ocfs2: support writing of unwritten extents

Update the write code to detect when the user is asking to write to an
unwritten extent. Like writing to a hole, we must zero the region between
the write and the cluster boundaries. Most of the existing cluster zeroing
logic can be re-used with some additional checks for the unwritten flag on
extent records.

Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>

diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 077583b50391ffbbcfb7c37c5843c942d0a49ac4..8af923316d2296086c50da946d6dee722d07258a 100644 (file)
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -782,8 +782,14 @@ struct ocfs2_write_cluster_desc {
         * filled.
         */
        unsigned        c_new;
+       unsigned        c_unwritten;
 };
 
+static inline int ocfs2_should_zero_cluster(struct ocfs2_write_cluster_desc *d)
+{
+       return d->c_new || d->c_unwritten;
+}
+
 struct ocfs2_write_ctxt {
        /* Logical cluster position / len of write */
        u32                             w_cpos;
@@ -829,6 +835,8 @@ struct ocfs2_write_ctxt {
        handle_t                        *w_handle;
 
        struct buffer_head              *w_di_bh;
+
+       struct ocfs2_cached_dealloc_ctxt w_dealloc;
 };
 
 static void ocfs2_free_write_ctxt(struct ocfs2_write_ctxt *wc)
@@ -868,6 +876,8 @@ static int ocfs2_alloc_write_ctxt(struct ocfs2_write_ctxt **wcp,
        else
                wc->w_large_pages = 0;
 
+       ocfs2_init_dealloc_ctxt(&wc->w_dealloc);
+
        *wcp = wc;
 
        return 0;
@@ -1103,16 +1113,19 @@ out:
  * Prepare a single cluster for write one cluster into the file.
  */
 static int ocfs2_write_cluster(struct address_space *mapping,
-                              u32 phys, struct ocfs2_alloc_context *data_ac,
+                              u32 phys, unsigned int unwritten,
+                              struct ocfs2_alloc_context *data_ac,
                               struct ocfs2_alloc_context *meta_ac,
                               struct ocfs2_write_ctxt *wc, u32 cpos,
                               loff_t user_pos, unsigned user_len)
 {
-       int ret, i, new;
+       int ret, i, new, should_zero = 0;
        u64 v_blkno, p_blkno;
        struct inode *inode = mapping->host;
 
        new = phys == 0 ? 1 : 0;
+       if (new || unwritten)
+               should_zero = 1;
 
        if (new) {
                u32 tmp_pos;
@@ -1142,11 +1155,20 @@ static int ocfs2_write_cluster(struct address_space *mapping,
                        mlog_errno(ret);
                        goto out;
                }
+       } else if (unwritten) {
+               ret = ocfs2_mark_extent_written(inode, wc->w_di_bh,
+                                               wc->w_handle, cpos, 1, phys,
+                                               meta_ac, &wc->w_dealloc);
+               if (ret < 0) {
+                       mlog_errno(ret);
+                       goto out;
+               }
+       }
 
+       if (should_zero)
                v_blkno = ocfs2_clusters_to_blocks(inode->i_sb, cpos);
-       } else {
+       else
                v_blkno = user_pos >> inode->i_sb->s_blocksize_bits;
-       }
 
        /*
         * The only reason this should fail is due to an inability to
@@ -1169,7 +1191,8 @@ static int ocfs2_write_cluster(struct address_space *mapping,
 
                tmpret = ocfs2_prepare_page_for_write(inode, &p_blkno, wc,
                                                      wc->w_pages[i], cpos,
-                                                     user_pos, user_len, new);
+                                                     user_pos, user_len,
+                                                     should_zero);
                if (tmpret) {
                        mlog_errno(tmpret);
                        if (ret == 0)
@@ -1200,8 +1223,9 @@ static int ocfs2_write_cluster_by_desc(struct address_space *mapping,
        for (i = 0; i < wc->w_clen; i++) {
                desc = &wc->w_desc[i];
 
-               ret = ocfs2_write_cluster(mapping, desc->c_phys, data_ac,
-                                         meta_ac, wc, desc->c_cpos, pos, len);
+               ret = ocfs2_write_cluster(mapping, desc->c_phys,
+                                         desc->c_unwritten, data_ac, meta_ac,
+                                         wc, desc->c_cpos, pos, len);
                if (ret) {
                        mlog_errno(ret);
                        goto out;
@@ -1242,19 +1266,19 @@ static void ocfs2_set_target_boundaries(struct ocfs2_super *osb,
        if (wc->w_large_pages) {
                /*
                 * We only care about the 1st and last cluster within
-                * our range and whether they are holes or not. Either
+                * our range and whether they should be zero'd or not. Either
                 * value may be extended out to the start/end of a
                 * newly allocated cluster.
                 */
                desc = &wc->w_desc[0];
-               if (desc->c_new)
+               if (ocfs2_should_zero_cluster(desc))
                        ocfs2_figure_cluster_boundaries(osb,
                                                        desc->c_cpos,
                                                        &wc->w_target_from,
                                                        NULL);
 
                desc = &wc->w_desc[wc->w_clen - 1];
-               if (desc->c_new)
+               if (ocfs2_should_zero_cluster(desc))
                        ocfs2_figure_cluster_boundaries(osb,
                                                        desc->c_cpos,
                                                        NULL,
@@ -1268,28 +1292,52 @@ static void ocfs2_set_target_boundaries(struct ocfs2_super *osb,
 /*
  * Populate each single-cluster write descriptor in the write context
  * with information about the i/o to be done.
+ *
+ * Returns the number of clusters that will have to be allocated, as
+ * well as a worst case estimate of the number of extent records that
+ * would have to be created during a write to an unwritten region.
  */
 static int ocfs2_populate_write_desc(struct inode *inode,
                                     struct ocfs2_write_ctxt *wc,
-                                    unsigned int *clusters_to_alloc)
+                                    unsigned int *clusters_to_alloc,
+                                    unsigned int *extents_to_split)
 {
        int ret;
        struct ocfs2_write_cluster_desc *desc;
        unsigned int num_clusters = 0;
+       unsigned int ext_flags = 0;
        u32 phys = 0;
        int i;
 
+       *clusters_to_alloc = 0;
+       *extents_to_split = 0;
+
        for (i = 0; i < wc->w_clen; i++) {
                desc = &wc->w_desc[i];
                desc->c_cpos = wc->w_cpos + i;
 
                if (num_clusters == 0) {
+                       /*
+                        * Need to look up the next extent record.
+                        */
                        ret = ocfs2_get_clusters(inode, desc->c_cpos, &phys,
-                                                &num_clusters, NULL);
+                                                &num_clusters, &ext_flags);
                        if (ret) {
                                mlog_errno(ret);
                                goto out;
                        }
+
+                       /*
+                        * Assume worst case - that we're writing in
+                        * the middle of the extent.
+                        *
+                        * We can assume that the write proceeds from
+                        * left to right, in which case the extent
+                        * insert code is smart enough to coalesce the
+                        * next splits into the previous records created.
+                        */
+                       if (ext_flags & OCFS2_EXT_UNWRITTEN)
+                               *extents_to_split = *extents_to_split + 2;
                } else if (phys) {
                        /*
                         * Only increment phys if it doesn't describe
@@ -1303,6 +1351,8 @@ static int ocfs2_populate_write_desc(struct inode *inode,
                        desc->c_new = 1;
                        *clusters_to_alloc = *clusters_to_alloc + 1;
                }
+               if (ext_flags & OCFS2_EXT_UNWRITTEN)
+                       desc->c_unwritten = 1;
 
                num_clusters--;
        }
@@ -1318,7 +1368,7 @@ int ocfs2_write_begin_nolock(struct address_space *mapping,
                             struct buffer_head *di_bh, struct page *mmap_page)
 {
        int ret, credits = OCFS2_INODE_UPDATE_CREDITS;
-       unsigned int clusters_to_alloc = 0;
+       unsigned int clusters_to_alloc, extents_to_split;
        struct ocfs2_write_ctxt *wc;
        struct inode *inode = mapping->host;
        struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
@@ -1333,7 +1383,8 @@ int ocfs2_write_begin_nolock(struct address_space *mapping,
                return ret;
        }
 
-       ret = ocfs2_populate_write_desc(inode, wc, &clusters_to_alloc);
+       ret = ocfs2_populate_write_desc(inode, wc, &clusters_to_alloc,
+                                       &extents_to_split);
        if (ret) {
                mlog_errno(ret);
                goto out;
@@ -1347,14 +1398,14 @@ int ocfs2_write_begin_nolock(struct address_space *mapping,
         * write out. An allocation requires that we write the entire
         * cluster range.
         */
-       if (clusters_to_alloc > 0) {
+       if (clusters_to_alloc || extents_to_split) {
                /*
                 * XXX: We are stretching the limits of
-                * ocfs2_lock_allocators(). It greately over-estimates
+                * ocfs2_lock_allocators(). It greatly over-estimates
                 * the work to be done.
                 */
                ret = ocfs2_lock_allocators(inode, di, clusters_to_alloc,
-                                           &data_ac, &meta_ac);
+                                           extents_to_split, &data_ac, &meta_ac);
                if (ret) {
                        mlog_errno(ret);
                        goto out;
@@ -1365,7 +1416,8 @@ int ocfs2_write_begin_nolock(struct address_space *mapping,
 
        }
 
-       ocfs2_set_target_boundaries(osb, wc, pos, len, clusters_to_alloc);
+       ocfs2_set_target_boundaries(osb, wc, pos, len,
+                                   clusters_to_alloc + extents_to_split);
 
        handle = ocfs2_start_trans(osb, credits);
        if (IS_ERR(handle)) {
@@ -1393,7 +1445,8 @@ int ocfs2_write_begin_nolock(struct address_space *mapping,
         * extent.
         */
        ret = ocfs2_grab_pages_for_write(mapping, wc, wc->w_cpos, pos,
-                                        clusters_to_alloc, mmap_page);
+                                        clusters_to_alloc + extents_to_split,
+                                        mmap_page);
        if (ret) {
                mlog_errno(ret);
                goto out_commit;
@@ -1538,11 +1591,12 @@ int ocfs2_write_end_nolock(struct address_space *mapping,
        inode->i_mtime = inode->i_ctime = CURRENT_TIME;
        di->i_mtime = di->i_ctime = cpu_to_le64(inode->i_mtime.tv_sec);
        di->i_mtime_nsec = di->i_ctime_nsec = cpu_to_le32(inode->i_mtime.tv_nsec);
-
        ocfs2_journal_dirty(handle, wc->w_di_bh);
 
        ocfs2_commit_trans(osb, handle);
 
+       ocfs2_run_deallocs(osb, &wc->w_dealloc);
+
        ocfs2_free_write_ctxt(wc);
 
        return copied;

diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index a80f31776d94beacc2383b53fe04dd7b1808077b..6745086da6fdae431cc35035d3371383cf223c89 100644 (file)
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -527,20 +527,21 @@ leave:
  * understand sparse inodes.
  */
 int ocfs2_lock_allocators(struct inode *inode, struct ocfs2_dinode *di,
-                         u32 clusters_to_add,
+                         u32 clusters_to_add, u32 extents_to_split,
                          struct ocfs2_alloc_context **data_ac,
                          struct ocfs2_alloc_context **meta_ac)
 {
        int ret, num_free_extents;
+       unsigned int max_recs_needed = clusters_to_add + 2 * extents_to_split;
        struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 
        *meta_ac = NULL;
        *data_ac = NULL;
 
        mlog(0, "extend inode %llu, i_size = %lld, di->i_clusters = %u, "
-            "clusters_to_add = %u\n",
+            "clusters_to_add = %u, extents_to_split = %u\n",
             (unsigned long long)OCFS2_I(inode)->ip_blkno, i_size_read(inode),
-            le32_to_cpu(di->i_clusters), clusters_to_add);
+            le32_to_cpu(di->i_clusters), clusters_to_add, extents_to_split);
 
        num_free_extents = ocfs2_num_free_extents(osb, inode, di);
        if (num_free_extents < 0) {
@@ -558,9 +559,12 @@ int ocfs2_lock_allocators(struct inode *inode, struct ocfs2_dinode *di,
         *
         * Most of the time we'll only be seeing this 1 cluster at a time
         * anyway.
+        *
+        * Always lock for any unwritten extents - we might want to
+        * add blocks during a split.
         */
        if (!num_free_extents ||
-           (ocfs2_sparse_alloc(osb) && num_free_extents < clusters_to_add)) {
+           (ocfs2_sparse_alloc(osb) && num_free_extents < max_recs_needed)) {
                ret = ocfs2_reserve_new_metadata(osb, di, meta_ac);
                if (ret < 0) {
                        if (ret != -ENOSPC)
@@ -641,7 +645,7 @@ restart_all:
        down_write(&OCFS2_I(inode)->ip_alloc_sem);
        drop_alloc_sem = 1;
 
-       status = ocfs2_lock_allocators(inode, fe, clusters_to_add, &data_ac,
+       status = ocfs2_lock_allocators(inode, fe, clusters_to_add, 0, &data_ac,
                                       &meta_ac);
        if (status) {
                mlog_errno(status);

diff --git a/fs/ocfs2/file.h b/fs/ocfs2/file.h
index a4dd1fa1822b04a9d3d643caa4535dbb780e9166..54df3c4bd2fd3800f96ff84c7321e4dfcf56844f 100644 (file)
--- a/fs/ocfs2/file.h
+++ b/fs/ocfs2/file.h
@@ -47,7 +47,7 @@ int ocfs2_do_extend_allocation(struct ocfs2_super *osb,
                               struct ocfs2_alloc_context *meta_ac,
                               enum ocfs2_alloc_restarted *reason);
 int ocfs2_lock_allocators(struct inode *inode, struct ocfs2_dinode *di,
-                         u32 clusters_to_add,
+                         u32 clusters_to_add, u32 extents_to_split,
                          struct ocfs2_alloc_context **data_ac,
                          struct ocfs2_alloc_context **meta_ac);
 int ocfs2_setattr(struct dentry *dentry, struct iattr *attr);