ocfs2: do not use ocfs2_zero_extend during direct IO
authorJoseph Qi <joseph.qi@huawei.com>
Tue, 14 Apr 2015 22:43:13 +0000 (15:43 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 14 Apr 2015 23:48:57 +0000 (16:48 -0700)
In ocfs2_direct_IO_write, we use ocfs2_zero_extend to zero allocated
clusters in case of cluster not aligned.  But ocfs2_zero_extend uses page
cache, this may happen that it clears the data which blockdev_direct_IO
has already written.

We should use blkdev_issue_zeroout instead of ocfs2_zero_extend during
direct IO.

So fix this issue by introducing ocfs2_direct_IO_zero_extend and
ocfs2_direct_IO_extend_no_holes.

Reported-by: Yiwen Jiang <jiangyiwen@huawei.com>
Signed-off-by: Joseph Qi <joseph.qi@huawei.com>
Tested-by: Yiwen Jiang <jiangyiwen@huawei.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/ocfs2/aops.c

index 973a636285d192a85251b6471f385470a2ef4661..1b0463a92b17cf7962fa7c5232c2eca21c0350d1 100644 (file)
@@ -663,6 +663,117 @@ static int ocfs2_is_overwrite(struct ocfs2_super *osb,
        return 0;
 }
 
+static int ocfs2_direct_IO_zero_extend(struct ocfs2_super *osb,
+               struct inode *inode, loff_t offset,
+               u64 zero_len, int cluster_align)
+{
+       u32 p_cpos = 0;
+       u32 v_cpos = ocfs2_bytes_to_clusters(osb->sb, i_size_read(inode));
+       unsigned int num_clusters = 0;
+       unsigned int ext_flags = 0;
+       int ret = 0;
+
+       if (offset <= i_size_read(inode) || cluster_align)
+               return 0;
+
+       ret = ocfs2_get_clusters(inode, v_cpos, &p_cpos, &num_clusters,
+                       &ext_flags);
+       if (ret < 0) {
+               mlog_errno(ret);
+               return ret;
+       }
+
+       if (p_cpos && !(ext_flags & OCFS2_EXT_UNWRITTEN)) {
+               u64 s = i_size_read(inode);
+               sector_t sector = (p_cpos << (osb->s_clustersize_bits - 9)) +
+                       (do_div(s, osb->s_clustersize) >> 9);
+
+               ret = blkdev_issue_zeroout(osb->sb->s_bdev, sector,
+                               zero_len >> 9, GFP_NOFS, false);
+               if (ret < 0)
+                       mlog_errno(ret);
+       }
+
+       return ret;
+}
+
+static int ocfs2_direct_IO_extend_no_holes(struct ocfs2_super *osb,
+               struct inode *inode, loff_t offset)
+{
+       u64 zero_start, zero_len, total_zero_len;
+       u32 p_cpos = 0, clusters_to_add;
+       u32 v_cpos = ocfs2_bytes_to_clusters(osb->sb, i_size_read(inode));
+       unsigned int num_clusters = 0;
+       unsigned int ext_flags = 0;
+       u32 size_div, offset_div;
+       int ret = 0;
+
+       {
+               u64 o = offset;
+               u64 s = i_size_read(inode);
+
+               offset_div = do_div(o, osb->s_clustersize);
+               size_div = do_div(s, osb->s_clustersize);
+       }
+
+       if (offset <= i_size_read(inode))
+               return 0;
+
+       clusters_to_add = ocfs2_bytes_to_clusters(inode->i_sb, offset) -
+               ocfs2_bytes_to_clusters(inode->i_sb, i_size_read(inode));
+       total_zero_len = offset - i_size_read(inode);
+       if (clusters_to_add)
+               total_zero_len -= offset_div;
+
+       /* Allocate clusters to fill out holes, and this is only needed
+        * when we add more than one clusters. Otherwise the cluster will
+        * be allocated during direct IO */
+       if (clusters_to_add > 1) {
+               ret = ocfs2_extend_allocation(inode,
+                               OCFS2_I(inode)->ip_clusters,
+                               clusters_to_add - 1, 0);
+               if (ret) {
+                       mlog_errno(ret);
+                       goto out;
+               }
+       }
+
+       while (total_zero_len) {
+               ret = ocfs2_get_clusters(inode, v_cpos, &p_cpos, &num_clusters,
+                               &ext_flags);
+               if (ret < 0) {
+                       mlog_errno(ret);
+                       goto out;
+               }
+
+               zero_start = ocfs2_clusters_to_bytes(osb->sb, p_cpos) +
+                       size_div;
+               zero_len = ocfs2_clusters_to_bytes(osb->sb, num_clusters) -
+                       size_div;
+               zero_len = min(total_zero_len, zero_len);
+
+               if (p_cpos && !(ext_flags & OCFS2_EXT_UNWRITTEN)) {
+                       ret = blkdev_issue_zeroout(osb->sb->s_bdev,
+                                       zero_start >> 9, zero_len >> 9,
+                                       GFP_NOFS, false);
+                       if (ret < 0) {
+                               mlog_errno(ret);
+                               goto out;
+                       }
+               }
+
+               total_zero_len -= zero_len;
+               v_cpos += ocfs2_bytes_to_clusters(osb->sb, zero_len + size_div);
+
+               /* Only at first iteration can be cluster not aligned.
+                * So set size_div to 0 for the rest */
+               size_div = 0;
+       }
+
+out:
+       return ret;
+}
+
 static ssize_t ocfs2_direct_IO_write(struct kiocb *iocb,
                struct iov_iter *iter,
                loff_t offset)
@@ -677,8 +788,8 @@ static ssize_t ocfs2_direct_IO_write(struct kiocb *iocb,
        struct buffer_head *di_bh = NULL;
        size_t count = iter->count;
        journal_t *journal = osb->journal->j_journal;
-       u32 zero_len;
-       int cluster_align;
+       u64 zero_len_head, zero_len_tail;
+       int cluster_align_head, cluster_align_tail;
        loff_t final_size = offset + count;
        int append_write = offset >= i_size_read(inode) ? 1 : 0;
        unsigned int num_clusters = 0;
@@ -686,9 +797,16 @@ static ssize_t ocfs2_direct_IO_write(struct kiocb *iocb,
 
        {
                u64 o = offset;
+               u64 s = i_size_read(inode);
+
+               zero_len_head = do_div(o, 1 << osb->s_clustersize_bits);
+               cluster_align_head = !zero_len_head;
 
-               zero_len = do_div(o, 1 << osb->s_clustersize_bits);
-               cluster_align = !zero_len;
+               zero_len_tail = osb->s_clustersize -
+                       do_div(s, osb->s_clustersize);
+               if ((offset - i_size_read(inode)) < zero_len_tail)
+                       zero_len_tail = offset - i_size_read(inode);
+               cluster_align_tail = !zero_len_tail;
        }
 
        /*
@@ -712,10 +830,13 @@ static ssize_t ocfs2_direct_IO_write(struct kiocb *iocb,
                        goto clean_orphan;
                }
 
+               /* zeroing out the previously allocated cluster tail
+                * that but not zeroed */
                if (ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb)))
-                       ret = ocfs2_zero_extend(inode, di_bh, offset);
+                       ret = ocfs2_direct_IO_zero_extend(osb, inode, offset,
+                                       zero_len_tail, cluster_align_tail);
                else
-                       ret = ocfs2_extend_no_holes(inode, di_bh, offset,
+                       ret = ocfs2_direct_IO_extend_no_holes(osb, inode,
                                        offset);
                if (ret < 0) {
                        mlog_errno(ret);
@@ -768,7 +889,8 @@ static ssize_t ocfs2_direct_IO_write(struct kiocb *iocb,
                                mlog_errno(ret);
                }
        } else if (written > 0 && append_write && !is_overwrite &&
-                       !cluster_align) {
+                       !cluster_align_head) {
+               /* zeroing out the allocated cluster head */
                u32 p_cpos = 0;
                u32 v_cpos = ocfs2_bytes_to_clusters(osb->sb, offset);
 
@@ -790,7 +912,7 @@ static ssize_t ocfs2_direct_IO_write(struct kiocb *iocb,
 
                ret = blkdev_issue_zeroout(osb->sb->s_bdev,
                                p_cpos << (osb->s_clustersize_bits - 9),
-                               zero_len >> 9, GFP_NOFS, false);
+                               zero_len_head >> 9, GFP_NOFS, false);
                if (ret < 0)
                        mlog_errno(ret);