fs: add a new user_write_streams() callback

fs/btrfs/extent_io.c

@@ -666,6 +666,7 @@ static void alloc_new_bio(struct btrfs_inode *inode,
 			       bio_ctrl->end_io_func, NULL);
 	bbio->bio.bi_iter.bi_sector = disk_bytenr >> SECTOR_SHIFT;
 	bbio->bio.bi_write_hint = inode->vfs_inode.i_write_hint;
+	bbio->bio.bi_write_stream = inode->vfs_inode.i_write_stream;
 	bbio->inode = inode;
 	bbio->file_offset = file_offset;
 	bio_ctrl->bbio = bbio;

fs/buffer.c

@@ -55,7 +55,8 @@
 
 static int fsync_buffers_list(spinlock_t *lock, struct list_head *list);
 static void submit_bh_wbc(blk_opf_t opf, struct buffer_head *bh,
-			  enum rw_hint hint, struct writeback_control *wbc);
+			  enum rw_hint hint, u8 write_stream,
+			  struct writeback_control *wbc);
 
 #define BH_ENTRY(list) list_entry((list), struct buffer_head, b_assoc_buffers)
 
@@ -1931,7 +1932,8 @@ int __block_write_full_folio(struct inode *inode, struct folio *folio,
 		struct buffer_head *next = bh->b_this_page;
 		if (buffer_async_write(bh)) {
 			submit_bh_wbc(REQ_OP_WRITE | write_flags, bh,
-				      inode->i_write_hint, wbc);
+				      inode->i_write_hint,
+				      inode->i_write_stream, wbc);
 			nr_underway++;
 		}
 		bh = next;
@@ -1986,7 +1988,8 @@ int __block_write_full_folio(struct inode *inode, struct folio *folio,
 		if (buffer_async_write(bh)) {
 			clear_buffer_dirty(bh);
 			submit_bh_wbc(REQ_OP_WRITE | write_flags, bh,
-				      inode->i_write_hint, wbc);
+				      inode->i_write_hint,
+				      inode->i_write_stream, wbc);
 			nr_underway++;
 		}
 		bh = next;
@@ -2778,7 +2781,7 @@ static void end_bio_bh_io_sync(struct bio *bio)
 }
 
 static void submit_bh_wbc(blk_opf_t opf, struct buffer_head *bh,
-			  enum rw_hint write_hint,
+			  enum rw_hint write_hint, u8 write_stream,
 			  struct writeback_control *wbc)
 {
 	const enum req_op op = opf & REQ_OP_MASK;
@@ -2807,6 +2810,7 @@ static void submit_bh_wbc(blk_opf_t opf, struct buffer_head *bh,
 
 	bio->bi_iter.bi_sector = bh->b_blocknr * (bh->b_size >> 9);
 	bio->bi_write_hint = write_hint;
+	bio->bi_write_stream = write_stream;
 
 	bio_add_folio_nofail(bio, bh->b_folio, bh->b_size, bh_offset(bh));
 
@@ -2826,7 +2830,7 @@ static void submit_bh_wbc(blk_opf_t opf, struct buffer_head *bh,
 
 void submit_bh(blk_opf_t opf, struct buffer_head *bh)
 {
-	submit_bh_wbc(opf, bh, WRITE_LIFE_NOT_SET, NULL);
+	submit_bh_wbc(opf, bh, WRITE_LIFE_NOT_SET, 0, NULL);
 }
 EXPORT_SYMBOL(submit_bh);
 

fs/direct-io.c

@@ -410,6 +410,7 @@ dio_bio_alloc(struct dio *dio, struct dio_submit *sdio,
 	if (dio->is_pinned)
 		bio_set_flag(bio, BIO_PAGE_PINNED);
 	bio->bi_write_hint = file_inode(dio->iocb->ki_filp)->i_write_hint;
+	bio->bi_write_stream = file_inode(dio->iocb->ki_filp)->i_write_stream;
 
 	sdio->bio = bio;
 	sdio->logical_offset_in_bio = sdio->cur_page_fs_offset;

fs/ext4/page-io.c

@@ -447,6 +447,7 @@ static void io_submit_add_bh(struct ext4_io_submit *io,
 	if (io->io_bio == NULL) {
 		io_submit_init_bio(io, bh);
 		io->io_bio->bi_write_hint = inode->i_write_hint;
+		io->io_bio->bi_write_stream = inode->i_write_stream;
 	}
 	if (!bio_add_folio(io->io_bio, io_folio, bh->b_size, bh_offset(bh)))
 		goto submit_and_retry;

fs/fcntl.c

@@ -27,6 +27,7 @@
 #include <linux/compat.h>
 #include <linux/mount.h>
 #include <linux/rw_hint.h>
+#include <linux/blkdev.h>
 
 #include <linux/poll.h>
 #include <asm/siginfo.h>
@@ -394,6 +395,60 @@ static long fcntl_set_rw_hint(struct file *file, unsigned int cmd,
 	return 0;
 }
 
+static u8 vfs_user_write_streams(struct inode *inode)
+{
+	struct super_block *sb;
+
+	if (S_ISBLK(inode->i_mode))
+		return bdev_max_write_streams(I_BDEV(inode));
+
+	sb = inode->i_sb;
+	/* If available, use per-mount/fs policy */
+	if (sb->s_op && sb->s_op->user_write_streams)
+		return sb->s_op->user_write_streams(sb);
+	/* otherwise, fallback to queue limit */
+	if (sb->s_bdev)
+		return bdev_max_write_streams(sb->s_bdev);
+	return 0;
+}
+
+static long fcntl_get_max_write_streams(struct file *file)
+{
+	struct inode *inode = file_inode(file);
+
+	if (S_ISBLK(inode->i_mode))
+		inode = file->f_mapping->host;
+
+	return vfs_user_write_streams(inode);
+}
+
+static long fcntl_get_write_stream(struct file *file)
+{
+	struct inode *inode = file_inode(file);
+
+	if (S_ISBLK(inode->i_mode))
+		inode = file->f_mapping->host;
+
+	return inode->i_write_stream;
+}
+
+static long fcntl_set_write_stream(struct file *file, unsigned long arg)
+{
+	struct inode *inode = file_inode(file);
+
+	if (!inode_owner_or_capable(file_mnt_idmap(file), inode))
+		return -EPERM;
+
+	if (S_ISBLK(inode->i_mode))
+		inode = file->f_mapping->host;
+
+	if (arg > vfs_user_write_streams(inode))
+		return -EINVAL;
+
+	WRITE_ONCE(inode->i_write_stream, arg);
+	return 0;
+}
+
 /* Is the file descriptor a dup of the file? */
 static long f_dupfd_query(int fd, struct file *filp)
 {
@@ -552,6 +607,15 @@ static long do_fcntl(int fd, unsigned int cmd, unsigned long arg,
 	case F_SET_RW_HINT:
 		err = fcntl_set_rw_hint(filp, cmd, arg);
 		break;
+	case F_GET_MAX_WRITE_STREAMS:
+		err = fcntl_get_max_write_streams(filp);
+		break;
+	case F_GET_WRITE_STREAM:
+		err = fcntl_get_write_stream(filp);
+		break;
+	case F_SET_WRITE_STREAM:
+		err = fcntl_set_write_stream(filp, arg);
+		break;
 	default:
 		break;
 	}

fs/inode.c

@@ -250,6 +250,7 @@ int inode_init_always_gfp(struct super_block *sb, struct inode *inode, gfp_t gfp
 	atomic_set(&inode->i_writecount, 0);
 	inode->i_size = 0;
 	inode->i_write_hint = WRITE_LIFE_NOT_SET;
+	inode->i_write_stream = 0;
 	inode->i_blocks = 0;
 	inode->i_bytes = 0;
 	inode->i_generation = 0;

fs/iomap/direct-io.c

@@ -430,6 +430,7 @@ static int iomap_dio_bio_iter(struct iomap_iter *iter, struct iomap_dio *dio)
 					  GFP_KERNEL);
 		bio->bi_iter.bi_sector = iomap_sector(iomap, pos);
 		bio->bi_write_hint = inode->i_write_hint;
+		bio->bi_write_stream = inode->i_write_stream;
 		bio->bi_ioprio = dio->iocb->ki_ioprio;
 		bio->bi_private = dio;
 		bio->bi_end_io = iomap_dio_bio_end_io;

fs/iomap/ioend.c

@@ -107,6 +107,7 @@ static struct iomap_ioend *iomap_alloc_ioend(struct iomap_writepage_ctx *wpc,
 			       GFP_NOFS, &iomap_ioend_bioset);
 	bio->bi_iter.bi_sector = iomap_sector(&wpc->iomap, pos);
 	bio->bi_write_hint = wpc->inode->i_write_hint;
+	bio->bi_write_stream = wpc->inode->i_write_stream;
 	wbc_init_bio(wpc->wbc, bio);
 	wpc->nr_folios = 0;
 	return iomap_init_ioend(wpc->inode, bio, pos, ioend_flags);

fs/mpage.c

@@ -595,6 +595,7 @@ static int mpage_write_folio(struct writeback_control *wbc, struct folio *folio,
 		bio->bi_iter.bi_sector = first_block << (blkbits - 9);
 		wbc_init_bio(wbc, bio);
 		bio->bi_write_hint = inode->i_write_hint;
+		bio->bi_write_stream = inode->i_write_stream;
 	}
 
 	/*

include/linux/fs.h

@@ -725,7 +725,8 @@ struct inode {
 
 	/* Misc */
 	u32			i_state;
-	/* 32-bit hole */
+	u8			i_write_stream;
+	/* 24-bit hole */
 	struct rw_semaphore	i_rwsem;
 
 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
@@ -2379,6 +2380,11 @@ struct super_operations {
 	 */
 	int (*remove_bdev)(struct super_block *sb, struct block_device *bdev);
 	void (*shutdown)(struct super_block *sb);
+	/*
+	 * Implement this callback if filesystem wants to control the
+	 * number of streams that are available to user space.
+	 */
+	u8 (*user_write_streams)(struct super_block *sb);
 };
 
 /*

include/uapi/linux/fcntl.h

@@ -61,6 +61,13 @@
 #define F_GET_FILE_RW_HINT	(F_LINUX_SPECIFIC_BASE + 13)
 #define F_SET_FILE_RW_HINT	(F_LINUX_SPECIFIC_BASE + 14)
 
+/*
+ *  Query available write streams
+ */
+#define F_GET_MAX_WRITE_STREAMS (F_LINUX_SPECIFIC_BASE + 15)
+#define F_GET_WRITE_STREAM	(F_LINUX_SPECIFIC_BASE + 16)
+#define F_SET_WRITE_STREAM	(F_LINUX_SPECIFIC_BASE + 17)
+
 /*
  * Valid hint values for F_{GET,SET}_RW_HINT. 0 is "not set", or can be
  * used to clear any hints previously set.