inode->i_mode))
+		return 0;
+
+	inode->i_mode &= ~current_umask();
+	return 0;
+}
+#endif
+
+#define NILFS_ATIME_DISABLE
+
+/* Flags that should be inherited by new inodes from their parent. */
+#define NILFS_FL_INHERITED						\
+	(FS_SECRM_FL | FS_UNRM_FL | FS_COMPR_FL | FS_SYNC_FL |		\
+	 FS_IMMUTABLE_FL | FS_APPEND_FL | FS_NODUMP_FL | FS_NOATIME_FL |\
+	 FS_COMPRBLK_FL | FS_NOCOMP_FL | FS_NOTAIL_FL | FS_DIRSYNC_FL)
+
+/* Mask out flags that are inappropriate for the given type of inode. */
+static inline __u32 nilfs_mask_flags(umode_t mode, __u32 flags)
+{
+	if (S_ISDIR(mode))
+		return flags;
+	else if (S_ISREG(mode))
+		return flags & ~(FS_DIRSYNC_FL | FS_TOPDIR_FL);
+	else
+		return flags & (FS_NODUMP_FL | FS_NOATIME_FL);
+}
+
+/* dir.c */
+int nilfs_add_link(struct dentry *, struct inode *);
+int nilfs_inode_by_name(struct inode *dir, const struct qstr *qstr, ino_t *ino);
+int nilfs_make_empty(struct inode *, struct inode *);
+struct nilfs_dir_entry *nilfs_find_entry(struct inode *, const struct qstr *,
+		struct folio **);
+int nilfs_delete_entry(struct nilfs_dir_entry *, struct folio *);
+int nilfs_empty_dir(struct inode *);
+struct nilfs_dir_entry *nilfs_dotdot(struct inode *, struct folio **);
+int nilfs_set_link(struct inode *dir, struct nilfs_dir_entry *de,
+		   struct folio *folio, struct inode *inode);
+
+/* file.c */
+extern int nilfs_sync_file(struct file *, loff_t, loff_t, int);
+
+/* ioctl.c */
+int nilfs_fileattr_get(struct dentry *dentry, struct file_kattr *m);
+int nilfs_fileattr_set(struct mnt_idmap *idmap,
+		       struct dentry *dentry, struct file_kattr *fa);
+long nilfs_ioctl(struct file *, unsigned int, unsigned long);
+long nilfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
+int nilfs_ioctl_prepare_clean_segments(struct the_nilfs *, struct nilfs_argv *,
+				       void **);
+
+/* inode.c */
+void nilfs_inode_add_blocks(struct inode *inode, int n);
+void nilfs_inode_sub_blocks(struct inode *inode, int n);
+extern struct inode *nilfs_new_inode(struct inode *, umode_t);
+extern int nilfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
+extern void nilfs_set_inode_flags(struct inode *);
+extern int nilfs_read_inode_common(struct inode *, struct nilfs_inode *);
+void nilfs_write_inode_common(struct inode *inode,
+			      struct nilfs_inode *raw_inode);
+struct inode *nilfs_ilookup(struct super_block *sb, struct nilfs_root *root,
+			    unsigned long ino);
+struct inode *nilfs_iget_locked(struct super_block *sb, struct nilfs_root *root,
+				unsigned long ino);
+struct inode *nilfs_iget(struct super_block *sb, struct nilfs_root *root,
+			 unsigned long ino);
+extern struct inode *nilfs_iget_for_gc(struct super_block *sb,
+				       unsigned long ino, __u64 cno);
+int nilfs_attach_btree_node_cache(struct inode *inode);
+void nilfs_detach_btree_node_cache(struct inode *inode);
+struct inode *nilfs_iget_for_shadow(struct inode *inode);
+extern void nilfs_update_inode(struct inode *, struct buffer_head *, int);
+extern void nilfs_truncate(struct inode *);
+extern void nilfs_evict_inode(struct inode *);
+extern int nilfs_setattr(struct mnt_idmap *, struct dentry *,
+			 struct iattr *);
+extern void nilfs_write_failed(struct address_space *mapping, loff_t to);
+int nilfs_permission(struct mnt_idmap *idmap, struct inode *inode,
+		     int mask);
+int nilfs_load_inode_block(struct inode *inode, struct buffer_head **pbh);
+extern int nilfs_inode_dirty(struct inode *);
+int nilfs_set_file_dirty(struct inode *inode, unsigned int nr_dirty);
+extern int __nilfs_mark_inode_dirty(struct inode *, int);
+extern void nilfs_dirty_inode(struct inode *, int flags);
+int nilfs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
+		 __u64 start, __u64 len);
+static inline int nilfs_mark_inode_dirty(struct inode *inode)
+{
+	return __nilfs_mark_inode_dirty(inode, I_DIRTY);
+}
+static inline int nilfs_mark_inode_dirty_sync(struct inode *inode)
+{
+	return __nilfs_mark_inode_dirty(inode, I_DIRTY_SYNC);
+}
+
+/* super.c */
+extern struct inode *nilfs_alloc_inode(struct super_block *);
+
+__printf(2, 3)
+void __nilfs_msg(struct super_block *sb, const char *fmt, ...);
+extern __printf(3, 4)
+void __nilfs_error(struct super_block *sb, const char *function,
+		   const char *fmt, ...);
+
+#ifdef CONFIG_PRINTK
+
+#define nilfs_msg(sb, level, fmt, ...)					\
+	__nilfs_msg(sb, level fmt, ##__VA_ARGS__)
+#define nilfs_error(sb, fmt, ...)					\
+	__nilfs_error(sb, __func__, fmt, ##__VA_ARGS__)
+
+#else
+
+#define nilfs_msg(sb, level, fmt, ...)					\
+	do {								\
+		no_printk(level fmt, ##__VA_ARGS__);			\
+		(void)(sb);						\
+	} while (0)
+#define nilfs_error(sb, fmt, ...)					\
+	do {								\
+		no_printk(fmt, ##__VA_ARGS__);				\
+		__nilfs_error(sb, "", " ");				\
+	} while (0)
+
+#endif /* CONFIG_PRINTK */
+
+#define nilfs_crit(sb, fmt, ...)					\
+	nilfs_msg(sb, KERN_CRIT, fmt, ##__VA_ARGS__)
+#define nilfs_err(sb, fmt, ...)						\
+	nilfs_msg(sb, KERN_ERR, fmt, ##__VA_ARGS__)
+#define nilfs_warn(sb, fmt, ...)					\
+	nilfs_msg(sb, KERN_WARNING, fmt, ##__VA_ARGS__)
+#define nilfs_info(sb, fmt, ...)					\
+	nilfs_msg(sb, KERN_INFO, fmt, ##__VA_ARGS__)
+
+extern struct nilfs_super_block *
+nilfs_read_super_block(struct super_block *, u64, int, struct buffer_head **);
+extern int nilfs_store_magic(struct super_block *sb,
+			     struct nilfs_super_block *sbp);
+extern int nilfs_check_feature_compatibility(struct super_block *,
+					     struct nilfs_super_block *);
+extern void nilfs_set_log_cursor(struct nilfs_super_block *,
+				 struct the_nilfs *);
+struct nilfs_super_block **nilfs_prepare_super(struct super_block *sb,
+					       int flip);
+int nilfs_commit_super(struct super_block *sb, int flag);
+int nilfs_cleanup_super(struct super_block *sb);
+int nilfs_resize_fs(struct super_block *sb, __u64 newsize);
+int nilfs_attach_checkpoint(struct super_block *sb, __u64 cno, int curr_mnt,
+			    struct nilfs_root **root);
+int nilfs_checkpoint_is_mounted(struct super_block *sb, __u64 cno);
+
+/* gcinode.c */
+int nilfs_gccache_submit_read_data(struct inode *, sector_t, sector_t, __u64,
+				   struct buffer_head **);
+int nilfs_gccache_submit_read_node(struct inode *, sector_t, __u64,
+				   struct buffer_head **);
+int nilfs_gccache_wait_and_mark_dirty(struct buffer_head *);
+int nilfs_init_gcinode(struct inode *inode);
+void nilfs_remove_all_gcinodes(struct the_nilfs *nilfs);
+
+/* sysfs.c */
+int __init nilfs_sysfs_init(void);
+void nilfs_sysfs_exit(void);
+int nilfs_sysfs_create_device_group(struct super_block *);
+void nilfs_sysfs_delete_device_group(struct the_nilfs *);
+int nilfs_sysfs_create_snapshot_group(struct nilfs_root *);
+void nilfs_sysfs_delete_snapshot_group(struct nilfs_root *);
+
+/*
+ * Inodes and files operations
+ */
+extern const struct file_operations nilfs_dir_operations;
+extern const struct inode_operations nilfs_file_inode_operations;
+extern const struct file_operations nilfs_file_operations;
+extern const struct address_space_operations nilfs_aops;
+extern const struct address_space_operations nilfs_buffer_cache_aops;
+extern const struct inode_operations nilfs_dir_inode_operations;
+extern const struct inode_operations nilfs_special_inode_operations;
+extern const struct inode_operations nilfs_symlink_inode_operations;
+
+/*
+ * filesystem type
+ */
+extern struct file_system_type nilfs_fs_type;
+
+
+#endif	/* _NILFS_H */

linux-7.2.3/fs/nilfs2/page.c

@@ -0,0 +1,567 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Buffer/page management specific to NILFS
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi and Seiji Kihara.
+ */
+
+#include <linux/pagemap.h>
+#include <linux/writeback.h>
+#include <linux/swap.h>
+#include <linux/bitops.h>
+#include <linux/page-flags.h>
+#include <linux/list.h>
+#include <linux/highmem.h>
+#include <linux/folio_batch.h>
+#include <linux/gfp.h>
+#include "nilfs.h"
+#include "page.h"
+#include "mdt.h"
+
+
+#define NILFS_BUFFER_INHERENT_BITS					\
+	(BIT(BH_Uptodate) | BIT(BH_Mapped) | BIT(BH_NILFS_Node) |	\
+	 BIT(BH_NILFS_Volatile) | BIT(BH_NILFS_Checked))
+
+static struct buffer_head *__nilfs_get_folio_block(struct folio *folio,
+		unsigned long block, pgoff_t index, int blkbits,
+		unsigned long b_state)
+
+{
+	unsigned long first_block;
+	struct buffer_head *bh = folio_buffers(folio);
+
+	if (!bh)
+		bh = create_empty_buffers(folio, 1 << blkbits, b_state);
+
+	first_block = (unsigned long)index << (PAGE_SHIFT - blkbits);
+	bh = get_nth_bh(bh, block - first_block);
+
+	wait_on_buffer(bh);
+	return bh;
+}
+
+struct buffer_head *nilfs_grab_buffer(struct inode *inode,
+				      struct address_space *mapping,
+				      unsigned long blkoff,
+				      unsigned long b_state)
+{
+	int blkbits = inode->i_blkbits;
+	pgoff_t index = blkoff >> (PAGE_SHIFT - blkbits);
+	struct folio *folio;
+	struct buffer_head *bh;
+
+	folio = filemap_grab_folio(mapping, index);
+	if (IS_ERR(folio))
+		return NULL;
+
+	bh = __nilfs_get_folio_block(folio, blkoff, index, blkbits, b_state);
+	if (unlikely(!bh)) {
+		folio_unlock(folio);
+		folio_put(folio);
+		return NULL;
+	}
+	bh->b_bdev = inode->i_sb->s_bdev;
+	return bh;
+}
+
+/**
+ * nilfs_forget_buffer - discard dirty state
+ * @bh: buffer head of the buffer to be discarded
+ */
+void nilfs_forget_buffer(struct buffer_head *bh)
+{
+	struct folio *folio = bh->b_folio;
+	const unsigned long clear_bits =
+		(BIT(BH_Uptodate) | BIT(BH_Dirty) | BIT(BH_Mapped) |
+		 BIT(BH_Async_Write) | BIT(BH_NILFS_Volatile) |
+		 BIT(BH_NILFS_Checked) | BIT(BH_NILFS_Redirected) |
+		 BIT(BH_Delay));
+
+	lock_buffer(bh);
+	set_mask_bits(&bh->b_state, clear_bits, 0);
+	if (nilfs_folio_buffers_clean(folio))
+		__nilfs_clear_folio_dirty(folio);
+
+	bh->b_blocknr = -1;
+	folio_clear_uptodate(folio);
+	folio_clear_mappedtodisk(folio);
+	unlock_buffer(bh);
+	brelse(bh);
+}
+
+/**
+ * nilfs_copy_buffer -- copy buffer data and flags
+ * @dbh: destination buffer
+ * @sbh: source buffer
+ */
+void nilfs_copy_buffer(struct buffer_head *dbh, struct buffer_head *sbh)
+{
+	void *saddr, *daddr;
+	unsigned long bits;
+	struct folio *sfolio = sbh->b_folio, *dfolio = dbh->b_folio;
+	struct buffer_head *bh;
+
+	saddr = kmap_local_folio(sfolio, bh_offset(sbh));
+	daddr = kmap_local_folio(dfolio, bh_offset(dbh));
+	memcpy(daddr, saddr, sbh->b_size);
+	kunmap_local(daddr);
+	kunmap_local(saddr);
+
+	dbh->b_state = sbh->b_state & NILFS_BUFFER_INHERENT_BITS;
+	dbh->b_blocknr = sbh->b_blocknr;
+	dbh->b_bdev = sbh->b_bdev;
+
+	bh = dbh;
+	bits = sbh->b_state & (BIT(BH_Uptodate) | BIT(BH_Mapped));
+	while ((bh = bh->b_this_page) != dbh) {
+		lock_buffer(bh);
+		bits &= bh->b_state;
+		unlock_buffer(bh);
+	}
+	if (bits & BIT(BH_Uptodate))
+		folio_mark_uptodate(dfolio);
+	else
+		folio_clear_uptodate(dfolio);
+	if (bits & BIT(BH_Mapped))
+		folio_set_mappedtodisk(dfolio);
+	else
+		folio_clear_mappedtodisk(dfolio);
+}
+
+/**
+ * nilfs_folio_buffers_clean - Check if a folio has dirty buffers or not.
+ * @folio: Folio to be checked.
+ *
+ * Return: false if the folio has dirty buffers, true otherwise.
+ */
+bool nilfs_folio_buffers_clean(struct folio *folio)
+{
+	struct buffer_head *bh, *head;
+
+	bh = head = folio_buffers(folio);
+	do {
+		if (buffer_dirty(bh))
+			return false;
+		bh = bh->b_this_page;
+	} while (bh != head);
+	return true;
+}
+
+void nilfs_folio_bug(struct folio *folio)
+{
+	struct buffer_head *bh, *head;
+	struct address_space *m;
+	unsigned long ino;
+
+	if (unlikely(!folio)) {
+		printk(KERN_CRIT "NILFS_FOLIO_BUG(NULL)\n");
+		return;
+	}
+
+	m = folio->mapping;
+	ino = m ? m->host->i_ino : 0;
+
+	printk(KERN_CRIT "NILFS_FOLIO_BUG(%p): cnt=%d index#=%llu flags=0x%lx "
+	       "mapping=%p ino=%lu\n",
+	       folio, folio_ref_count(folio),
+	       (unsigned long long)folio->index, folio->flags.f, m, ino);
+
+	head = folio_buffers(folio);
+	if (head) {
+		int i = 0;
+
+		bh = head;
+		do {
+			printk(KERN_CRIT
+			       " BH[%d] %p: cnt=%d block#=%llu state=0x%lx\n",
+			       i++, bh, atomic_read(&bh->b_count),
+			       (unsigned long long)bh->b_blocknr, bh->b_state);
+			bh = bh->b_this_page;
+		} while (bh != head);
+	}
+}
+
+/**
+ * nilfs_copy_folio -- copy the folio with buffers
+ * @dst: destination folio
+ * @src: source folio
+ * @copy_dirty: flag whether to copy dirty states on the folio's buffer heads.
+ *
+ * This function is for both data folios and btnode folios.  The dirty flag
+ * should be treated by caller.  The folio must not be under i/o.
+ * Both src and dst folio must be locked
+ */
+static void nilfs_copy_folio(struct folio *dst, struct folio *src,
+		bool copy_dirty)
+{
+	struct buffer_head *dbh, *dbufs, *sbh;
+	unsigned long mask = NILFS_BUFFER_INHERENT_BITS;
+
+	BUG_ON(folio_test_writeback(dst));
+
+	sbh = folio_buffers(src);
+	dbh = folio_buffers(dst);
+	if (!dbh)
+		dbh = create_empty_buffers(dst, sbh->b_size, 0);
+
+	if (copy_dirty)
+		mask |= BIT(BH_Dirty);
+
+	dbufs = dbh;
+	do {
+		lock_buffer(sbh);
+		lock_buffer(dbh);
+		dbh->b_state = sbh->b_state & mask;
+		dbh->b_blocknr = sbh->b_blocknr;
+		dbh->b_bdev = sbh->b_bdev;
+		sbh = sbh->b_this_page;
+		dbh = dbh->b_this_page;
+	} while (dbh != dbufs);
+
+	folio_copy(dst, src);
+
+	if (folio_test_uptodate(src) && !folio_test_uptodate(dst))
+		folio_mark_uptodate(dst);
+	else if (!folio_test_uptodate(src) && folio_test_uptodate(dst))
+		folio_clear_uptodate(dst);
+	if (folio_test_mappedtodisk(src) && !folio_test_mappedtodisk(dst))
+		folio_set_mappedtodisk(dst);
+	else if (!folio_test_mappedtodisk(src) && folio_test_mappedtodisk(dst))
+		folio_clear_mappedtodisk(dst);
+
+	do {
+		unlock_buffer(sbh);
+		unlock_buffer(dbh);
+		sbh = sbh->b_this_page;
+		dbh = dbh->b_this_page;
+	} while (dbh != dbufs);
+}
+
+int nilfs_copy_dirty_pages(struct address_space *dmap,
+			   struct address_space *smap)
+{
+	struct folio_batch fbatch;
+	unsigned int i;
+	pgoff_t index = 0;
+	int err = 0;
+
+	folio_batch_init(&fbatch);
+repeat:
+	if (!filemap_get_folios_tag(smap, &index, (pgoff_t)-1,
+				PAGECACHE_TAG_DIRTY, &fbatch))
+		return 0;
+
+	for (i = 0; i < folio_batch_count(&fbatch); i++) {
+		struct folio *folio = fbatch.folios[i], *dfolio;
+
+		folio_lock(folio);
+		if (unlikely(!folio_test_dirty(folio)))
+			NILFS_FOLIO_BUG(folio, "inconsistent dirty state");
+
+		dfolio = filemap_grab_folio(dmap, folio->index);
+		if (IS_ERR(dfolio)) {
+			/* No empty page is added to the page cache */
+			folio_unlock(folio);
+			err = PTR_ERR(dfolio);
+			break;
+		}
+		if (unlikely(!folio_buffers(folio)))
+			NILFS_FOLIO_BUG(folio,
+				       "found empty page in dat page cache");
+
+		nilfs_copy_folio(dfolio, folio, true);
+		filemap_dirty_folio(folio_mapping(dfolio), dfolio);
+
+		folio_unlock(dfolio);
+		folio_put(dfolio);
+		folio_unlock(folio);
+	}
+	folio_batch_release(&fbatch);
+	cond_resched();
+
+	if (likely(!err))
+		goto repeat;
+	return err;
+}
+
+/**
+ * nilfs_copy_back_pages -- copy back pages to original cache from shadow cache
+ * @dmap: destination page cache
+ * @smap: source page cache
+ *
+ * No pages must be added to the cache during this process.
+ * This must be ensured by the caller.
+ */
+void nilfs_copy_back_pages(struct address_space *dmap,
+			   struct address_space *smap)
+{
+	struct folio_batch fbatch;
+	unsigned int i, n;
+	pgoff_t start = 0;
+
+	folio_batch_init(&fbatch);
+repeat:
+	n = filemap_get_folios(smap, &start, ~0UL, &fbatch);
+	if (!n)
+		return;
+
+	for (i = 0; i < folio_batch_count(&fbatch); i++) {
+		struct folio *folio = fbatch.folios[i], *dfolio;
+		pgoff_t index = folio->index;
+
+		folio_lock(folio);
+		dfolio = filemap_lock_folio(dmap, index);
+		if (!IS_ERR(dfolio)) {
+			/* overwrite existing folio in the destination cache */
+			WARN_ON(folio_test_dirty(dfolio));
+			nilfs_copy_folio(dfolio, folio, false);
+			folio_unlock(dfolio);
+			folio_put(dfolio);
+			/* Do we not need to remove folio from smap here? */
+		} else {
+			struct folio *f;
+
+			/* move the folio to the destination cache */
+			xa_lock_irq(&smap->i_pages);
+			f = __xa_erase(&smap->i_pages, index);
+			WARN_ON(folio != f);
+			smap->nrpages--;
+			xa_unlock_irq(&smap->i_pages);
+
+			xa_lock_irq(&dmap->i_pages);
+			f = __xa_store(&dmap->i_pages, index, folio, GFP_NOFS);
+			if (unlikely(f)) {
+				/* Probably -ENOMEM */
+				folio->mapping = NULL;
+				folio_put(folio);
+			} else {
+				folio->mapping = dmap;
+				dmap->nrpages++;
+				if (folio_test_dirty(folio))
+					__xa_set_mark(&dmap->i_pages, index,
+							PAGECACHE_TAG_DIRTY);
+			}
+			xa_unlock_irq(&dmap->i_pages);
+		}
+		folio_unlock(folio);
+	}
+	folio_batch_release(&fbatch);
+	cond_resched();
+
+	goto repeat;
+}
+
+/**
+ * nilfs_clear_dirty_pages - discard dirty pages in address space
+ * @mapping: address space with dirty pages for discarding
+ */
+void nilfs_clear_dirty_pages(struct address_space *mapping)
+{
+	struct folio_batch fbatch;
+	unsigned int i;
+	pgoff_t index = 0;
+
+	folio_batch_init(&fbatch);
+
+	while (filemap_get_folios_tag(mapping, &index, (pgoff_t)-1,
+				PAGECACHE_TAG_DIRTY, &fbatch)) {
+		for (i = 0; i < folio_batch_count(&fbatch); i++) {
+			struct folio *folio = fbatch.folios[i];
+
+			folio_lock(folio);
+
+			/*
+			 * This folio may have been removed from the address
+			 * space by truncation or invalidation when the lock
+			 * was acquired.  Skip processing in that case.
+			 */
+			if (likely(folio->mapping == mapping))
+				nilfs_clear_folio_dirty(folio);
+
+			folio_unlock(folio);
+		}
+		folio_batch_release(&fbatch);
+		cond_resched();
+	}
+}
+
+/**
+ * nilfs_clear_folio_dirty - discard dirty folio
+ * @folio: dirty folio that will be discarded
+ *
+ * nilfs_clear_folio_dirty() clears working states including dirty state for
+ * the folio and its buffers.  If the folio has buffers, clear only if it is
+ * confirmed that none of the buffer heads are busy (none have valid
+ * references and none are locked).
+ */
+void nilfs_clear_folio_dirty(struct folio *folio)
+{
+	struct buffer_head *bh, *head;
+
+	BUG_ON(!folio_test_locked(folio));
+
+	head = folio_buffers(folio);
+	if (head) {
+		const unsigned long clear_bits =
+			(BIT(BH_Uptodate) | BIT(BH_Dirty) | BIT(BH_Mapped) |
+			 BIT(BH_Async_Write) | BIT(BH_NILFS_Volatile) |
+			 BIT(BH_NILFS_Checked) | BIT(BH_NILFS_Redirected) |
+			 BIT(BH_Delay));
+		bool busy, invalidated = false;
+
+recheck_buffers:
+		busy = false;
+		bh = head;
+		do {
+			if (atomic_read(&bh->b_count) | buffer_locked(bh)) {
+				busy = true;
+				break;
+			}
+		} while (bh = bh->b_this_page, bh != head);
+
+		if (busy) {
+			if (invalidated)
+				return;
+			invalidate_bh_lrus();
+			invalidated = true;
+			goto recheck_buffers;
+		}
+
+		bh = head;
+		do {
+			lock_buffer(bh);
+			set_mask_bits(&bh->b_state, clear_bits, 0);
+			unlock_buffer(bh);
+		} while (bh = bh->b_this_page, bh != head);
+	}
+
+	folio_clear_uptodate(folio);
+	folio_clear_mappedtodisk(folio);
+	folio_clear_checked(folio);
+	__nilfs_clear_folio_dirty(folio);
+}
+
+unsigned int nilfs_page_count_clean_buffers(struct folio *folio,
+					    unsigned int from, unsigned int to)
+{
+	unsigned int block_start, block_end;
+	struct buffer_head *bh, *head;
+	unsigned int nc = 0;
+
+	for (bh = head = folio_buffers(folio), block_start = 0;
+	     bh != head || !block_start;
+	     block_start = block_end, bh = bh->b_this_page) {
+		block_end = block_start + bh->b_size;
+		if (block_end > from && block_start < to && !buffer_dirty(bh))
+			nc++;
+	}
+	return nc;
+}
+
+/*
+ * NILFS2 needs clear_page_dirty() in the following two cases:
+ *
+ * 1) For B-tree node pages and data pages of DAT file, NILFS2 clears dirty
+ *    flag of pages when it copies back pages from shadow cache to the
+ *    original cache.
+ *
+ * 2) Some B-tree operations like insertion or deletion may dispose buffers
+ *    in dirty state, and this needs to cancel the dirty state of their pages.
+ */
+void __nilfs_clear_folio_dirty(struct folio *folio)
+{
+	struct address_space *mapping = folio->mapping;
+
+	if (mapping) {
+		xa_lock_irq(&mapping->i_pages);
+		if (folio_test_dirty(folio)) {
+			__xa_clear_mark(&mapping->i_pages, folio->index,
+					     PAGECACHE_TAG_DIRTY);
+			xa_unlock_irq(&mapping->i_pages);
+			folio_clear_dirty_for_io(folio);
+			return;
+		}
+		xa_unlock_irq(&mapping->i_pages);
+		return;
+	}
+	folio_clear_dirty(folio);
+}
+
+/**
+ * nilfs_find_uncommitted_extent - find extent of uncommitted data
+ * @inode: inode
+ * @start_blk: start block offset (in)
+ * @blkoff: start offset of the found extent (out)
+ *
+ * This function searches an extent of buffers marked "delayed" which
+ * starts from a block offset equal to or larger than @start_blk.  If
+ * such an extent was found, this will store the start offset in
+ * @blkoff and return its length in blocks.
+ *
+ * Return: Length in blocks of found extent, 0 otherwise.
+ */
+unsigned long nilfs_find_uncommitted_extent(struct inode *inode,
+					    sector_t start_blk,
+					    sector_t *blkoff)
+{
+	unsigned int i, nr_folios;
+	pgoff_t index;
+	unsigned long length = 0;
+	struct folio_batch fbatch;
+	struct folio *folio;
+
+	if (inode->i_mapping->nrpages == 0)
+		return 0;
+
+	index = start_blk >> (PAGE_SHIFT - inode->i_blkbits);
+
+	folio_batch_init(&fbatch);
+
+repeat:
+	nr_folios = filemap_get_folios_contig(inode->i_mapping, &index, ULONG_MAX,
+			&fbatch);
+	if (nr_folios == 0)
+		return length;
+
+	i = 0;
+	do {
+		folio = fbatch.folios[i];
+
+		folio_lock(folio);
+		if (folio_buffers(folio)) {
+			struct buffer_head *bh, *head;
+			sector_t b;
+
+			b = folio->index << (PAGE_SHIFT - inode->i_blkbits);
+			bh = head = folio_buffers(folio);
+			do {
+				if (b < start_blk)
+					continue;
+				if (buffer_delay(bh)) {
+					if (length == 0)
+						*blkoff = b;
+					length++;
+				} else if (length > 0) {
+					goto out_locked;
+				}
+			} while (++b, bh = bh->b_this_page, bh != head);
+		} else {
+			if (length > 0)
+				goto out_locked;
+		}
+		folio_unlock(folio);
+
+	} while (++i < nr_folios);
+
+	folio_batch_release(&fbatch);
+	cond_resched();
+	goto repeat;
+
+out_locked:
+	folio_unlock(folio);
+	folio_batch_release(&fbatch);
+	return length;
+}

linux-7.2.3/fs/nilfs2/page.h

@@ -0,0 +1,55 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Buffer/page management specific to NILFS
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi and Seiji Kihara.
+ */
+
+#ifndef _NILFS_PAGE_H
+#define _NILFS_PAGE_H
+
+#include <linux/buffer_head.h>
+#include "nilfs.h"
+
+/*
+ * Extended buffer state bits
+ */
+enum {
+	BH_NILFS_Allocated = BH_PrivateStart,
+	BH_NILFS_Node,
+	BH_NILFS_Volatile,
+	BH_NILFS_Checked,
+	BH_NILFS_Redirected,
+};
+
+BUFFER_FNS(NILFS_Node, nilfs_node)		/* nilfs node buffers */
+BUFFER_FNS(NILFS_Volatile, nilfs_volatile)
+BUFFER_FNS(NILFS_Checked, nilfs_checked)	/* buffer is verified */
+BUFFER_FNS(NILFS_Redirected, nilfs_redirected)	/* redirected to a copy */
+
+
+void __nilfs_clear_folio_dirty(struct folio *);
+
+struct buffer_head *nilfs_grab_buffer(struct inode *, struct address_space *,
+				      unsigned long, unsigned long);
+void nilfs_forget_buffer(struct buffer_head *);
+void nilfs_copy_buffer(struct buffer_head *, struct buffer_head *);
+bool nilfs_folio_buffers_clean(struct folio *);
+void nilfs_folio_bug(struct folio *);
+
+int nilfs_copy_dirty_pages(struct address_space *, struct address_space *);
+void nilfs_copy_back_pages(struct address_space *, struct address_space *);
+void nilfs_clear_folio_dirty(struct folio *folio);
+void nilfs_clear_dirty_pages(struct address_space *mapping);
+unsigned int nilfs_page_count_clean_buffers(struct folio *folio,
+		unsigned int from, unsigned int to);
+unsigned long nilfs_find_uncommitted_extent(struct inode *inode,
+					    sector_t start_blk,
+					    sector_t *blkoff);
+
+#define NILFS_FOLIO_BUG(folio, m, a...) \
+	do { nilfs_folio_bug(folio); BUG(); } while (0)
+
+#endif /* _NILFS_PAGE_H */

linux-7.2.3/fs/nilfs2/segbuf.c

@@ -0,0 +1,472 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS segment buffer
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi.
+ *
+ */
+
+#include <linux/buffer_head.h>
+#include <linux/writeback.h>
+#include <linux/crc32.h>
+#include <linux/backing-dev.h>
+#include <linux/slab.h>
+#include "page.h"
+#include "segbuf.h"
+
+
+struct nilfs_write_info {
+	struct the_nilfs       *nilfs;
+	struct bio	       *bio;
+	int			start, end; /* The region to be submitted */
+	int			rest_blocks;
+	int			max_pages;
+	int			nr_vecs;
+	sector_t		blocknr;
+};
+
+static int nilfs_segbuf_write(struct nilfs_segment_buffer *segbuf,
+			      struct the_nilfs *nilfs);
+static int nilfs_segbuf_wait(struct nilfs_segment_buffer *segbuf);
+
+struct nilfs_segment_buffer *nilfs_segbuf_new(struct super_block *sb)
+{
+	struct nilfs_segment_buffer *segbuf;
+
+	segbuf = kmem_cache_alloc(nilfs_segbuf_cachep, GFP_NOFS);
+	if (unlikely(!segbuf))
+		return NULL;
+
+	segbuf->sb_super = sb;
+	INIT_LIST_HEAD(&segbuf->sb_list);
+	INIT_LIST_HEAD(&segbuf->sb_segsum_buffers);
+	INIT_LIST_HEAD(&segbuf->sb_payload_buffers);
+	segbuf->sb_super_root = NULL;
+
+	init_completion(&segbuf->sb_bio_event);
+	atomic_set(&segbuf->sb_err, 0);
+	segbuf->sb_nbio = 0;
+
+	return segbuf;
+}
+
+void nilfs_segbuf_free(struct nilfs_segment_buffer *segbuf)
+{
+	kmem_cache_free(nilfs_segbuf_cachep, segbuf);
+}
+
+void nilfs_segbuf_map(struct nilfs_segment_buffer *segbuf, __u64 segnum,
+		     unsigned long offset, struct the_nilfs *nilfs)
+{
+	segbuf->sb_segnum = segnum;
+	nilfs_get_segment_range(nilfs, segnum, &segbuf->sb_fseg_start,
+				&segbuf->sb_fseg_end);
+
+	segbuf->sb_pseg_start = segbuf->sb_fseg_start + offset;
+	segbuf->sb_rest_blocks =
+		segbuf->sb_fseg_end - segbuf->sb_pseg_start + 1;
+}
+
+/**
+ * nilfs_segbuf_map_cont - map a new log behind a given log
+ * @segbuf: new segment buffer
+ * @prev: segment buffer containing a log to be continued
+ */
+void nilfs_segbuf_map_cont(struct nilfs_segment_buffer *segbuf,
+			   struct nilfs_segment_buffer *prev)
+{
+	segbuf->sb_segnum = prev->sb_segnum;
+	segbuf->sb_fseg_start = prev->sb_fseg_start;
+	segbuf->sb_fseg_end = prev->sb_fseg_end;
+	segbuf->sb_pseg_start = prev->sb_pseg_start + prev->sb_sum.nblocks;
+	segbuf->sb_rest_blocks =
+		segbuf->sb_fseg_end - segbuf->sb_pseg_start + 1;
+}
+
+void nilfs_segbuf_set_next_segnum(struct nilfs_segment_buffer *segbuf,
+				  __u64 nextnum, struct the_nilfs *nilfs)
+{
+	segbuf->sb_nextnum = nextnum;
+	segbuf->sb_sum.next = nilfs_get_segment_start_blocknr(nilfs, nextnum);
+}
+
+int nilfs_segbuf_extend_segsum(struct nilfs_segment_buffer *segbuf)
+{
+	struct buffer_head *bh;
+
+	bh = sb_getblk(segbuf->sb_super,
+		       segbuf->sb_pseg_start + segbuf->sb_sum.nsumblk);
+	if (unlikely(!bh))
+		return -ENOMEM;
+
+	lock_buffer(bh);
+	if (!buffer_uptodate(bh)) {
+		memset(bh->b_data, 0, bh->b_size);
+		set_buffer_uptodate(bh);
+	}
+	unlock_buffer(bh);
+	nilfs_segbuf_add_segsum_buffer(segbuf, bh);
+	return 0;
+}
+
+int nilfs_segbuf_extend_payload(struct nilfs_segment_buffer *segbuf,
+				struct buffer_head **bhp)
+{
+	struct buffer_head *bh;
+
+	bh = sb_getblk(segbuf->sb_super,
+		       segbuf->sb_pseg_start + segbuf->sb_sum.nblocks);
+	if (unlikely(!bh))
+		return -ENOMEM;
+
+	nilfs_segbuf_add_payload_buffer(segbuf, bh);
+	*bhp = bh;
+	return 0;
+}
+
+int nilfs_segbuf_reset(struct nilfs_segment_buffer *segbuf, unsigned int flags,
+		       time64_t ctime, __u64 cno)
+{
+	int err;
+
+	segbuf->sb_sum.nblocks = segbuf->sb_sum.nsumblk = 0;
+	err = nilfs_segbuf_extend_segsum(segbuf);
+	if (unlikely(err))
+		return err;
+
+	segbuf->sb_sum.flags = flags;
+	segbuf->sb_sum.sumbytes = sizeof(struct nilfs_segment_summary);
+	segbuf->sb_sum.nfinfo = segbuf->sb_sum.nfileblk = 0;
+	segbuf->sb_sum.ctime = ctime;
+	segbuf->sb_sum.cno = cno;
+	return 0;
+}
+
+/*
+ * Setup segment summary
+ */
+void nilfs_segbuf_fill_in_segsum(struct nilfs_segment_buffer *segbuf)
+{
+	struct nilfs_segment_summary *raw_sum;
+	struct buffer_head *bh_sum;
+
+	bh_sum = list_entry(segbuf->sb_segsum_buffers.next,
+			    struct buffer_head, b_assoc_buffers);
+	raw_sum = (struct nilfs_segment_summary *)bh_sum->b_data;
+
+	raw_sum->ss_magic    = cpu_to_le32(NILFS_SEGSUM_MAGIC);
+	raw_sum->ss_bytes    = cpu_to_le16(sizeof(*raw_sum));
+	raw_sum->ss_flags    = cpu_to_le16(segbuf->sb_sum.flags);
+	raw_sum->ss_seq      = cpu_to_le64(segbuf->sb_sum.seg_seq);
+	raw_sum->ss_create   = cpu_to_le64(segbuf->sb_sum.ctime);
+	raw_sum->ss_next     = cpu_to_le64(segbuf->sb_sum.next);
+	raw_sum->ss_nblocks  = cpu_to_le32(segbuf->sb_sum.nblocks);
+	raw_sum->ss_nfinfo   = cpu_to_le32(segbuf->sb_sum.nfinfo);
+	raw_sum->ss_sumbytes = cpu_to_le32(segbuf->sb_sum.sumbytes);
+	raw_sum->ss_pad      = 0;
+	raw_sum->ss_cno      = cpu_to_le64(segbuf->sb_sum.cno);
+}
+
+/*
+ * CRC calculation routines
+ */
+static void
+nilfs_segbuf_fill_in_segsum_crc(struct nilfs_segment_buffer *segbuf, u32 seed)
+{
+	struct buffer_head *bh;
+	struct nilfs_segment_summary *raw_sum;
+	unsigned long size, bytes = segbuf->sb_sum.sumbytes;
+	u32 crc;
+
+	bh = list_entry(segbuf->sb_segsum_buffers.next, struct buffer_head,
+			b_assoc_buffers);
+
+	raw_sum = (struct nilfs_segment_summary *)bh->b_data;
+	size = min_t(unsigned long, bytes, bh->b_size);
+	crc = crc32_le(seed,
+		       (unsigned char *)raw_sum +
+		       sizeof(raw_sum->ss_datasum) + sizeof(raw_sum->ss_sumsum),
+		       size - (sizeof(raw_sum->ss_datasum) +
+			       sizeof(raw_sum->ss_sumsum)));
+
+	list_for_each_entry_continue(bh, &segbuf->sb_segsum_buffers,
+				     b_assoc_buffers) {
+		bytes -= size;
+		size = min_t(unsigned long, bytes, bh->b_size);
+		crc = crc32_le(crc, bh->b_data, size);
+	}
+	raw_sum->ss_sumsum = cpu_to_le32(crc);
+}
+
+static void nilfs_segbuf_fill_in_data_crc(struct nilfs_segment_buffer *segbuf,
+					  u32 seed)
+{
+	struct buffer_head *bh;
+	struct nilfs_segment_summary *raw_sum;
+	u32 crc;
+
+	bh = list_entry(segbuf->sb_segsum_buffers.next, struct buffer_head,
+			b_assoc_buffers);
+	raw_sum = (struct nilfs_segment_summary *)bh->b_data;
+	crc = crc32_le(seed,
+		       (unsigned char *)raw_sum + sizeof(raw_sum->ss_datasum),
+		       bh->b_size - sizeof(raw_sum->ss_datasum));
+
+	list_for_each_entry_continue(bh, &segbuf->sb_segsum_buffers,
+				     b_assoc_buffers) {
+		crc = crc32_le(crc, bh->b_data, bh->b_size);
+	}
+	list_for_each_entry(bh, &segbuf->sb_payload_buffers, b_assoc_buffers) {
+		size_t offset = offset_in_folio(bh->b_folio, bh->b_data);
+		unsigned char *from;
+
+		/* Do not support block sizes larger than PAGE_SIZE */
+		from = kmap_local_folio(bh->b_folio, offset);
+		crc = crc32_le(crc, from, bh->b_size);
+		kunmap_local(from);
+	}
+	raw_sum->ss_datasum = cpu_to_le32(crc);
+}
+
+static void
+nilfs_segbuf_fill_in_super_root_crc(struct nilfs_segment_buffer *segbuf,
+				    u32 seed)
+{
+	struct nilfs_super_root *raw_sr;
+	struct the_nilfs *nilfs = segbuf->sb_super->s_fs_info;
+	unsigned int srsize;
+	u32 crc;
+
+	raw_sr = (struct nilfs_super_root *)segbuf->sb_super_root->b_data;
+	srsize = NILFS_SR_BYTES(nilfs->ns_inode_size);
+	crc = crc32_le(seed,
+		       (unsigned char *)raw_sr + sizeof(raw_sr->sr_sum),
+		       srsize - sizeof(raw_sr->sr_sum));
+	raw_sr->sr_sum = cpu_to_le32(crc);
+}
+
+static void nilfs_release_buffers(struct list_head *list)
+{
+	struct buffer_head *bh, *n;
+
+	list_for_each_entry_safe(bh, n, list, b_assoc_buffers) {
+		list_del_init(&bh->b_assoc_buffers);
+		brelse(bh);
+	}
+}
+
+static void nilfs_segbuf_clear(struct nilfs_segment_buffer *segbuf)
+{
+	nilfs_release_buffers(&segbuf->sb_segsum_buffers);
+	nilfs_release_buffers(&segbuf->sb_payload_buffers);
+	segbuf->sb_super_root = NULL;
+}
+
+/*
+ * Iterators for segment buffers
+ */
+void nilfs_clear_logs(struct list_head *logs)
+{
+	struct nilfs_segment_buffer *segbuf;
+
+	list_for_each_entry(segbuf, logs, sb_list)
+		nilfs_segbuf_clear(segbuf);
+}
+
+void nilfs_truncate_logs(struct list_head *logs,
+			 struct nilfs_segment_buffer *last)
+{
+	struct nilfs_segment_buffer *n, *segbuf;
+
+	segbuf = list_prepare_entry(last, logs, sb_list);
+	list_for_each_entry_safe_continue(segbuf, n, logs, sb_list) {
+		list_del_init(&segbuf->sb_list);
+		nilfs_segbuf_clear(segbuf);
+		nilfs_segbuf_free(segbuf);
+	}
+}
+
+int nilfs_write_logs(struct list_head *logs, struct the_nilfs *nilfs)
+{
+	struct nilfs_segment_buffer *segbuf;
+	int ret = 0;
+
+	list_for_each_entry(segbuf, logs, sb_list) {
+		ret = nilfs_segbuf_write(segbuf, nilfs);
+		if (ret)
+			break;
+	}
+	return ret;
+}
+
+int nilfs_wait_on_logs(struct list_head *logs)
+{
+	struct nilfs_segment_buffer *segbuf;
+	int err, ret = 0;
+
+	list_for_each_entry(segbuf, logs, sb_list) {
+		err = nilfs_segbuf_wait(segbuf);
+		if (err && !ret)
+			ret = err;
+	}
+	return ret;
+}
+
+/**
+ * nilfs_add_checksums_on_logs - add checksums on the logs
+ * @logs: list of segment buffers storing target logs
+ * @seed: checksum seed value
+ */
+void nilfs_add_checksums_on_logs(struct list_head *logs, u32 seed)
+{
+	struct nilfs_segment_buffer *segbuf;
+
+	list_for_each_entry(segbuf, logs, sb_list) {
+		if (segbuf->sb_super_root)
+			nilfs_segbuf_fill_in_super_root_crc(segbuf, seed);
+		nilfs_segbuf_fill_in_segsum_crc(segbuf, seed);
+		nilfs_segbuf_fill_in_data_crc(segbuf, seed);
+	}
+}
+
+/*
+ * BIO operations
+ */
+static void nilfs_end_bio_write(struct bio *bio)
+{
+	struct nilfs_segment_buffer *segbuf = bio->bi_private;
+
+	if (bio->bi_status)
+		atomic_inc(&segbuf->sb_err);
+
+	bio_put(bio);
+	complete(&segbuf->sb_bio_event);
+}
+
+static int nilfs_segbuf_submit_bio(struct nilfs_segment_buffer *segbuf,
+				   struct nilfs_write_info *wi)
+{
+	struct bio *bio = wi->bio;
+
+	bio->bi_end_io = nilfs_end_bio_write;
+	bio->bi_private = segbuf;
+	submit_bio(bio);
+	segbuf->sb_nbio++;
+
+	wi->bio = NULL;
+	wi->rest_blocks -= wi->end - wi->start;
+	wi->nr_vecs = min(wi->max_pages, wi->rest_blocks);
+	wi->start = wi->end;
+	return 0;
+}
+
+static void nilfs_segbuf_prepare_write(struct nilfs_segment_buffer *segbuf,
+				       struct nilfs_write_info *wi)
+{
+	wi->bio = NULL;
+	wi->rest_blocks = segbuf->sb_sum.nblocks;
+	wi->max_pages = BIO_MAX_VECS;
+	wi->nr_vecs = min(wi->max_pages, wi->rest_blocks);
+	wi->start = wi->end = 0;
+	wi->blocknr = segbuf->sb_pseg_start;
+}
+
+static int nilfs_segbuf_submit_bh(struct nilfs_segment_buffer *segbuf,
+				  struct nilfs_write_info *wi,
+				  struct buffer_head *bh)
+{
+	int err;
+
+	BUG_ON(wi->nr_vecs <= 0);
+ repeat:
+	if (!wi->bio) {
+		wi->bio = bio_alloc(wi->nilfs->ns_bdev, wi->nr_vecs,
+				    REQ_OP_WRITE, GFP_NOIO);
+		wi->bio->bi_iter.bi_sector = (wi->blocknr + wi->end) <<
+			(wi->nilfs->ns_blocksize_bits - 9);
+	}
+
+	if (bio_add_folio(wi->bio, bh->b_folio, bh->b_size,
+			  offset_in_folio(bh->b_folio, bh->b_data))) {
+		wi->end++;
+		return 0;
+	}
+	/* bio is FULL */
+	err = nilfs_segbuf_submit_bio(segbuf, wi);
+	/* never submit current bh */
+	if (likely(!err))
+		goto repeat;
+	return err;
+}
+
+/**
+ * nilfs_segbuf_write - submit write requests of a log
+ * @segbuf: buffer storing a log to be written
+ * @nilfs: nilfs object
+ *
+ * Return: Always 0.
+ */
+static int nilfs_segbuf_write(struct nilfs_segment_buffer *segbuf,
+			      struct the_nilfs *nilfs)
+{
+	struct nilfs_write_info wi;
+	struct buffer_head *bh;
+	int res = 0;
+
+	wi.nilfs = nilfs;
+	nilfs_segbuf_prepare_write(segbuf, &wi);
+
+	list_for_each_entry(bh, &segbuf->sb_segsum_buffers, b_assoc_buffers) {
+		res = nilfs_segbuf_submit_bh(segbuf, &wi, bh);
+		if (unlikely(res))
+			goto failed_bio;
+	}
+
+	list_for_each_entry(bh, &segbuf->sb_payload_buffers, b_assoc_buffers) {
+		res = nilfs_segbuf_submit_bh(segbuf, &wi, bh);
+		if (unlikely(res))
+			goto failed_bio;
+	}
+
+	if (wi.bio) {
+		/*
+		 * Last BIO is always sent through the following
+		 * submission.
+		 */
+		wi.bio->bi_opf |= REQ_SYNC;
+		res = nilfs_segbuf_submit_bio(segbuf, &wi);
+	}
+
+ failed_bio:
+	return res;
+}
+
+/**
+ * nilfs_segbuf_wait - wait for completion of requested BIOs
+ * @segbuf: segment buffer
+ *
+ * Return: 0 on success, or %-EIO if I/O error is detected.
+ */
+static int nilfs_segbuf_wait(struct nilfs_segment_buffer *segbuf)
+{
+	int err = 0;
+
+	if (!segbuf->sb_nbio)
+		return 0;
+
+	do {
+		wait_for_completion(&segbuf->sb_bio_event);
+	} while (--segbuf->sb_nbio > 0);
+
+	if (unlikely(atomic_read(&segbuf->sb_err) > 0)) {
+		nilfs_err(segbuf->sb_super,
+			  "I/O error writing log (start-blocknr=%llu, block-count=%lu) in segment %llu",
+			  (unsigned long long)segbuf->sb_pseg_start,
+			  segbuf->sb_sum.nblocks,
+			  (unsigned long long)segbuf->sb_segnum);
+		err = -EIO;
+	}
+	return err;
+}

linux-7.2.3/fs/nilfs2/segbuf.h

@@ -0,0 +1,172 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS Segment buffer prototypes and definitions
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi.
+ *
+ */
+#ifndef _NILFS_SEGBUF_H
+#define _NILFS_SEGBUF_H
+
+#include <linux/fs.h>
+#include <linux/buffer_head.h>
+#include <linux/bio.h>
+#include <linux/completion.h>
+
+/**
+ * struct nilfs_segsum_info - On-memory segment summary
+ * @flags: Flags
+ * @nfinfo: Number of file information structures
+ * @nblocks: Number of blocks included in the partial segment
+ * @nsumblk: Number of summary blocks
+ * @sumbytes: Byte count of segment summary
+ * @nfileblk: Total number of file blocks
+ * @seg_seq: Segment sequence number
+ * @cno: Checkpoint number
+ * @ctime: Creation time
+ * @next: Block number of the next full segment
+ */
+struct nilfs_segsum_info {
+	unsigned int		flags;
+	unsigned long		nfinfo;
+	unsigned long		nblocks;
+	unsigned long		nsumblk;
+	unsigned long		sumbytes;
+	unsigned long		nfileblk;
+	u64			seg_seq;
+	__u64			cno;
+	time64_t		ctime;
+	sector_t		next;
+};
+
+/**
+ * struct nilfs_segment_buffer - Segment buffer
+ * @sb_super: back pointer to a superblock struct
+ * @sb_list: List head to chain this structure
+ * @sb_sum: On-memory segment summary
+ * @sb_segnum: Index number of the full segment
+ * @sb_nextnum: Index number of the next full segment
+ * @sb_fseg_start: Start block number of the full segment
+ * @sb_fseg_end: End block number of the full segment
+ * @sb_pseg_start: Disk block number of partial segment
+ * @sb_rest_blocks: Number of residual blocks in the current segment
+ * @sb_segsum_buffers: List of buffers for segment summaries
+ * @sb_payload_buffers: List of buffers for segment payload
+ * @sb_super_root: Pointer to buffer storing a super root block (if exists)
+ * @sb_nbio: Number of flying bio requests
+ * @sb_err: I/O error status
+ * @sb_bio_event: Completion event of log writing
+ */
+struct nilfs_segment_buffer {
+	struct super_block     *sb_super;
+	struct list_head	sb_list;
+
+	/* Segment information */
+	struct nilfs_segsum_info sb_sum;
+	__u64			sb_segnum;
+	__u64			sb_nextnum;
+	sector_t		sb_fseg_start, sb_fseg_end;
+	sector_t		sb_pseg_start;
+	unsigned int		sb_rest_blocks;
+
+	/* Buffers */
+	struct list_head	sb_segsum_buffers;
+	struct list_head	sb_payload_buffers; /* including super root */
+	struct buffer_head     *sb_super_root;
+
+	/* io status */
+	int			sb_nbio;
+	atomic_t		sb_err;
+	struct completion	sb_bio_event;
+};
+
+#define NILFS_LIST_SEGBUF(head)  \
+	list_entry((head), struct nilfs_segment_buffer, sb_list)
+#define NILFS_NEXT_SEGBUF(segbuf)  NILFS_LIST_SEGBUF((segbuf)->sb_list.next)
+#define NILFS_PREV_SEGBUF(segbuf)  NILFS_LIST_SEGBUF((segbuf)->sb_list.prev)
+#define NILFS_LAST_SEGBUF(head)    NILFS_LIST_SEGBUF((head)->prev)
+#define NILFS_FIRST_SEGBUF(head)   NILFS_LIST_SEGBUF((head)->next)
+#define NILFS_SEGBUF_IS_LAST(segbuf, head)  ((segbuf)->sb_list.next == (head))
+
+#define nilfs_for_each_segbuf_before(s, t, h) \
+	for ((s) = NILFS_FIRST_SEGBUF(h); (s) != (t); \
+	     (s) = NILFS_NEXT_SEGBUF(s))
+
+#define NILFS_SEGBUF_FIRST_BH(head)  \
+	(list_entry((head)->next, struct buffer_head, b_assoc_buffers))
+#define NILFS_SEGBUF_NEXT_BH(bh)  \
+	(list_entry((bh)->b_assoc_buffers.next, struct buffer_head, \
+		    b_assoc_buffers))
+#define NILFS_SEGBUF_BH_IS_LAST(bh, head)  ((bh)->b_assoc_buffers.next == head)
+
+extern struct kmem_cache *nilfs_segbuf_cachep;
+
+struct nilfs_segment_buffer *nilfs_segbuf_new(struct super_block *);
+void nilfs_segbuf_free(struct nilfs_segment_buffer *);
+void nilfs_segbuf_map(struct nilfs_segment_buffer *, __u64, unsigned long,
+		      struct the_nilfs *);
+void nilfs_segbuf_map_cont(struct nilfs_segment_buffer *segbuf,
+			   struct nilfs_segment_buffer *prev);
+void nilfs_segbuf_set_next_segnum(struct nilfs_segment_buffer *, __u64,
+				  struct the_nilfs *);
+int nilfs_segbuf_reset(struct nilfs_segment_buffer *, unsigned int, time64_t,
+		       __u64);
+int nilfs_segbuf_extend_segsum(struct nilfs_segment_buffer *);
+int nilfs_segbuf_extend_payload(struct nilfs_segment_buffer *,
+				struct buffer_head **);
+void nilfs_segbuf_fill_in_segsum(struct nilfs_segment_buffer *);
+
+static inline int nilfs_segbuf_simplex(struct nilfs_segment_buffer *segbuf)
+{
+	unsigned int flags = segbuf->sb_sum.flags;
+
+	return (flags & (NILFS_SS_LOGBGN | NILFS_SS_LOGEND)) ==
+		(NILFS_SS_LOGBGN | NILFS_SS_LOGEND);
+}
+
+static inline int nilfs_segbuf_empty(struct nilfs_segment_buffer *segbuf)
+{
+	return segbuf->sb_sum.nblocks == segbuf->sb_sum.nsumblk;
+}
+
+static inline void
+nilfs_segbuf_add_segsum_buffer(struct nilfs_segment_buffer *segbuf,
+			       struct buffer_head *bh)
+{
+	list_add_tail(&bh->b_assoc_buffers, &segbuf->sb_segsum_buffers);
+	segbuf->sb_sum.nblocks++;
+	segbuf->sb_sum.nsumblk++;
+}
+
+static inline void
+nilfs_segbuf_add_payload_buffer(struct nilfs_segment_buffer *segbuf,
+				struct buffer_head *bh)
+{
+	list_add_tail(&bh->b_assoc_buffers, &segbuf->sb_payload_buffers);
+	segbuf->sb_sum.nblocks++;
+}
+
+static inline void
+nilfs_segbuf_add_file_buffer(struct nilfs_segment_buffer *segbuf,
+			     struct buffer_head *bh)
+{
+	get_bh(bh);
+	nilfs_segbuf_add_payload_buffer(segbuf, bh);
+	segbuf->sb_sum.nfileblk++;
+}
+
+void nilfs_clear_logs(struct list_head *logs);
+void nilfs_truncate_logs(struct list_head *logs,
+			 struct nilfs_segment_buffer *last);
+int nilfs_write_logs(struct list_head *logs, struct the_nilfs *nilfs);
+int nilfs_wait_on_logs(struct list_head *logs);
+void nilfs_add_checksums_on_logs(struct list_head *logs, u32 seed);
+
+static inline void nilfs_destroy_logs(struct list_head *logs)
+{
+	nilfs_truncate_logs(logs, NULL);
+}
+
+#endif /* _NILFS_SEGBUF_H */

linux-7.2.3/fs/nilfs2/segment.h

@@ -0,0 +1,244 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS Segment constructor prototypes and definitions
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi.
+ *
+ */
+#ifndef _NILFS_SEGMENT_H
+#define _NILFS_SEGMENT_H
+
+#include <linux/types.h>
+#include <linux/fs.h>
+#include <linux/buffer_head.h>
+#include <linux/workqueue.h>
+#include "nilfs.h"
+
+struct nilfs_root;
+
+/**
+ * struct nilfs_recovery_info - Recovery information
+ * @ri_need_recovery: Recovery status
+ * @ri_super_root: Block number of the last super root
+ * @ri_cno: Number of the last checkpoint
+ * @ri_lsegs_start: Region for roll-forwarding (start block number)
+ * @ri_lsegs_end: Region for roll-forwarding (end block number)
+ * @ri_lsegs_start_seq: Sequence value of the segment at ri_lsegs_start
+ * @ri_used_segments: List of segments to be mark active
+ * @ri_pseg_start: Block number of the last partial segment
+ * @ri_seq: Sequence number on the last partial segment
+ * @ri_segnum: Segment number on the last partial segment
+ * @ri_nextnum: Next segment number on the last partial segment
+ */
+struct nilfs_recovery_info {
+	int			ri_need_recovery;
+	sector_t		ri_super_root;
+	__u64			ri_cno;
+
+	sector_t		ri_lsegs_start;
+	sector_t		ri_lsegs_end;
+	u64			ri_lsegs_start_seq;
+	struct list_head	ri_used_segments;
+	sector_t		ri_pseg_start;
+	u64			ri_seq;
+	__u64			ri_segnum;
+	__u64			ri_nextnum;
+};
+
+/* ri_need_recovery */
+#define NILFS_RECOVERY_SR_UPDATED	 1  /* The super root was updated */
+#define NILFS_RECOVERY_ROLLFORWARD_DONE	 2  /* Rollforward was carried out */
+
+/**
+ * struct nilfs_cstage - Context of collection stage
+ * @scnt: Stage count, must be accessed via wrappers:
+ *        nilfs_sc_cstage_inc(), nilfs_sc_cstage_set(), nilfs_sc_cstage_get()
+ * @flags: State flags
+ * @dirty_file_ptr: Pointer on dirty_files list, or inode of a target file
+ * @gc_inode_ptr: Pointer on the list of gc-inodes
+ */
+struct nilfs_cstage {
+	int			scnt;
+	unsigned int		flags;
+	struct nilfs_inode_info *dirty_file_ptr;
+	struct nilfs_inode_info *gc_inode_ptr;
+};
+
+struct nilfs_segment_buffer;
+
+struct nilfs_segsum_pointer {
+	struct buffer_head     *bh;
+	unsigned int		offset; /* offset in bytes */
+};
+
+/**
+ * struct nilfs_sc_info - Segment constructor information
+ * @sc_super: Back pointer to super_block struct
+ * @sc_root: root object of the current filesystem tree
+ * @sc_nblk_inc: Block count of current generation
+ * @sc_dirty_files: List of files to be written
+ * @sc_gc_inodes: List of GC inodes having blocks to be written
+ * @sc_iput_queue: list of inodes for which iput should be done
+ * @sc_iput_work: work struct to defer iput call
+ * @sc_freesegs: array of segment numbers to be freed
+ * @sc_nfreesegs: number of segments on @sc_freesegs
+ * @sc_dsync_inode: inode whose data pages are written for a sync operation
+ * @sc_dsync_start: start byte offset of data pages
+ * @sc_dsync_end: end byte offset of data pages (inclusive)
+ * @sc_segbufs: List of segment buffers
+ * @sc_write_logs: List of segment buffers to hold logs under writing
+ * @sc_segbuf_nblocks: Number of available blocks in segment buffers.
+ * @sc_curseg: Current segment buffer
+ * @sc_stage: Collection stage
+ * @sc_finfo_ptr: pointer to the current finfo struct in the segment summary
+ * @sc_binfo_ptr: pointer to the current binfo struct in the segment summary
+ * @sc_blk_cnt:	Block count of a file
+ * @sc_datablk_cnt: Data block count of a file
+ * @sc_nblk_this_inc: Number of blocks included in the current logical segment
+ * @sc_seg_ctime: Creation time
+ * @sc_cno: checkpoint number of current log
+ * @sc_flags: Internal flags
+ * @sc_state_lock: spinlock for sc_state and so on
+ * @sc_state: Segctord state flags
+ * @sc_flush_request: inode bitmap of metadata files to be flushed
+ * @sc_wait_request: Client request queue
+ * @sc_wait_daemon: Daemon wait queue
+ * @sc_seq_request: Request counter
+ * @sc_seq_accepted: Accepted request count
+ * @sc_seq_done: Completion counter
+ * @sc_sync: Request of explicit sync operation
+ * @sc_interval: Timeout value of background construction
+ * @sc_mjcp_freq: Frequency of creating checkpoints
+ * @sc_lseg_stime: Start time of the latest logical segment
+ * @sc_watermark: Watermark for the number of dirty buffers
+ * @sc_timer: Timer for segctord
+ * @sc_task: current thread of segctord
+ */
+struct nilfs_sc_info {
+	struct super_block     *sc_super;
+	struct nilfs_root      *sc_root;
+
+	unsigned long		sc_nblk_inc;
+
+	struct list_head	sc_dirty_files;
+	struct list_head	sc_gc_inodes;
+	struct list_head	sc_iput_queue;
+	struct work_struct	sc_iput_work;
+
+	__u64		       *sc_freesegs;
+	size_t			sc_nfreesegs;
+
+	struct nilfs_inode_info *sc_dsync_inode;
+	loff_t			sc_dsync_start;
+	loff_t			sc_dsync_end;
+
+	/* Segment buffers */
+	struct list_head	sc_segbufs;
+	struct list_head	sc_write_logs;
+	unsigned long		sc_segbuf_nblocks;
+	struct nilfs_segment_buffer *sc_curseg;
+
+	struct nilfs_cstage	sc_stage;
+
+	struct nilfs_segsum_pointer sc_finfo_ptr;
+	struct nilfs_segsum_pointer sc_binfo_ptr;
+	unsigned long		sc_blk_cnt;
+	unsigned long		sc_datablk_cnt;
+	unsigned long		sc_nblk_this_inc;
+	time64_t		sc_seg_ctime;
+	__u64			sc_cno;
+	unsigned long		sc_flags;
+
+	spinlock_t		sc_state_lock;
+	unsigned long		sc_state;
+	unsigned long		sc_flush_request;
+
+	wait_queue_head_t	sc_wait_request;
+	wait_queue_head_t	sc_wait_daemon;
+
+	__u32			sc_seq_request;
+	__u32			sc_seq_accepted;
+	__u32			sc_seq_done;
+
+	int			sc_sync;
+	unsigned long		sc_interval;
+	unsigned long		sc_mjcp_freq;
+	unsigned long		sc_lseg_stime;	/* in 1/HZ seconds */
+	unsigned long		sc_watermark;
+
+	struct timer_list	sc_timer;
+	struct task_struct     *sc_task;
+};
+
+/* sc_flags */
+enum {
+	NILFS_SC_DIRTY,		/* One or more dirty meta-data blocks exist */
+	NILFS_SC_UNCLOSED,	/* Logical segment is not closed */
+	NILFS_SC_SUPER_ROOT,	/* The latest segment has a super root */
+	NILFS_SC_PRIOR_FLUSH,	/*
+				 * Requesting immediate flush without making a
+				 * checkpoint
+				 */
+	NILFS_SC_HAVE_DELTA,	/*
+				 * Next checkpoint will have update of files
+				 * other than DAT, cpfile, sufile, or files
+				 * moved by GC.
+				 */
+};
+
+/* sc_state */
+#define NILFS_SEGCTOR_COMMIT	    0x0004  /* committed transaction exists */
+
+/*
+ * Constant parameters
+ */
+#define NILFS_SC_CLEANUP_RETRY	    3  /*
+					* Retry count of construction when
+					* destroying segctord
+					*/
+
+/*
+ * Default values of timeout, in seconds.
+ */
+#define NILFS_SC_DEFAULT_TIMEOUT    5   /*
+					 * Timeout value of dirty blocks.
+					 * It triggers construction of a
+					 * logical segment with a super root.
+					 */
+#define NILFS_SC_DEFAULT_SR_FREQ    30  /*
+					 * Maximum frequency of super root
+					 * creation
+					 */
+
+/*
+ * The default threshold amount of data, in block counts.
+ */
+#define NILFS_SC_DEFAULT_WATERMARK  3600
+
+/* super.c */
+extern struct kmem_cache *nilfs_transaction_cachep;
+
+/* segment.c */
+extern void nilfs_relax_pressure_in_lock(struct super_block *);
+
+extern int nilfs_construct_segment(struct super_block *);
+extern int nilfs_construct_dsync_segment(struct super_block *, struct inode *,
+					 loff_t, loff_t);
+extern int nilfs_clean_segments(struct super_block *, struct nilfs_argv *,
+				void **);
+
+int nilfs_attach_log_writer(struct super_block *sb, struct nilfs_root *root);
+void nilfs_detach_log_writer(struct super_block *sb);
+
+/* recovery.c */
+extern int nilfs_read_super_root_block(struct the_nilfs *, sector_t,
+				       struct buffer_head **, int);
+extern int nilfs_search_super_root(struct the_nilfs *,
+				   struct nilfs_recovery_info *);
+int nilfs_salvage_orphan_logs(struct the_nilfs *nilfs, struct super_block *sb,
+			      struct nilfs_recovery_info *ri);
+extern void nilfs_dispose_segment_list(struct list_head *);
+
+#endif /* _NILFS_SEGMENT_H */

linux-7.2.3/fs/nilfs2/sufile.h

@@ -0,0 +1,134 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * NILFS segment usage file.
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Koji Sato.
+ */
+
+#ifndef _NILFS_SUFILE_H
+#define _NILFS_SUFILE_H
+
+#include <linux/fs.h>
+#include <linux/buffer_head.h>
+#include "mdt.h"
+
+
+static inline unsigned long nilfs_sufile_get_nsegments(struct inode *sufile)
+{
+	return ((struct the_nilfs *)sufile->i_sb->s_fs_info)->ns_nsegments;
+}
+
+unsigned long nilfs_sufile_get_ncleansegs(struct inode *sufile);
+
+int nilfs_sufile_set_alloc_range(struct inode *sufile, __u64 start, __u64 end);
+int nilfs_sufile_alloc(struct inode *, __u64 *);
+int nilfs_sufile_mark_dirty(struct inode *sufile, __u64 segnum);
+int nilfs_sufile_set_segment_usage(struct inode *sufile, __u64 segnum,
+				   unsigned long nblocks, time64_t modtime);
+int nilfs_sufile_get_stat(struct inode *, struct nilfs_sustat *);
+ssize_t nilfs_sufile_get_suinfo(struct inode *, __u64, void *, unsigned int,
+				size_t);
+ssize_t nilfs_sufile_set_suinfo(struct inode *, void *, unsigned int, size_t);
+
+int nilfs_sufile_updatev(struct inode *, __u64 *, size_t, int, size_t *,
+			 void (*dofunc)(struct inode *, __u64,
+					struct buffer_head *,
+					struct buffer_head *));
+int nilfs_sufile_update(struct inode *, __u64, int,
+			void (*dofunc)(struct inode *, __u64,
+				       struct buffer_head *,
+				       struct buffer_head *));
+void nilfs_sufile_do_scrap(struct inode *, __u64, struct buffer_head *,
+			   struct buffer_head *);
+void nilfs_sufile_do_free(struct inode *, __u64, struct buffer_head *,
+			  struct buffer_head *);
+void nilfs_sufile_do_cancel_free(struct inode *, __u64, struct buffer_head *,
+				 struct buffer_head *);
+void nilfs_sufile_do_set_error(struct inode *, __u64, struct buffer_head *,
+			       struct buffer_head *);
+
+int nilfs_sufile_resize(struct inode *sufile, __u64 newnsegs);
+int nilfs_sufile_read(struct super_block *sb, size_t susize,
+		      struct nilfs_inode *raw_inode, struct inode **inodep);
+int nilfs_sufile_trim_fs(struct inode *sufile, struct fstrim_range *range);
+
+/**
+ * nilfs_sufile_scrap - make a segment garbage
+ * @sufile: inode of segment usage file
+ * @segnum: segment number to be freed
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static inline int nilfs_sufile_scrap(struct inode *sufile, __u64 segnum)
+{
+	return nilfs_sufile_update(sufile, segnum, 1, nilfs_sufile_do_scrap);
+}
+
+/**
+ * nilfs_sufile_free - free segment
+ * @sufile: inode of segment usage file
+ * @segnum: segment number to be freed
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static inline int nilfs_sufile_free(struct inode *sufile, __u64 segnum)
+{
+	return nilfs_sufile_update(sufile, segnum, 0, nilfs_sufile_do_free);
+}
+
+/**
+ * nilfs_sufile_freev - free segments
+ * @sufile: inode of segment usage file
+ * @segnumv: array of segment numbers
+ * @nsegs: size of @segnumv array
+ * @ndone: place to store the number of freed segments
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static inline int nilfs_sufile_freev(struct inode *sufile, __u64 *segnumv,
+				     size_t nsegs, size_t *ndone)
+{
+	return nilfs_sufile_updatev(sufile, segnumv, nsegs, 0, ndone,
+				    nilfs_sufile_do_free);
+}
+
+/**
+ * nilfs_sufile_cancel_freev - reallocate freeing segments
+ * @sufile: inode of segment usage file
+ * @segnumv: array of segment numbers
+ * @nsegs: size of @segnumv array
+ * @ndone: place to store the number of cancelled segments
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static inline int nilfs_sufile_cancel_freev(struct inode *sufile,
+					    __u64 *segnumv, size_t nsegs,
+					    size_t *ndone)
+{
+	return nilfs_sufile_updatev(sufile, segnumv, nsegs, 0, ndone,
+				    nilfs_sufile_do_cancel_free);
+}
+
+/**
+ * nilfs_sufile_set_error - mark a segment as erroneous
+ * @sufile: inode of segment usage file
+ * @segnum: segment number
+ *
+ * Description: nilfs_sufile_set_error() marks the segment specified by
+ * @segnum as erroneous. The error segment will never be used again.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EINVAL	- Invalid segment usage number.
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+static inline int nilfs_sufile_set_error(struct inode *sufile, __u64 segnum)
+{
+	return nilfs_sufile_update(sufile, segnum, 0,
+				   nilfs_sufile_do_set_error);
+}
+
+#endif	/* _NILFS_SUFILE_H */

linux-7.2.3/fs/nilfs2/sysfs.h

@@ -0,0 +1,167 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Sysfs support declarations.
+ *
+ * Copyright (C) 2005-2014 Nippon Telegraph and Telephone Corporation.
+ * Copyright (C) 2014 HGST, Inc., a Western Digital Company.
+ *
+ * Written by Vyacheslav Dubeyko <Vyacheslav.Dubeyko@hgst.com>
+ */
+
+#ifndef _NILFS_SYSFS_H
+#define _NILFS_SYSFS_H
+
+#include <linux/sysfs.h>
+
+#define NILFS_ROOT_GROUP_NAME	"nilfs2"
+
+/*
+ * struct nilfs_sysfs_dev_subgroups - device subgroup kernel objects
+ * @sg_superblock_kobj: /sys/fs/<nilfs>/<device>/superblock
+ * @sg_superblock_kobj_unregister: completion state
+ * @sg_segctor_kobj: /sys/fs/<nilfs>/<device>/segctor
+ * @sg_segctor_kobj_unregister: completion state
+ * @sg_mounted_snapshots_kobj: /sys/fs/<nilfs>/<device>/mounted_snapshots
+ * @sg_mounted_snapshots_kobj_unregister: completion state
+ * @sg_checkpoints_kobj: /sys/fs/<nilfs>/<device>/checkpoints
+ * @sg_checkpoints_kobj_unregister: completion state
+ * @sg_segments_kobj: /sys/fs/<nilfs>/<device>/segments
+ * @sg_segments_kobj_unregister: completion state
+ */
+struct nilfs_sysfs_dev_subgroups {
+	/* /sys/fs/<nilfs>/<device>/superblock */
+	struct kobject sg_superblock_kobj;
+	struct completion sg_superblock_kobj_unregister;
+
+	/* /sys/fs/<nilfs>/<device>/segctor */
+	struct kobject sg_segctor_kobj;
+	struct completion sg_segctor_kobj_unregister;
+
+	/* /sys/fs/<nilfs>/<device>/mounted_snapshots */
+	struct kobject sg_mounted_snapshots_kobj;
+	struct completion sg_mounted_snapshots_kobj_unregister;
+
+	/* /sys/fs/<nilfs>/<device>/checkpoints */
+	struct kobject sg_checkpoints_kobj;
+	struct completion sg_checkpoints_kobj_unregister;
+
+	/* /sys/fs/<nilfs>/<device>/segments */
+	struct kobject sg_segments_kobj;
+	struct completion sg_segments_kobj_unregister;
+};
+
+#define NILFS_KOBJ_ATTR_STRUCT(name) \
+struct nilfs_##name##_attr { \
+	struct attribute attr; \
+	ssize_t (*show)(struct kobject *, struct kobj_attribute *, \
+			char *); \
+	ssize_t (*store)(struct kobject *, struct kobj_attribute *, \
+			 const char *, size_t); \
+}
+
+NILFS_KOBJ_ATTR_STRUCT(feature);
+
+#define NILFS_DEV_ATTR_STRUCT(name) \
+struct nilfs_##name##_attr { \
+	struct attribute attr; \
+	ssize_t (*show)(struct nilfs_##name##_attr *, struct the_nilfs *, \
+			char *); \
+	ssize_t (*store)(struct nilfs_##name##_attr *, struct the_nilfs *, \
+			 const char *, size_t); \
+}
+
+NILFS_DEV_ATTR_STRUCT(dev);
+NILFS_DEV_ATTR_STRUCT(segments);
+NILFS_DEV_ATTR_STRUCT(mounted_snapshots);
+NILFS_DEV_ATTR_STRUCT(checkpoints);
+NILFS_DEV_ATTR_STRUCT(superblock);
+NILFS_DEV_ATTR_STRUCT(segctor);
+
+#define NILFS_CP_ATTR_STRUCT(name) \
+struct nilfs_##name##_attr { \
+	struct attribute attr; \
+	ssize_t (*show)(struct nilfs_##name##_attr *, struct nilfs_root *, \
+			char *); \
+	ssize_t (*store)(struct nilfs_##name##_attr *, struct nilfs_root *, \
+			 const char *, size_t); \
+}
+
+NILFS_CP_ATTR_STRUCT(snapshot);
+
+#define NILFS_ATTR(type, name, mode, show, store) \
+	static struct nilfs_##type##_attr nilfs_##type##_attr_##name = \
+		__ATTR(name, mode, show, store)
+
+#define NILFS_INFO_ATTR(type, name) \
+	NILFS_ATTR(type, name, 0444, NULL, NULL)
+#define NILFS_RO_ATTR(type, name) \
+	NILFS_ATTR(type, name, 0444, nilfs_##type##_##name##_show, NULL)
+#define NILFS_RW_ATTR(type, name) \
+	NILFS_ATTR(type, name, 0644, \
+		    nilfs_##type##_##name##_show, \
+		    nilfs_##type##_##name##_store)
+
+#define NILFS_FEATURE_INFO_ATTR(name) \
+	NILFS_INFO_ATTR(feature, name)
+#define NILFS_FEATURE_RO_ATTR(name) \
+	NILFS_RO_ATTR(feature, name)
+#define NILFS_FEATURE_RW_ATTR(name) \
+	NILFS_RW_ATTR(feature, name)
+
+#define NILFS_DEV_INFO_ATTR(name) \
+	NILFS_INFO_ATTR(dev, name)
+#define NILFS_DEV_RO_ATTR(name) \
+	NILFS_RO_ATTR(dev, name)
+#define NILFS_DEV_RW_ATTR(name) \
+	NILFS_RW_ATTR(dev, name)
+
+#define NILFS_SEGMENTS_RO_ATTR(name) \
+	NILFS_RO_ATTR(segments, name)
+#define NILFS_SEGMENTS_RW_ATTR(name) \
+	NILFS_RW_ATTR(segs_info, name)
+
+#define NILFS_MOUNTED_SNAPSHOTS_RO_ATTR(name) \
+	NILFS_RO_ATTR(mounted_snapshots, name)
+
+#define NILFS_CHECKPOINTS_RO_ATTR(name) \
+	NILFS_RO_ATTR(checkpoints, name)
+#define NILFS_CHECKPOINTS_RW_ATTR(name) \
+	NILFS_RW_ATTR(checkpoints, name)
+
+#define NILFS_SNAPSHOT_INFO_ATTR(name) \
+	NILFS_INFO_ATTR(snapshot, name)
+#define NILFS_SNAPSHOT_RO_ATTR(name) \
+	NILFS_RO_ATTR(snapshot, name)
+#define NILFS_SNAPSHOT_RW_ATTR(name) \
+	NILFS_RW_ATTR(snapshot, name)
+
+#define NILFS_SUPERBLOCK_RO_ATTR(name) \
+	NILFS_RO_ATTR(superblock, name)
+#define NILFS_SUPERBLOCK_RW_ATTR(name) \
+	NILFS_RW_ATTR(superblock, name)
+
+#define NILFS_SEGCTOR_INFO_ATTR(name) \
+	NILFS_INFO_ATTR(segctor, name)
+#define NILFS_SEGCTOR_RO_ATTR(name) \
+	NILFS_RO_ATTR(segctor, name)
+#define NILFS_SEGCTOR_RW_ATTR(name) \
+	NILFS_RW_ATTR(segctor, name)
+
+#define NILFS_FEATURE_ATTR_LIST(name) \
+	(&nilfs_feature_attr_##name.attr)
+#define NILFS_DEV_ATTR_LIST(name) \
+	(&nilfs_dev_attr_##name.attr)
+#define NILFS_SEGMENTS_ATTR_LIST(name) \
+	(&nilfs_segments_attr_##name.attr)
+#define NILFS_MOUNTED_SNAPSHOTS_ATTR_LIST(name) \
+	(&nilfs_mounted_snapshots_attr_##name.attr)
+#define NILFS_CHECKPOINTS_ATTR_LIST(name) \
+	(&nilfs_checkpoints_attr_##name.attr)
+#define NILFS_SNAPSHOT_ATTR_LIST(name) \
+	(&nilfs_snapshot_attr_##name.attr)
+#define NILFS_SUPERBLOCK_ATTR_LIST(name) \
+	(&nilfs_superblock_attr_##name.attr)
+#define NILFS_SEGCTOR_ATTR_LIST(name) \
+	(&nilfs_segctor_attr_##name.attr)
+
+#endif /* _NILFS_SYSFS_H */

linux-7.2.3/fs/nilfs2/the_nilfs.c

@@ -0,0 +1,939 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * the_nilfs shared structure.
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi.
+ *
+ */
+
+#include <linux/buffer_head.h>
+#include <linux/slab.h>
+#include <linux/blkdev.h>
+#include <linux/backing-dev.h>
+#include <linux/log2.h>
+#include <linux/crc32.h>
+#include "nilfs.h"
+#include "segment.h"
+#include "alloc.h"
+#include "cpfile.h"
+#include "sufile.h"
+#include "dat.h"
+#include "segbuf.h"
+
+
+static int nilfs_valid_sb(struct nilfs_super_block *sbp);
+
+void nilfs_set_last_segment(struct the_nilfs *nilfs,
+			    sector_t start_blocknr, u64 seq, __u64 cno)
+{
+	spin_lock(&nilfs->ns_last_segment_lock);
+	nilfs->ns_last_pseg = start_blocknr;
+	nilfs->ns_last_seq = seq;
+	nilfs->ns_last_cno = cno;
+
+	if (!nilfs_sb_dirty(nilfs)) {
+		if (nilfs->ns_prev_seq == nilfs->ns_last_seq)
+			goto stay_cursor;
+
+		set_nilfs_sb_dirty(nilfs);
+	}
+	nilfs->ns_prev_seq = nilfs->ns_last_seq;
+
+ stay_cursor:
+	spin_unlock(&nilfs->ns_last_segment_lock);
+}
+
+/**
+ * alloc_nilfs - allocate a nilfs object
+ * @sb: super block instance
+ *
+ * Return: a pointer to the allocated nilfs object on success, or NULL on
+ * failure.
+ */
+struct the_nilfs *alloc_nilfs(struct super_block *sb)
+{
+	struct the_nilfs *nilfs;
+
+	nilfs = kzalloc_obj(*nilfs);
+	if (!nilfs)
+		return NULL;
+
+	nilfs->ns_sb = sb;
+	nilfs->ns_bdev = sb->s_bdev;
+	atomic_set(&nilfs->ns_ndirtyblks, 0);
+	init_rwsem(&nilfs->ns_sem);
+	mutex_init(&nilfs->ns_snapshot_mount_mutex);
+	INIT_LIST_HEAD(&nilfs->ns_dirty_files);
+	INIT_LIST_HEAD(&nilfs->ns_gc_inodes);
+	spin_lock_init(&nilfs->ns_inode_lock);
+	spin_lock_init(&nilfs->ns_last_segment_lock);
+	nilfs->ns_cptree = RB_ROOT;
+	spin_lock_init(&nilfs->ns_cptree_lock);
+	init_rwsem(&nilfs->ns_segctor_sem);
+	nilfs->ns_sb_update_freq = NILFS_SB_FREQ;
+
+	return nilfs;
+}
+
+/**
+ * destroy_nilfs - destroy nilfs object
+ * @nilfs: nilfs object to be released
+ */
+void destroy_nilfs(struct the_nilfs *nilfs)
+{
+	might_sleep();
+	if (nilfs_init(nilfs)) {
+		brelse(nilfs->ns_sbh[0]);
+		brelse(nilfs->ns_sbh[1]);
+	}
+	kfree(nilfs);
+}
+
+static int nilfs_load_super_root(struct the_nilfs *nilfs,
+				 struct super_block *sb, sector_t sr_block)
+{
+	struct buffer_head *bh_sr;
+	struct nilfs_super_root *raw_sr;
+	struct nilfs_super_block **sbp = nilfs->ns_sbp;
+	struct nilfs_inode *rawi;
+	unsigned int dat_entry_size, segment_usage_size, checkpoint_size;
+	unsigned int inode_size;
+	int err;
+
+	err = nilfs_read_super_root_block(nilfs, sr_block, &bh_sr, 1);
+	if (unlikely(err))
+		return err;
+
+	down_read(&nilfs->ns_sem);
+	dat_entry_size = le16_to_cpu(sbp[0]->s_dat_entry_size);
+	checkpoint_size = le16_to_cpu(sbp[0]->s_checkpoint_size);
+	segment_usage_size = le16_to_cpu(sbp[0]->s_segment_usage_size);
+	up_read(&nilfs->ns_sem);
+
+	inode_size = nilfs->ns_inode_size;
+
+	rawi = (void *)bh_sr->b_data + NILFS_SR_DAT_OFFSET(inode_size);
+	err = nilfs_dat_read(sb, dat_entry_size, rawi, &nilfs->ns_dat);
+	if (err)
+		goto failed;
+
+	rawi = (void *)bh_sr->b_data + NILFS_SR_CPFILE_OFFSET(inode_size);
+	err = nilfs_cpfile_read(sb, checkpoint_size, rawi, &nilfs->ns_cpfile);
+	if (err)
+		goto failed_dat;
+
+	rawi = (void *)bh_sr->b_data + NILFS_SR_SUFILE_OFFSET(inode_size);
+	err = nilfs_sufile_read(sb, segment_usage_size, rawi,
+				&nilfs->ns_sufile);
+	if (err)
+		goto failed_cpfile;
+
+	raw_sr = (struct nilfs_super_root *)bh_sr->b_data;
+	nilfs->ns_nongc_ctime = le64_to_cpu(raw_sr->sr_nongc_ctime);
+
+ failed:
+	brelse(bh_sr);
+	return err;
+
+ failed_cpfile:
+	iput(nilfs->ns_cpfile);
+
+ failed_dat:
+	iput(nilfs->ns_dat);
+	goto failed;
+}
+
+static void nilfs_init_recovery_info(struct nilfs_recovery_info *ri)
+{
+	memset(ri, 0, sizeof(*ri));
+	INIT_LIST_HEAD(&ri->ri_used_segments);
+}
+
+static void nilfs_clear_recovery_info(struct nilfs_recovery_info *ri)
+{
+	nilfs_dispose_segment_list(&ri->ri_used_segments);
+}
+
+/**
+ * nilfs_store_log_cursor - load log cursor from a super block
+ * @nilfs: nilfs object
+ * @sbp: buffer storing super block to be read
+ *
+ * nilfs_store_log_cursor() reads the last position of the log
+ * containing a super root from a given super block, and initializes
+ * relevant information on the nilfs object preparatory for log
+ * scanning and recovery.
+ *
+ * Return: 0 on success, or %-EINVAL if current segment number is out
+ * of range.
+ */
+static int nilfs_store_log_cursor(struct the_nilfs *nilfs,
+				  struct nilfs_super_block *sbp)
+{
+	int ret = 0;
+
+	nilfs->ns_last_pseg = le64_to_cpu(sbp->s_last_pseg);
+	nilfs->ns_last_cno = le64_to_cpu(sbp->s_last_cno);
+	nilfs->ns_last_seq = le64_to_cpu(sbp->s_last_seq);
+
+	nilfs->ns_prev_seq = nilfs->ns_last_seq;
+	nilfs->ns_seg_seq = nilfs->ns_last_seq;
+	nilfs->ns_segnum =
+		nilfs_get_segnum_of_block(nilfs, nilfs->ns_last_pseg);
+	nilfs->ns_cno = nilfs->ns_last_cno + 1;
+	if (nilfs->ns_segnum >= nilfs->ns_nsegments) {
+		nilfs_err(nilfs->ns_sb,
+			  "pointed segment number is out of range: segnum=%llu, nsegments=%lu",
+			  (unsigned long long)nilfs->ns_segnum,
+			  nilfs->ns_nsegments);
+		ret = -EINVAL;
+	}
+	return ret;
+}
+
+/**
+ * nilfs_get_blocksize - get block size from raw superblock data
+ * @sb: super block instance
+ * @sbp: superblock raw data buffer
+ * @blocksize: place to store block size
+ *
+ * nilfs_get_blocksize() calculates the block size from the block size
+ * exponent information written in @sbp and stores it in @blocksize,
+ * or aborts with an error message if it's too large.
+ *
+ * Return: 0 on success, or %-EINVAL if the block size is too large.
+ */
+static int nilfs_get_blocksize(struct super_block *sb,
+			       struct nilfs_super_block *sbp, int *blocksize)
+{
+	unsigned int shift_bits = le32_to_cpu(sbp->s_log_block_size);
+
+	if (unlikely(shift_bits >
+		     ilog2(NILFS_MAX_BLOCK_SIZE) - BLOCK_SIZE_BITS)) {
+		nilfs_err(sb, "too large filesystem blocksize: 2 ^ %u KiB",
+			  shift_bits);
+		return -EINVAL;
+	}
+	*blocksize = BLOCK_SIZE << shift_bits;
+	return 0;
+}
+
+/**
+ * load_nilfs - load and recover the nilfs
+ * @nilfs: the_nilfs structure to be released
+ * @sb: super block instance used to recover past segment
+ *
+ * load_nilfs() searches and load the latest super root,
+ * attaches the last segment, and does recovery if needed.
+ * The caller must call this exclusively for simultaneous mounts.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EINVAL	- No valid segment found.
+ * * %-EIO	- I/O error.
+ * * %-ENOMEM	- Insufficient memory available.
+ * * %-EROFS	- Read only device or RO compat mode (if recovery is required)
+ */
+int load_nilfs(struct the_nilfs *nilfs, struct super_block *sb)
+{
+	struct nilfs_recovery_info ri;
+	unsigned int s_flags = sb->s_flags;
+	int really_read_only = bdev_read_only(nilfs->ns_bdev);
+	int valid_fs = nilfs_valid_fs(nilfs);
+	int err;
+
+	if (!valid_fs) {
+		nilfs_warn(sb, "mounting unchecked fs");
+		if (s_flags & SB_RDONLY) {
+			nilfs_info(sb,
+				   "recovery required for readonly filesystem");
+			nilfs_info(sb,
+				   "write access will be enabled during recovery");
+		}
+	}
+
+	nilfs_init_recovery_info(&ri);
+
+	err = nilfs_search_super_root(nilfs, &ri);
+	if (unlikely(err)) {
+		struct nilfs_super_block **sbp = nilfs->ns_sbp;
+		int blocksize;
+
+		if (err != -EINVAL)
+			goto scan_error;
+
+		if (!nilfs_valid_sb(sbp[1])) {
+			nilfs_warn(sb,
+				   "unable to fall back to spare super block");
+			goto scan_error;
+		}
+		nilfs_info(sb, "trying rollback from an earlier position");
+
+		/*
+		 * restore super block with its spare and reconfigure
+		 * relevant states of the nilfs object.
+		 */
+		memcpy(sbp[0], sbp[1], nilfs->ns_sbsize);
+		nilfs->ns_crc_seed = le32_to_cpu(sbp[0]->s_crc_seed);
+		nilfs->ns_sbwtime = le64_to_cpu(sbp[0]->s_wtime);
+
+		/* verify consistency between two super blocks */
+		err = nilfs_get_blocksize(sb, sbp[0], &blocksize);
+		if (err)
+			goto scan_error;
+
+		if (blocksize != nilfs->ns_blocksize) {
+			nilfs_warn(sb,
+				   "blocksize differs between two super blocks (%d != %d)",
+				   blocksize, nilfs->ns_blocksize);
+			err = -EINVAL;
+			goto scan_error;
+		}
+
+		err = nilfs_store_log_cursor(nilfs, sbp[0]);
+		if (err)
+			goto scan_error;
+
+		/* drop clean flag to allow roll-forward and recovery */
+		nilfs->ns_mount_state &= ~NILFS_VALID_FS;
+		valid_fs = 0;
+
+		err = nilfs_search_super_root(nilfs, &ri);
+		if (err)
+			goto scan_error;
+	}
+
+	err = nilfs_load_super_root(nilfs, sb, ri.ri_super_root);
+	if (unlikely(err)) {
+		nilfs_err(sb, "error %d while loading super root", err);
+		goto failed;
+	}
+
+	err = nilfs_sysfs_create_device_group(sb);
+	if (unlikely(err))
+		goto sysfs_error;
+
+	if (valid_fs)
+		goto skip_recovery;
+
+	if (s_flags & SB_RDONLY) {
+		__u64 features;
+
+		if (nilfs_test_opt(nilfs, NORECOVERY)) {
+			nilfs_info(sb,
+				   "norecovery option specified, skipping roll-forward recovery");
+			goto skip_recovery;
+		}
+		features = le64_to_cpu(nilfs->ns_sbp[0]->s_feature_compat_ro) &
+			~NILFS_FEATURE_COMPAT_RO_SUPP;
+		if (features) {
+			nilfs_err(sb,
+				  "couldn't proceed with recovery because of unsupported optional features (%llx)",
+				  (unsigned long long)features);
+			err = -EROFS;
+			goto failed_unload;
+		}
+		if (really_read_only) {
+			nilfs_err(sb,
+				  "write access unavailable, cannot proceed");
+			err = -EROFS;
+			goto failed_unload;
+		}
+		sb->s_flags &= ~SB_RDONLY;
+	} else if (nilfs_test_opt(nilfs, NORECOVERY)) {
+		nilfs_err(sb,
+			  "recovery cancelled because norecovery option was specified for a read/write mount");
+		err = -EINVAL;
+		goto failed_unload;
+	}
+
+	err = nilfs_salvage_orphan_logs(nilfs, sb, &ri);
+	if (err)
+		goto failed_unload;
+
+	down_write(&nilfs->ns_sem);
+	nilfs->ns_mount_state |= NILFS_VALID_FS; /* set "clean" flag */
+	err = nilfs_cleanup_super(sb);
+	up_write(&nilfs->ns_sem);
+
+	if (err) {
+		nilfs_err(sb,
+			  "error %d updating super block. recovery unfinished.",
+			  err);
+		goto failed_unload;
+	}
+	nilfs_info(sb, "recovery complete");
+
+ skip_recovery:
+	nilfs_clear_recovery_info(&ri);
+	sb->s_flags = s_flags;
+	return 0;
+
+ scan_error:
+	nilfs_err(sb, "error %d while searching super root", err);
+	goto failed;
+
+ failed_unload:
+	nilfs_sysfs_delete_device_group(nilfs);
+
+ sysfs_error:
+	iput(nilfs->ns_cpfile);
+	iput(nilfs->ns_sufile);
+	iput(nilfs->ns_dat);
+
+ failed:
+	nilfs_clear_recovery_info(&ri);
+	sb->s_flags = s_flags;
+	return err;
+}
+
+static unsigned long long nilfs_max_size(unsigned int blkbits)
+{
+	unsigned int max_bits;
+	unsigned long long res = MAX_LFS_FILESIZE; /* page cache limit */
+
+	max_bits = blkbits + NILFS_BMAP_KEY_BIT; /* bmap size limit */
+	if (max_bits < 64)
+		res = min_t(unsigned long long, res, (1ULL << max_bits) - 1);
+	return res;
+}
+
+/**
+ * nilfs_nrsvsegs - calculate the number of reserved segments
+ * @nilfs: nilfs object
+ * @nsegs: total number of segments
+ *
+ * Return: Number of reserved segments.
+ */
+unsigned long nilfs_nrsvsegs(struct the_nilfs *nilfs, unsigned long nsegs)
+{
+	return max_t(unsigned long, NILFS_MIN_NRSVSEGS,
+		     DIV_ROUND_UP(nsegs * nilfs->ns_r_segments_percentage,
+				  100));
+}
+
+/**
+ * nilfs_max_segment_count - calculate the maximum number of segments
+ * @nilfs: nilfs object
+ *
+ * Return: Maximum number of segments
+ */
+static u64 nilfs_max_segment_count(struct the_nilfs *nilfs)
+{
+	u64 max_count = U64_MAX;
+
+	max_count = div64_ul(max_count, nilfs->ns_blocks_per_segment);
+	return min_t(u64, max_count, ULONG_MAX);
+}
+
+void nilfs_set_nsegments(struct the_nilfs *nilfs, unsigned long nsegs)
+{
+	nilfs->ns_nsegments = nsegs;
+	nilfs->ns_nrsvsegs = nilfs_nrsvsegs(nilfs, nsegs);
+}
+
+static int nilfs_store_disk_layout(struct the_nilfs *nilfs,
+				   struct nilfs_super_block *sbp)
+{
+	u64 nsegments, nblocks;
+
+	if (le32_to_cpu(sbp->s_rev_level) < NILFS_MIN_SUPP_REV) {
+		nilfs_err(nilfs->ns_sb,
+			  "unsupported revision (superblock rev.=%d.%d, current rev.=%d.%d). Please check the version of mkfs.nilfs(2).",
+			  le32_to_cpu(sbp->s_rev_level),
+			  le16_to_cpu(sbp->s_minor_rev_level),
+			  NILFS_CURRENT_REV, NILFS_MINOR_REV);
+		return -EINVAL;
+	}
+	nilfs->ns_sbsize = le16_to_cpu(sbp->s_bytes);
+	if (nilfs->ns_sbsize > BLOCK_SIZE)
+		return -EINVAL;
+
+	nilfs->ns_inode_size = le16_to_cpu(sbp->s_inode_size);
+	if (nilfs->ns_inode_size > nilfs->ns_blocksize) {
+		nilfs_err(nilfs->ns_sb, "too large inode size: %d bytes",
+			  nilfs->ns_inode_size);
+		return -EINVAL;
+	} else if (nilfs->ns_inode_size < NILFS_MIN_INODE_SIZE) {
+		nilfs_err(nilfs->ns_sb, "too small inode size: %d bytes",
+			  nilfs->ns_inode_size);
+		return -EINVAL;
+	}
+
+	nilfs->ns_first_ino = le32_to_cpu(sbp->s_first_ino);
+	if (nilfs->ns_first_ino < NILFS_USER_INO) {
+		nilfs_err(nilfs->ns_sb,
+			  "too small lower limit for non-reserved inode numbers: %u",
+			  nilfs->ns_first_ino);
+		return -EINVAL;
+	}
+
+	nilfs->ns_blocks_per_segment = le32_to_cpu(sbp->s_blocks_per_segment);
+	if (nilfs->ns_blocks_per_segment < NILFS_SEG_MIN_BLOCKS) {
+		nilfs_err(nilfs->ns_sb, "too short segment: %lu blocks",
+			  nilfs->ns_blocks_per_segment);
+		return -EINVAL;
+	}
+
+	nilfs->ns_first_data_block = le64_to_cpu(sbp->s_first_data_block);
+	nilfs->ns_r_segments_percentage =
+		le32_to_cpu(sbp->s_r_segments_percentage);
+	if (nilfs->ns_r_segments_percentage < 1 ||
+	    nilfs->ns_r_segments_percentage > 99) {
+		nilfs_err(nilfs->ns_sb,
+			  "invalid reserved segments percentage: %lu",
+			  nilfs->ns_r_segments_percentage);
+		return -EINVAL;
+	}
+
+	nsegments = le64_to_cpu(sbp->s_nsegments);
+	if (nsegments > nilfs_max_segment_count(nilfs)) {
+		nilfs_err(nilfs->ns_sb,
+			  "segment count %llu exceeds upper limit (%llu segments)",
+			  (unsigned long long)nsegments,
+			  (unsigned long long)nilfs_max_segment_count(nilfs));
+		return -EINVAL;
+	}
+
+	nblocks = sb_bdev_nr_blocks(nilfs->ns_sb);
+	if (nblocks) {
+		u64 min_block_count = nsegments * nilfs->ns_blocks_per_segment;
+		/*
+		 * To avoid failing to mount early device images without a
+		 * second superblock, exclude that block count from the
+		 * "min_block_count" calculation.
+		 */
+
+		if (nblocks < min_block_count) {
+			nilfs_err(nilfs->ns_sb,
+				  "total number of segment blocks %llu exceeds device size (%llu blocks)",
+				  (unsigned long long)min_block_count,
+				  (unsigned long long)nblocks);
+			return -EINVAL;
+		}
+	}
+
+	nilfs_set_nsegments(nilfs, nsegments);
+	nilfs->ns_crc_seed = le32_to_cpu(sbp->s_crc_seed);
+	return 0;
+}
+
+static int nilfs_valid_sb(struct nilfs_super_block *sbp)
+{
+	static unsigned char sum[4];
+	const int sumoff = offsetof(struct nilfs_super_block, s_sum);
+	size_t bytes;
+	u32 crc;
+
+	if (!sbp || le16_to_cpu(sbp->s_magic) != NILFS_SUPER_MAGIC)
+		return 0;
+	bytes = le16_to_cpu(sbp->s_bytes);
+	if (bytes < sumoff + 4 || bytes > BLOCK_SIZE)
+		return 0;
+	crc = crc32_le(le32_to_cpu(sbp->s_crc_seed), (unsigned char *)sbp,
+		       sumoff);
+	crc = crc32_le(crc, sum, 4);
+	crc = crc32_le(crc, (unsigned char *)sbp + sumoff + 4,
+		       bytes - sumoff - 4);
+	return crc == le32_to_cpu(sbp->s_sum);
+}
+
+/**
+ * nilfs_sb2_bad_offset - check the location of the second superblock
+ * @sbp: superblock raw data buffer
+ * @offset: byte offset of second superblock calculated from device size
+ *
+ * nilfs_sb2_bad_offset() checks if the position on the second
+ * superblock is valid or not based on the filesystem parameters
+ * stored in @sbp.  If @offset points to a location within the segment
+ * area, or if the parameters themselves are not normal, it is
+ * determined to be invalid.
+ *
+ * Return: true if invalid, false if valid.
+ */
+static bool nilfs_sb2_bad_offset(struct nilfs_super_block *sbp, u64 offset)
+{
+	unsigned int shift_bits = le32_to_cpu(sbp->s_log_block_size);
+	u32 blocks_per_segment = le32_to_cpu(sbp->s_blocks_per_segment);
+	u64 nsegments = le64_to_cpu(sbp->s_nsegments);
+	u64 index;
+
+	if (blocks_per_segment < NILFS_SEG_MIN_BLOCKS ||
+	    shift_bits > ilog2(NILFS_MAX_BLOCK_SIZE) - BLOCK_SIZE_BITS)
+		return true;
+
+	index = offset >> (shift_bits + BLOCK_SIZE_BITS);
+	do_div(index, blocks_per_segment);
+	return index < nsegments;
+}
+
+static void nilfs_release_super_block(struct the_nilfs *nilfs)
+{
+	int i;
+
+	for (i = 0; i < 2; i++) {
+		if (nilfs->ns_sbp[i]) {
+			brelse(nilfs->ns_sbh[i]);
+			nilfs->ns_sbh[i] = NULL;
+			nilfs->ns_sbp[i] = NULL;
+		}
+	}
+}
+
+void nilfs_fall_back_super_block(struct the_nilfs *nilfs)
+{
+	brelse(nilfs->ns_sbh[0]);
+	nilfs->ns_sbh[0] = nilfs->ns_sbh[1];
+	nilfs->ns_sbp[0] = nilfs->ns_sbp[1];
+	nilfs->ns_sbh[1] = NULL;
+	nilfs->ns_sbp[1] = NULL;
+}
+
+void nilfs_swap_super_block(struct the_nilfs *nilfs)
+{
+	struct buffer_head *tsbh = nilfs->ns_sbh[0];
+	struct nilfs_super_block *tsbp = nilfs->ns_sbp[0];
+
+	nilfs->ns_sbh[0] = nilfs->ns_sbh[1];
+	nilfs->ns_sbp[0] = nilfs->ns_sbp[1];
+	nilfs->ns_sbh[1] = tsbh;
+	nilfs->ns_sbp[1] = tsbp;
+}
+
+static int nilfs_load_super_block(struct the_nilfs *nilfs,
+				  struct super_block *sb, int blocksize,
+				  struct nilfs_super_block **sbpp)
+{
+	struct nilfs_super_block **sbp = nilfs->ns_sbp;
+	struct buffer_head **sbh = nilfs->ns_sbh;
+	u64 sb2off, devsize = bdev_nr_bytes(nilfs->ns_bdev);
+	int valid[2], swp = 0, older;
+
+	if (devsize < NILFS_SEG_MIN_BLOCKS * NILFS_MIN_BLOCK_SIZE + 4096) {
+		nilfs_err(sb, "device size too small");
+		return -EINVAL;
+	}
+	sb2off = NILFS_SB2_OFFSET_BYTES(devsize);
+
+	sbp[0] = nilfs_read_super_block(sb, NILFS_SB_OFFSET_BYTES, blocksize,
+					&sbh[0]);
+	sbp[1] = nilfs_read_super_block(sb, sb2off, blocksize, &sbh[1]);
+
+	if (!sbp[0]) {
+		if (!sbp[1]) {
+			nilfs_err(sb, "unable to read superblock");
+			return -EIO;
+		}
+		nilfs_warn(sb,
+			   "unable to read primary superblock (blocksize = %d)",
+			   blocksize);
+	} else if (!sbp[1]) {
+		nilfs_warn(sb,
+			   "unable to read secondary superblock (blocksize = %d)",
+			   blocksize);
+	}
+
+	/*
+	 * Compare two super blocks and set 1 in swp if the secondary
+	 * super block is valid and newer.  Otherwise, set 0 in swp.
+	 */
+	valid[0] = nilfs_valid_sb(sbp[0]);
+	valid[1] = nilfs_valid_sb(sbp[1]);
+	swp = valid[1] && (!valid[0] ||
+			   le64_to_cpu(sbp[1]->s_last_cno) >
+			   le64_to_cpu(sbp[0]->s_last_cno));
+
+	if (valid[swp] && nilfs_sb2_bad_offset(sbp[swp], sb2off)) {
+		brelse(sbh[1]);
+		sbh[1] = NULL;
+		sbp[1] = NULL;
+		valid[1] = 0;
+		swp = 0;
+	}
+	if (!valid[swp]) {
+		nilfs_release_super_block(nilfs);
+		nilfs_err(sb, "couldn't find nilfs on the device");
+		return -EINVAL;
+	}
+
+	if (!valid[!swp])
+		nilfs_warn(sb,
+			   "broken superblock, retrying with spare superblock (blocksize = %d)",
+			   blocksize);
+	if (swp)
+		nilfs_swap_super_block(nilfs);
+
+	/*
+	 * Calculate the array index of the older superblock data.
+	 * If one has been dropped, set index 0 pointing to the remaining one,
+	 * otherwise set index 1 pointing to the old one (including if both
+	 * are the same).
+	 *
+	 *  Divided case             valid[0]  valid[1]  swp  ->  older
+	 *  -------------------------------------------------------------
+	 *  Both SBs are invalid        0         0       N/A (Error)
+	 *  SB1 is invalid              0         1       1         0
+	 *  SB2 is invalid              1         0       0         0
+	 *  SB2 is newer                1         1       1         0
+	 *  SB2 is older or the same    1         1       0         1
+	 */
+	older = valid[1] ^ swp;
+
+	nilfs->ns_sbwcount = 0;
+	nilfs->ns_sbwtime = le64_to_cpu(sbp[0]->s_wtime);
+	nilfs->ns_prot_seq = le64_to_cpu(sbp[older]->s_last_seq);
+	*sbpp = sbp[0];
+	return 0;
+}
+
+/**
+ * init_nilfs - initialize a NILFS instance.
+ * @nilfs: the_nilfs structure
+ * @sb: super block
+ *
+ * init_nilfs() performs common initialization per block device (e.g.
+ * reading the super block, getting disk layout information, initializing
+ * shared fields in the_nilfs).
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int init_nilfs(struct the_nilfs *nilfs, struct super_block *sb)
+{
+	struct nilfs_super_block *sbp;
+	int blocksize;
+	int err;
+
+	blocksize = sb_min_blocksize(sb, NILFS_MIN_BLOCK_SIZE);
+	if (!blocksize) {
+		nilfs_err(sb, "unable to set blocksize");
+		err = -EINVAL;
+		goto out;
+	}
+	err = nilfs_load_super_block(nilfs, sb, blocksize, &sbp);
+	if (err)
+		goto out;
+
+	err = nilfs_store_magic(sb, sbp);
+	if (err)
+		goto failed_sbh;
+
+	err = nilfs_check_feature_compatibility(sb, sbp);
+	if (err)
+		goto failed_sbh;
+
+	err = nilfs_get_blocksize(sb, sbp, &blocksize);
+	if (err)
+		goto failed_sbh;
+
+	if (blocksize < NILFS_MIN_BLOCK_SIZE) {
+		nilfs_err(sb,
+			  "couldn't mount because of unsupported filesystem blocksize %d",
+			  blocksize);
+		err = -EINVAL;
+		goto failed_sbh;
+	}
+	if (sb->s_blocksize != blocksize) {
+		int hw_blocksize = bdev_logical_block_size(sb->s_bdev);
+
+		if (blocksize < hw_blocksize) {
+			nilfs_err(sb,
+				  "blocksize %d too small for device (sector-size = %d)",
+				  blocksize, hw_blocksize);
+			err = -EINVAL;
+			goto failed_sbh;
+		}
+		nilfs_release_super_block(nilfs);
+		if (!sb_set_blocksize(sb, blocksize)) {
+			nilfs_err(sb, "bad blocksize %d", blocksize);
+			err = -EINVAL;
+			goto out;
+		}
+
+		err = nilfs_load_super_block(nilfs, sb, blocksize, &sbp);
+		if (err)
+			goto out;
+			/*
+			 * Not to failed_sbh; sbh is released automatically
+			 * when reloading fails.
+			 */
+	}
+	nilfs->ns_blocksize_bits = sb->s_blocksize_bits;
+	nilfs->ns_blocksize = blocksize;
+
+	err = nilfs_store_disk_layout(nilfs, sbp);
+	if (err)
+		goto failed_sbh;
+
+	sb->s_maxbytes = nilfs_max_size(sb->s_blocksize_bits);
+
+	nilfs->ns_mount_state = le16_to_cpu(sbp->s_state);
+
+	err = nilfs_store_log_cursor(nilfs, sbp);
+	if (err)
+		goto failed_sbh;
+
+	set_nilfs_init(nilfs);
+	err = 0;
+ out:
+	return err;
+
+ failed_sbh:
+	nilfs_release_super_block(nilfs);
+	goto out;
+}
+
+int nilfs_discard_segments(struct the_nilfs *nilfs, __u64 *segnump,
+			    size_t nsegs)
+{
+	sector_t seg_start, seg_end;
+	sector_t start = 0, nblocks = 0;
+	unsigned int sects_per_block;
+	__u64 *sn;
+	int ret = 0;
+
+	sects_per_block = (1 << nilfs->ns_blocksize_bits) /
+		bdev_logical_block_size(nilfs->ns_bdev);
+	for (sn = segnump; sn < segnump + nsegs; sn++) {
+		nilfs_get_segment_range(nilfs, *sn, &seg_start, &seg_end);
+
+		if (!nblocks) {
+			start = seg_start;
+			nblocks = seg_end - seg_start + 1;
+		} else if (start + nblocks == seg_start) {
+			nblocks += seg_end - seg_start + 1;
+		} else {
+			ret = blkdev_issue_discard(nilfs->ns_bdev,
+						   start * sects_per_block,
+						   nblocks * sects_per_block,
+						   GFP_NOFS);
+			if (ret < 0)
+				return ret;
+			nblocks = 0;
+		}
+	}
+	if (nblocks)
+		ret = blkdev_issue_discard(nilfs->ns_bdev,
+					   start * sects_per_block,
+					   nblocks * sects_per_block,
+					   GFP_NOFS);
+	return ret;
+}
+
+int nilfs_count_free_blocks(struct the_nilfs *nilfs, sector_t *nblocks)
+{
+	unsigned long ncleansegs;
+
+	ncleansegs = nilfs_sufile_get_ncleansegs(nilfs->ns_sufile);
+	*nblocks = (sector_t)ncleansegs * nilfs->ns_blocks_per_segment;
+	return 0;
+}
+
+int nilfs_near_disk_full(struct the_nilfs *nilfs)
+{
+	unsigned long ncleansegs, nincsegs;
+
+	ncleansegs = nilfs_sufile_get_ncleansegs(nilfs->ns_sufile);
+	nincsegs = atomic_read(&nilfs->ns_ndirtyblks) /
+		nilfs->ns_blocks_per_segment + 1;
+
+	return ncleansegs <= nilfs->ns_nrsvsegs + nincsegs;
+}
+
+struct nilfs_root *nilfs_lookup_root(struct the_nilfs *nilfs, __u64 cno)
+{
+	struct rb_node *n;
+	struct nilfs_root *root;
+
+	spin_lock(&nilfs->ns_cptree_lock);
+	n = nilfs->ns_cptree.rb_node;
+	while (n) {
+		root = rb_entry(n, struct nilfs_root, rb_node);
+
+		if (cno < root->cno) {
+			n = n->rb_left;
+		} else if (cno > root->cno) {
+			n = n->rb_right;
+		} else {
+			refcount_inc(&root->count);
+			spin_unlock(&nilfs->ns_cptree_lock);
+			return root;
+		}
+	}
+	spin_unlock(&nilfs->ns_cptree_lock);
+
+	return NULL;
+}
+
+struct nilfs_root *
+nilfs_find_or_create_root(struct the_nilfs *nilfs, __u64 cno)
+{
+	struct rb_node **p, *parent;
+	struct nilfs_root *root, *new;
+	int err;
+
+	root = nilfs_lookup_root(nilfs, cno);
+	if (root)
+		return root;
+
+	new = kzalloc_obj(*root);
+	if (!new)
+		return NULL;
+
+	spin_lock(&nilfs->ns_cptree_lock);
+
+	p = &nilfs->ns_cptree.rb_node;
+	parent = NULL;
+
+	while (*p) {
+		parent = *p;
+		root = rb_entry(parent, struct nilfs_root, rb_node);
+
+		if (cno < root->cno) {
+			p = &(*p)->rb_left;
+		} else if (cno > root->cno) {
+			p = &(*p)->rb_right;
+		} else {
+			refcount_inc(&root->count);
+			spin_unlock(&nilfs->ns_cptree_lock);
+			kfree(new);
+			return root;
+		}
+	}
+
+	new->cno = cno;
+	new->ifile = NULL;
+	new->nilfs = nilfs;
+	refcount_set(&new->count, 1);
+	atomic64_set(&new->inodes_count, 0);
+	atomic64_set(&new->blocks_count, 0);
+
+	rb_link_node(&new->rb_node, parent, p);
+	rb_insert_color(&new->rb_node, &nilfs->ns_cptree);
+
+	spin_unlock(&nilfs->ns_cptree_lock);
+
+	err = nilfs_sysfs_create_snapshot_group(new);
+	if (err) {
+		kfree(new);
+		new = NULL;
+	}
+
+	return new;
+}
+
+void nilfs_put_root(struct nilfs_root *root)
+{
+	struct the_nilfs *nilfs = root->nilfs;
+
+	if (refcount_dec_and_lock(&root->count, &nilfs->ns_cptree_lock)) {
+		rb_erase(&root->rb_node, &nilfs->ns_cptree);
+		spin_unlock(&nilfs->ns_cptree_lock);
+
+		nilfs_sysfs_delete_snapshot_group(root);
+		iput(root->ifile);
+
+		kfree(root);
+	}
+}

linux-7.2.3/fs/nilfs2/the_nilfs.h

@@ -0,0 +1,376 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * the_nilfs shared structure.
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi.
+ *
+ */
+
+#ifndef _THE_NILFS_H
+#define _THE_NILFS_H
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include <linux/rbtree.h>
+#include <linux/fs.h>
+#include <linux/blkdev.h>
+#include <linux/backing-dev.h>
+#include <linux/slab.h>
+#include <linux/refcount.h>
+
+struct nilfs_sc_info;
+struct nilfs_sysfs_dev_subgroups;
+
+/* the_nilfs struct */
+enum {
+	THE_NILFS_INIT = 0,     /* Information from super_block is set */
+	THE_NILFS_DISCONTINUED,	/* 'next' pointer chain has broken */
+	THE_NILFS_GC_RUNNING,	/* gc process is running */
+	THE_NILFS_SB_DIRTY,	/* super block is dirty */
+	THE_NILFS_PURGING,	/* disposing dirty files for cleanup */
+};
+
+/**
+ * struct the_nilfs - struct to supervise multiple nilfs mount points
+ * @ns_flags: flags
+ * @ns_flushed_device: flag indicating if all volatile data was flushed
+ * @ns_sb: back pointer to super block instance
+ * @ns_bdev: block device
+ * @ns_sem: semaphore for shared states
+ * @ns_snapshot_mount_mutex: mutex to protect snapshot mounts
+ * @ns_sbh: buffer heads of on-disk super blocks
+ * @ns_sbp: pointers to super block data
+ * @ns_sbwtime: previous write time of super block
+ * @ns_sbwcount: write count of super block
+ * @ns_sbsize: size of valid data in super block
+ * @ns_mount_state: file system state
+ * @ns_sb_update_freq: interval of periodical update of superblocks (in seconds)
+ * @ns_seg_seq: segment sequence counter
+ * @ns_segnum: index number of the latest full segment.
+ * @ns_nextnum: index number of the full segment index to be used next
+ * @ns_pseg_offset: offset of next partial segment in the current full segment
+ * @ns_cno: next checkpoint number
+ * @ns_ctime: write time of the last segment
+ * @ns_nongc_ctime: write time of the last segment not for cleaner operation
+ * @ns_ndirtyblks: Number of dirty data blocks
+ * @ns_last_segment_lock: lock protecting fields for the latest segment
+ * @ns_last_pseg: start block number of the latest segment
+ * @ns_last_seq: sequence value of the latest segment
+ * @ns_last_cno: checkpoint number of the latest segment
+ * @ns_prot_seq: least sequence number of segments which must not be reclaimed
+ * @ns_prev_seq: base sequence number used to decide if advance log cursor
+ * @ns_writer: log writer
+ * @ns_segctor_sem: semaphore protecting log write
+ * @ns_dat: DAT file inode
+ * @ns_cpfile: checkpoint file inode
+ * @ns_sufile: segusage file inode
+ * @ns_cptree: rb-tree of all mounted checkpoints (nilfs_root)
+ * @ns_cptree_lock: lock protecting @ns_cptree
+ * @ns_dirty_files: list of dirty files
+ * @ns_inode_lock: lock protecting @ns_dirty_files
+ * @ns_gc_inodes: dummy inodes to keep live blocks
+ * @ns_mount_opt: mount options
+ * @ns_resuid: uid for reserved blocks
+ * @ns_resgid: gid for reserved blocks
+ * @ns_interval: checkpoint creation interval
+ * @ns_watermark: watermark for the number of dirty buffers
+ * @ns_blocksize_bits: bit length of block size
+ * @ns_blocksize: block size
+ * @ns_nsegments: number of segments in filesystem
+ * @ns_blocks_per_segment: number of blocks per segment
+ * @ns_r_segments_percentage: reserved segments percentage
+ * @ns_nrsvsegs: number of reserved segments
+ * @ns_first_data_block: block number of first data block
+ * @ns_inode_size: size of on-disk inode
+ * @ns_first_ino: first not-special inode number
+ * @ns_crc_seed: seed value of CRC32 calculation
+ * @ns_dev_kobj: /sys/fs/<nilfs>/<device>
+ * @ns_dev_kobj_unregister: completion state
+ * @ns_dev_subgroups: <device> subgroups pointer
+ */
+struct the_nilfs {
+	unsigned long		ns_flags;
+	int			ns_flushed_device;
+
+	struct super_block     *ns_sb;
+	struct block_device    *ns_bdev;
+	struct rw_semaphore	ns_sem;
+	struct mutex		ns_snapshot_mount_mutex;
+
+	/*
+	 * used for
+	 * - loading the latest checkpoint exclusively.
+	 * - allocating a new full segment.
+	 */
+	struct buffer_head     *ns_sbh[2];
+	struct nilfs_super_block *ns_sbp[2];
+	time64_t		ns_sbwtime;
+	unsigned int		ns_sbwcount;
+	unsigned int		ns_sbsize;
+	unsigned int		ns_mount_state;
+	unsigned int		ns_sb_update_freq;
+
+	/*
+	 * The following fields are updated by a writable FS-instance.
+	 * These fields are protected by ns_segctor_sem outside load_nilfs().
+	 */
+	u64			ns_seg_seq;
+	__u64			ns_segnum;
+	__u64			ns_nextnum;
+	unsigned long		ns_pseg_offset;
+	__u64			ns_cno;
+	time64_t		ns_ctime;
+	time64_t		ns_nongc_ctime;
+	atomic_t		ns_ndirtyblks;
+
+	/*
+	 * The following fields hold information on the latest partial segment
+	 * written to disk with a super root.  These fields are protected by
+	 * ns_last_segment_lock.
+	 */
+	spinlock_t		ns_last_segment_lock;
+	sector_t		ns_last_pseg;
+	u64			ns_last_seq;
+	__u64			ns_last_cno;
+	u64			ns_prot_seq;
+	u64			ns_prev_seq;
+
+	struct nilfs_sc_info   *ns_writer;
+	struct rw_semaphore	ns_segctor_sem;
+
+	/*
+	 * Following fields are lock free except for the period before
+	 * the_nilfs is initialized.
+	 */
+	struct inode	       *ns_dat;
+	struct inode	       *ns_cpfile;
+	struct inode	       *ns_sufile;
+
+	/* Checkpoint tree */
+	struct rb_root		ns_cptree;
+	spinlock_t		ns_cptree_lock;
+
+	/* Dirty inode list */
+	struct list_head	ns_dirty_files;
+	spinlock_t		ns_inode_lock;
+
+	/* GC inode list */
+	struct list_head	ns_gc_inodes;
+
+	/* Mount options */
+	unsigned long		ns_mount_opt;
+
+	uid_t			ns_resuid;
+	gid_t			ns_resgid;
+	unsigned long		ns_interval;
+	unsigned long		ns_watermark;
+
+	/* Disk layout information (static) */
+	unsigned int		ns_blocksize_bits;
+	unsigned int		ns_blocksize;
+	unsigned long		ns_nsegments;
+	unsigned long		ns_blocks_per_segment;
+	unsigned long		ns_r_segments_percentage;
+	unsigned long		ns_nrsvsegs;
+	unsigned long		ns_first_data_block;
+	int			ns_inode_size;
+	unsigned int		ns_first_ino;
+	u32			ns_crc_seed;
+
+	/* /sys/fs/<nilfs>/<device> */
+	struct kobject ns_dev_kobj;
+	struct completion ns_dev_kobj_unregister;
+	struct nilfs_sysfs_dev_subgroups *ns_dev_subgroups;
+};
+
+#define THE_NILFS_FNS(bit, name)					\
+static inline void set_nilfs_##name(struct the_nilfs *nilfs)		\
+{									\
+	set_bit(THE_NILFS_##bit, &(nilfs)->ns_flags);			\
+}									\
+static inline void clear_nilfs_##name(struct the_nilfs *nilfs)		\
+{									\
+	clear_bit(THE_NILFS_##bit, &(nilfs)->ns_flags);			\
+}									\
+static inline int nilfs_##name(struct the_nilfs *nilfs)			\
+{									\
+	return test_bit(THE_NILFS_##bit, &(nilfs)->ns_flags);		\
+}
+
+THE_NILFS_FNS(INIT, init)
+THE_NILFS_FNS(DISCONTINUED, discontinued)
+THE_NILFS_FNS(GC_RUNNING, gc_running)
+THE_NILFS_FNS(SB_DIRTY, sb_dirty)
+THE_NILFS_FNS(PURGING, purging)
+
+/*
+ * Mount option operations
+ */
+#define nilfs_clear_opt(nilfs, opt)  \
+	((nilfs)->ns_mount_opt &= ~NILFS_MOUNT_##opt)
+#define nilfs_set_opt(nilfs, opt)  \
+	((nilfs)->ns_mount_opt |= NILFS_MOUNT_##opt)
+#define nilfs_test_opt(nilfs, opt) ((nilfs)->ns_mount_opt & NILFS_MOUNT_##opt)
+
+/**
+ * struct nilfs_root - nilfs root object
+ * @cno: checkpoint number
+ * @rb_node: red-black tree node
+ * @count: refcount of this structure
+ * @nilfs: nilfs object
+ * @ifile: inode file
+ * @inodes_count: number of inodes
+ * @blocks_count: number of blocks
+ * @snapshot_kobj: /sys/fs/<nilfs>/<device>/mounted_snapshots/<snapshot>
+ * @snapshot_kobj_unregister: completion state for kernel object
+ */
+struct nilfs_root {
+	__u64 cno;
+	struct rb_node rb_node;
+
+	refcount_t count;
+	struct the_nilfs *nilfs;
+	struct inode *ifile;
+
+	atomic64_t inodes_count;
+	atomic64_t blocks_count;
+
+	/* /sys/fs/<nilfs>/<device>/mounted_snapshots/<snapshot> */
+	struct kobject snapshot_kobj;
+	struct completion snapshot_kobj_unregister;
+};
+
+/* Special checkpoint number */
+#define NILFS_CPTREE_CURRENT_CNO	0
+
+/* Minimum interval of periodical update of superblocks (in seconds) */
+#define NILFS_SB_FREQ		10
+
+static inline int nilfs_sb_need_update(struct the_nilfs *nilfs)
+{
+	u64 t = ktime_get_real_seconds();
+
+	return t < nilfs->ns_sbwtime ||
+		t > nilfs->ns_sbwtime + nilfs->ns_sb_update_freq;
+}
+
+static inline int nilfs_sb_will_flip(struct the_nilfs *nilfs)
+{
+	int flip_bits = nilfs->ns_sbwcount & 0x0FL;
+
+	return (flip_bits != 0x08 && flip_bits != 0x0F);
+}
+
+void nilfs_set_last_segment(struct the_nilfs *, sector_t, u64, __u64);
+struct the_nilfs *alloc_nilfs(struct super_block *sb);
+void destroy_nilfs(struct the_nilfs *nilfs);
+int init_nilfs(struct the_nilfs *nilfs, struct super_block *sb);
+int load_nilfs(struct the_nilfs *nilfs, struct super_block *sb);
+unsigned long nilfs_nrsvsegs(struct the_nilfs *nilfs, unsigned long nsegs);
+void nilfs_set_nsegments(struct the_nilfs *nilfs, unsigned long nsegs);
+int nilfs_discard_segments(struct the_nilfs *, __u64 *, size_t);
+int nilfs_count_free_blocks(struct the_nilfs *, sector_t *);
+struct nilfs_root *nilfs_lookup_root(struct the_nilfs *nilfs, __u64 cno);
+struct nilfs_root *nilfs_find_or_create_root(struct the_nilfs *nilfs,
+					     __u64 cno);
+void nilfs_put_root(struct nilfs_root *root);
+int nilfs_near_disk_full(struct the_nilfs *);
+void nilfs_fall_back_super_block(struct the_nilfs *);
+void nilfs_swap_super_block(struct the_nilfs *);
+
+
+static inline void nilfs_get_root(struct nilfs_root *root)
+{
+	refcount_inc(&root->count);
+}
+
+static inline int nilfs_valid_fs(struct the_nilfs *nilfs)
+{
+	unsigned int valid_fs;
+
+	down_read(&nilfs->ns_sem);
+	valid_fs = (nilfs->ns_mount_state & NILFS_VALID_FS);
+	up_read(&nilfs->ns_sem);
+	return valid_fs;
+}
+
+static inline void
+nilfs_get_segment_range(struct the_nilfs *nilfs, __u64 segnum,
+			sector_t *seg_start, sector_t *seg_end)
+{
+	*seg_start = (sector_t)nilfs->ns_blocks_per_segment * segnum;
+	*seg_end = *seg_start + nilfs->ns_blocks_per_segment - 1;
+	if (segnum == 0)
+		*seg_start = nilfs->ns_first_data_block;
+}
+
+static inline sector_t
+nilfs_get_segment_start_blocknr(struct the_nilfs *nilfs, __u64 segnum)
+{
+	return (segnum == 0) ? nilfs->ns_first_data_block :
+		(sector_t)nilfs->ns_blocks_per_segment * segnum;
+}
+
+static inline __u64
+nilfs_get_segnum_of_block(struct the_nilfs *nilfs, sector_t blocknr)
+{
+	sector_t segnum = blocknr;
+
+	sector_div(segnum, nilfs->ns_blocks_per_segment);
+	return segnum;
+}
+
+static inline void
+nilfs_terminate_segment(struct the_nilfs *nilfs, sector_t seg_start,
+			sector_t seg_end)
+{
+	/* terminate the current full segment (used in case of I/O-error) */
+	nilfs->ns_pseg_offset = seg_end - seg_start + 1;
+}
+
+static inline void nilfs_shift_to_next_segment(struct the_nilfs *nilfs)
+{
+	/* move forward with a full segment */
+	nilfs->ns_segnum = nilfs->ns_nextnum;
+	nilfs->ns_pseg_offset = 0;
+	nilfs->ns_seg_seq++;
+}
+
+static inline __u64 nilfs_last_cno(struct the_nilfs *nilfs)
+{
+	__u64 cno;
+
+	spin_lock(&nilfs->ns_last_segment_lock);
+	cno = nilfs->ns_last_cno;
+	spin_unlock(&nilfs->ns_last_segment_lock);
+	return cno;
+}
+
+static inline int nilfs_segment_is_active(struct the_nilfs *nilfs, __u64 n)
+{
+	return n == nilfs->ns_segnum || n == nilfs->ns_nextnum;
+}
+
+static inline int nilfs_flush_device(struct the_nilfs *nilfs)
+{
+	int err;
+
+	if (!nilfs_test_opt(nilfs, BARRIER) || nilfs->ns_flushed_device)
+		return 0;
+
+	nilfs->ns_flushed_device = 1;
+	/*
+	 * the store to ns_flushed_device must not be reordered after
+	 * blkdev_issue_flush().
+	 */
+	smp_wmb();
+
+	err = blkdev_issue_flush(nilfs->ns_bdev);
+	if (err != -EIO)
+		err = 0;
+	return err;
+}
+
+#endif /* _THE_NILFS_H */