#include "sysfile.h"
#include "dlmglue.h"
#include "extent_map.h"
+#include "aops.h"
+#include "xattr.h"
+#include "namei.h"
+
+#include <linux/bio.h>
+#include <linux/blkdev.h>
+#include <linux/gfp.h>
+#include <linux/slab.h>
+#include <linux/writeback.h>
+#include <linux/pagevec.h>
+#include <linux/swap.h>
+#include <linux/security.h>
+#include <linux/fsnotify.h>
+#include <linux/quotaops.h>
+#include <linux/namei.h>
+#include <linux/mount.h>
+
+struct ocfs2_cow_context {
+ struct inode *inode;
+ u32 cow_start;
+ u32 cow_len;
+ struct ocfs2_extent_tree data_et;
+ struct ocfs2_refcount_tree *ref_tree;
+ struct buffer_head *ref_root_bh;
+ struct ocfs2_alloc_context *meta_ac;
+ struct ocfs2_alloc_context *data_ac;
+ struct ocfs2_cached_dealloc_ctxt dealloc;
+ void *cow_object;
+ struct ocfs2_post_refcount *post_refcount;
+ int extra_credits;
+ int (*get_clusters)(struct ocfs2_cow_context *context,
+ u32 v_cluster, u32 *p_cluster,
+ u32 *num_clusters,
+ unsigned int *extent_flags);
+ int (*cow_duplicate_clusters)(handle_t *handle,
+ struct ocfs2_cow_context *context,
+ u32 cpos, u32 old_cluster,
+ u32 new_cluster, u32 new_len);
+};
static inline struct ocfs2_refcount_tree *
cache_info_to_refcount(struct ocfs2_caching_info *ci)
}
/*
+ * Try to remove refcount tree. The mechanism is:
+ * 1) Check whether i_clusters == 0, if no, exit.
+ * 2) check whether we have i_xattr_loc in dinode. if yes, exit.
+ * 3) Check whether we have inline xattr stored outside, if yes, exit.
+ * 4) Remove the tree.
+ */
+int ocfs2_try_remove_refcount_tree(struct inode *inode,
+ struct buffer_head *di_bh)
+{
+ int ret;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+
+ down_write(&oi->ip_xattr_sem);
+ down_write(&oi->ip_alloc_sem);
+
+ if (oi->ip_clusters)
+ goto out;
+
+ if ((oi->ip_dyn_features & OCFS2_HAS_XATTR_FL) && di->i_xattr_loc)
+ goto out;
+
+ if (oi->ip_dyn_features & OCFS2_INLINE_XATTR_FL &&
+ ocfs2_has_inline_xattr_value_outside(inode, di))
+ goto out;
+
+ ret = ocfs2_remove_refcount_tree(inode, di_bh);
+ if (ret)
+ mlog_errno(ret);
+out:
+ up_write(&oi->ip_alloc_sem);
+ up_write(&oi->ip_xattr_sem);
+ return 0;
+}
+
+/*
* Given a cpos and len, try to find the refcount record which contains cpos.
* 1. If cpos can be found in one refcount record, return the record.
* 2. If cpos can't be found, return a fake record which start from cpos
static int ocfs2_change_refcount_rec(handle_t *handle,
struct ocfs2_caching_info *ci,
struct buffer_head *ref_leaf_bh,
- int index, int change)
+ int index, int merge, int change)
{
int ret;
struct ocfs2_refcount_block *rb =
}
le16_add_cpu(&rl->rl_used, -1);
- } else
+ } else if (merge)
ocfs2_refcount_rec_merge(rb, index);
ret = ocfs2_journal_dirty(handle, ref_leaf_bh);
struct buffer_head *ref_root_bh,
struct buffer_head *ref_leaf_bh,
struct ocfs2_refcount_rec *rec,
- int index,
+ int index, int merge,
struct ocfs2_alloc_context *meta_ac)
{
int ret;
le16_add_cpu(&rf_list->rl_used, 1);
- ocfs2_refcount_rec_merge(rb, index);
+ if (merge)
+ ocfs2_refcount_rec_merge(rb, index);
ret = ocfs2_journal_dirty(handle, ref_leaf_bh);
if (ret) {
struct buffer_head *ref_root_bh,
struct buffer_head *ref_leaf_bh,
struct ocfs2_refcount_rec *split_rec,
- int index,
+ int index, int merge,
struct ocfs2_alloc_context *meta_ac,
struct ocfs2_cached_dealloc_ctxt *dealloc)
{
le32_to_cpu(split_rec->r_refcount),
(unsigned long long)ref_leaf_bh->b_blocknr, index);
- ocfs2_refcount_rec_merge(rb, index);
+ if (merge)
+ ocfs2_refcount_rec_merge(rb, index);
}
ret = ocfs2_journal_dirty(handle, ref_leaf_bh);
static int __ocfs2_increase_refcount(handle_t *handle,
struct ocfs2_caching_info *ci,
struct buffer_head *ref_root_bh,
- u64 cpos, u32 len,
+ u64 cpos, u32 len, int merge,
struct ocfs2_alloc_context *meta_ac,
struct ocfs2_cached_dealloc_ctxt *dealloc)
{
"count %u\n", (unsigned long long)cpos, set_len,
le32_to_cpu(rec.r_refcount));
ret = ocfs2_change_refcount_rec(handle, ci,
- ref_leaf_bh, index, 1);
+ ref_leaf_bh, index,
+ merge, 1);
if (ret) {
mlog_errno(ret);
goto out;
set_len);
ret = ocfs2_insert_refcount_rec(handle, ci, ref_root_bh,
ref_leaf_bh,
- &rec, index, meta_ac);
+ &rec, index,
+ merge, meta_ac);
if (ret) {
mlog_errno(ret);
goto out;
set_len, le32_to_cpu(rec.r_refcount));
ret = ocfs2_split_refcount_rec(handle, ci,
ref_root_bh, ref_leaf_bh,
- &rec, index,
+ &rec, index, merge,
meta_ac, dealloc);
if (ret) {
mlog_errno(ret);
return ret;
}
+int ocfs2_increase_refcount(handle_t *handle,
+ struct ocfs2_caching_info *ci,
+ struct buffer_head *ref_root_bh,
+ u64 cpos, u32 len,
+ struct ocfs2_alloc_context *meta_ac,
+ struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+ return __ocfs2_increase_refcount(handle, ci, ref_root_bh,
+ cpos, len, 1,
+ meta_ac, dealloc);
+}
+
static int ocfs2_decrease_refcount_rec(handle_t *handle,
struct ocfs2_caching_info *ci,
struct buffer_head *ref_root_bh,
if (cpos == le64_to_cpu(rec->r_cpos) &&
len == le32_to_cpu(rec->r_clusters))
ret = ocfs2_change_refcount_rec(handle, ci,
- ref_leaf_bh, index, -1);
+ ref_leaf_bh, index, 1, -1);
else {
struct ocfs2_refcount_rec split = *rec;
split.r_cpos = cpu_to_le64(cpos);
le32_to_cpu(rec->r_clusters));
ret = ocfs2_split_refcount_rec(handle, ci,
ref_root_bh, ref_leaf_bh,
- &split, index,
+ &split, index, 1,
meta_ac, dealloc);
}
struct buffer_head *ref_root_bh,
u64 cpos, u32 len,
struct ocfs2_alloc_context *meta_ac,
- struct ocfs2_cached_dealloc_ctxt *dealloc)
+ struct ocfs2_cached_dealloc_ctxt *dealloc,
+ int delete)
{
int ret = 0, index = 0;
struct ocfs2_refcount_rec rec;
struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
struct buffer_head *ref_leaf_bh = NULL;
- mlog(0, "Tree owner %llu, decrease refcount start %llu, len %u\n",
+ mlog(0, "Tree owner %llu, decrease refcount start %llu, "
+ "len %u, delete %u\n",
(unsigned long long)ocfs2_metadata_cache_owner(ci),
- (unsigned long long)cpos, len);
+ (unsigned long long)cpos, len, delete);
while (len) {
ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
r_count = le32_to_cpu(rec.r_refcount);
BUG_ON(r_count == 0);
+ if (!delete)
+ BUG_ON(r_count > 1);
r_len = min((u64)(cpos + len), le64_to_cpu(rec.r_cpos) +
le32_to_cpu(rec.r_clusters)) - cpos;
goto out;
}
- if (le32_to_cpu(rec.r_refcount) == 1) {
+ if (le32_to_cpu(rec.r_refcount) == 1 && delete) {
ret = ocfs2_cache_cluster_dealloc(dealloc,
ocfs2_clusters_to_blocks(sb, cpos),
r_len);
int ocfs2_decrease_refcount(struct inode *inode,
handle_t *handle, u32 cpos, u32 len,
struct ocfs2_alloc_context *meta_ac,
- struct ocfs2_cached_dealloc_ctxt *dealloc)
+ struct ocfs2_cached_dealloc_ctxt *dealloc,
+ int delete)
{
int ret;
u64 ref_blkno;
}
ret = __ocfs2_decrease_refcount(handle, &tree->rf_ci, ref_root_bh,
- cpos, len, meta_ac, dealloc);
+ cpos, len, meta_ac, dealloc, delete);
+ if (ret)
+ mlog_errno(ret);
+out:
+ brelse(ref_root_bh);
+ return ret;
+}
+
+/*
+ * Mark the already-existing extent at cpos as refcounted for len clusters.
+ * This adds the refcount extent flag.
+ *
+ * If the existing extent is larger than the request, initiate a
+ * split. An attempt will be made at merging with adjacent extents.
+ *
+ * The caller is responsible for passing down meta_ac if we'll need it.
+ */
+static int ocfs2_mark_extent_refcounted(struct inode *inode,
+ struct ocfs2_extent_tree *et,
+ handle_t *handle, u32 cpos,
+ u32 len, u32 phys,
+ struct ocfs2_alloc_context *meta_ac,
+ struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+ int ret;
+
+ mlog(0, "Inode %lu refcount tree cpos %u, len %u, phys cluster %u\n",
+ inode->i_ino, cpos, len, phys);
+
+ if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb))) {
+ ocfs2_error(inode->i_sb, "Inode %lu want to use refcount "
+ "tree, but the feature bit is not set in the "
+ "super block.", inode->i_ino);
+ ret = -EROFS;
+ goto out;
+ }
+
+ ret = ocfs2_change_extent_flag(handle, et, cpos,
+ len, phys, meta_ac, dealloc,
+ OCFS2_EXT_REFCOUNTED, 0);
if (ret)
mlog_errno(ret);
+
+out:
+ return ret;
+}
+
+/*
+ * Given some contiguous physical clusters, calculate what we need
+ * for modifying their refcount.
+ */
+static int ocfs2_calc_refcount_meta_credits(struct super_block *sb,
+ struct ocfs2_caching_info *ci,
+ struct buffer_head *ref_root_bh,
+ u64 start_cpos,
+ u32 clusters,
+ int *meta_add,
+ int *credits)
+{
+ int ret = 0, index, ref_blocks = 0, recs_add = 0;
+ u64 cpos = start_cpos;
+ struct ocfs2_refcount_block *rb;
+ struct ocfs2_refcount_rec rec;
+ struct buffer_head *ref_leaf_bh = NULL, *prev_bh = NULL;
+ u32 len;
+
+ mlog(0, "start_cpos %llu, clusters %u\n",
+ (unsigned long long)start_cpos, clusters);
+ while (clusters) {
+ ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
+ cpos, clusters, &rec,
+ &index, &ref_leaf_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ if (ref_leaf_bh != prev_bh) {
+ /*
+ * Now we encounter a new leaf block, so calculate
+ * whether we need to extend the old leaf.
+ */
+ if (prev_bh) {
+ rb = (struct ocfs2_refcount_block *)
+ prev_bh->b_data;
+
+ if (le64_to_cpu(rb->rf_records.rl_used) +
+ recs_add >
+ le16_to_cpu(rb->rf_records.rl_count))
+ ref_blocks++;
+ }
+
+ recs_add = 0;
+ *credits += 1;
+ brelse(prev_bh);
+ prev_bh = ref_leaf_bh;
+ get_bh(prev_bh);
+ }
+
+ rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+
+ mlog(0, "recs_add %d,cpos %llu, clusters %u, rec->r_cpos %llu,"
+ "rec->r_clusters %u, rec->r_refcount %u, index %d\n",
+ recs_add, (unsigned long long)cpos, clusters,
+ (unsigned long long)le64_to_cpu(rec.r_cpos),
+ le32_to_cpu(rec.r_clusters),
+ le32_to_cpu(rec.r_refcount), index);
+
+ len = min((u64)cpos + clusters, le64_to_cpu(rec.r_cpos) +
+ le32_to_cpu(rec.r_clusters)) - cpos;
+ /*
+ * If the refcount rec already exist, cool. We just need
+ * to check whether there is a split. Otherwise we just need
+ * to increase the refcount.
+ * If we will insert one, increases recs_add.
+ *
+ * We record all the records which will be inserted to the
+ * same refcount block, so that we can tell exactly whether
+ * we need a new refcount block or not.
+ */
+ if (rec.r_refcount) {
+ /* Check whether we need a split at the beginning. */
+ if (cpos == start_cpos &&
+ cpos != le64_to_cpu(rec.r_cpos))
+ recs_add++;
+
+ /* Check whether we need a split in the end. */
+ if (cpos + clusters < le64_to_cpu(rec.r_cpos) +
+ le32_to_cpu(rec.r_clusters))
+ recs_add++;
+ } else
+ recs_add++;
+
+ brelse(ref_leaf_bh);
+ ref_leaf_bh = NULL;
+ clusters -= len;
+ cpos += len;
+ }
+
+ if (prev_bh) {
+ rb = (struct ocfs2_refcount_block *)prev_bh->b_data;
+
+ if (le64_to_cpu(rb->rf_records.rl_used) + recs_add >
+ le16_to_cpu(rb->rf_records.rl_count))
+ ref_blocks++;
+
+ *credits += 1;
+ }
+
+ if (!ref_blocks)
+ goto out;
+
+ mlog(0, "we need ref_blocks %d\n", ref_blocks);
+ *meta_add += ref_blocks;
+ *credits += ref_blocks;
+
+ /*
+ * So we may need ref_blocks to insert into the tree.
+ * That also means we need to change the b-tree and add that number
+ * of records since we never merge them.
+ * We need one more block for expansion since the new created leaf
+ * block is also full and needs split.
+ */
+ rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+ if (le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL) {
+ struct ocfs2_extent_tree et;
+
+ ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);
+ *meta_add += ocfs2_extend_meta_needed(et.et_root_el);
+ *credits += ocfs2_calc_extend_credits(sb,
+ et.et_root_el,
+ ref_blocks);
+ } else {
+ *credits += OCFS2_EXPAND_REFCOUNT_TREE_CREDITS;
+ *meta_add += 1;
+ }
+
+out:
+ brelse(ref_leaf_bh);
+ brelse(prev_bh);
+ return ret;
+}
+
+/*
+ * For refcount tree, we will decrease some contiguous clusters
+ * refcount count, so just go through it to see how many blocks
+ * we gonna touch and whether we need to create new blocks.
+ *
+ * Normally the refcount blocks store these refcount should be
+ * contiguous also, so that we can get the number easily.
+ * As for meta_ac, we will at most add split 2 refcount record and
+ * 2 more refcount block, so just check it in a rough way.
+ *
+ * Caller must hold refcount tree lock.
+ */
+int ocfs2_prepare_refcount_change_for_del(struct inode *inode,
+ struct buffer_head *di_bh,
+ u64 phys_blkno,
+ u32 clusters,
+ int *credits,
+ struct ocfs2_alloc_context **meta_ac)
+{
+ int ret, ref_blocks = 0;
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ struct buffer_head *ref_root_bh = NULL;
+ struct ocfs2_refcount_tree *tree;
+ u64 start_cpos = ocfs2_blocks_to_clusters(inode->i_sb, phys_blkno);
+
+ if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb))) {
+ ocfs2_error(inode->i_sb, "Inode %lu want to use refcount "
+ "tree, but the feature bit is not set in the "
+ "super block.", inode->i_ino);
+ ret = -EROFS;
+ goto out;
+ }
+
+ BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+
+ ret = ocfs2_get_refcount_tree(OCFS2_SB(inode->i_sb),
+ le64_to_cpu(di->i_refcount_loc), &tree);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_read_refcount_block(&tree->rf_ci,
+ le64_to_cpu(di->i_refcount_loc),
+ &ref_root_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_calc_refcount_meta_credits(inode->i_sb,
+ &tree->rf_ci,
+ ref_root_bh,
+ start_cpos, clusters,
+ &ref_blocks, credits);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ mlog(0, "reserve new metadata %d, credits = %d\n",
+ ref_blocks, *credits);
+
+ if (ref_blocks) {
+ ret = ocfs2_reserve_new_metadata_blocks(OCFS2_SB(inode->i_sb),
+ ref_blocks, meta_ac);
+ if (ret)
+ mlog_errno(ret);
+ }
+
out:
brelse(ref_root_bh);
return ret;
}
+
+#define MAX_CONTIG_BYTES 1048576
+
+static inline unsigned int ocfs2_cow_contig_clusters(struct super_block *sb)
+{
+ return ocfs2_clusters_for_bytes(sb, MAX_CONTIG_BYTES);
+}
+
+static inline unsigned int ocfs2_cow_contig_mask(struct super_block *sb)
+{
+ return ~(ocfs2_cow_contig_clusters(sb) - 1);
+}
+
+/*
+ * Given an extent that starts at 'start' and an I/O that starts at 'cpos',
+ * find an offset (start + (n * contig_clusters)) that is closest to cpos
+ * while still being less than or equal to it.
+ *
+ * The goal is to break the extent at a multiple of contig_clusters.
+ */
+static inline unsigned int ocfs2_cow_align_start(struct super_block *sb,
+ unsigned int start,
+ unsigned int cpos)
+{
+ BUG_ON(start > cpos);
+
+ return start + ((cpos - start) & ocfs2_cow_contig_mask(sb));
+}
+
+/*
+ * Given a cluster count of len, pad it out so that it is a multiple
+ * of contig_clusters.
+ */
+static inline unsigned int ocfs2_cow_align_length(struct super_block *sb,
+ unsigned int len)
+{
+ unsigned int padded =
+ (len + (ocfs2_cow_contig_clusters(sb) - 1)) &
+ ocfs2_cow_contig_mask(sb);
+
+ /* Did we wrap? */
+ if (padded < len)
+ padded = UINT_MAX;
+
+ return padded;
+}
+
+/*
+ * Calculate out the start and number of virtual clusters we need to to CoW.
+ *
+ * cpos is vitual start cluster position we want to do CoW in a
+ * file and write_len is the cluster length.
+ * max_cpos is the place where we want to stop CoW intentionally.
+ *
+ * Normal we will start CoW from the beginning of extent record cotaining cpos.
+ * We try to break up extents on boundaries of MAX_CONTIG_BYTES so that we
+ * get good I/O from the resulting extent tree.
+ */
+static int ocfs2_refcount_cal_cow_clusters(struct inode *inode,
+ struct ocfs2_extent_list *el,
+ u32 cpos,
+ u32 write_len,
+ u32 max_cpos,
+ u32 *cow_start,
+ u32 *cow_len)
+{
+ int ret = 0;
+ int tree_height = le16_to_cpu(el->l_tree_depth), i;
+ struct buffer_head *eb_bh = NULL;
+ struct ocfs2_extent_block *eb = NULL;
+ struct ocfs2_extent_rec *rec;
+ unsigned int want_clusters, rec_end = 0;
+ int contig_clusters = ocfs2_cow_contig_clusters(inode->i_sb);
+ int leaf_clusters;
+
+ BUG_ON(cpos + write_len > max_cpos);
+
+ if (tree_height > 0) {
+ ret = ocfs2_find_leaf(INODE_CACHE(inode), el, cpos, &eb_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ eb = (struct ocfs2_extent_block *) eb_bh->b_data;
+ el = &eb->h_list;
+
+ if (el->l_tree_depth) {
+ ocfs2_error(inode->i_sb,
+ "Inode %lu has non zero tree depth in "
+ "leaf block %llu\n", inode->i_ino,
+ (unsigned long long)eb_bh->b_blocknr);
+ ret = -EROFS;
+ goto out;
+ }
+ }
+
+ *cow_len = 0;
+ for (i = 0; i < le16_to_cpu(el->l_next_free_rec); i++) {
+ rec = &el->l_recs[i];
+
+ if (ocfs2_is_empty_extent(rec)) {
+ mlog_bug_on_msg(i != 0, "Inode %lu has empty record in "
+ "index %d\n", inode->i_ino, i);
+ continue;
+ }
+
+ if (le32_to_cpu(rec->e_cpos) +
+ le16_to_cpu(rec->e_leaf_clusters) <= cpos)
+ continue;
+
+ if (*cow_len == 0) {
+ /*
+ * We should find a refcounted record in the
+ * first pass.
+ */
+ BUG_ON(!(rec->e_flags & OCFS2_EXT_REFCOUNTED));
+ *cow_start = le32_to_cpu(rec->e_cpos);
+ }
+
+ /*
+ * If we encounter a hole, a non-refcounted record or
+ * pass the max_cpos, stop the search.
+ */
+ if ((!(rec->e_flags & OCFS2_EXT_REFCOUNTED)) ||
+ (*cow_len && rec_end != le32_to_cpu(rec->e_cpos)) ||
+ (max_cpos <= le32_to_cpu(rec->e_cpos)))
+ break;
+
+ leaf_clusters = le16_to_cpu(rec->e_leaf_clusters);
+ rec_end = le32_to_cpu(rec->e_cpos) + leaf_clusters;
+ if (rec_end > max_cpos) {
+ rec_end = max_cpos;
+ leaf_clusters = rec_end - le32_to_cpu(rec->e_cpos);
+ }
+
+ /*
+ * How many clusters do we actually need from
+ * this extent? First we see how many we actually
+ * need to complete the write. If that's smaller
+ * than contig_clusters, we try for contig_clusters.
+ */
+ if (!*cow_len)
+ want_clusters = write_len;
+ else
+ want_clusters = (cpos + write_len) -
+ (*cow_start + *cow_len);
+ if (want_clusters < contig_clusters)
+ want_clusters = contig_clusters;
+
+ /*
+ * If the write does not cover the whole extent, we
+ * need to calculate how we're going to split the extent.
+ * We try to do it on contig_clusters boundaries.
+ *
+ * Any extent smaller than contig_clusters will be
+ * CoWed in its entirety.
+ */
+ if (leaf_clusters <= contig_clusters)
+ *cow_len += leaf_clusters;
+ else if (*cow_len || (*cow_start == cpos)) {
+ /*
+ * This extent needs to be CoW'd from its
+ * beginning, so all we have to do is compute
+ * how many clusters to grab. We align
+ * want_clusters to the edge of contig_clusters
+ * to get better I/O.
+ */
+ want_clusters = ocfs2_cow_align_length(inode->i_sb,
+ want_clusters);
+
+ if (leaf_clusters < want_clusters)
+ *cow_len += leaf_clusters;
+ else
+ *cow_len += want_clusters;
+ } else if ((*cow_start + contig_clusters) >=
+ (cpos + write_len)) {
+ /*
+ * Breaking off contig_clusters at the front
+ * of the extent will cover our write. That's
+ * easy.
+ */
+ *cow_len = contig_clusters;
+ } else if ((rec_end - cpos) <= contig_clusters) {
+ /*
+ * Breaking off contig_clusters at the tail of
+ * this extent will cover cpos.
+ */
+ *cow_start = rec_end - contig_clusters;
+ *cow_len = contig_clusters;
+ } else if ((rec_end - cpos) <= want_clusters) {
+ /*
+ * While we can't fit the entire write in this
+ * extent, we know that the write goes from cpos
+ * to the end of the extent. Break that off.
+ * We try to break it at some multiple of
+ * contig_clusters from the front of the extent.
+ * Failing that (ie, cpos is within
+ * contig_clusters of the front), we'll CoW the
+ * entire extent.
+ */
+ *cow_start = ocfs2_cow_align_start(inode->i_sb,
+ *cow_start, cpos);
+ *cow_len = rec_end - *cow_start;
+ } else {
+ /*
+ * Ok, the entire write lives in the middle of
+ * this extent. Let's try to slice the extent up
+ * nicely. Optimally, our CoW region starts at
+ * m*contig_clusters from the beginning of the
+ * extent and goes for n*contig_clusters,
+ * covering the entire write.
+ */
+ *cow_start = ocfs2_cow_align_start(inode->i_sb,
+ *cow_start, cpos);
+
+ want_clusters = (cpos + write_len) - *cow_start;
+ want_clusters = ocfs2_cow_align_length(inode->i_sb,
+ want_clusters);
+ if (*cow_start + want_clusters <= rec_end)
+ *cow_len = want_clusters;
+ else
+ *cow_len = rec_end - *cow_start;
+ }
+
+ /* Have we covered our entire write yet? */
+ if ((*cow_start + *cow_len) >= (cpos + write_len))
+ break;
+
+ /*
+ * If we reach the end of the extent block and don't get enough
+ * clusters, continue with the next extent block if possible.
+ */
+ if (i + 1 == le16_to_cpu(el->l_next_free_rec) &&
+ eb && eb->h_next_leaf_blk) {
+ brelse(eb_bh);
+ eb_bh = NULL;
+
+ ret = ocfs2_read_extent_block(INODE_CACHE(inode),
+ le64_to_cpu(eb->h_next_leaf_blk),
+ &eb_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ eb = (struct ocfs2_extent_block *) eb_bh->b_data;
+ el = &eb->h_list;
+ i = -1;
+ }
+ }
+
+out:
+ brelse(eb_bh);
+ return ret;
+}
+
+/*
+ * Prepare meta_ac, data_ac and calculate credits when we want to add some
+ * num_clusters in data_tree "et" and change the refcount for the old
+ * clusters(starting form p_cluster) in the refcount tree.
+ *
+ * Note:
+ * 1. since we may split the old tree, so we at most will need num_clusters + 2
+ * more new leaf records.
+ * 2. In some case, we may not need to reserve new clusters(e.g, reflink), so
+ * just give data_ac = NULL.
+ */
+static int ocfs2_lock_refcount_allocators(struct super_block *sb,
+ u32 p_cluster, u32 num_clusters,
+ struct ocfs2_extent_tree *et,
+ struct ocfs2_caching_info *ref_ci,
+ struct buffer_head *ref_root_bh,
+ struct ocfs2_alloc_context **meta_ac,
+ struct ocfs2_alloc_context **data_ac,
+ int *credits)
+{
+ int ret = 0, meta_add = 0;
+ int num_free_extents = ocfs2_num_free_extents(OCFS2_SB(sb), et);
+
+ if (num_free_extents < 0) {
+ ret = num_free_extents;
+ mlog_errno(ret);
+ goto out;
+ }
+
+ if (num_free_extents < num_clusters + 2)
+ meta_add =
+ ocfs2_extend_meta_needed(et->et_root_el);
+
+ *credits += ocfs2_calc_extend_credits(sb, et->et_root_el,
+ num_clusters + 2);
+
+ ret = ocfs2_calc_refcount_meta_credits(sb, ref_ci, ref_root_bh,
+ p_cluster, num_clusters,
+ &meta_add, credits);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ mlog(0, "reserve new metadata %d, clusters %u, credits = %d\n",
+ meta_add, num_clusters, *credits);
+ ret = ocfs2_reserve_new_metadata_blocks(OCFS2_SB(sb), meta_add,
+ meta_ac);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ if (data_ac) {
+ ret = ocfs2_reserve_clusters(OCFS2_SB(sb), num_clusters,
+ data_ac);
+ if (ret)
+ mlog_errno(ret);
+ }
+
+out:
+ if (ret) {
+ if (*meta_ac) {
+ ocfs2_free_alloc_context(*meta_ac);
+ *meta_ac = NULL;
+ }
+ }
+
+ return ret;
+}
+
+static int ocfs2_clear_cow_buffer(handle_t *handle, struct buffer_head *bh)
+{
+ BUG_ON(buffer_dirty(bh));
+
+ clear_buffer_mapped(bh);
+
+ return 0;
+}
+
+static int ocfs2_duplicate_clusters_by_page(handle_t *handle,
+ struct ocfs2_cow_context *context,
+ u32 cpos, u32 old_cluster,
+ u32 new_cluster, u32 new_len)
+{
+ int ret = 0, partial;
+ struct ocfs2_caching_info *ci = context->data_et.et_ci;
+ struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
+ u64 new_block = ocfs2_clusters_to_blocks(sb, new_cluster);
+ struct page *page;
+ pgoff_t page_index;
+ unsigned int from, to;
+ loff_t offset, end, map_end;
+ struct address_space *mapping = context->inode->i_mapping;
+
+ mlog(0, "old_cluster %u, new %u, len %u at offset %u\n", old_cluster,
+ new_cluster, new_len, cpos);
+
+ offset = ((loff_t)cpos) << OCFS2_SB(sb)->s_clustersize_bits;
+ end = offset + (new_len << OCFS2_SB(sb)->s_clustersize_bits);
+
+ while (offset < end) {
+ page_index = offset >> PAGE_CACHE_SHIFT;
+ map_end = (page_index + 1) << PAGE_CACHE_SHIFT;
+ if (map_end > end)
+ map_end = end;
+
+ /* from, to is the offset within the page. */
+ from = offset & (PAGE_CACHE_SIZE - 1);
+ to = PAGE_CACHE_SIZE;
+ if (map_end & (PAGE_CACHE_SIZE - 1))
+ to = map_end & (PAGE_CACHE_SIZE - 1);
+
+ page = grab_cache_page(mapping, page_index);
+
+ /* This page can't be dirtied before we CoW it out. */
+ BUG_ON(PageDirty(page));
+
+ if (!PageUptodate(page)) {
+ ret = block_read_full_page(page, ocfs2_get_block);
+ if (ret) {
+ mlog_errno(ret);
+ goto unlock;
+ }
+ lock_page(page);
+ }
+
+ if (page_has_buffers(page)) {
+ ret = walk_page_buffers(handle, page_buffers(page),
+ from, to, &partial,
+ ocfs2_clear_cow_buffer);
+ if (ret) {
+ mlog_errno(ret);
+ goto unlock;
+ }
+ }
+
+ ocfs2_map_and_dirty_page(context->inode,
+ handle, from, to,
+ page, 0, &new_block);
+ mark_page_accessed(page);
+unlock:
+ unlock_page(page);
+ page_cache_release(page);
+ page = NULL;
+ offset = map_end;
+ if (ret)
+ break;
+ }
+
+ return ret;
+}
+
+static int ocfs2_duplicate_clusters_by_jbd(handle_t *handle,
+ struct ocfs2_cow_context *context,
+ u32 cpos, u32 old_cluster,
+ u32 new_cluster, u32 new_len)
+{
+ int ret = 0;
+ struct super_block *sb = context->inode->i_sb;
+ struct ocfs2_caching_info *ci = context->data_et.et_ci;
+ int i, blocks = ocfs2_clusters_to_blocks(sb, new_len);
+ u64 old_block = ocfs2_clusters_to_blocks(sb, old_cluster);
+ u64 new_block = ocfs2_clusters_to_blocks(sb, new_cluster);
+ struct ocfs2_super *osb = OCFS2_SB(sb);
+ struct buffer_head *old_bh = NULL;
+ struct buffer_head *new_bh = NULL;
+
+ mlog(0, "old_cluster %u, new %u, len %u\n", old_cluster,
+ new_cluster, new_len);
+
+ for (i = 0; i < blocks; i++, old_block++, new_block++) {
+ new_bh = sb_getblk(osb->sb, new_block);
+ if (new_bh == NULL) {
+ ret = -EIO;
+ mlog_errno(ret);
+ break;
+ }
+
+ ocfs2_set_new_buffer_uptodate(ci, new_bh);
+
+ ret = ocfs2_read_block(ci, old_block, &old_bh, NULL);
+ if (ret) {
+ mlog_errno(ret);
+ break;
+ }
+
+ ret = ocfs2_journal_access(handle, ci, new_bh,
+ OCFS2_JOURNAL_ACCESS_CREATE);
+ if (ret) {
+ mlog_errno(ret);
+ break;
+ }
+
+ memcpy(new_bh->b_data, old_bh->b_data, sb->s_blocksize);
+ ret = ocfs2_journal_dirty(handle, new_bh);
+ if (ret) {
+ mlog_errno(ret);
+ break;
+ }
+
+ brelse(new_bh);
+ brelse(old_bh);
+ new_bh = NULL;
+ old_bh = NULL;
+ }
+
+ brelse(new_bh);
+ brelse(old_bh);
+ return ret;
+}
+
+static int ocfs2_clear_ext_refcount(handle_t *handle,
+ struct ocfs2_extent_tree *et,
+ u32 cpos, u32 p_cluster, u32 len,
+ unsigned int ext_flags,
+ struct ocfs2_alloc_context *meta_ac,
+ struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+ int ret, index;
+ struct ocfs2_extent_rec replace_rec;
+ struct ocfs2_path *path = NULL;
+ struct ocfs2_extent_list *el;
+ struct super_block *sb = ocfs2_metadata_cache_get_super(et->et_ci);
+ u64 ino = ocfs2_metadata_cache_owner(et->et_ci);
+
+ mlog(0, "inode %llu cpos %u, len %u, p_cluster %u, ext_flags %u\n",
+ (unsigned long long)ino, cpos, len, p_cluster, ext_flags);
+
+ memset(&replace_rec, 0, sizeof(replace_rec));
+ replace_rec.e_cpos = cpu_to_le32(cpos);
+ replace_rec.e_leaf_clusters = cpu_to_le16(len);
+ replace_rec.e_blkno = cpu_to_le64(ocfs2_clusters_to_blocks(sb,
+ p_cluster));
+ replace_rec.e_flags = ext_flags;
+ replace_rec.e_flags &= ~OCFS2_EXT_REFCOUNTED;
+
+ path = ocfs2_new_path_from_et(et);
+ if (!path) {
+ ret = -ENOMEM;
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_find_path(et->et_ci, path, cpos);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ el = path_leaf_el(path);
+
+ index = ocfs2_search_extent_list(el, cpos);
+ if (index == -1 || index >= le16_to_cpu(el->l_next_free_rec)) {
+ ocfs2_error(sb,
+ "Inode %llu has an extent at cpos %u which can no "
+ "longer be found.\n",
+ (unsigned long long)ino, cpos);
+ ret = -EROFS;
+ goto out;
+ }
+
+ ret = ocfs2_split_extent(handle, et, path, index,
+ &replace_rec, meta_ac, dealloc);
+ if (ret)
+ mlog_errno(ret);
+
+out:
+ ocfs2_free_path(path);
+ return ret;
+}
+
+static int ocfs2_replace_clusters(handle_t *handle,
+ struct ocfs2_cow_context *context,
+ u32 cpos, u32 old,
+ u32 new, u32 len,
+ unsigned int ext_flags)
+{
+ int ret;
+ struct ocfs2_caching_info *ci = context->data_et.et_ci;
+ u64 ino = ocfs2_metadata_cache_owner(ci);
+
+ mlog(0, "inode %llu, cpos %u, old %u, new %u, len %u, ext_flags %u\n",
+ (unsigned long long)ino, cpos, old, new, len, ext_flags);
+
+ /*If the old clusters is unwritten, no need to duplicate. */
+ if (!(ext_flags & OCFS2_EXT_UNWRITTEN)) {
+ ret = context->cow_duplicate_clusters(handle, context, cpos,
+ old, new, len);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+ }
+
+ ret = ocfs2_clear_ext_refcount(handle, &context->data_et,
+ cpos, new, len, ext_flags,
+ context->meta_ac, &context->dealloc);
+ if (ret)
+ mlog_errno(ret);
+out:
+ return ret;
+}
+
+static int ocfs2_cow_sync_writeback(struct super_block *sb,
+ struct ocfs2_cow_context *context,
+ u32 cpos, u32 num_clusters)
+{
+ int ret = 0;
+ loff_t offset, end, map_end;
+ pgoff_t page_index;
+ struct page *page;
+
+ if (ocfs2_should_order_data(context->inode))
+ return 0;
+
+ offset = ((loff_t)cpos) << OCFS2_SB(sb)->s_clustersize_bits;
+ end = offset + (num_clusters << OCFS2_SB(sb)->s_clustersize_bits);
+
+ ret = filemap_fdatawrite_range(context->inode->i_mapping,
+ offset, end - 1);
+ if (ret < 0) {
+ mlog_errno(ret);
+ return ret;
+ }
+
+ while (offset < end) {
+ page_index = offset >> PAGE_CACHE_SHIFT;
+ map_end = (page_index + 1) << PAGE_CACHE_SHIFT;
+ if (map_end > end)
+ map_end = end;
+
+ page = grab_cache_page(context->inode->i_mapping, page_index);
+ BUG_ON(!page);
+
+ wait_on_page_writeback(page);
+ if (PageError(page)) {
+ ret = -EIO;
+ mlog_errno(ret);
+ } else
+ mark_page_accessed(page);
+
+ unlock_page(page);
+ page_cache_release(page);
+ page = NULL;
+ offset = map_end;
+ if (ret)
+ break;
+ }
+
+ return ret;
+}
+
+static int ocfs2_di_get_clusters(struct ocfs2_cow_context *context,
+ u32 v_cluster, u32 *p_cluster,
+ u32 *num_clusters,
+ unsigned int *extent_flags)
+{
+ return ocfs2_get_clusters(context->inode, v_cluster, p_cluster,
+ num_clusters, extent_flags);
+}
+
+static int ocfs2_make_clusters_writable(struct super_block *sb,
+ struct ocfs2_cow_context *context,
+ u32 cpos, u32 p_cluster,
+ u32 num_clusters, unsigned int e_flags)
+{
+ int ret, delete, index, credits = 0;
+ u32 new_bit, new_len;
+ unsigned int set_len;
+ struct ocfs2_super *osb = OCFS2_SB(sb);
+ handle_t *handle;
+ struct buffer_head *ref_leaf_bh = NULL;
+ struct ocfs2_caching_info *ref_ci = &context->ref_tree->rf_ci;
+ struct ocfs2_refcount_rec rec;
+
+ mlog(0, "cpos %u, p_cluster %u, num_clusters %u, e_flags %u\n",
+ cpos, p_cluster, num_clusters, e_flags);
+
+ ret = ocfs2_lock_refcount_allocators(sb, p_cluster, num_clusters,
+ &context->data_et,
+ ref_ci,
+ context->ref_root_bh,
+ &context->meta_ac,
+ &context->data_ac, &credits);
+ if (ret) {
+ mlog_errno(ret);
+ return ret;
+ }
+
+ if (context->post_refcount)
+ credits += context->post_refcount->credits;
+
+ credits += context->extra_credits;
+ handle = ocfs2_start_trans(osb, credits);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ mlog_errno(ret);
+ goto out;
+ }
+
+ while (num_clusters) {
+ ret = ocfs2_get_refcount_rec(ref_ci, context->ref_root_bh,
+ p_cluster, num_clusters,
+ &rec, &index, &ref_leaf_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ BUG_ON(!rec.r_refcount);
+ set_len = min((u64)p_cluster + num_clusters,
+ le64_to_cpu(rec.r_cpos) +
+ le32_to_cpu(rec.r_clusters)) - p_cluster;
+
+ /*
+ * There are many different situation here.
+ * 1. If refcount == 1, remove the flag and don't COW.
+ * 2. If refcount > 1, allocate clusters.
+ * Here we may not allocate r_len once at a time, so continue
+ * until we reach num_clusters.
+ */
+ if (le32_to_cpu(rec.r_refcount) == 1) {
+ delete = 0;
+ ret = ocfs2_clear_ext_refcount(handle,
+ &context->data_et,
+ cpos, p_cluster,
+ set_len, e_flags,
+ context->meta_ac,
+ &context->dealloc);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+ } else {
+ delete = 1;
+
+ ret = __ocfs2_claim_clusters(osb, handle,
+ context->data_ac,
+ 1, set_len,
+ &new_bit, &new_len);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ ret = ocfs2_replace_clusters(handle, context,
+ cpos, p_cluster, new_bit,
+ new_len, e_flags);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+ set_len = new_len;
+ }
+
+ ret = __ocfs2_decrease_refcount(handle, ref_ci,
+ context->ref_root_bh,
+ p_cluster, set_len,
+ context->meta_ac,
+ &context->dealloc, delete);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ cpos += set_len;
+ p_cluster += set_len;
+ num_clusters -= set_len;
+ brelse(ref_leaf_bh);
+ ref_leaf_bh = NULL;
+ }
+
+ /* handle any post_cow action. */
+ if (context->post_refcount && context->post_refcount->func) {
+ ret = context->post_refcount->func(context->inode, handle,
+ context->post_refcount->para);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+ }
+
+ /*
+ * Here we should write the new page out first if we are
+ * in write-back mode.
+ */
+ if (context->get_clusters == ocfs2_di_get_clusters) {
+ ret = ocfs2_cow_sync_writeback(sb, context, cpos, num_clusters);
+ if (ret)
+ mlog_errno(ret);
+ }
+
+out_commit:
+ ocfs2_commit_trans(osb, handle);
+
+out:
+ if (context->data_ac) {
+ ocfs2_free_alloc_context(context->data_ac);
+ context->data_ac = NULL;
+ }
+ if (context->meta_ac) {
+ ocfs2_free_alloc_context(context->meta_ac);
+ context->meta_ac = NULL;
+ }
+ brelse(ref_leaf_bh);
+
+ return ret;
+}
+
+static int ocfs2_replace_cow(struct ocfs2_cow_context *context)
+{
+ int ret = 0;
+ struct inode *inode = context->inode;
+ u32 cow_start = context->cow_start, cow_len = context->cow_len;
+ u32 p_cluster, num_clusters;
+ unsigned int ext_flags;
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+
+ if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb))) {
+ ocfs2_error(inode->i_sb, "Inode %lu want to use refcount "
+ "tree, but the feature bit is not set in the "
+ "super block.", inode->i_ino);
+ return -EROFS;
+ }
+
+ ocfs2_init_dealloc_ctxt(&context->dealloc);
+
+ while (cow_len) {
+ ret = context->get_clusters(context, cow_start, &p_cluster,
+ &num_clusters, &ext_flags);
+ if (ret) {
+ mlog_errno(ret);
+ break;
+ }
+
+ BUG_ON(!(ext_flags & OCFS2_EXT_REFCOUNTED));
+
+ if (cow_len < num_clusters)
+ num_clusters = cow_len;
+
+ ret = ocfs2_make_clusters_writable(inode->i_sb, context,
+ cow_start, p_cluster,
+ num_clusters, ext_flags);
+ if (ret) {
+ mlog_errno(ret);
+ break;
+ }
+
+ cow_len -= num_clusters;
+ cow_start += num_clusters;
+ }
+
+ if (ocfs2_dealloc_has_cluster(&context->dealloc)) {
+ ocfs2_schedule_truncate_log_flush(osb, 1);
+ ocfs2_run_deallocs(osb, &context->dealloc);
+ }
+
+ return ret;
+}
+
+/*
+ * Starting at cpos, try to CoW write_len clusters. Don't CoW
+ * past max_cpos. This will stop when it runs into a hole or an
+ * unrefcounted extent.
+ */
+static int ocfs2_refcount_cow_hunk(struct inode *inode,
+ struct buffer_head *di_bh,
+ u32 cpos, u32 write_len, u32 max_cpos)
+{
+ int ret;
+ u32 cow_start = 0, cow_len = 0;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+ struct buffer_head *ref_root_bh = NULL;
+ struct ocfs2_refcount_tree *ref_tree;
+ struct ocfs2_cow_context *context = NULL;
+
+ BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+
+ ret = ocfs2_refcount_cal_cow_clusters(inode, &di->id2.i_list,
+ cpos, write_len, max_cpos,
+ &cow_start, &cow_len);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ mlog(0, "CoW inode %lu, cpos %u, write_len %u, cow_start %u, "
+ "cow_len %u\n", inode->i_ino,
+ cpos, write_len, cow_start, cow_len);
+
+ BUG_ON(cow_len == 0);
+
+ context = kzalloc(sizeof(struct ocfs2_cow_context), GFP_NOFS);
+ if (!context) {
+ ret = -ENOMEM;
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_lock_refcount_tree(osb, le64_to_cpu(di->i_refcount_loc),
+ 1, &ref_tree, &ref_root_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ context->inode = inode;
+ context->cow_start = cow_start;
+ context->cow_len = cow_len;
+ context->ref_tree = ref_tree;
+ context->ref_root_bh = ref_root_bh;
+ context->cow_duplicate_clusters = ocfs2_duplicate_clusters_by_page;
+ context->get_clusters = ocfs2_di_get_clusters;
+
+ ocfs2_init_dinode_extent_tree(&context->data_et,
+ INODE_CACHE(inode), di_bh);
+
+ ret = ocfs2_replace_cow(context);
+ if (ret)
+ mlog_errno(ret);
+
+ /*
+ * truncate the extent map here since no matter whether we meet with
+ * any error during the action, we shouldn't trust cached extent map
+ * any more.
+ */
+ ocfs2_extent_map_trunc(inode, cow_start);
+
+ ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+ brelse(ref_root_bh);
+out:
+ kfree(context);
+ return ret;
+}
+
+/*
+ * CoW any and all clusters between cpos and cpos+write_len.
+ * Don't CoW past max_cpos. If this returns successfully, all
+ * clusters between cpos and cpos+write_len are safe to modify.
+ */
+int ocfs2_refcount_cow(struct inode *inode,
+ struct buffer_head *di_bh,
+ u32 cpos, u32 write_len, u32 max_cpos)
+{
+ int ret = 0;
+ u32 p_cluster, num_clusters;
+ unsigned int ext_flags;
+
+ while (write_len) {
+ ret = ocfs2_get_clusters(inode, cpos, &p_cluster,
+ &num_clusters, &ext_flags);
+ if (ret) {
+ mlog_errno(ret);
+ break;
+ }
+
+ if (write_len < num_clusters)
+ num_clusters = write_len;
+
+ if (ext_flags & OCFS2_EXT_REFCOUNTED) {
+ ret = ocfs2_refcount_cow_hunk(inode, di_bh, cpos,
+ num_clusters, max_cpos);
+ if (ret) {
+ mlog_errno(ret);
+ break;
+ }
+ }
+
+ write_len -= num_clusters;
+ cpos += num_clusters;
+ }
+
+ return ret;
+}
+
+static int ocfs2_xattr_value_get_clusters(struct ocfs2_cow_context *context,
+ u32 v_cluster, u32 *p_cluster,
+ u32 *num_clusters,
+ unsigned int *extent_flags)
+{
+ struct inode *inode = context->inode;
+ struct ocfs2_xattr_value_root *xv = context->cow_object;
+
+ return ocfs2_xattr_get_clusters(inode, v_cluster, p_cluster,
+ num_clusters, &xv->xr_list,
+ extent_flags);
+}
+
+/*
+ * Given a xattr value root, calculate the most meta/credits we need for
+ * refcount tree change if we truncate it to 0.
+ */
+int ocfs2_refcounted_xattr_delete_need(struct inode *inode,
+ struct ocfs2_caching_info *ref_ci,
+ struct buffer_head *ref_root_bh,
+ struct ocfs2_xattr_value_root *xv,
+ int *meta_add, int *credits)
+{
+ int ret = 0, index, ref_blocks = 0;
+ u32 p_cluster, num_clusters;
+ u32 cpos = 0, clusters = le32_to_cpu(xv->xr_clusters);
+ struct ocfs2_refcount_block *rb;
+ struct ocfs2_refcount_rec rec;
+ struct buffer_head *ref_leaf_bh = NULL;
+
+ while (cpos < clusters) {
+ ret = ocfs2_xattr_get_clusters(inode, cpos, &p_cluster,
+ &num_clusters, &xv->xr_list,
+ NULL);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ cpos += num_clusters;
+
+ while (num_clusters) {
+ ret = ocfs2_get_refcount_rec(ref_ci, ref_root_bh,
+ p_cluster, num_clusters,
+ &rec, &index,
+ &ref_leaf_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ BUG_ON(!rec.r_refcount);
+
+ rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
+
+ /*
+ * We really don't know whether the other clusters is in
+ * this refcount block or not, so just take the worst
+ * case that all the clusters are in this block and each
+ * one will split a refcount rec, so totally we need
+ * clusters * 2 new refcount rec.
+ */
+ if (le64_to_cpu(rb->rf_records.rl_used) + clusters * 2 >
+ le16_to_cpu(rb->rf_records.rl_count))
+ ref_blocks++;
+
+ *credits += 1;
+ brelse(ref_leaf_bh);
+ ref_leaf_bh = NULL;
+
+ if (num_clusters <= le32_to_cpu(rec.r_clusters))
+ break;
+ else
+ num_clusters -= le32_to_cpu(rec.r_clusters);
+ p_cluster += num_clusters;
+ }
+ }
+
+ *meta_add += ref_blocks;
+ if (!ref_blocks)
+ goto out;
+
+ rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+ if (le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL)
+ *credits += OCFS2_EXPAND_REFCOUNT_TREE_CREDITS;
+ else {
+ struct ocfs2_extent_tree et;
+
+ ocfs2_init_refcount_extent_tree(&et, ref_ci, ref_root_bh);
+ *credits += ocfs2_calc_extend_credits(inode->i_sb,
+ et.et_root_el,
+ ref_blocks);
+ }
+
+out:
+ brelse(ref_leaf_bh);
+ return ret;
+}
+
+/*
+ * Do CoW for xattr.
+ */
+int ocfs2_refcount_cow_xattr(struct inode *inode,
+ struct ocfs2_dinode *di,
+ struct ocfs2_xattr_value_buf *vb,
+ struct ocfs2_refcount_tree *ref_tree,
+ struct buffer_head *ref_root_bh,
+ u32 cpos, u32 write_len,
+ struct ocfs2_post_refcount *post)
+{
+ int ret;
+ struct ocfs2_xattr_value_root *xv = vb->vb_xv;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ struct ocfs2_cow_context *context = NULL;
+ u32 cow_start, cow_len;
+
+ BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+
+ ret = ocfs2_refcount_cal_cow_clusters(inode, &xv->xr_list,
+ cpos, write_len, UINT_MAX,
+ &cow_start, &cow_len);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ BUG_ON(cow_len == 0);
+
+ context = kzalloc(sizeof(struct ocfs2_cow_context), GFP_NOFS);
+ if (!context) {
+ ret = -ENOMEM;
+ mlog_errno(ret);
+ goto out;
+ }
+
+ context->inode = inode;
+ context->cow_start = cow_start;
+ context->cow_len = cow_len;
+ context->ref_tree = ref_tree;
+ context->ref_root_bh = ref_root_bh;;
+ context->cow_object = xv;
+
+ context->cow_duplicate_clusters = ocfs2_duplicate_clusters_by_jbd;
+ /* We need the extra credits for duplicate_clusters by jbd. */
+ context->extra_credits =
+ ocfs2_clusters_to_blocks(inode->i_sb, 1) * cow_len;
+ context->get_clusters = ocfs2_xattr_value_get_clusters;
+ context->post_refcount = post;
+
+ ocfs2_init_xattr_value_extent_tree(&context->data_et,
+ INODE_CACHE(inode), vb);
+
+ ret = ocfs2_replace_cow(context);
+ if (ret)
+ mlog_errno(ret);
+
+out:
+ kfree(context);
+ return ret;
+}
+
+/*
+ * Insert a new extent into refcount tree and mark a extent rec
+ * as refcounted in the dinode tree.
+ */
+int ocfs2_add_refcount_flag(struct inode *inode,
+ struct ocfs2_extent_tree *data_et,
+ struct ocfs2_caching_info *ref_ci,
+ struct buffer_head *ref_root_bh,
+ u32 cpos, u32 p_cluster, u32 num_clusters,
+ struct ocfs2_cached_dealloc_ctxt *dealloc,
+ struct ocfs2_post_refcount *post)
+{
+ int ret;
+ handle_t *handle;
+ int credits = 1, ref_blocks = 0;
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+ struct ocfs2_alloc_context *meta_ac = NULL;
+
+ ret = ocfs2_calc_refcount_meta_credits(inode->i_sb,
+ ref_ci, ref_root_bh,
+ p_cluster, num_clusters,
+ &ref_blocks, &credits);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ mlog(0, "reserve new metadata %d, credits = %d\n",
+ ref_blocks, credits);
+
+ if (ref_blocks) {
+ ret = ocfs2_reserve_new_metadata_blocks(OCFS2_SB(inode->i_sb),
+ ref_blocks, &meta_ac);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+ }
+
+ if (post)
+ credits += post->credits;
+
+ handle = ocfs2_start_trans(osb, credits);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_mark_extent_refcounted(inode, data_et, handle,
+ cpos, num_clusters, p_cluster,
+ meta_ac, dealloc);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ ret = __ocfs2_increase_refcount(handle, ref_ci, ref_root_bh,
+ p_cluster, num_clusters, 0,
+ meta_ac, dealloc);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ if (post && post->func) {
+ ret = post->func(inode, handle, post->para);
+ if (ret)
+ mlog_errno(ret);
+ }
+
+out_commit:
+ ocfs2_commit_trans(osb, handle);
+out:
+ if (meta_ac)
+ ocfs2_free_alloc_context(meta_ac);
+ return ret;
+}
+
+static int ocfs2_change_ctime(struct inode *inode,
+ struct buffer_head *di_bh)
+{
+ int ret;
+ handle_t *handle;
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+
+ handle = ocfs2_start_trans(OCFS2_SB(inode->i_sb),
+ OCFS2_INODE_UPDATE_CREDITS);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
+ OCFS2_JOURNAL_ACCESS_WRITE);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ inode->i_ctime = CURRENT_TIME;
+ di->i_ctime = cpu_to_le64(inode->i_ctime.tv_sec);
+ di->i_ctime_nsec = cpu_to_le32(inode->i_ctime.tv_nsec);
+
+ ocfs2_journal_dirty(handle, di_bh);
+
+out_commit:
+ ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle);
+out:
+ return ret;
+}
+
+static int ocfs2_attach_refcount_tree(struct inode *inode,
+ struct buffer_head *di_bh)
+{
+ int ret, data_changed = 0;
+ struct buffer_head *ref_root_bh = NULL;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+ struct ocfs2_refcount_tree *ref_tree;
+ unsigned int ext_flags;
+ loff_t size;
+ u32 cpos, num_clusters, clusters, p_cluster;
+ struct ocfs2_cached_dealloc_ctxt dealloc;
+ struct ocfs2_extent_tree di_et;
+
+ ocfs2_init_dealloc_ctxt(&dealloc);
+
+ if (!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL)) {
+ ret = ocfs2_create_refcount_tree(inode, di_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+ }
+
+ BUG_ON(!di->i_refcount_loc);
+ ret = ocfs2_lock_refcount_tree(osb,
+ le64_to_cpu(di->i_refcount_loc), 1,
+ &ref_tree, &ref_root_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ocfs2_init_dinode_extent_tree(&di_et, INODE_CACHE(inode), di_bh);
+
+ size = i_size_read(inode);
+ clusters = ocfs2_clusters_for_bytes(inode->i_sb, size);
+
+ cpos = 0;
+ while (cpos < clusters) {
+ ret = ocfs2_get_clusters(inode, cpos, &p_cluster,
+ &num_clusters, &ext_flags);
+
+ if (p_cluster && !(ext_flags & OCFS2_EXT_REFCOUNTED)) {
+ ret = ocfs2_add_refcount_flag(inode, &di_et,
+ &ref_tree->rf_ci,
+ ref_root_bh, cpos,
+ p_cluster, num_clusters,
+ &dealloc, NULL);
+ if (ret) {
+ mlog_errno(ret);
+ goto unlock;
+ }
+
+ data_changed = 1;
+ }
+ cpos += num_clusters;
+ }
+
+ if (oi->ip_dyn_features & OCFS2_HAS_XATTR_FL) {
+ ret = ocfs2_xattr_attach_refcount_tree(inode, di_bh,
+ &ref_tree->rf_ci,
+ ref_root_bh,
+ &dealloc);
+ if (ret) {
+ mlog_errno(ret);
+ goto unlock;
+ }
+ }
+
+ if (data_changed) {
+ ret = ocfs2_change_ctime(inode, di_bh);
+ if (ret)
+ mlog_errno(ret);
+ }
+
+unlock:
+ ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+ brelse(ref_root_bh);
+
+ if (!ret && ocfs2_dealloc_has_cluster(&dealloc)) {
+ ocfs2_schedule_truncate_log_flush(osb, 1);
+ ocfs2_run_deallocs(osb, &dealloc);
+ }
+out:
+ /*
+ * Empty the extent map so that we may get the right extent
+ * record from the disk.
+ */
+ ocfs2_extent_map_trunc(inode, 0);
+
+ return ret;
+}
+
+static int ocfs2_add_refcounted_extent(struct inode *inode,
+ struct ocfs2_extent_tree *et,
+ struct ocfs2_caching_info *ref_ci,
+ struct buffer_head *ref_root_bh,
+ u32 cpos, u32 p_cluster, u32 num_clusters,
+ unsigned int ext_flags,
+ struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+ int ret;
+ handle_t *handle;
+ int credits = 0;
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+ struct ocfs2_alloc_context *meta_ac = NULL;
+
+ ret = ocfs2_lock_refcount_allocators(inode->i_sb,
+ p_cluster, num_clusters,
+ et, ref_ci,
+ ref_root_bh, &meta_ac,
+ NULL, &credits);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ handle = ocfs2_start_trans(osb, credits);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_insert_extent(handle, et, cpos,
+ cpu_to_le64(ocfs2_clusters_to_blocks(inode->i_sb,
+ p_cluster)),
+ num_clusters, ext_flags, meta_ac);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ ret = ocfs2_increase_refcount(handle, ref_ci, ref_root_bh,
+ p_cluster, num_clusters,
+ meta_ac, dealloc);
+ if (ret)
+ mlog_errno(ret);
+
+out_commit:
+ ocfs2_commit_trans(osb, handle);
+out:
+ if (meta_ac)
+ ocfs2_free_alloc_context(meta_ac);
+ return ret;
+}
+
+static int ocfs2_duplicate_extent_list(struct inode *s_inode,
+ struct inode *t_inode,
+ struct buffer_head *t_bh,
+ struct ocfs2_caching_info *ref_ci,
+ struct buffer_head *ref_root_bh,
+ struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+ int ret = 0;
+ u32 p_cluster, num_clusters, clusters, cpos;
+ loff_t size;
+ unsigned int ext_flags;
+ struct ocfs2_extent_tree et;
+
+ ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(t_inode), t_bh);
+
+ size = i_size_read(s_inode);
+ clusters = ocfs2_clusters_for_bytes(s_inode->i_sb, size);
+
+ cpos = 0;
+ while (cpos < clusters) {
+ ret = ocfs2_get_clusters(s_inode, cpos, &p_cluster,
+ &num_clusters, &ext_flags);
+
+ if (p_cluster) {
+ ret = ocfs2_add_refcounted_extent(t_inode, &et,
+ ref_ci, ref_root_bh,
+ cpos, p_cluster,
+ num_clusters,
+ ext_flags,
+ dealloc);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+ }
+
+ cpos += num_clusters;
+ }
+
+out:
+ return ret;
+}
+
+/*
+ * change the new file's attributes to the src.
+ *
+ * reflink creates a snapshot of a file, that means the attributes
+ * must be identical except for three exceptions - nlink, ino, and ctime.
+ */
+static int ocfs2_complete_reflink(struct inode *s_inode,
+ struct buffer_head *s_bh,
+ struct inode *t_inode,
+ struct buffer_head *t_bh,
+ bool preserve)
+{
+ int ret;
+ handle_t *handle;
+ struct ocfs2_dinode *s_di = (struct ocfs2_dinode *)s_bh->b_data;
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)t_bh->b_data;
+ loff_t size = i_size_read(s_inode);
+
+ handle = ocfs2_start_trans(OCFS2_SB(t_inode->i_sb),
+ OCFS2_INODE_UPDATE_CREDITS);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ mlog_errno(ret);
+ return ret;
+ }
+
+ ret = ocfs2_journal_access_di(handle, INODE_CACHE(t_inode), t_bh,
+ OCFS2_JOURNAL_ACCESS_WRITE);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ spin_lock(&OCFS2_I(t_inode)->ip_lock);
+ OCFS2_I(t_inode)->ip_clusters = OCFS2_I(s_inode)->ip_clusters;
+ OCFS2_I(t_inode)->ip_attr = OCFS2_I(s_inode)->ip_attr;
+ OCFS2_I(t_inode)->ip_dyn_features = OCFS2_I(s_inode)->ip_dyn_features;
+ spin_unlock(&OCFS2_I(t_inode)->ip_lock);
+ i_size_write(t_inode, size);
+
+ di->i_xattr_inline_size = s_di->i_xattr_inline_size;
+ di->i_clusters = s_di->i_clusters;
+ di->i_size = s_di->i_size;
+ di->i_dyn_features = s_di->i_dyn_features;
+ di->i_attr = s_di->i_attr;
+
+ if (preserve) {
+ di->i_uid = s_di->i_uid;
+ di->i_gid = s_di->i_gid;
+ di->i_mode = s_di->i_mode;
+
+ /*
+ * update time.
+ * we want mtime to appear identical to the source and
+ * update ctime.
+ */
+ t_inode->i_ctime = CURRENT_TIME;
+
+ di->i_ctime = cpu_to_le64(t_inode->i_ctime.tv_sec);
+ di->i_ctime_nsec = cpu_to_le32(t_inode->i_ctime.tv_nsec);
+
+ t_inode->i_mtime = s_inode->i_mtime;
+ di->i_mtime = s_di->i_mtime;
+ di->i_mtime_nsec = s_di->i_mtime_nsec;
+ }
+
+ ocfs2_journal_dirty(handle, t_bh);
+
+out_commit:
+ ocfs2_commit_trans(OCFS2_SB(t_inode->i_sb), handle);
+ return ret;
+}
+
+static int ocfs2_create_reflink_node(struct inode *s_inode,
+ struct buffer_head *s_bh,
+ struct inode *t_inode,
+ struct buffer_head *t_bh,
+ bool preserve)
+{
+ int ret;
+ struct buffer_head *ref_root_bh = NULL;
+ struct ocfs2_cached_dealloc_ctxt dealloc;
+ struct ocfs2_super *osb = OCFS2_SB(s_inode->i_sb);
+ struct ocfs2_refcount_block *rb;
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)s_bh->b_data;
+ struct ocfs2_refcount_tree *ref_tree;
+
+ ocfs2_init_dealloc_ctxt(&dealloc);
+
+ ret = ocfs2_set_refcount_tree(t_inode, t_bh,
+ le64_to_cpu(di->i_refcount_loc));
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_lock_refcount_tree(osb, le64_to_cpu(di->i_refcount_loc),
+ 1, &ref_tree, &ref_root_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+ rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+
+ ret = ocfs2_duplicate_extent_list(s_inode, t_inode, t_bh,
+ &ref_tree->rf_ci, ref_root_bh,
+ &dealloc);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_unlock_refcount;
+ }
+
+ ret = ocfs2_complete_reflink(s_inode, s_bh, t_inode, t_bh, preserve);
+ if (ret)
+ mlog_errno(ret);
+
+out_unlock_refcount:
+ ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+ brelse(ref_root_bh);
+out:
+ if (ocfs2_dealloc_has_cluster(&dealloc)) {
+ ocfs2_schedule_truncate_log_flush(osb, 1);
+ ocfs2_run_deallocs(osb, &dealloc);
+ }
+
+ return ret;
+}
+
+static int __ocfs2_reflink(struct dentry *old_dentry,
+ struct buffer_head *old_bh,
+ struct inode *new_inode,
+ bool preserve)
+{
+ int ret;
+ struct inode *inode = old_dentry->d_inode;
+ struct buffer_head *new_bh = NULL;
+
+ ret = filemap_fdatawrite(inode->i_mapping);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_attach_refcount_tree(inode, old_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ mutex_lock(&new_inode->i_mutex);
+ ret = ocfs2_inode_lock(new_inode, &new_bh, 1);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_unlock;
+ }
+
+ ret = ocfs2_create_reflink_node(inode, old_bh,
+ new_inode, new_bh, preserve);
+ if (ret) {
+ mlog_errno(ret);
+ goto inode_unlock;
+ }
+
+ if (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_XATTR_FL) {
+ ret = ocfs2_reflink_xattrs(inode, old_bh,
+ new_inode, new_bh,
+ preserve);
+ if (ret)
+ mlog_errno(ret);
+ }
+inode_unlock:
+ ocfs2_inode_unlock(new_inode, 1);
+ brelse(new_bh);
+out_unlock:
+ mutex_unlock(&new_inode->i_mutex);
+out:
+ if (!ret) {
+ ret = filemap_fdatawait(inode->i_mapping);
+ if (ret)
+ mlog_errno(ret);
+ }
+ return ret;
+}
+
+static int ocfs2_reflink(struct dentry *old_dentry, struct inode *dir,
+ struct dentry *new_dentry, bool preserve)
+{
+ int error;
+ struct inode *inode = old_dentry->d_inode;
+ struct buffer_head *old_bh = NULL;
+ struct inode *new_orphan_inode = NULL;
+
+ if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb)))
+ return -EOPNOTSUPP;
+
+ error = ocfs2_create_inode_in_orphan(dir, inode->i_mode,
+ &new_orphan_inode);
+ if (error) {
+ mlog_errno(error);
+ goto out;
+ }
+
+ error = ocfs2_inode_lock(inode, &old_bh, 1);
+ if (error) {
+ mlog_errno(error);
+ goto out;
+ }
+
+ down_write(&OCFS2_I(inode)->ip_xattr_sem);
+ down_write(&OCFS2_I(inode)->ip_alloc_sem);
+ error = __ocfs2_reflink(old_dentry, old_bh,
+ new_orphan_inode, preserve);
+ up_write(&OCFS2_I(inode)->ip_alloc_sem);
+ up_write(&OCFS2_I(inode)->ip_xattr_sem);
+
+ ocfs2_inode_unlock(inode, 1);
+ brelse(old_bh);
+
+ if (error) {
+ mlog_errno(error);
+ goto out;
+ }
+
+ /* If the security isn't preserved, we need to re-initialize them. */
+ if (!preserve) {
+ error = ocfs2_init_security_and_acl(dir, new_orphan_inode);
+ if (error)
+ mlog_errno(error);
+ }
+out:
+ if (!error) {
+ error = ocfs2_mv_orphaned_inode_to_new(dir, new_orphan_inode,
+ new_dentry);
+ if (error)
+ mlog_errno(error);
+ }
+
+ if (new_orphan_inode) {
+ /*
+ * We need to open_unlock the inode no matter whether we
+ * succeed or not, so that other nodes can delete it later.
+ */
+ ocfs2_open_unlock(new_orphan_inode);
+ if (error)
+ iput(new_orphan_inode);
+ }
+
+ return error;
+}
+
+/*
+ * Below here are the bits used by OCFS2_IOC_REFLINK() to fake
+ * sys_reflink(). This will go away when vfs_reflink() exists in
+ * fs/namei.c.
+ */
+
+/* copied from may_create in VFS. */
+static inline int ocfs2_may_create(struct inode *dir, struct dentry *child)
+{
+ if (child->d_inode)
+ return -EEXIST;
+ if (IS_DEADDIR(dir))
+ return -ENOENT;
+ return inode_permission(dir, MAY_WRITE | MAY_EXEC);
+}
+
+/* copied from user_path_parent. */
+static int ocfs2_user_path_parent(const char __user *path,
+ struct nameidata *nd, char **name)
+{
+ char *s = getname(path);
+ int error;
+
+ if (IS_ERR(s))
+ return PTR_ERR(s);
+
+ error = path_lookup(s, LOOKUP_PARENT, nd);
+ if (error)
+ putname(s);
+ else
+ *name = s;
+
+ return error;
+}
+
+/**
+ * ocfs2_vfs_reflink - Create a reference-counted link
+ *
+ * @old_dentry: source dentry + inode
+ * @dir: directory to create the target
+ * @new_dentry: target dentry
+ * @preserve: if true, preserve all file attributes
+ */
+int ocfs2_vfs_reflink(struct dentry *old_dentry, struct inode *dir,
+ struct dentry *new_dentry, bool preserve)
+{
+ struct inode *inode = old_dentry->d_inode;
+ int error;
+
+ if (!inode)
+ return -ENOENT;
+
+ error = ocfs2_may_create(dir, new_dentry);
+ if (error)
+ return error;
+
+ if (dir->i_sb != inode->i_sb)
+ return -EXDEV;
+
+ /*
+ * A reflink to an append-only or immutable file cannot be created.
+ */
+ if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
+ return -EPERM;
+
+ /* Only regular files can be reflinked. */
+ if (!S_ISREG(inode->i_mode))
+ return -EPERM;
+
+ /*
+ * If the caller wants to preserve ownership, they require the
+ * rights to do so.
+ */
+ if (preserve) {
+ if ((current_fsuid() != inode->i_uid) && !capable(CAP_CHOWN))
+ return -EPERM;
+ if (!in_group_p(inode->i_gid) && !capable(CAP_CHOWN))
+ return -EPERM;
+ }
+
+ /*
+ * If the caller is modifying any aspect of the attributes, they
+ * are not creating a snapshot. They need read permission on the
+ * file.
+ */
+ if (!preserve) {
+ error = inode_permission(inode, MAY_READ);
+ if (error)
+ return error;
+ }
+
+ mutex_lock(&inode->i_mutex);
+ vfs_dq_init(dir);
+ error = ocfs2_reflink(old_dentry, dir, new_dentry, preserve);
+ mutex_unlock(&inode->i_mutex);
+ if (!error)
+ fsnotify_create(dir, new_dentry);
+ return error;
+}
+/*
+ * Most codes are copied from sys_linkat.
+ */
+int ocfs2_reflink_ioctl(struct inode *inode,
+ const char __user *oldname,
+ const char __user *newname,
+ bool preserve)
+{
+ struct dentry *new_dentry;
+ struct nameidata nd;
+ struct path old_path;
+ int error;
+ char *to = NULL;
+
+ if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb)))
+ return -EOPNOTSUPP;
+
+ error = user_path_at(AT_FDCWD, oldname, 0, &old_path);
+ if (error) {
+ mlog_errno(error);
+ return error;
+ }
+
+ error = ocfs2_user_path_parent(newname, &nd, &to);
+ if (error) {
+ mlog_errno(error);
+ goto out;
+ }
+
+ error = -EXDEV;
+ if (old_path.mnt != nd.path.mnt)
+ goto out_release;
+ new_dentry = lookup_create(&nd, 0);
+ error = PTR_ERR(new_dentry);
+ if (IS_ERR(new_dentry)) {
+ mlog_errno(error);
+ goto out_unlock;
+ }
+
+ error = mnt_want_write(nd.path.mnt);
+ if (error) {
+ mlog_errno(error);
+ goto out_dput;
+ }
+
+ error = ocfs2_vfs_reflink(old_path.dentry,
+ nd.path.dentry->d_inode,
+ new_dentry, preserve);
+ mnt_drop_write(nd.path.mnt);
+out_dput:
+ dput(new_dentry);
+out_unlock:
+ mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
+out_release:
+ path_put(&nd.path);
+ putname(to);
+out:
+ path_put(&old_path);
+
+ return error;
+}
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff