ocfs2: add ocfs2_write_type_t type to identify the caller of write

Patchset: fix ocfs2 direct io code patch to support sparse file and data
ordering semantics

The idea is to use buffer io(more precisely use the interface
ocfs2_write_begin_nolock & ocfs2_write_end_nolock) to do the zero work
beyond block size.  And clear UNWRITTEN flag until direct io data has
been written to disk, which can prevent data corruption when system
crashed during direct write.

And we will also archive a better performance: eg.  dd direct write new
file with block size 4KB: before this patchset:
  2.5 MB/s
after this patchset:
  66.4 MB/s

This patch (of 8):

To support direct io in ocfs2_write_begin_nolock &
ocfs2_write_end_nolock.

Remove unused args filp & flags.  Add new arg type.  The type is one of
buffer/direct/mmap.  Indicate 3 way to perform write.  buffer/mmap type
has implemented.  direct type will be implemented later.

Signed-off-by: Ryan Ding <ryan.ding@oracle.com>
Reviewed-by: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <joseph.qi@huawei.com>
Cc: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 043110e5212dd8c6ab2e7118ac8f17746448dac0..69d7392f56d8322e92964270d922a11982955579 100644 (file)
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -1224,6 +1224,9 @@ struct ocfs2_write_ctxt {
        /* First cluster allocated in a nonsparse extend */
        u32                             w_first_new_cpos;
 
+       /* Type of caller. Must be one of buffer, mmap, direct.  */
+       ocfs2_write_type_t              w_type;
+
        struct ocfs2_write_cluster_desc w_desc[OCFS2_MAX_CLUSTERS_PER_PAGE];
 
        /*
@@ -1319,7 +1322,8 @@ static void ocfs2_free_write_ctxt(struct ocfs2_write_ctxt *wc)
 
 static int ocfs2_alloc_write_ctxt(struct ocfs2_write_ctxt **wcp,
                                  struct ocfs2_super *osb, loff_t pos,
-                                 unsigned len, struct buffer_head *di_bh)
+                                 unsigned len, ocfs2_write_type_t type,
+                                 struct buffer_head *di_bh)
 {
        u32 cend;
        struct ocfs2_write_ctxt *wc;
@@ -1334,6 +1338,7 @@ static int ocfs2_alloc_write_ctxt(struct ocfs2_write_ctxt **wcp,
        wc->w_clen = cend - wc->w_cpos + 1;
        get_bh(di_bh);
        wc->w_di_bh = di_bh;
+       wc->w_type = type;
 
        if (unlikely(PAGE_CACHE_SHIFT > osb->s_clustersize_bits))
                wc->w_large_pages = 1;
@@ -2077,9 +2082,8 @@ out:
        return ret;
 }
 
-int ocfs2_write_begin_nolock(struct file *filp,
-                            struct address_space *mapping,
-                            loff_t pos, unsigned len, unsigned flags,
+int ocfs2_write_begin_nolock(struct address_space *mapping,
+                            loff_t pos, unsigned len, ocfs2_write_type_t type,
                             struct page **pagep, void **fsdata,
                             struct buffer_head *di_bh, struct page *mmap_page)
 {
@@ -2096,7 +2100,7 @@ int ocfs2_write_begin_nolock(struct file *filp,
        int try_free = 1, ret1;
 
 try_again:
-       ret = ocfs2_alloc_write_ctxt(&wc, osb, pos, len, di_bh);
+       ret = ocfs2_alloc_write_ctxt(&wc, osb, pos, len, type, di_bh);
        if (ret) {
                mlog_errno(ret);
                return ret;
@@ -2153,7 +2157,7 @@ try_again:
                        (unsigned long long)OCFS2_I(inode)->ip_blkno,
                        (long long)i_size_read(inode),
                        le32_to_cpu(di->i_clusters),
-                       pos, len, flags, mmap_page,
+                       pos, len, type, mmap_page,
                        clusters_to_alloc, extents_to_split);
 
        /*
@@ -2323,8 +2327,8 @@ static int ocfs2_write_begin(struct file *file, struct address_space *mapping,
         */
        down_write(&OCFS2_I(inode)->ip_alloc_sem);
 
-       ret = ocfs2_write_begin_nolock(file, mapping, pos, len, flags, pagep,
-                                      fsdata, di_bh, NULL);
+       ret = ocfs2_write_begin_nolock(mapping, pos, len, OCFS2_WRITE_BUFFER,
+                                      pagep, fsdata, di_bh, NULL);
        if (ret) {
                mlog_errno(ret);
                goto out_fail;

diff --git a/fs/ocfs2/aops.h b/fs/ocfs2/aops.h
index 24e496d6bdcdba9036dbc67d5f118aa9d10f431a..d06b80f58f83d162ffdeb757a80e5b047fe13911 100644 (file)
--- a/fs/ocfs2/aops.h
+++ b/fs/ocfs2/aops.h
@@ -47,9 +47,14 @@ int ocfs2_write_end_nolock(struct address_space *mapping,
                           loff_t pos, unsigned len, unsigned copied,
                           struct page *page, void *fsdata);
 
-int ocfs2_write_begin_nolock(struct file *filp,
-                            struct address_space *mapping,
-                            loff_t pos, unsigned len, unsigned flags,
+typedef enum {
+       OCFS2_WRITE_BUFFER = 0,
+       OCFS2_WRITE_DIRECT,
+       OCFS2_WRITE_MMAP,
+} ocfs2_write_type_t;
+
+int ocfs2_write_begin_nolock(struct address_space *mapping,
+                            loff_t pos, unsigned len, ocfs2_write_type_t type,
                             struct page **pagep, void **fsdata,
                             struct buffer_head *di_bh, struct page *mmap_page);
 

diff --git a/fs/ocfs2/mmap.c b/fs/ocfs2/mmap.c
index 77ebc2bc1cca112056501fe4dc160d48cc7069cd..9ea081f4e6e4613a18a51c21f95de5a3f14183fb 100644 (file)
--- a/fs/ocfs2/mmap.c
+++ b/fs/ocfs2/mmap.c
@@ -104,8 +104,8 @@ static int __ocfs2_page_mkwrite(struct file *file, struct buffer_head *di_bh,
        if (page->index == last_index)
                len = ((size - 1) & ~PAGE_CACHE_MASK) + 1;
 
-       ret = ocfs2_write_begin_nolock(file, mapping, pos, len, 0, &locked_page,
-                                      &fsdata, di_bh, page);
+       ret = ocfs2_write_begin_nolock(mapping, pos, len, OCFS2_WRITE_MMAP,
+                                      &locked_page, &fsdata, di_bh, page);
        if (ret) {
                if (ret != -ENOSPC)
                        mlog_errno(ret);