remove SWRITE* I/O types

These flags aren't real I/O types, but tell ll_rw_block to always
lock the buffer instead of giving up on a failed trylock.

Instead add a new write_dirty_buffer helper that implements this semantic
and use it from the existing SWRITE* callers.  Note that the ll_rw_block
code had a bug where it didn't promote WRITE_SYNC_PLUG properly, which
this patch fixes.

In the ufs code clean up the helper that used to call ll_rw_block
to mirror sync_dirty_buffer, which is the function it implements for
compound buffers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

fs/buffer.c

@@ -770,11 +770,12 @@ static int fsync_buffers_list(spinlock_t *lock, struct list_head *list)
 				spin_unlock(lock);
 				/*
 				 * Ensure any pending I/O completes so that
-				 * ll_rw_block() actually writes the current
+				 * write_dirty_buffer() actually writes the
-				 * contents - it is a noop if I/O is still in
+				 * current contents - it is a noop if I/O is
-				 * flight on potentially older contents.
+				 * still in flight on potentially older
+				 * contents.
 				 */
-				ll_rw_block(SWRITE_SYNC_PLUG, 1, &bh);
+				write_dirty_buffer(bh, WRITE_SYNC_PLUG);
 
 				/*
 				 * Kick off IO for the previous mapping. Note
@@ -2949,22 +2950,21 @@ EXPORT_SYMBOL(submit_bh);
 
 /**
  * ll_rw_block: low-level access to block devices (DEPRECATED)
- * @rw: whether to %READ or %WRITE or %SWRITE or maybe %READA (readahead)
+ * @rw: whether to %READ or %WRITE or maybe %READA (readahead)
  * @nr: number of &struct buffer_heads in the array
  * @bhs: array of pointers to &struct buffer_head
  *
  * ll_rw_block() takes an array of pointers to &struct buffer_heads, and
  * requests an I/O operation on them, either a %READ or a %WRITE.  The third
- * %SWRITE is like %WRITE only we make sure that the *current* data in buffers
+ * %READA option is described in the documentation for generic_make_request()
- * are sent to disk. The fourth %READA option is described in the documentation
+ * which ll_rw_block() calls.
- * for generic_make_request() which ll_rw_block() calls.
  *
  * This function drops any buffer that it cannot get a lock on (with the
- * BH_Lock state bit) unless SWRITE is required, any buffer that appears to be
+ * BH_Lock state bit), any buffer that appears to be clean when doing a write
- * clean when doing a write request, and any buffer that appears to be
+ * request, and any buffer that appears to be up-to-date when doing read
- * up-to-date when doing read request.  Further it marks as clean buffers that
+ * request.  Further it marks as clean buffers that are processed for
- * are processed for writing (the buffer cache won't assume that they are
+ * writing (the buffer cache won't assume that they are actually clean
- * actually clean until the buffer gets unlocked).
+ * until the buffer gets unlocked).
  *
  * ll_rw_block sets b_end_io to simple completion handler that marks
  * the buffer up-to-date (if approriate), unlocks the buffer and wakes
@@ -2980,20 +2980,13 @@ void ll_rw_block(int rw, int nr, struct buffer_head *bhs[])
 	for (i = 0; i < nr; i++) {
 		struct buffer_head *bh = bhs[i];
 
-		if (rw == SWRITE || rw == SWRITE_SYNC || rw == SWRITE_SYNC_PLUG)
+		if (!trylock_buffer(bh))
-			lock_buffer(bh);
-		else if (!trylock_buffer(bh))
 			continue;
-
+		if (rw == WRITE) {
-		if (rw == WRITE || rw == SWRITE || rw == SWRITE_SYNC ||
-		    rw == SWRITE_SYNC_PLUG) {
 			if (test_clear_buffer_dirty(bh)) {
 				bh->b_end_io = end_buffer_write_sync;
 				get_bh(bh);
-				if (rw == SWRITE_SYNC)
+				submit_bh(WRITE, bh);
-					submit_bh(WRITE_SYNC, bh);
-				else
-					submit_bh(WRITE, bh);
 				continue;
 			}
 		} else {
@@ -3009,6 +3002,19 @@ void ll_rw_block(int rw, int nr, struct buffer_head *bhs[])
 }
 EXPORT_SYMBOL(ll_rw_block);
 
+void write_dirty_buffer(struct buffer_head *bh, int rw)
+{
+	lock_buffer(bh);
+	if (!test_clear_buffer_dirty(bh)) {
+		unlock_buffer(bh);
+		return;
+	}
+	bh->b_end_io = end_buffer_write_sync;
+	get_bh(bh);
+	submit_bh(rw, bh);
+}
+EXPORT_SYMBOL(write_dirty_buffer);
+
 /*
  * For a data-integrity writeout, we need to wait upon any in-progress I/O
  * and then start new I/O and then wait upon it.  The caller must have a ref on

fs/fat/misc.c

@@ -250,7 +250,9 @@ int fat_sync_bhs(struct buffer_head **bhs, int nr_bhs)
 {
 	int i, err = 0;
 
-	ll_rw_block(SWRITE, nr_bhs, bhs);
+	for (i = 0; i < nr_bhs; i++)
+		write_dirty_buffer(bhs[i], WRITE);
+
 	for (i = 0; i < nr_bhs; i++) {
 		wait_on_buffer(bhs[i]);
 		if (buffer_eopnotsupp(bhs[i])) {

fs/jbd/checkpoint.c

@@ -254,7 +254,9 @@ __flush_batch(journal_t *journal, struct buffer_head **bhs, int *batch_count)
 {
 	int i;
 
-	ll_rw_block(SWRITE, *batch_count, bhs);
+	for (i = 0; i < *batch_count; i++)
+		write_dirty_buffer(bhs[i], WRITE);
+
 	for (i = 0; i < *batch_count; i++) {
 		struct buffer_head *bh = bhs[i];
 		clear_buffer_jwrite(bh);

fs/jbd/journal.c

@@ -1024,7 +1024,7 @@ void journal_update_superblock(journal_t *journal, int wait)
 	if (wait)
 		sync_dirty_buffer(bh);
 	else
-		ll_rw_block(SWRITE, 1, &bh);
+		write_dirty_buffer(bh, WRITE);
 
 out:
 	/* If we have just flushed the log (by marking s_start==0), then

fs/jbd/revoke.c

@@ -617,7 +617,7 @@ static void flush_descriptor(journal_t *journal,
 	set_buffer_jwrite(bh);
 	BUFFER_TRACE(bh, "write");
 	set_buffer_dirty(bh);
-	ll_rw_block((write_op == WRITE) ? SWRITE : SWRITE_SYNC_PLUG, 1, &bh);
+	write_dirty_buffer(bh, write_op);
 }
 #endif
 

fs/jbd2/checkpoint.c

@@ -255,7 +255,9 @@ __flush_batch(journal_t *journal, int *batch_count)
 {
 	int i;
 
-	ll_rw_block(SWRITE, *batch_count, journal->j_chkpt_bhs);
+	for (i = 0; i < *batch_count; i++)
+		write_dirty_buffer(journal->j_chkpt_bhs[i], WRITE);
+
 	for (i = 0; i < *batch_count; i++) {
 		struct buffer_head *bh = journal->j_chkpt_bhs[i];
 		clear_buffer_jwrite(bh);

fs/jbd2/journal.c

@@ -1124,7 +1124,7 @@ void jbd2_journal_update_superblock(journal_t *journal, int wait)
 			set_buffer_uptodate(bh);
 		}
 	} else
-		ll_rw_block(SWRITE, 1, &bh);
+		write_dirty_buffer(bh, WRITE);
 
 out:
 	/* If we have just flushed the log (by marking s_start==0), then

fs/jbd2/revoke.c

@@ -625,7 +625,7 @@ static void flush_descriptor(journal_t *journal,
 	set_buffer_jwrite(bh);
 	BUFFER_TRACE(bh, "write");
 	set_buffer_dirty(bh);
-	ll_rw_block((write_op == WRITE) ? SWRITE : SWRITE_SYNC_PLUG, 1, &bh);
+	write_dirty_buffer(bh, write_op);
 }
 #endif
 

fs/reiserfs/journal.c

@@ -2311,7 +2311,7 @@ static int journal_read_transaction(struct super_block *sb,
 	/* flush out the real blocks */
 	for (i = 0; i < get_desc_trans_len(desc); i++) {
 		set_buffer_dirty(real_blocks[i]);
-		ll_rw_block(SWRITE, 1, real_blocks + i);
+		write_dirty_buffer(real_blocks[i], WRITE);
 	}
 	for (i = 0; i < get_desc_trans_len(desc); i++) {
 		wait_on_buffer(real_blocks[i]);

fs/ufs/balloc.c

@@ -114,10 +114,8 @@ void ufs_free_fragments(struct inode *inode, u64 fragment, unsigned count)
 	
 	ubh_mark_buffer_dirty (USPI_UBH(uspi));
 	ubh_mark_buffer_dirty (UCPI_UBH(ucpi));
-	if (sb->s_flags & MS_SYNCHRONOUS) {
+	if (sb->s_flags & MS_SYNCHRONOUS)
-		ubh_ll_rw_block(SWRITE, UCPI_UBH(ucpi));
+		ubh_sync_block(UCPI_UBH(ucpi));
-		ubh_wait_on_buffer (UCPI_UBH(ucpi));
-	}
 	sb->s_dirt = 1;
 	
 	unlock_super (sb);
@@ -207,10 +205,8 @@ do_more:
 
 	ubh_mark_buffer_dirty (USPI_UBH(uspi));
 	ubh_mark_buffer_dirty (UCPI_UBH(ucpi));
-	if (sb->s_flags & MS_SYNCHRONOUS) {
+	if (sb->s_flags & MS_SYNCHRONOUS)
-		ubh_ll_rw_block(SWRITE, UCPI_UBH(ucpi));
+		ubh_sync_block(UCPI_UBH(ucpi));
-		ubh_wait_on_buffer (UCPI_UBH(ucpi));
-	}
 
 	if (overflow) {
 		fragment += count;
@@ -558,10 +554,8 @@ static u64 ufs_add_fragments(struct inode *inode, u64 fragment,
 	
 	ubh_mark_buffer_dirty (USPI_UBH(uspi));
 	ubh_mark_buffer_dirty (UCPI_UBH(ucpi));
-	if (sb->s_flags & MS_SYNCHRONOUS) {
+	if (sb->s_flags & MS_SYNCHRONOUS)
-		ubh_ll_rw_block(SWRITE, UCPI_UBH(ucpi));
+		ubh_sync_block(UCPI_UBH(ucpi));
-		ubh_wait_on_buffer (UCPI_UBH(ucpi));
-	}
 	sb->s_dirt = 1;
 
 	UFSD("EXIT, fragment %llu\n", (unsigned long long)fragment);
@@ -680,10 +674,8 @@ cg_found:
 succed:
 	ubh_mark_buffer_dirty (USPI_UBH(uspi));
 	ubh_mark_buffer_dirty (UCPI_UBH(ucpi));
-	if (sb->s_flags & MS_SYNCHRONOUS) {
+	if (sb->s_flags & MS_SYNCHRONOUS)
-		ubh_ll_rw_block(SWRITE, UCPI_UBH(ucpi));
+		ubh_sync_block(UCPI_UBH(ucpi));
-		ubh_wait_on_buffer (UCPI_UBH(ucpi));
-	}
 	sb->s_dirt = 1;
 
 	result += cgno * uspi->s_fpg;

fs/ufs/ialloc.c

@@ -113,10 +113,8 @@ void ufs_free_inode (struct inode * inode)
 
 	ubh_mark_buffer_dirty (USPI_UBH(uspi));
 	ubh_mark_buffer_dirty (UCPI_UBH(ucpi));
-	if (sb->s_flags & MS_SYNCHRONOUS) {
+	if (sb->s_flags & MS_SYNCHRONOUS)
-		ubh_ll_rw_block(SWRITE, UCPI_UBH(ucpi));
+		ubh_sync_block(UCPI_UBH(ucpi));
-		ubh_wait_on_buffer (UCPI_UBH(ucpi));
-	}
 	
 	sb->s_dirt = 1;
 	unlock_super (sb);
@@ -156,10 +154,8 @@ static void ufs2_init_inodes_chunk(struct super_block *sb,
 
 	fs32_add(sb, &ucg->cg_u.cg_u2.cg_initediblk, uspi->s_inopb);
 	ubh_mark_buffer_dirty(UCPI_UBH(ucpi));
-	if (sb->s_flags & MS_SYNCHRONOUS) {
+	if (sb->s_flags & MS_SYNCHRONOUS)
-		ubh_ll_rw_block(SWRITE, UCPI_UBH(ucpi));
+		ubh_sync_block(UCPI_UBH(ucpi));
-		ubh_wait_on_buffer(UCPI_UBH(ucpi));
-	}
 
 	UFSD("EXIT\n");
 }
@@ -290,10 +286,8 @@ cg_found:
 	}
 	ubh_mark_buffer_dirty (USPI_UBH(uspi));
 	ubh_mark_buffer_dirty (UCPI_UBH(ucpi));
-	if (sb->s_flags & MS_SYNCHRONOUS) {
+	if (sb->s_flags & MS_SYNCHRONOUS)
-		ubh_ll_rw_block(SWRITE, UCPI_UBH(ucpi));
+		ubh_sync_block(UCPI_UBH(ucpi));
-		ubh_wait_on_buffer (UCPI_UBH(ucpi));
-	}
 	sb->s_dirt = 1;
 
 	inode->i_ino = cg * uspi->s_ipg + bit;

fs/ufs/truncate.c

@@ -243,10 +243,8 @@ static int ufs_trunc_indirect(struct inode *inode, u64 offset, void *p)
 		ubh_bforget(ind_ubh);
 		ind_ubh = NULL;
 	}
-	if (IS_SYNC(inode) && ind_ubh && ubh_buffer_dirty(ind_ubh)) {
+	if (IS_SYNC(inode) && ind_ubh && ubh_buffer_dirty(ind_ubh))
-		ubh_ll_rw_block(SWRITE, ind_ubh);
+		ubh_sync_block(ind_ubh);
-		ubh_wait_on_buffer (ind_ubh);
-	}
 	ubh_brelse (ind_ubh);
 	
 	UFSD("EXIT: ino %lu\n", inode->i_ino);
@@ -307,10 +305,8 @@ static int ufs_trunc_dindirect(struct inode *inode, u64 offset, void *p)
 		ubh_bforget(dind_bh);
 		dind_bh = NULL;
 	}
-	if (IS_SYNC(inode) && dind_bh && ubh_buffer_dirty(dind_bh)) {
+	if (IS_SYNC(inode) && dind_bh && ubh_buffer_dirty(dind_bh))
-		ubh_ll_rw_block(SWRITE, dind_bh);
+		ubh_sync_block(dind_bh);
-		ubh_wait_on_buffer (dind_bh);
-	}
 	ubh_brelse (dind_bh);
 	
 	UFSD("EXIT: ino %lu\n", inode->i_ino);
@@ -367,10 +363,8 @@ static int ufs_trunc_tindirect(struct inode *inode)
 		ubh_bforget(tind_bh);
 		tind_bh = NULL;
 	}
-	if (IS_SYNC(inode) && tind_bh && ubh_buffer_dirty(tind_bh)) {
+	if (IS_SYNC(inode) && tind_bh && ubh_buffer_dirty(tind_bh))
-		ubh_ll_rw_block(SWRITE, tind_bh);
+		ubh_sync_block(tind_bh);
-		ubh_wait_on_buffer (tind_bh);
-	}
 	ubh_brelse (tind_bh);
 	
 	UFSD("EXIT: ino %lu\n", inode->i_ino);

fs/ufs/util.c

@@ -113,21 +113,17 @@ void ubh_mark_buffer_uptodate (struct ufs_buffer_head * ubh, int flag)
 	}
 }
 
-void ubh_ll_rw_block(int rw, struct ufs_buffer_head *ubh)
+void ubh_sync_block(struct ufs_buffer_head *ubh)
 {
-	if (!ubh)
+	if (ubh) {
-		return;
+		unsigned i;
 
-	ll_rw_block(rw, ubh->count, ubh->bh);
+		for (i = 0; i < ubh->count; i++)
-}
+			write_dirty_buffer(ubh->bh[i], WRITE);
 
-void ubh_wait_on_buffer (struct ufs_buffer_head * ubh)
+		for (i = 0; i < ubh->count; i++)
-{
+			wait_on_buffer(ubh->bh[i]);
-	unsigned i;
+	}
-	if (!ubh)
-		return;
-	for ( i = 0; i < ubh->count; i++ )
-		wait_on_buffer (ubh->bh[i]);
 }
 
 void ubh_bforget (struct ufs_buffer_head * ubh)

fs/ufs/util.h

@@ -269,8 +269,7 @@ extern void ubh_brelse (struct ufs_buffer_head *);
 extern void ubh_brelse_uspi (struct ufs_sb_private_info *);
 extern void ubh_mark_buffer_dirty (struct ufs_buffer_head *);
 extern void ubh_mark_buffer_uptodate (struct ufs_buffer_head *, int);
-extern void ubh_ll_rw_block(int, struct ufs_buffer_head *);
+extern void ubh_sync_block(struct ufs_buffer_head *);
-extern void ubh_wait_on_buffer (struct ufs_buffer_head *);
 extern void ubh_bforget (struct ufs_buffer_head *);
 extern int  ubh_buffer_dirty (struct ufs_buffer_head *);
 #define ubh_ubhcpymem(mem,ubh,size) _ubh_ubhcpymem_(uspi,mem,ubh,size)

include/linux/buffer_head.h

@@ -182,6 +182,7 @@ void __lock_buffer(struct buffer_head *bh);
 void ll_rw_block(int, int, struct buffer_head * bh[]);
 int sync_dirty_buffer(struct buffer_head *bh);
 int __sync_dirty_buffer(struct buffer_head *bh, int rw);
+void write_dirty_buffer(struct buffer_head *bh, int rw);
 int submit_bh(int, struct buffer_head *);
 void write_boundary_block(struct block_device *bdev,
 			sector_t bblock, unsigned blocksize);

include/linux/fs.h

@@ -125,9 +125,6 @@ struct inodes_stat_t {
  *			block layer could (in theory) choose to ignore this
  *			request if it runs into resource problems.
  * WRITE		A normal async write. Device will be plugged.
- * SWRITE		Like WRITE, but a special case for ll_rw_block() that
- *			tells it to lock the buffer first. Normally a buffer
- *			must be locked before doing IO.
  * WRITE_SYNC_PLUG	Synchronous write. Identical to WRITE, but passes down
  *			the hint that someone will be waiting on this IO
  *			shortly. The device must still be unplugged explicitly,
@@ -138,9 +135,6 @@ struct inodes_stat_t {
  *			immediately after submission. The write equivalent
  *			of READ_SYNC.
  * WRITE_ODIRECT_PLUG	Special case write for O_DIRECT only.
- * SWRITE_SYNC
- * SWRITE_SYNC_PLUG	Like WRITE_SYNC/WRITE_SYNC_PLUG, but locks the buffer.
- *			See SWRITE.
  * WRITE_BARRIER	Like WRITE_SYNC, but tells the block layer that all
  *			previously submitted writes must be safely on storage
  *			before this one is started. Also guarantees that when
@@ -155,7 +149,6 @@ struct inodes_stat_t {
 #define READ			0
 #define WRITE			RW_MASK
 #define READA			RWA_MASK
-#define SWRITE			(WRITE | READA)
 
 #define READ_SYNC		(READ | REQ_SYNC | REQ_UNPLUG)
 #define READ_META		(READ | REQ_META)
@@ -165,8 +158,6 @@ struct inodes_stat_t {
 #define WRITE_META		(WRITE | REQ_META)
 #define WRITE_BARRIER		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG | \
 				 REQ_HARDBARRIER)
-#define SWRITE_SYNC_PLUG	(SWRITE | REQ_SYNC | REQ_NOIDLE)
-#define SWRITE_SYNC		(SWRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG)
 
 /*
  * These aren't really reads or writes, they pass down information about