[Ocfs2-devel] [PATCH 2/3] Ocfs2: Fix punching hole codes to correctly do CoW during cluster zeroing.
Tao Ma
tao.ma at oracle.com
Thu Feb 4 18:29:11 PST 2010
Hi tristan,
Tristan Ye wrote:
> Based on the former patch of truncating optimization, bugfix for refcount on
> punching holes can be fairly easy and straightforward since most of work we
> should take into account for refcounting have been completed already in func
> ocfs2_remove_btree_range(), which is also being used by our truncating codes.
>
> The patch just did CoW for reflinks when a hole is being punched whose start
> and end offset were within one cluster, which means partial zeroing for a cluster
> will be performed soon.
>
> The patch has been tested fixing the following bug:
>
> http://oss.oracle.com/bugzilla/show_bug.cgi?id=1216
>
> Signed-off-by: Tristan Ye <tristan.ye at oracle.com>
> ---
> fs/ocfs2/file.c | 56 +++++++++++++++++++++++++++++++++++++++++++++++++++++-
> 1 files changed, 54 insertions(+), 2 deletions(-)
>
> diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
> index e0c9d1c..85911a9 100644
> --- a/fs/ocfs2/file.c
> +++ b/fs/ocfs2/file.c
> @@ -1427,12 +1427,17 @@ static int ocfs2_remove_inode_range(struct inode *inode,
> struct buffer_head *di_bh, u64 byte_start,
> u64 byte_len)
> {
> - int ret = 0;
> + int ret = 0, flags = 0, i;
> u32 trunc_start, trunc_len, cpos, phys_cpos, alloc_size;
> struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
> struct ocfs2_cached_dealloc_ctxt dealloc;
> struct address_space *mapping = inode->i_mapping;
> struct ocfs2_extent_tree et;
> + struct ocfs2_path *path = NULL;
> + struct ocfs2_extent_list *el = NULL;
> + struct ocfs2_extent_rec *rec = NULL;
> + struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
> + u64 refcount_loc = le64_to_cpu(di->i_refcount_loc);
>
> ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode), di_bh);
> ocfs2_init_dealloc_ctxt(&dealloc);
> @@ -1458,6 +1463,27 @@ static int ocfs2_remove_inode_range(struct inode *inode,
> goto out;
> }
>
> + /*
> + * For reflinks, we may need to CoW 2 clusters which might be
> + * partially zero'd later, if hole's start and end offset were
> + * within one cluster(means is not exactly aligned to clustersize).
> + */
> +
> + if (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL) {
> +
> + ret = ocfs2_cow_file_pos(inode, di_bh, byte_start);
> + if (ret) {
> + mlog_errno(ret);
> + goto out;
> + }
> +
> + ret = ocfs2_cow_file_pos(inode, di_bh, byte_start + byte_len);
> + if (ret) {
> + mlog_errno(ret);
> + goto out;
> + }
> + }
> +
> trunc_start = ocfs2_clusters_for_bytes(osb->sb, byte_start);
> trunc_len = (byte_start + byte_len) >> osb->s_clustersize_bits;
> if (trunc_len >= trunc_start)
> @@ -1476,6 +1502,13 @@ static int ocfs2_remove_inode_range(struct inode *inode,
> goto out;
> }
>
> + path = ocfs2_new_path_from_et(&et);
> + if (!path) {
> + ret = -ENOMEM;
> + mlog_errno(ret);
> + goto out;
> + }
> +
> cpos = trunc_start;
> while (trunc_len) {
> ret = ocfs2_get_clusters(inode, cpos, &phys_cpos,
> @@ -1488,11 +1521,28 @@ static int ocfs2_remove_inode_range(struct inode *inode,
> if (alloc_size > trunc_len)
> alloc_size = trunc_len;
>
> + ret = ocfs2_find_path(INODE_CACHE(inode), path, cpos);
> + if (ret) {
> + mlog_errno(ret);
> + goto out;
> + }
> +
> + el = path_leaf_el(path);
> +
> + for (i = le16_to_cpu(el->l_next_free_rec) - 1; i >= 0; i--) {
> + rec = &el->l_recs[i];
> + if (le32_to_cpu(rec->e_cpos) <= cpos)
> + break;
> + }
> +
> + flags = rec->e_flags;
> +
uh, you implement all these while only wanting to get the flags for the
record I guess?
ocfs2_get_clusters have a parameter *extent_flags, just pass it in and
It will give you the flags you want.
> /* Only do work for non-holes */
> if (phys_cpos != 0) {
> ret = ocfs2_remove_btree_range(inode, &et, cpos,
> phys_cpos, alloc_size,
> - &dealloc, 0, 0);
> + &dealloc, refcount_loc,
> + flags);
> if (ret) {
> mlog_errno(ret);
> goto out;
Regards,
Tao
More information about the Ocfs2-devel
mailing list