diff mbox series

[-next,RFC,01/14] block: add some bdev apis

Message ID 20231205123728.1866699-2-yukuai1@huaweicloud.com (mailing list archive)
State Superseded
Headers show
Series block: don't access bd_inode directly from other modules | expand

Commit Message

Yu Kuai Dec. 5, 2023, 12:37 p.m. UTC
From: Yu Kuai <yukuai3@huawei.com>

Those apis will be used for other modules, so that bd_inode won't be
accessed directly from other modules.

Signed-off-by: Yu Kuai <yukuai3@huawei.com>
---
 block/bdev.c           | 116 +++++++++++++++++++++++++++++++++++++++++
 block/bio.c            |   1 +
 block/blk.h            |   2 -
 include/linux/blkdev.h |  27 ++++++++++
 4 files changed, 144 insertions(+), 2 deletions(-)

Comments

Bart Van Assche Dec. 5, 2023, 5:03 p.m. UTC | #1
On 12/5/23 04:37, Yu Kuai wrote:
> +static inline u8 block_bits(struct block_device *bdev)
> +{
> +	return bdev->bd_inode->i_blkbits;
> +}

This function needs a name that's more descriptive.

Thanks,

Bart.
Christoph Hellwig Dec. 6, 2023, 6:14 a.m. UTC | #2
> +void invalidate_bdev_range(struct block_device *bdev, pgoff_t start,
> +			   pgoff_t end)
> +{
> +	invalidate_mapping_pages(bdev->bd_inode->i_mapping, start, end);
> +}
> +EXPORT_SYMBOL_GPL(invalidate_bdev_range);

All these could probably use kerneldoc comments.

For this one I really don't like it existing at all, but we'll have to
discuss that in the btrfs patch.

> +loff_t bdev_size(struct block_device *bdev)
> +{
> +	loff_t size;
> +
> +	spin_lock(&bdev->bd_size_lock);
> +	size = i_size_read(bdev->bd_inode);
> +	spin_unlock(&bdev->bd_size_lock);
> +
> +	return size;
> +}
> +EXPORT_SYMBOL_GPL(bdev_size);

No need for this one.  The callers can simply use bdev_nr_bytes.

> +struct folio *bdev_read_folio(struct block_device *bdev, pgoff_t index)
> +{
> +	return read_mapping_folio(bdev->bd_inode->i_mapping, index, NULL);
> +}
> +EXPORT_SYMBOL_GPL(bdev_read_folio);
> +
> +struct folio *bdev_read_folio_gfp(struct block_device *bdev, pgoff_t index,
> +				  gfp_t gfp)
> +{
> +	return mapping_read_folio_gfp(bdev->bd_inode->i_mapping, index, gfp);
> +}
> +EXPORT_SYMBOL_GPL(bdev_read_folio_gfp);

I think we can just drop bdev_read_folio_gfp. Half of the callers simply
pass GPK_KERNEL, and the other half passes GFP_NOFS and could just use
memalloc_nofs_save().

> +void bdev_balance_dirty_pages_ratelimited(struct block_device *bdev)
> +{
> +	return balance_dirty_pages_ratelimited(bdev->bd_inode->i_mapping);
> +}
> +EXPORT_SYMBOL_GPL(bdev_balance_dirty_pages_ratelimited);

Hmm, this is just used for block2mtd, and feels a little too low-level
to me, as block2mtd really should be using the normal fileread/write
APIs.  I guess we'll have to live with it for now if we want to expedite
killing off bd_inode.

> +void bdev_correlate_mapping(struct block_device *bdev,
> +			    struct address_space *mapping)
> +{
> +	mapping->host = bdev->bd_inode;
> +}
> +EXPORT_SYMBOL_GPL(bdev_correlate_mapping);

Maybe associated insted of correlate?  Either way this basically
fully exposes the bdev inode again :(

> +gfp_t bdev_gfp_constraint(struct block_device *bdev, gfp_t gfp)
> +{
> +	return mapping_gfp_constraint(bdev->bd_inode->i_mapping, gfp);
> +}
> +EXPORT_SYMBOL_GPL(bdev_gfp_constraint);

The right fix here is to:

 - use memalloc_nofs_save in extet instead of using
   mapping_gfp_constraint to clear it from the mapping flags
 - remove __ext4_sb_bread_gfp and just have buffer.c helper that does
   the right thing (either by changing the calling conventions of an
   existing one, or adding a new one).

> +/*
> + * The del_gendisk() function uninitializes the disk-specific data
> + * structures, including the bdi structure, without telling anyone
> + * else.  Once this happens, any attempt to call mark_buffer_dirty()
> + * (for example, by ext4_commit_super), will cause a kernel OOPS.
> + * This is a kludge to prevent these oops until we can put in a proper
> + * hook in del_gendisk() to inform the VFS and file system layers.
> + */
> +int bdev_ejected(struct block_device *bdev)
> +{
> +	struct backing_dev_info *bdi = inode_to_bdi(bdev->bd_inode);
> +
> +	return bdi->dev == NULL;
> +}
> +EXPORT_SYMBOL_GPL(bdev_ejected);

And this code in ext4 should just go away entirely.  The bdi should
always be valid for a live bdev for years.

> --- a/block/bio.c
> +++ b/block/bio.c
> @@ -1119,6 +1119,7 @@ void bio_add_folio_nofail(struct bio *bio, struct folio *folio, size_t len,
>  	WARN_ON_ONCE(off > UINT_MAX);
>  	__bio_add_page(bio, &folio->page, len, off);
>  }
> +EXPORT_SYMBOL_GPL(bio_add_folio_nofail);

How is this realted?  The export is fine, but really should be a
separate, well-documented commit.

>  
> +static inline u8 block_bits(struct block_device *bdev)
> +{
> +	return bdev->bd_inode->i_blkbits;
> +}

Not sure we should need this.  i_blkbits comes from the blocksize
the fs set, so it should have other ways to get at it.
Yu Kuai Dec. 6, 2023, 6:50 a.m. UTC | #3
Hi,

在 2023/12/06 14:14, Christoph Hellwig 写道:
>> +void invalidate_bdev_range(struct block_device *bdev, pgoff_t start,
>> +			   pgoff_t end)
>> +{
>> +	invalidate_mapping_pages(bdev->bd_inode->i_mapping, start, end);
>> +}
>> +EXPORT_SYMBOL_GPL(invalidate_bdev_range);
> 
> All these could probably use kerneldoc comments.

Ok, and thanks for reviewing the patchset!
> 
> For this one I really don't like it existing at all, but we'll have to
> discuss that in the btrfs patch.
> 
>> +loff_t bdev_size(struct block_device *bdev)
>> +{
>> +	loff_t size;
>> +
>> +	spin_lock(&bdev->bd_size_lock);
>> +	size = i_size_read(bdev->bd_inode);
>> +	spin_unlock(&bdev->bd_size_lock);
>> +
>> +	return size;
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_size);
> 
> No need for this one.  The callers can simply use bdev_nr_bytes.

Ok, I'll replace it with bdev_nr_bytes.
> 
>> +struct folio *bdev_read_folio(struct block_device *bdev, pgoff_t index)
>> +{
>> +	return read_mapping_folio(bdev->bd_inode->i_mapping, index, NULL);
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_read_folio);
>> +
>> +struct folio *bdev_read_folio_gfp(struct block_device *bdev, pgoff_t index,
>> +				  gfp_t gfp)
>> +{
>> +	return mapping_read_folio_gfp(bdev->bd_inode->i_mapping, index, gfp);
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_read_folio_gfp);
> 
> I think we can just drop bdev_read_folio_gfp. Half of the callers simply
> pass GPK_KERNEL, and the other half passes GFP_NOFS and could just use
> memalloc_nofs_save().

I'm a litter confused, so there are 3 use cases:
1) use GFP_USER, default gfp from bdev_alloc.
2) use GFP_KERNEL
3) use GFP_NOFS

I understand that you're suggesting memalloc_nofs_save() to distinguish
2 and 3, but how can I distinguish 1?
> 
>> +void bdev_balance_dirty_pages_ratelimited(struct block_device *bdev)
>> +{
>> +	return balance_dirty_pages_ratelimited(bdev->bd_inode->i_mapping);
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_balance_dirty_pages_ratelimited);
> 
> Hmm, this is just used for block2mtd, and feels a little too low-level
> to me, as block2mtd really should be using the normal fileread/write
> APIs.  I guess we'll have to live with it for now if we want to expedite
> killing off bd_inode.
> 
>> +void bdev_correlate_mapping(struct block_device *bdev,
>> +			    struct address_space *mapping)
>> +{
>> +	mapping->host = bdev->bd_inode;
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_correlate_mapping);
> 
> Maybe associated insted of correlate?  Either way this basically
> fully exposes the bdev inode again :(
> 
>> +gfp_t bdev_gfp_constraint(struct block_device *bdev, gfp_t gfp)
>> +{
>> +	return mapping_gfp_constraint(bdev->bd_inode->i_mapping, gfp);
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_gfp_constraint);
> 
> The right fix here is to:
> 
>   - use memalloc_nofs_save in extet instead of using
>     mapping_gfp_constraint to clear it from the mapping flags
>   - remove __ext4_sb_bread_gfp and just have buffer.c helper that does
>     the right thing (either by changing the calling conventions of an
>     existing one, or adding a new one).

Thanks for the suggestions, but I'm not sure how to do this yet, I must
read more ext4 code.
> 
>> +/*
>> + * The del_gendisk() function uninitializes the disk-specific data
>> + * structures, including the bdi structure, without telling anyone
>> + * else.  Once this happens, any attempt to call mark_buffer_dirty()
>> + * (for example, by ext4_commit_super), will cause a kernel OOPS.
>> + * This is a kludge to prevent these oops until we can put in a proper
>> + * hook in del_gendisk() to inform the VFS and file system layers.
>> + */
>> +int bdev_ejected(struct block_device *bdev)
>> +{
>> +	struct backing_dev_info *bdi = inode_to_bdi(bdev->bd_inode);
>> +
>> +	return bdi->dev == NULL;
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_ejected);
> 
> And this code in ext4 should just go away entirely.  The bdi should
> always be valid for a live bdev for years.
Sounds good, I was confused about this code as well.

> 
>> --- a/block/bio.c
>> +++ b/block/bio.c
>> @@ -1119,6 +1119,7 @@ void bio_add_folio_nofail(struct bio *bio, struct folio *folio, size_t len,
>>   	WARN_ON_ONCE(off > UINT_MAX);
>>   	__bio_add_page(bio, &folio->page, len, off);
>>   }
>> +EXPORT_SYMBOL_GPL(bio_add_folio_nofail);
> 
> How is this realted?  The export is fine, but really should be a
> separate, well-documented commit.

This is used to replace __bio_add_page() in btrfs while converting page
to folio, please let me know if I should keep this, if so, I'll split
this into a new commit.
> 
>>   
>> +static inline u8 block_bits(struct block_device *bdev)
>> +{
>> +	return bdev->bd_inode->i_blkbits;
>> +}
> 
> Not sure we should need this.  i_blkbits comes from the blocksize
> the fs set, so it should have other ways to get at it.

Yes, this is now only used for erofs, and erofs do call
sb_set_blocksize() while initializing, hence it's right there is other
way to get blkbits and this helper is not needed.

Thanks,
Kuai

> .
>
Christoph Hellwig Dec. 6, 2023, 7:20 a.m. UTC | #4
On Wed, Dec 06, 2023 at 02:50:56PM +0800, Yu Kuai wrote:
> I'm a litter confused, so there are 3 use cases:
> 1) use GFP_USER, default gfp from bdev_alloc.
> 2) use GFP_KERNEL
> 3) use GFP_NOFS
> 
> I understand that you're suggesting memalloc_nofs_save() to distinguish
> 2 and 3, but how can I distinguish 1?

You shouldn't.  Diverging from the default flags except for clearing
the FS or IO flags is simply a bug.  Note that things like block2mtd
should probably also ensure a noio allocation if they aren't doing that
yet.

> >   - use memalloc_nofs_save in extet instead of using
> >     mapping_gfp_constraint to clear it from the mapping flags
> >   - remove __ext4_sb_bread_gfp and just have buffer.c helper that does
> >     the right thing (either by changing the calling conventions of an
> >     existing one, or adding a new one).
> 
> Thanks for the suggestions, but I'm not sure how to do this yet, I must
> read more ext4 code.

the nofs save part should be trivial.  You can just skip the rest for
now as it's not needed for this patch series.
Matthew Wilcox Dec. 6, 2023, 2:58 p.m. UTC | #5
On Tue, Dec 05, 2023 at 08:37:15PM +0800, Yu Kuai wrote:
> +struct folio *bdev_read_folio(struct block_device *bdev, pgoff_t index)
> +{
> +	return read_mapping_folio(bdev->bd_inode->i_mapping, index, NULL);
> +}
> +EXPORT_SYMBOL_GPL(bdev_read_folio);

I'm coming to the opinion that 'index' is the wrong parameter here.
Looking through all the callers of bdev_read_folio() in this patchset,
they all have a position in bytes, and they all convert it to
index for this call.  The API should probably be:

struct folio *bdev_read_folio(struct block_device *bdev, loff_t pos)
{
	return read_mapping_folio(bdev->bd_inode->i_mapping,
			pos / PAGE_SIZE, NULL);
}

... and at some point, we'll get round to converting read_mapping_folio()
to take its argument in loff_t.

Similiarly for these two APIs:

> +struct folio *bdev_read_folio_gfp(struct block_device *bdev, pgoff_t index,
> +				  gfp_t gfp)
> +struct folio *bdev_get_folio(struct block_device *bdev, pgoff_t index)

> +struct folio *bdev_find_or_create_folio(struct block_device *bdev,
> +					pgoff_t index, gfp_t gfp)
> +{
> +	return __filemap_get_folio(bdev->bd_inode->i_mapping, index,
> +				   FGP_LOCK | FGP_ACCESSED | FGP_CREAT, gfp);
> +}
> +EXPORT_SYMBOL_GPL(bdev_find_or_create_folio);

This one probably shouldn't exist.  I've been converting callers of
find_or_create_page() to call __filemap_get_folio; I suspect we
should expose a __bdev_get_folio and have the callers use the FGP
arguments directly, but I'm open to other opinions here.

> +void bdev_sync_readahead(struct block_device *bdev, struct file_ra_state *ra,
> +			 struct file *file, pgoff_t index,
> +			 unsigned long req_count)
> +{
> +	struct file_ra_state tmp_ra = {};
> +
> +	if (!ra) {
> +		ra = &tmp_ra;
> +		file_ra_state_init(ra, bdev->bd_inode->i_mapping);
> +	}
> +	page_cache_sync_readahead(bdev->bd_inode->i_mapping, ra, file, index,
> +				  req_count);
> +}

I think the caller should always be passing in a valid file_ra_state.
It's only cramfs that doesn't have one, and it really should!
Not entirely sure about the arguments here; part of me says "bytes",
but this is weird enough to maybe take arguments in pages.
Theodore Ts'o Dec. 6, 2023, 5:50 p.m. UTC | #6
On Tue, Dec 05, 2023 at 10:14:00PM -0800, Christoph Hellwig wrote:
> > +/*
> > + * The del_gendisk() function uninitializes the disk-specific data
> > + * structures, including the bdi structure, without telling anyone
> > + * else.  Once this happens, any attempt to call mark_buffer_dirty()
> > + * (for example, by ext4_commit_super), will cause a kernel OOPS.
> > + * This is a kludge to prevent these oops until we can put in a proper
> > + * hook in del_gendisk() to inform the VFS and file system layers.
> > + */
> > +int bdev_ejected(struct block_device *bdev)
> > +{
> > +	struct backing_dev_info *bdi = inode_to_bdi(bdev->bd_inode);
> > +
> > +	return bdi->dev == NULL;
> > +}
> > +EXPORT_SYMBOL_GPL(bdev_ejected);
> 
> And this code in ext4 should just go away entirely.  The bdi should
> always be valid for a live bdev for years.

This was added because pulling a mounted a USB thumb drive (or a HDD
drops off the SATA bus) while the file system is mounted and actively
in use, would result in a kernel OOPS.  If that's no longer true,
that's great, but it would be good to test to make sure this is the
case....

If we really want to remove it, I'd suggest doing this as a separate
commit, so that after we see syzbot reports, or users complaining
about kernel crashes, we can revert the removal if necessary.

Cheers,

					- Ted
Christoph Hellwig Dec. 6, 2023, 5:57 p.m. UTC | #7
On Wed, Dec 06, 2023 at 12:50:38PM -0500, Theodore Ts'o wrote:
> This was added because pulling a mounted a USB thumb drive (or a HDD
> drops off the SATA bus) while the file system is mounted and actively
> in use, would result in a kernel OOPS.  If that's no longer true,
> that's great, but it would be good to test to make sure this is the
> case....

And, surprise, surprise - that didn't just affect ext4.  So I ended
up fixing this properly in the block layer.

> If we really want to remove it, I'd suggest doing this as a separate
> commit, so that after we see syzbot reports, or users complaining
> about kernel crashes, we can revert the removal if necessary.

Yes, this should of course be separate, well documented commit.
Yu Kuai Dec. 7, 2023, 2:45 a.m. UTC | #8
Hi,

在 2023/12/06 22:58, Matthew Wilcox 写道:
> On Tue, Dec 05, 2023 at 08:37:15PM +0800, Yu Kuai wrote:
>> +struct folio *bdev_read_folio(struct block_device *bdev, pgoff_t index)
>> +{
>> +	return read_mapping_folio(bdev->bd_inode->i_mapping, index, NULL);
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_read_folio);
> 
> I'm coming to the opinion that 'index' is the wrong parameter here.
> Looking through all the callers of bdev_read_folio() in this patchset,
> they all have a position in bytes, and they all convert it to
> index for this call.  The API should probably be:
> 
> struct folio *bdev_read_folio(struct block_device *bdev, loff_t pos)
> {
> 	return read_mapping_folio(bdev->bd_inode->i_mapping,
> 			pos / PAGE_SIZE, NULL);
> }

Thanks for reviewing this patchset! Okay, I'll convert to pass in "pos"
in v2.
> 
> ... and at some point, we'll get round to converting read_mapping_folio()
> to take its argument in loff_t.
> 
> Similiarly for these two APIs:
> 
>> +struct folio *bdev_read_folio_gfp(struct block_device *bdev, pgoff_t index,
>> +				  gfp_t gfp)
>> +struct folio *bdev_get_folio(struct block_device *bdev, pgoff_t index)
> 
>> +struct folio *bdev_find_or_create_folio(struct block_device *bdev,
>> +					pgoff_t index, gfp_t gfp)
>> +{
>> +	return __filemap_get_folio(bdev->bd_inode->i_mapping, index,
>> +				   FGP_LOCK | FGP_ACCESSED | FGP_CREAT, gfp);
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_find_or_create_folio);
> 
> This one probably shouldn't exist.  I've been converting callers of
> find_or_create_page() to call __filemap_get_folio; I suspect we
> should expose a __bdev_get_folio and have the callers use the FGP
> arguments directly, but I'm open to other opinions here.

If nobody against this, I will expose single __bdev_get_folio() to use
in v2.
> 
>> +void bdev_sync_readahead(struct block_device *bdev, struct file_ra_state *ra,
>> +			 struct file *file, pgoff_t index,
>> +			 unsigned long req_count)
>> +{
>> +	struct file_ra_state tmp_ra = {};
>> +
>> +	if (!ra) {
>> +		ra = &tmp_ra;
>> +		file_ra_state_init(ra, bdev->bd_inode->i_mapping);
>> +	}
>> +	page_cache_sync_readahead(bdev->bd_inode->i_mapping, ra, file, index,
>> +				  req_count);
>> +}
> 
> I think the caller should always be passing in a valid file_ra_state.
> It's only cramfs that doesn't have one, and it really should!
> Not entirely sure about the arguments here; part of me says "bytes",
> but this is weird enough to maybe take arguments in pages.

In fact, bdev_sync_readahead() is only called for cramfs and ext4.

For ext4 it's used in ext4_readdir() so there is valid file_ra_state.

Hoever, for cramfs it's used in cramfs_read(), and cramfs_read() is used
for:

1) cramfs_read_folio
2) cramfs_readdir
3) cramfs_lookup
4) cramfs_read_super

Looks like it's easy to pass in valid file_ra_state() for 1) and 2),
however, I don't see an easy way to do this for 3) and 4).

Thanks,
Kuai

> 
> .
>
diff mbox series

Patch

diff --git a/block/bdev.c b/block/bdev.c
index 6f73b02d549c..fcba5c1bd113 100644
--- a/block/bdev.c
+++ b/block/bdev.c
@@ -92,6 +92,13 @@  void invalidate_bdev(struct block_device *bdev)
 }
 EXPORT_SYMBOL(invalidate_bdev);
 
+void invalidate_bdev_range(struct block_device *bdev, pgoff_t start,
+			   pgoff_t end)
+{
+	invalidate_mapping_pages(bdev->bd_inode->i_mapping, start, end);
+}
+EXPORT_SYMBOL_GPL(invalidate_bdev_range);
+
 /*
  * Drop all buffers & page cache for given bdev range. This function bails
  * with error if bdev has other exclusive owner (such as filesystem).
@@ -124,6 +131,7 @@  int truncate_bdev_range(struct block_device *bdev, blk_mode_t mode,
 					     lstart >> PAGE_SHIFT,
 					     lend >> PAGE_SHIFT);
 }
+EXPORT_SYMBOL_GPL(truncate_bdev_range);
 
 static void set_init_blocksize(struct block_device *bdev)
 {
@@ -138,6 +146,18 @@  static void set_init_blocksize(struct block_device *bdev)
 	bdev->bd_inode->i_blkbits = blksize_bits(bsize);
 }
 
+loff_t bdev_size(struct block_device *bdev)
+{
+	loff_t size;
+
+	spin_lock(&bdev->bd_size_lock);
+	size = i_size_read(bdev->bd_inode);
+	spin_unlock(&bdev->bd_size_lock);
+
+	return size;
+}
+EXPORT_SYMBOL_GPL(bdev_size);
+
 int set_blocksize(struct block_device *bdev, int size)
 {
 	/* Size must be a power of two, and between 512 and PAGE_SIZE */
@@ -1144,3 +1164,99 @@  static int __init setup_bdev_allow_write_mounted(char *str)
 	return 1;
 }
 __setup("bdev_allow_write_mounted=", setup_bdev_allow_write_mounted);
+
+struct folio *bdev_read_folio(struct block_device *bdev, pgoff_t index)
+{
+	return read_mapping_folio(bdev->bd_inode->i_mapping, index, NULL);
+}
+EXPORT_SYMBOL_GPL(bdev_read_folio);
+
+struct folio *bdev_read_folio_gfp(struct block_device *bdev, pgoff_t index,
+				  gfp_t gfp)
+{
+	return mapping_read_folio_gfp(bdev->bd_inode->i_mapping, index, gfp);
+}
+EXPORT_SYMBOL_GPL(bdev_read_folio_gfp);
+
+struct folio *bdev_get_folio(struct block_device *bdev, pgoff_t index)
+{
+	return filemap_get_folio(bdev->bd_inode->i_mapping, index);
+}
+EXPORT_SYMBOL_GPL(bdev_get_folio);
+
+struct folio *bdev_find_or_create_folio(struct block_device *bdev,
+					pgoff_t index, gfp_t gfp)
+{
+	return __filemap_get_folio(bdev->bd_inode->i_mapping, index,
+				   FGP_LOCK | FGP_ACCESSED | FGP_CREAT, gfp);
+}
+EXPORT_SYMBOL_GPL(bdev_find_or_create_folio);
+
+int bdev_wb_err_check(struct block_device *bdev, errseq_t since)
+{
+	return errseq_check(&bdev->bd_inode->i_mapping->wb_err, since);
+}
+EXPORT_SYMBOL_GPL(bdev_wb_err_check);
+
+int bdev_wb_err_check_and_advance(struct block_device *bdev, errseq_t *since)
+{
+	return errseq_check_and_advance(&bdev->bd_inode->i_mapping->wb_err,
+					since);
+}
+EXPORT_SYMBOL_GPL(bdev_wb_err_check_and_advance);
+
+void bdev_balance_dirty_pages_ratelimited(struct block_device *bdev)
+{
+	return balance_dirty_pages_ratelimited(bdev->bd_inode->i_mapping);
+}
+EXPORT_SYMBOL_GPL(bdev_balance_dirty_pages_ratelimited);
+
+void bdev_sync_readahead(struct block_device *bdev, struct file_ra_state *ra,
+			 struct file *file, pgoff_t index,
+			 unsigned long req_count)
+{
+	struct file_ra_state tmp_ra = {};
+
+	if (!ra) {
+		ra = &tmp_ra;
+		file_ra_state_init(ra, bdev->bd_inode->i_mapping);
+	}
+	page_cache_sync_readahead(bdev->bd_inode->i_mapping, ra, file, index,
+				  req_count);
+}
+EXPORT_SYMBOL_GPL(bdev_sync_readahead);
+
+void bdev_attach_wb(struct block_device *bdev)
+{
+	inode_attach_wb(bdev->bd_inode, NULL);
+}
+EXPORT_SYMBOL_GPL(bdev_attach_wb);
+
+void bdev_correlate_mapping(struct block_device *bdev,
+			    struct address_space *mapping)
+{
+	mapping->host = bdev->bd_inode;
+}
+EXPORT_SYMBOL_GPL(bdev_correlate_mapping);
+
+gfp_t bdev_gfp_constraint(struct block_device *bdev, gfp_t gfp)
+{
+	return mapping_gfp_constraint(bdev->bd_inode->i_mapping, gfp);
+}
+EXPORT_SYMBOL_GPL(bdev_gfp_constraint);
+
+/*
+ * The del_gendisk() function uninitializes the disk-specific data
+ * structures, including the bdi structure, without telling anyone
+ * else.  Once this happens, any attempt to call mark_buffer_dirty()
+ * (for example, by ext4_commit_super), will cause a kernel OOPS.
+ * This is a kludge to prevent these oops until we can put in a proper
+ * hook in del_gendisk() to inform the VFS and file system layers.
+ */
+int bdev_ejected(struct block_device *bdev)
+{
+	struct backing_dev_info *bdi = inode_to_bdi(bdev->bd_inode);
+
+	return bdi->dev == NULL;
+}
+EXPORT_SYMBOL_GPL(bdev_ejected);
diff --git a/block/bio.c b/block/bio.c
index 816d412c06e9..f7123ad9b4ee 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -1119,6 +1119,7 @@  void bio_add_folio_nofail(struct bio *bio, struct folio *folio, size_t len,
 	WARN_ON_ONCE(off > UINT_MAX);
 	__bio_add_page(bio, &folio->page, len, off);
 }
+EXPORT_SYMBOL_GPL(bio_add_folio_nofail);
 
 /**
  * bio_add_folio - Attempt to add part of a folio to a bio.
diff --git a/block/blk.h b/block/blk.h
index 08a358bc0919..da4becd4f7e9 100644
--- a/block/blk.h
+++ b/block/blk.h
@@ -467,8 +467,6 @@  extern struct device_attribute dev_attr_events_poll_msecs;
 extern struct attribute_group blk_trace_attr_group;
 
 blk_mode_t file_to_blk_mode(struct file *file);
-int truncate_bdev_range(struct block_device *bdev, blk_mode_t mode,
-		loff_t lstart, loff_t lend);
 long blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg);
 long compat_blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg);
 
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 3f8a21cd9233..a55db77274a4 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -1342,6 +1342,11 @@  static inline unsigned int block_size(struct block_device *bdev)
 	return 1 << bdev->bd_inode->i_blkbits;
 }
 
+static inline u8 block_bits(struct block_device *bdev)
+{
+	return bdev->bd_inode->i_blkbits;
+}
+
 int kblockd_schedule_work(struct work_struct *work);
 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
 
@@ -1515,6 +1520,28 @@  struct block_device *blkdev_get_no_open(dev_t dev);
 void blkdev_put_no_open(struct block_device *bdev);
 
 struct block_device *I_BDEV(struct inode *inode);
+loff_t bdev_size(struct block_device *bdev);
+void invalidate_bdev_range(struct block_device *bdev, pgoff_t start,
+			   pgoff_t end);
+int truncate_bdev_range(struct block_device *bdev, blk_mode_t mode,
+		loff_t lstart, loff_t lend);
+struct folio *bdev_get_folio(struct block_device *bdev, pgoff_t index);
+struct folio *bdev_find_or_create_folio(struct block_device *bdev,
+					pgoff_t index, gfp_t gfp);
+struct folio *bdev_read_folio(struct block_device *bdev, pgoff_t index);
+struct folio *bdev_read_folio_gfp(struct block_device *bdev, pgoff_t index,
+				  gfp_t gfp);
+int bdev_wb_err_check(struct block_device *bdev, errseq_t since);
+int bdev_wb_err_check_and_advance(struct block_device *bdev, errseq_t *since);
+void bdev_balance_dirty_pages_ratelimited(struct block_device *bdev);
+void bdev_sync_readahead(struct block_device *bdev, struct file_ra_state *ra,
+			 struct file *file, pgoff_t index,
+			 unsigned long req_count);
+void bdev_attach_wb(struct block_device *bdev);
+void bdev_correlate_mapping(struct block_device *bdev,
+			    struct address_space *mapping);
+gfp_t bdev_gfp_constraint(struct block_device *bdev, gfp_t gfp);
+int bdev_ejected(struct block_device *bdev);
 
 #ifdef CONFIG_BLOCK
 void invalidate_bdev(struct block_device *bdev);