[v3,08/10] block: add support to pass user meta buffer

Commit Message

Anuj Gupta Aug. 23, 2024, 10:38 a.m. UTC

From: Kanchan Joshi <[e-mail]>

If iocb contains the meta, extract that and prepare the bip.
Based on flags specified by the user, set corresponding guard/app/ref
tags to be checked in bip. Introduce BIP_INTEGRITY_USER flag to
indicate integrity payload is user address. Make sure that
->prepare_fn and ->complete_fn are skipped for user-owned meta buffer.

Signed-off-by: Anuj Gupta <[e-mail]>
Signed-off-by: Kanchan Joshi <[e-mail]>
---
 block/bio-integrity.c         | 45 ++++++++++++++++++++++++++++++++++-
 block/fops.c                  | 25 +++++++++++++++++++
 block/t10-pi.c                |  6 +++++
 include/linux/bio-integrity.h | 13 +++++++++-
 4 files changed, 87 insertions(+), 2 deletions(-)

Comments

Christoph Hellwig Aug. 24, 2024, 8:44 a.m. UTC | #1

On Fri, Aug 23, 2024 at 04:08:09PM +0530, Anuj Gupta wrote:
> From: Kanchan Joshi <[e-mail]>
> 
> If iocb contains the meta, extract that and prepare the bip.

If an iocb contains metadata, ...

> --- a/block/fops.c
> +++ b/block/fops.c
> @@ -154,6 +154,9 @@ static void blkdev_bio_end_io(struct bio *bio)
>  		}
>  	}
>  
> +	if (bio_integrity(bio) && (dio->iocb->ki_flags & IOCB_HAS_META))
> +		bio_integrity_unmap_user(bio);

How could bio_integrity() be true here without the iocb flag?

> +		if (!is_sync && unlikely(iocb->ki_flags & IOCB_HAS_META)) {

unlikely is actively harmful here, as the code is likely if you use
the feature..

> +			ret = bio_integrity_map_iter(bio, iocb->private);
> +			if (unlikely(ret)) {
> +				bio_release_pages(bio, false);
> +				bio_clear_flag(bio, BIO_REFFED);
> +				bio_put(bio);
> +				blk_finish_plug(&plug);
> +				return ret;
> +			}

This duplicates the error handling done just above.  Please add a
goto label to de-duplicate it.

> +	if (unlikely(iocb->ki_flags & IOCB_HAS_META)) {
> +		ret = bio_integrity_map_iter(bio, iocb->private);
> +		WRITE_ONCE(iocb->private, NULL);
> +		if (unlikely(ret)) {
> +			bio_put(bio);
> +			return ret;

This probably also wants an out_bio_put label even if the duplication
is minimal so far.

You probably also want a WARN_ON for the iocb meta flag in
__blkdev_direct_IO_simple so that we don't get caught off guard
if someone adds a synchronous path using PI.

> diff --git a/block/t10-pi.c b/block/t10-pi.c
> index e7052a728966..cb7bc4a88380 100644
> --- a/block/t10-pi.c
> +++ b/block/t10-pi.c
> @@ -139,6 +139,8 @@ static void t10_pi_type1_prepare(struct request *rq)
>  		/* Already remapped? */
>  		if (bip->bip_flags & BIP_MAPPED_INTEGRITY)
>  			break;
> +		if (bip->bip_flags & BIP_INTEGRITY_USER)
> +			break;

This is wrong.  When submitting metadata on a partition the ref tag
does need to be remapped.  Please also add a tests that tests submitting
metadata on a partition so that we have a regression test for this.

> +	BIP_INTEGRITY_USER      = 1 << 9, /* Integrity payload is user
> +					   * address
> +					   */

.. and with the above fix this flag should not be needed.

>  };
>  
>  struct bio_integrity_payload {
> @@ -24,6 +27,7 @@ struct bio_integrity_payload {
>  	unsigned short		bip_vcnt;	/* # of integrity bio_vecs */
>  	unsigned short		bip_max_vcnt;	/* integrity bio_vec slots */
>  	unsigned short		bip_flags;	/* control flags */
> +	u16			app_tag;

Please document the field even if it seems obvious.

Patch

diff --git a/block/bio-integrity.c b/block/bio-integrity.c
index 7fbf8c307a36..02b766c2e57d 100644
--- a/block/bio-integrity.c
+++ b/block/bio-integrity.c
@@ -12,6 +12,7 @@ 
 #include <linux/bio.h>
 #include <linux/workqueue.h>
 #include <linux/slab.h>
+#include <uapi/linux/io_uring.h>
 #include "blk.h"
 
 static struct kmem_cache *bip_slab;
@@ -252,7 +253,7 @@  static int bio_integrity_copy_user(struct bio *bio, struct bio_vec *bvec,
 		goto free_bip;
 	}
 
-	bip->bip_flags |= BIP_COPY_USER;
+	bip->bip_flags |= BIP_INTEGRITY_USER | BIP_COPY_USER;
 	bip->bip_iter.bi_sector = seed;
 	bip->bip_vcnt = nr_vecs;
 	bip->bio_iter = bio->bi_iter;
@@ -274,6 +275,7 @@  static int bio_integrity_init_user(struct bio *bio, struct bio_vec *bvec,
 		return PTR_ERR(bip);
 
 	memcpy(bip->bip_vec, bvec, nr_vecs * sizeof(*bvec));
+	bip->bip_flags |= BIP_INTEGRITY_USER;
 	bip->bip_iter.bi_sector = seed;
 	bip->bip_iter.bi_size = len;
 	bip->bip_vcnt = nr_vecs;
@@ -310,6 +312,47 @@  static unsigned int bvec_from_pages(struct bio_vec *bvec, struct page **pages,
 	return nr_bvecs;
 }
 
+static void bio_uio_meta_to_bip(struct bio *bio, struct uio_meta *meta)
+{
+	struct bio_integrity_payload *bip = bio_integrity(bio);
+
+	if (meta->flags & INTEGRITY_CHK_GUARD)
+		bip->bip_flags |= BIP_CHECK_GUARD;
+	if (meta->flags & INTEGRITY_CHK_APPTAG)
+		bip->bip_flags |= BIP_CHECK_APPTAG;
+	if (meta->flags & INTEGRITY_CHK_REFTAG)
+		bip->bip_flags |= BIP_CHECK_REFTAG;
+
+	bip->app_tag = meta->app_tag;
+}
+
+int bio_integrity_map_iter(struct bio *bio, struct uio_meta *meta)
+{
+	struct blk_integrity *bi = blk_get_integrity(bio->bi_bdev->bd_disk);
+	unsigned int integrity_bytes;
+	int ret;
+	struct iov_iter it;
+
+	if (!bi)
+		return -EINVAL;
+	/*
+	 * original meta iterator can be bigger.
+	 * process integrity info corresponding to current data buffer only.
+	 */
+	it = meta->iter;
+	integrity_bytes = bio_integrity_bytes(bi, bio_sectors(bio));
+	if (it.count < integrity_bytes)
+		return -EINVAL;
+
+	it.count = integrity_bytes;
+	ret = bio_integrity_map_user(bio, &it, 0);
+	if (!ret) {
+		bio_uio_meta_to_bip(bio, meta);
+		iov_iter_advance(&meta->iter, integrity_bytes);
+	}
+	return ret;
+}
+
 int bio_integrity_map_user(struct bio *bio, struct iov_iter *iter,
 			   u32 seed)
 {
diff --git a/block/fops.c b/block/fops.c
index 9825c1713a49..5c18676c17ab 100644
--- a/block/fops.c
+++ b/block/fops.c
@@ -154,6 +154,9 @@  static void blkdev_bio_end_io(struct bio *bio)
 		}
 	}
 
+	if (bio_integrity(bio) && (dio->iocb->ki_flags & IOCB_HAS_META))
+		bio_integrity_unmap_user(bio);
+
 	if (should_dirty) {
 		bio_check_pages_dirty(bio);
 	} else {
@@ -231,6 +234,16 @@  static ssize_t __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
 			}
 			bio->bi_opf |= REQ_NOWAIT;
 		}
+		if (!is_sync && unlikely(iocb->ki_flags & IOCB_HAS_META)) {
+			ret = bio_integrity_map_iter(bio, iocb->private);
+			if (unlikely(ret)) {
+				bio_release_pages(bio, false);
+				bio_clear_flag(bio, BIO_REFFED);
+				bio_put(bio);
+				blk_finish_plug(&plug);
+				return ret;
+			}
+		}
 
 		if (is_read) {
 			if (dio->flags & DIO_SHOULD_DIRTY)
@@ -288,6 +301,9 @@  static void blkdev_bio_end_io_async(struct bio *bio)
 		ret = blk_status_to_errno(bio->bi_status);
 	}
 
+	if (bio_integrity(bio) && (iocb->ki_flags & IOCB_HAS_META))
+		bio_integrity_unmap_user(bio);
+
 	iocb->ki_complete(iocb, ret);
 
 	if (dio->flags & DIO_SHOULD_DIRTY) {
@@ -348,6 +364,15 @@  static ssize_t __blkdev_direct_IO_async(struct kiocb *iocb,
 		task_io_account_write(bio->bi_iter.bi_size);
 	}
 
+	if (unlikely(iocb->ki_flags & IOCB_HAS_META)) {
+		ret = bio_integrity_map_iter(bio, iocb->private);
+		WRITE_ONCE(iocb->private, NULL);
+		if (unlikely(ret)) {
+			bio_put(bio);
+			return ret;
+		}
+	}
+
 	if (iocb->ki_flags & IOCB_ATOMIC)
 		bio->bi_opf |= REQ_ATOMIC;
 
diff --git a/block/t10-pi.c b/block/t10-pi.c
index e7052a728966..cb7bc4a88380 100644
--- a/block/t10-pi.c
+++ b/block/t10-pi.c
@@ -139,6 +139,8 @@  static void t10_pi_type1_prepare(struct request *rq)
 		/* Already remapped? */
 		if (bip->bip_flags & BIP_MAPPED_INTEGRITY)
 			break;
+		if (bip->bip_flags & BIP_INTEGRITY_USER)
+			break;
 
 		bip_for_each_vec(iv, bip, iter) {
 			unsigned int j;
@@ -188,6 +190,8 @@  static void t10_pi_type1_complete(struct request *rq, unsigned int nr_bytes)
 		struct bio_vec iv;
 		struct bvec_iter iter;
 
+		if (bip->bip_flags & BIP_INTEGRITY_USER)
+			break;
 		bip_for_each_vec(iv, bip, iter) {
 			unsigned int j;
 			void *p;
@@ -313,6 +317,8 @@  static void ext_pi_type1_prepare(struct request *rq)
 		/* Already remapped? */
 		if (bip->bip_flags & BIP_MAPPED_INTEGRITY)
 			break;
+		if (bip->bip_flags & BIP_INTEGRITY_USER)
+			break;
 
 		bip_for_each_vec(iv, bip, iter) {
 			unsigned int j;
diff --git a/include/linux/bio-integrity.h b/include/linux/bio-integrity.h
index c7c0121689e1..22ff2ae16444 100644
--- a/include/linux/bio-integrity.h
+++ b/include/linux/bio-integrity.h
@@ -14,6 +14,9 @@  enum bip_flags {
 	BIP_CHECK_GUARD		= 1 << 6,
 	BIP_CHECK_REFTAG	= 1 << 7,
 	BIP_CHECK_APPTAG	= 1 << 8,
+	BIP_INTEGRITY_USER      = 1 << 9, /* Integrity payload is user
+					   * address
+					   */
 };
 
 struct bio_integrity_payload {
@@ -24,6 +27,7 @@  struct bio_integrity_payload {
 	unsigned short		bip_vcnt;	/* # of integrity bio_vecs */
 	unsigned short		bip_max_vcnt;	/* integrity bio_vec slots */
 	unsigned short		bip_flags;	/* control flags */
+	u16			app_tag;
 
 	struct bvec_iter	bio_iter;	/* for rewinding parent bio */
 
@@ -44,7 +48,8 @@  struct uio_meta {
 
 #define BIP_CLONE_FLAGS (BIP_MAPPED_INTEGRITY | BIP_CTRL_NOCHECK | \
 			 BIP_DISK_NOCHECK | BIP_IP_CHECKSUM | \
-			 BIP_CHECK_GUARD | BIP_CHECK_REFTAG | BIP_CHECK_APPTAG)
+			 BIP_CHECK_GUARD | BIP_CHECK_REFTAG | \
+			 BIP_CHECK_APPTAG | BIP_INTEGRITY_USER)
 
 #ifdef CONFIG_BLK_DEV_INTEGRITY
 
@@ -89,6 +94,7 @@  struct bio_integrity_payload *bio_integrity_alloc(struct bio *bio, gfp_t gfp,
 int bio_integrity_add_page(struct bio *bio, struct page *page, unsigned int len,
 		unsigned int offset);
 int bio_integrity_map_user(struct bio *bio, struct iov_iter *iter, u32 seed);
+int bio_integrity_map_iter(struct bio *bio, struct uio_meta *meta);
 void bio_integrity_unmap_user(struct bio *bio);
 bool bio_integrity_prep(struct bio *bio);
 void bio_integrity_advance(struct bio *bio, unsigned int bytes_done);
@@ -120,6 +126,11 @@  static inline int bio_integrity_map_user(struct bio *bio, struct iov_iter *iter,
 	return -EINVAL;
 }
 
+static inline int bio_integrity_map_iter(struct bio *bio, struct uio_meta *meta)
+{
+	return -EINVAL;
+}
+
 static inline void bio_integrity_unmap_user(struct bio *bio)
 {
 }