block: use an on-stack bio in blkdev_issue_flush

There is no point in allocating memory for a synchronous flush.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Acked-by: Damien Le Moal <damien.lemoal@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block/blk-flush.c

@@ -432,23 +432,18 @@ void blk_insert_flush(struct request *rq)
 /**
  * blkdev_issue_flush - queue a flush
  * @bdev:	blockdev to issue flush for
- * @gfp_mask:	memory allocation flags (for bio_alloc)
  *
  * Description:
  *    Issue a flush for the block device in question.
  */
-int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
+int blkdev_issue_flush(struct block_device *bdev)
 {
-	struct bio *bio;
-	int ret = 0;
+	struct bio bio;
 
-	bio = bio_alloc(gfp_mask, 0);
-	bio_set_dev(bio, bdev);
-	bio->bi_opf = REQ_OP_WRITE | REQ_PREFLUSH;
-
-	ret = submit_bio_wait(bio);
-	bio_put(bio);
-	return ret;
+	bio_init(&bio, NULL, 0);
+	bio_set_dev(&bio, bdev);
+	bio.bi_opf = REQ_OP_WRITE | REQ_PREFLUSH;
+	return submit_bio_wait(&bio);
 }
 EXPORT_SYMBOL(blkdev_issue_flush);
 

drivers/md/dm-zoned-metadata.c

@@ -819,7 +819,7 @@ static int dmz_write_sb(struct dmz_metadata *zmd, unsigned int set)
 	ret = dmz_rdwr_block(dev, REQ_OP_WRITE, zmd->sb[set].block,
 			     mblk->page);
 	if (ret == 0)
-		ret = blkdev_issue_flush(dev->bdev, GFP_NOIO);
+		ret = blkdev_issue_flush(dev->bdev);
 
 	return ret;
 }
@@ -862,7 +862,7 @@ static int dmz_write_dirty_mblocks(struct dmz_metadata *zmd,
 
 	/* Flush drive cache (this will also sync data) */
 	if (ret == 0)
-		ret = blkdev_issue_flush(dev->bdev, GFP_NOIO);
+		ret = blkdev_issue_flush(dev->bdev);
 
 	return ret;
 }
@@ -933,7 +933,7 @@ int dmz_flush_metadata(struct dmz_metadata *zmd)
 
 	/* If there are no dirty metadata blocks, just flush the device cache */
 	if (list_empty(&write_list)) {
-		ret = blkdev_issue_flush(dev->bdev, GFP_NOIO);
+		ret = blkdev_issue_flush(dev->bdev);
 		goto err;
 	}
 

drivers/md/raid5-ppl.c

@@ -1037,7 +1037,7 @@ static int ppl_recover(struct ppl_log *log, struct ppl_header *pplhdr,
 	}
 
 	/* flush the disk cache after recovery if necessary */
-	ret = blkdev_issue_flush(rdev->bdev, GFP_KERNEL);
+	ret = blkdev_issue_flush(rdev->bdev);
 out:
 	__free_page(page);
 	return ret;

drivers/nvme/target/io-cmd-bdev.c

@@ -333,7 +333,7 @@ static void nvmet_bdev_execute_flush(struct nvmet_req *req)
 
 u16 nvmet_bdev_flush(struct nvmet_req *req)
 {
-	if (blkdev_issue_flush(req->ns->bdev, GFP_KERNEL))
+	if (blkdev_issue_flush(req->ns->bdev))
 		return NVME_SC_INTERNAL | NVME_SC_DNR;
 	return 0;
 }

fs/block_dev.c

@@ -680,7 +680,7 @@ int blkdev_fsync(struct file *filp, loff_t start, loff_t end, int datasync)
 	 * i_mutex and doing so causes performance issues with concurrent
 	 * O_SYNC writers to a block device.
 	 */
-	error = blkdev_issue_flush(bdev, GFP_KERNEL);
+	error = blkdev_issue_flush(bdev);
 	if (error == -EOPNOTSUPP)
 		error = 0;
 

fs/exfat/file.c

@@ -361,7 +361,7 @@ int exfat_file_fsync(struct file *filp, loff_t start, loff_t end, int datasync)
 	if (err)
 		return err;
 
-	return blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL);
+	return blkdev_issue_flush(inode->i_sb->s_bdev);
 }
 
 const struct file_operations exfat_file_operations = {

fs/ext4/fast_commit.c

@@ -1076,7 +1076,7 @@ static int ext4_fc_perform_commit(journal_t *journal)
 	 * flush before we start writing fast commit blocks.
 	 */
 	if (journal->j_fs_dev != journal->j_dev)
-		blkdev_issue_flush(journal->j_fs_dev, GFP_NOFS);
+		blkdev_issue_flush(journal->j_fs_dev);
 
 	blk_start_plug(&plug);
 	if (sbi->s_fc_bytes == 0) {
@@ -1535,7 +1535,7 @@ static int ext4_fc_replay_inode(struct super_block *sb, struct ext4_fc_tl *tl)
 out:
 	iput(inode);
 	if (!ret)
-		blkdev_issue_flush(sb->s_bdev, GFP_KERNEL);
+		blkdev_issue_flush(sb->s_bdev);
 
 	return 0;
 }

fs/ext4/fsync.c

@@ -174,7 +174,7 @@ int ext4_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 		ret = ext4_fsync_journal(inode, datasync, &needs_barrier);
 
 	if (needs_barrier) {
-		err = blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL);
+		err = blkdev_issue_flush(inode->i_sb->s_bdev);
 		if (!ret)
 			ret = err;
 	}

fs/ext4/ialloc.c

@@ -1583,7 +1583,7 @@ int ext4_init_inode_table(struct super_block *sb, ext4_group_t group,
 	if (ret < 0)
 		goto err_out;
 	if (barrier)
-		blkdev_issue_flush(sb->s_bdev, GFP_NOFS);
+		blkdev_issue_flush(sb->s_bdev);
 
 skip_zeroout:
 	ext4_lock_group(sb, group);

fs/ext4/super.c

@@ -5709,7 +5709,7 @@ static int ext4_sync_fs(struct super_block *sb, int wait)
 		needs_barrier = true;
 	if (needs_barrier) {
 		int err;
-		err = blkdev_issue_flush(sb->s_bdev, GFP_KERNEL);
+		err = blkdev_issue_flush(sb->s_bdev);
 		if (!ret)
 			ret = err;
 	}

fs/fat/file.c

@@ -195,7 +195,7 @@ int fat_file_fsync(struct file *filp, loff_t start, loff_t end, int datasync)
 	if (err)
 		return err;
 
-	return blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL);
+	return blkdev_issue_flush(inode->i_sb->s_bdev);
 }
 
 

fs/hfsplus/inode.c

@@ -340,7 +340,7 @@ int hfsplus_file_fsync(struct file *file, loff_t start, loff_t end,
 	}
 
 	if (!test_bit(HFSPLUS_SB_NOBARRIER, &sbi->flags))
-		blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL);
+		blkdev_issue_flush(inode->i_sb->s_bdev);
 
 	inode_unlock(inode);
 

fs/hfsplus/super.c

@@ -239,7 +239,7 @@ static int hfsplus_sync_fs(struct super_block *sb, int wait)
 	mutex_unlock(&sbi->vh_mutex);
 
 	if (!test_bit(HFSPLUS_SB_NOBARRIER, &sbi->flags))
-		blkdev_issue_flush(sb->s_bdev, GFP_KERNEL);
+		blkdev_issue_flush(sb->s_bdev);
 
 	return error;
 }

fs/jbd2/checkpoint.c

@@ -416,7 +416,7 @@ int jbd2_cleanup_journal_tail(journal_t *journal)
 	 * jbd2_cleanup_journal_tail() doesn't get called all that often.
 	 */
 	if (journal->j_flags & JBD2_BARRIER)
-		blkdev_issue_flush(journal->j_fs_dev, GFP_NOFS);
+		blkdev_issue_flush(journal->j_fs_dev);
 
 	return __jbd2_update_log_tail(journal, first_tid, blocknr);
 }

fs/jbd2/commit.c

@@ -825,7 +825,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 	if (commit_transaction->t_need_data_flush &&
 	    (journal->j_fs_dev != journal->j_dev) &&
 	    (journal->j_flags & JBD2_BARRIER))
-		blkdev_issue_flush(journal->j_fs_dev, GFP_NOFS);
+		blkdev_issue_flush(journal->j_fs_dev);
 
 	/* Done it all: now write the commit record asynchronously. */
 	if (jbd2_has_feature_async_commit(journal)) {
@@ -932,7 +932,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 	stats.run.rs_blocks_logged++;
 	if (jbd2_has_feature_async_commit(journal) &&
 	    journal->j_flags & JBD2_BARRIER) {
-		blkdev_issue_flush(journal->j_dev, GFP_NOFS);
+		blkdev_issue_flush(journal->j_dev);
 	}
 
 	if (err)

fs/jbd2/recovery.c

@@ -326,7 +326,7 @@ int jbd2_journal_recover(journal_t *journal)
 		err = err2;
 	/* Make sure all replayed data is on permanent storage */
 	if (journal->j_flags & JBD2_BARRIER) {
-		err2 = blkdev_issue_flush(journal->j_fs_dev, GFP_KERNEL);
+		err2 = blkdev_issue_flush(journal->j_fs_dev);
 		if (!err)
 			err = err2;
 	}

fs/libfs.c

@@ -1117,7 +1117,7 @@ int generic_file_fsync(struct file *file, loff_t start, loff_t end,
 	err = __generic_file_fsync(file, start, end, datasync);
 	if (err)
 		return err;
-	return blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL);
+	return blkdev_issue_flush(inode->i_sb->s_bdev);
 }
 EXPORT_SYMBOL(generic_file_fsync);
 

fs/nilfs2/the_nilfs.h

@@ -375,7 +375,7 @@ static inline int nilfs_flush_device(struct the_nilfs *nilfs)
 	 */
 	smp_wmb();
 
-	err = blkdev_issue_flush(nilfs->ns_bdev, GFP_KERNEL);
+	err = blkdev_issue_flush(nilfs->ns_bdev);
 	if (err != -EIO)
 		err = 0;
 	return err;

fs/ocfs2/file.c

@@ -194,7 +194,7 @@ static int ocfs2_sync_file(struct file *file, loff_t start, loff_t end,
 		needs_barrier = true;
 	err = jbd2_complete_transaction(journal, commit_tid);
 	if (needs_barrier) {
-		ret = blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL);
+		ret = blkdev_issue_flush(inode->i_sb->s_bdev);
 		if (!err)
 			err = ret;
 	}

fs/reiserfs/file.c

@@ -159,7 +159,7 @@ static int reiserfs_sync_file(struct file *filp, loff_t start, loff_t end,
 	barrier_done = reiserfs_commit_for_inode(inode);
 	reiserfs_write_unlock(inode->i_sb);
 	if (barrier_done != 1 && reiserfs_barrier_flush(inode->i_sb))
-		blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL);
+		blkdev_issue_flush(inode->i_sb->s_bdev);
 	inode_unlock(inode);
 	if (barrier_done < 0)
 		return barrier_done;

fs/xfs/xfs_super.c

@@ -342,7 +342,7 @@ void
 xfs_blkdev_issue_flush(
 	xfs_buftarg_t		*buftarg)
 {
-	blkdev_issue_flush(buftarg->bt_bdev, GFP_NOFS);
+	blkdev_issue_flush(buftarg->bt_bdev);
 }
 
 STATIC void

fs/zonefs/super.c

@@ -541,7 +541,7 @@ static int zonefs_file_fsync(struct file *file, loff_t start, loff_t end,
 	if (ZONEFS_I(inode)->i_ztype == ZONEFS_ZTYPE_CNV)
 		ret = file_write_and_wait_range(file, start, end);
 	if (!ret)
-		ret = blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL);
+		ret = blkdev_issue_flush(inode->i_sb->s_bdev);
 
 	if (ret)
 		zonefs_io_error(inode, true);

include/linux/blkdev.h

@@ -1288,7 +1288,7 @@ static inline bool blk_needs_flush_plug(struct task_struct *tsk)
 		 !list_empty(&plug->cb_list));
 }
 
-int blkdev_issue_flush(struct block_device *, gfp_t);
+int blkdev_issue_flush(struct block_device *bdev);
 long nr_blockdev_pages(void);
 #else /* CONFIG_BLOCK */
 struct blk_plug {
@@ -1316,7 +1316,7 @@ static inline bool blk_needs_flush_plug(struct task_struct *tsk)
 	return false;
 }
 
-static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask)
+static inline int blkdev_issue_flush(struct block_device *bdev)
 {
 	return 0;
 }