ext4/jbd2: data=journal: write-protect pages on transaction commit

index 6d2da8ad0e6f..f79b86b4241f 100644
@@ -187,19 +187,17 @@ static int journal_wait_on_commit_record(journal_t *journal,
  * use writepages() because with delayed allocation we may be doing
  * block allocation in writepages().
  */
-static int journal_submit_inode_data_buffers(struct address_space *mapping,
-		loff_t dirty_start, loff_t dirty_end)
+int jbd2_journal_submit_inode_data_buffers(struct jbd2_inode *jinode)
 {
-	int ret;
+	struct address_space *mapping = jinode->i_vfs_inode->i_mapping;
 	struct writeback_control wbc = {
 		.sync_mode =  WB_SYNC_ALL,
 		.nr_to_write = mapping->nrpages * 2,
-		.range_start = dirty_start,
-		.range_end = dirty_end,
+		.range_start = jinode->i_dirty_start,
+		.range_end = jinode->i_dirty_end,
 	};
 
-	ret = generic_writepages(mapping, &wbc);
-	return ret;
+	return generic_writepages(mapping, &wbc);
 }
 
 /*

@@ -215,16 +213,11 @@ static int journal_submit_data_buffers(journal_t *journal,
 {
 	struct jbd2_inode *jinode;
 	int err, ret = 0;
-	struct address_space *mapping;
 
 	spin_lock(&journal->j_list_lock);
 	list_for_each_entry(jinode, &commit_transaction->t_inode_list, i_list) {
-		loff_t dirty_start = jinode->i_dirty_start;
-		loff_t dirty_end = jinode->i_dirty_end;
-
 		if (!(jinode->i_flags & JI_WRITE_DATA))
 			continue;
-		mapping = jinode->i_vfs_inode->i_mapping;
 		jinode->i_flags |= JI_COMMIT_RUNNING;
 		spin_unlock(&journal->j_list_lock);
 		/*

@@ -234,8 +227,7 @@ static int journal_submit_data_buffers(journal_t *journal,
 		 * only allocated blocks here.
 		 */
 		trace_jbd2_submit_inode_data(jinode->i_vfs_inode);
-		err = journal_submit_inode_data_buffers(mapping, dirty_start,
-				dirty_end);
+		err = jbd2_journal_submit_inode_data_buffers(jinode);
 		if (!ret)
 			ret = err;
 		spin_lock(&journal->j_list_lock);

@@ -248,6 +240,15 @@ static int journal_submit_data_buffers(journal_t *journal,
 	return ret;
 }
 
+int jbd2_journal_finish_inode_data_buffers(struct jbd2_inode *jinode)
+{
+	struct address_space *mapping = jinode->i_vfs_inode->i_mapping;
+
+	return filemap_fdatawait_range_keep_errors(mapping,
+						   jinode->i_dirty_start,
+						   jinode->i_dirty_end);
+}
+
 /*
  * Wait for data submitted for writeout, refile inodes to proper
  * transaction if needed.

@@ -262,16 +263,11 @@ static int journal_finish_inode_data_buffers(journal_t *journal,
 	/* For locking, see the comment in journal_submit_data_buffers() */
 	spin_lock(&journal->j_list_lock);
 	list_for_each_entry(jinode, &commit_transaction->t_inode_list, i_list) {
-		loff_t dirty_start = jinode->i_dirty_start;
-		loff_t dirty_end = jinode->i_dirty_end;
-
 		if (!(jinode->i_flags & JI_WAIT_DATA))
 			continue;
 		jinode->i_flags |= JI_COMMIT_RUNNING;
 		spin_unlock(&journal->j_list_lock);
-		err = filemap_fdatawait_range_keep_errors(
-				jinode->i_vfs_inode->i_mapping, dirty_start,
-				dirty_end);
+		err = jbd2_journal_finish_inode_data_buffers(jinode);
 		if (!ret)
 			ret = err;
 		spin_lock(&journal->j_list_lock);

index 17fdc482f554..c0600405e7a2 100644
@@ -91,6 +91,8 @@ EXPORT_SYMBOL(jbd2_journal_try_to_free_buffers);
 EXPORT_SYMBOL(jbd2_journal_force_commit);
 EXPORT_SYMBOL(jbd2_journal_inode_ranged_write);
 EXPORT_SYMBOL(jbd2_journal_inode_ranged_wait);
+EXPORT_SYMBOL(jbd2_journal_submit_inode_data_buffers);
+EXPORT_SYMBOL(jbd2_journal_finish_inode_data_buffers);
 EXPORT_SYMBOL(jbd2_journal_init_jbd_inode);
 EXPORT_SYMBOL(jbd2_journal_release_jbd_inode);
 EXPORT_SYMBOL(jbd2_journal_begin_ordered_truncate);

index 08f904943ab2..2865a5475888 100644
@@ -1421,6 +1421,10 @@ extern int	   jbd2_journal_inode_ranged_write(handle_t *handle,
 extern int	   jbd2_journal_inode_ranged_wait(handle_t *handle,
 			struct jbd2_inode *inode, loff_t start_byte,
 			loff_t length);
+extern int	   jbd2_journal_submit_inode_data_buffers(
+			struct jbd2_inode *jinode);
+extern int	   jbd2_journal_finish_inode_data_buffers(
+			struct jbd2_inode *jinode);
 extern int	   jbd2_journal_begin_ordered_truncate(journal_t *journal,
 				struct jbd2_inode *inode, loff_t new_size);
 extern void	   jbd2_journal_init_jbd_inode(struct jbd2_inode *jinode, struct inode *inode);

index ea425b49b345..a14c1ed39aa3 100644
@@ -4646,6 +4646,10 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 	set_task_ioprio(sbi->s_journal->j_task, journal_ioprio);
 
 	sbi->s_journal->j_commit_callback = ext4_journal_commit_callback;
+	sbi->s_journal->j_submit_inode_data_buffers =
+		jbd2_journal_submit_inode_data_buffers;
+	sbi->s_journal->j_finish_inode_data_buffers =
+		jbd2_journal_finish_inode_data_buffers;
 
 no_journal:
 	if (!test_opt(sb, NO_MBCACHE)) {

index f79b86b4241f..6252b4c50666 100644
@@ -197,6 +197,12 @@ int jbd2_journal_submit_inode_data_buffers(struct jbd2_inode *jinode)
 		.range_end = jinode->i_dirty_end,
 	};
 
+	/*
+	 * submit the inode data buffers. We use writepage
+	 * instead of writepages. Because writepages can do
+	 * block allocation with delalloc. We need to write
+	 * only allocated blocks here.
+	 */
 	return generic_writepages(mapping, &wbc);
 }
 

@@ -220,16 +226,13 @@ static int journal_submit_data_buffers(journal_t *journal,
 			continue;
 		jinode->i_flags |= JI_COMMIT_RUNNING;
 		spin_unlock(&journal->j_list_lock);
-		/*
-		 * submit the inode data buffers. We use writepage
-		 * instead of writepages. Because writepages can do
-		 * block allocation  with delalloc. We need to write
-		 * only allocated blocks here.
-		 */
+		/* submit the inode data buffers. */
 		trace_jbd2_submit_inode_data(jinode->i_vfs_inode);
-		err = jbd2_journal_submit_inode_data_buffers(jinode);
-		if (!ret)
-			ret = err;
+		if (journal->j_submit_inode_data_buffers) {
+			err = journal->j_submit_inode_data_buffers(jinode);
+			if (!ret)
+				ret = err;
+		}
 		spin_lock(&journal->j_list_lock);
 		J_ASSERT(jinode->i_transaction == commit_transaction);
 		jinode->i_flags &= ~JI_COMMIT_RUNNING;

@@ -267,9 +270,12 @@ static int journal_finish_inode_data_buffers(journal_t *journal,
 			continue;
 		jinode->i_flags |= JI_COMMIT_RUNNING;
 		spin_unlock(&journal->j_list_lock);
-		err = jbd2_journal_finish_inode_data_buffers(jinode);
-		if (!ret)
-			ret = err;
+		/* wait for the inode data buffers writeout. */
+		if (journal->j_finish_inode_data_buffers) {
+			err = journal->j_finish_inode_data_buffers(jinode);
+			if (!ret)
+				ret = err;
+		}
 		spin_lock(&journal->j_list_lock);
 		jinode->i_flags &= ~JI_COMMIT_RUNNING;
 		smp_mb();

index b425f0b01dce..b9a9d69dde7e 100644
@@ -883,6 +883,10 @@ int ocfs2_journal_init(struct ocfs2_journal *journal, int *dirty)
 		  OCFS2_JOURNAL_DIRTY_FL);
 
 	journal->j_journal = j_journal;
+	journal->j_journal->j_submit_inode_data_buffers =
+		jbd2_journal_submit_inode_data_buffers;
+	journal->j_journal->j_finish_inode_data_buffers =
+		jbd2_journal_finish_inode_data_buffers;
 	journal->j_inode = inode;
 	journal->j_bh = bh;
 

index 2865a5475888..4aaa408c0ca7 100644
@@ -629,7 +629,9 @@ struct transaction_s
 	struct journal_head	*t_shadow_list;
 
 	/*
-	 * List of inodes whose data we've modified in data=ordered mode.
+	 * List of inodes associated with the transaction; e.g., ext4 uses
+	 * this to track inodes in data=ordered and data=journal mode that
+	 * need special handling on transaction commit; also used by ocfs2.
 	 * [j_list_lock]
 	 */
 	struct list_head	t_inode_list;

@@ -1111,6 +1113,27 @@ struct journal_s
 	void			(*j_commit_callback)(journal_t *,
 						     transaction_t *);
 
+	/**
+	 * @j_submit_inode_data_buffers:
+	 *
+	 * This function is called for all inodes associated with the
+	 * committing transaction marked with JI_WRITE_DATA flag
+	 * before we start to write out the transaction to the journal.
+	 */
+	int			(*j_submit_inode_data_buffers)
+					(struct jbd2_inode *);
+
+	/**
+	 * @j_finish_inode_data_buffers:
+	 *
+	 * This function is called for all inodes associated with the
+	 * committing transaction marked with JI_WAIT_DATA flag
+	 * after we have written the transaction to the journal
+	 * but before we write out the commit block.
+	 */
+	int			(*j_finish_inode_data_buffers)
+					(struct jbd2_inode *);
+
 	/*
 	 * Journal statistics
 	 */

index bf596467c234..ac153e340a6f 100644
@@ -5977,9 +5977,17 @@ vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf)
 	if (err)
 		goto out_ret;
 
+	/*
+	 * On data journalling we skip straight to the transaction handle:
+	 * there's no delalloc; page truncated will be checked later; the
+	 * early return w/ all buffers mapped (calculates size/len) can't
+	 * be used; and there's no dioread_nolock, so only ext4_get_block.
+	 */
+	if (ext4_should_journal_data(inode))
+		goto retry_alloc;
+
 	/* Delalloc case is easy... */
 	if (test_opt(inode->i_sb, DELALLOC) &&
-	    !ext4_should_journal_data(inode) &&
 	    !ext4_nonda_switch(inode->i_sb)) {
 		do {
 			err = block_page_mkwrite(vma, vmf,

@@ -6005,6 +6013,9 @@ vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf)
 	/*
 	 * Return if we have all the buffers mapped. This avoids the need to do
 	 * journal_start/journal_stop which can block and take a long time
+	 *
+	 * This cannot be done for data journalling, as we have to add the
+	 * inode to the transaction's list to writeprotect pages on commit.
 	 */
 	if (page_has_buffers(page)) {
 		if (!ext4_walk_page_buffers(NULL, page_buffers(page),

@@ -6029,16 +6040,42 @@ vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf)
 		ret = VM_FAULT_SIGBUS;
 		goto out;
 	}
-	err = block_page_mkwrite(vma, vmf, get_block);
-	if (!err && ext4_should_journal_data(inode)) {
-		if (ext4_walk_page_buffers(handle, page_buffers(page), 0,
-			  PAGE_SIZE, NULL, do_journal_get_write_access)) {
+	/*
+	 * Data journalling can't use block_page_mkwrite() because it
+	 * will set_buffer_dirty() before do_journal_get_write_access()
+	 * thus might hit warning messages for dirty metadata buffers.
+	 */
+	if (!ext4_should_journal_data(inode)) {
+		err = block_page_mkwrite(vma, vmf, get_block);
+	} else {
+		lock_page(page);
+		size = i_size_read(inode);
+		/* Page got truncated from under us? */
+		if (page->mapping != mapping || page_offset(page) > size) {
 			unlock_page(page);
-			ret = VM_FAULT_SIGBUS;
+			ret = VM_FAULT_NOPAGE;
 			ext4_journal_stop(handle);
 			goto out;
 		}
-		ext4_set_inode_state(inode, EXT4_STATE_JDATA);
+
+		if (page->index == size >> PAGE_SHIFT)
+			len = size & ~PAGE_MASK;
+		else
+			len = PAGE_SIZE;
+
+		err = __block_write_begin(page, 0, len, ext4_get_block);
+		if (!err) {
+			if (ext4_walk_page_buffers(handle, page_buffers(page),
+					0, len, NULL, do_journal_get_write_access)) {
+				unlock_page(page);
+				ret = VM_FAULT_SIGBUS;
+				ext4_journal_stop(handle);
+				goto out;
+			}
+			ext4_set_inode_state(inode, EXT4_STATE_JDATA);
+		} else {
+			unlock_page(page);
+		}
 	}
 	ext4_journal_stop(handle);
 	if (err == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))

index ac153e340a6f..af5de62c1214 100644
@@ -1910,6 +1910,9 @@ static int __ext4_journalled_writepage(struct page *page,
 		err = ext4_walk_page_buffers(handle, page_bufs, 0, len, NULL,
 					     write_end_fn);
 	}
+	if (ret == 0)
+		ret = err;
+	err = ext4_jbd2_inode_add_write(handle, inode, 0, len);
 	if (ret == 0)
 		ret = err;
 	EXT4_I(inode)->i_datasync_tid = handle->h_transaction->t_tid;

@@ -6052,10 +6055,8 @@ vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf)
 		size = i_size_read(inode);
 		/* Page got truncated from under us? */
 		if (page->mapping != mapping || page_offset(page) > size) {
-			unlock_page(page);
 			ret = VM_FAULT_NOPAGE;
-			ext4_journal_stop(handle);
-			goto out;
+			goto out_error;
 		}
 
 		if (page->index == size >> PAGE_SHIFT)

@@ -6065,13 +6066,15 @@ vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf)
 
 		err = __block_write_begin(page, 0, len, ext4_get_block);
 		if (!err) {
+			ret = VM_FAULT_SIGBUS;
 			if (ext4_walk_page_buffers(handle, page_buffers(page),
-					0, len, NULL, do_journal_get_write_access)) {
-				unlock_page(page);
-				ret = VM_FAULT_SIGBUS;
-				ext4_journal_stop(handle);
-				goto out;
-			}
+					0, len, NULL, do_journal_get_write_access))
+				goto out_error;
+			if (ext4_walk_page_buffers(handle, page_buffers(page),
+					0, len, NULL, write_end_fn))
+				goto out_error;
+			if (ext4_jbd2_inode_add_write(handle, inode, 0, len))
+				goto out_error;
 			ext4_set_inode_state(inode, EXT4_STATE_JDATA);
 		} else {
 			unlock_page(page);

@@ -6086,6 +6089,10 @@ vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf)
 	up_read(&EXT4_I(inode)->i_mmap_sem);
 	sb_end_pagefault(inode->i_sb);
 	return ret;
+out_error:
+	unlock_page(page);
+	ext4_journal_stop(handle);
+	goto out;
 }
 
 vm_fault_t ext4_filemap_fault(struct vm_fault *vmf)

100 lines not on this page · raw message