* [PATCH] block: unmap and free user mapped integrity via submitter
[not found] <CGME20240510095142epcas5p4fde65328020139931417f83ccedbce90@epcas5p4.samsung.com>
@ 2024-05-10 9:44 ` Anuj Gupta
2024-05-10 13:45 ` Jens Axboe
2024-05-10 18:56 ` Keith Busch
0 siblings, 2 replies; 5+ messages in thread
From: Anuj Gupta @ 2024-05-10 9:44 UTC (permalink / raw)
To: axboe, hch, kbusch
Cc: linux-nvme, linux-block, martin.petersen, Anuj Gupta,
Kanchan Joshi
The user mapped intergity is copied back and unpinned by
bio_integrity_free which is a low-level routine. Do it via the submitter
rather than doing it in the low-level block layer code, to split the
submitter side from the consumer side of the bio.
Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Signed-off-by: Kanchan Joshi <joshi.k@samsung.com>
---
block/bio-integrity.c | 25 +++++++++++++++++++++++--
drivers/nvme/host/ioctl.c | 20 ++++++++++++++++----
include/linux/bio.h | 4 ++++
3 files changed, 43 insertions(+), 6 deletions(-)
diff --git a/block/bio-integrity.c b/block/bio-integrity.c
index 2e3e8e04961e..82a0fa91bb8f 100644
--- a/block/bio-integrity.c
+++ b/block/bio-integrity.c
@@ -144,16 +144,37 @@ void bio_integrity_free(struct bio *bio)
struct bio_integrity_payload *bip = bio_integrity(bio);
struct bio_set *bs = bio->bi_pool;
+ if (bip->bip_flags & BIP_INTEGRITY_USER)
+ return;
if (bip->bip_flags & BIP_BLOCK_INTEGRITY)
kfree(bvec_virt(bip->bip_vec));
- else if (bip->bip_flags & BIP_INTEGRITY_USER)
- bio_integrity_unmap_user(bip);
__bio_integrity_free(bs, bip);
bio->bi_integrity = NULL;
bio->bi_opf &= ~REQ_INTEGRITY;
}
+/**
+ * bio_integrity_unmap_free_user - Unmap and free bio user integrity payload
+ * @bio: bio containing bip to be unmapped and freed
+ *
+ * Description: Used to unmap and free the user mapped integrity portion of a
+ * bio. Submitter attaching the user integrity buffer is responsible for
+ * unmapping and freeing it during completion.
+ */
+void bio_integrity_unmap_free_user(struct bio *bio)
+{
+ struct bio_integrity_payload *bip = bio_integrity(bio);
+ struct bio_set *bs = bio->bi_pool;
+
+ WARN_ON(!(bip->bip_flags & BIP_INTEGRITY_USER));
+ bio_integrity_unmap_user(bip);
+ __bio_integrity_free(bs, bip);
+ bio->bi_integrity = NULL;
+ bio->bi_opf &= ~REQ_INTEGRITY;
+}
+EXPORT_SYMBOL(bio_integrity_unmap_free_user);
+
/**
* bio_integrity_add_page - Attach integrity metadata
* @bio: bio to update
diff --git a/drivers/nvme/host/ioctl.c b/drivers/nvme/host/ioctl.c
index 499a8bb7cac7..8016dcf23ad1 100644
--- a/drivers/nvme/host/ioctl.c
+++ b/drivers/nvme/host/ioctl.c
@@ -156,8 +156,11 @@ static int nvme_map_user_request(struct request *req, u64 ubuffer,
return ret;
out_unmap:
- if (bio)
+ if (bio) {
+ if (bio_integrity(bio))
+ bio_integrity_unmap_free_user(bio);
blk_rq_unmap_user(bio);
+ }
out:
blk_mq_free_request(req);
return ret;
@@ -194,8 +197,11 @@ static int nvme_submit_user_cmd(struct request_queue *q,
ret = nvme_execute_rq(req, false);
if (result)
*result = le64_to_cpu(nvme_req(req)->result.u64);
- if (bio)
+ if (bio) {
+ if (bio_integrity(bio))
+ bio_integrity_unmap_free_user(bio);
blk_rq_unmap_user(bio);
+ }
blk_mq_free_request(req);
if (effects)
@@ -405,8 +411,11 @@ static void nvme_uring_task_cb(struct io_uring_cmd *ioucmd,
{
struct nvme_uring_cmd_pdu *pdu = nvme_uring_cmd_pdu(ioucmd);
- if (pdu->bio)
+ if (pdu->bio) {
+ if (bio_integrity(pdu->bio))
+ bio_integrity_unmap_free_user(pdu->bio);
blk_rq_unmap_user(pdu->bio);
+ }
io_uring_cmd_done(ioucmd, pdu->status, pdu->result, issue_flags);
}
@@ -431,8 +440,11 @@ static enum rq_end_io_ret nvme_uring_cmd_end_io(struct request *req,
* Otherwise, move the completion to task work.
*/
if (blk_rq_is_poll(req)) {
- if (pdu->bio)
+ if (pdu->bio) {
+ if (bio_integrity(pdu->bio))
+ bio_integrity_unmap_free_user(pdu->bio);
blk_rq_unmap_user(pdu->bio);
+ }
io_uring_cmd_iopoll_done(ioucmd, pdu->result, pdu->status);
} else {
io_uring_cmd_do_in_task_lazy(ioucmd, nvme_uring_task_cb);
diff --git a/include/linux/bio.h b/include/linux/bio.h
index d5379548d684..818e93612947 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -731,6 +731,7 @@ static inline bool bioset_initialized(struct bio_set *bs)
bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
int bio_integrity_map_user(struct bio *bio, void __user *ubuf, ssize_t len, u32 seed);
+void bio_integrity_unmap_free_user(struct bio *bio);
extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
extern bool bio_integrity_prep(struct bio *);
@@ -807,6 +808,9 @@ static inline int bio_integrity_map_user(struct bio *bio, void __user *ubuf,
{
return -EINVAL;
}
+static inline void bio_integrity_unmap_free_user(struct bio *bio)
+{
+}
#endif /* CONFIG_BLK_DEV_INTEGRITY */
--
2.25.1
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH] block: unmap and free user mapped integrity via submitter
2024-05-10 9:44 ` [PATCH] block: unmap and free user mapped integrity via submitter Anuj Gupta
@ 2024-05-10 13:45 ` Jens Axboe
2024-05-10 18:42 ` Anuj gupta
2024-05-10 18:56 ` Keith Busch
1 sibling, 1 reply; 5+ messages in thread
From: Jens Axboe @ 2024-05-10 13:45 UTC (permalink / raw)
To: Anuj Gupta, hch, kbusch
Cc: linux-nvme, linux-block, martin.petersen, Kanchan Joshi
On 5/10/24 3:44 AM, Anuj Gupta wrote:
> +/**
> + * bio_integrity_unmap_free_user - Unmap and free bio user integrity payload
> + * @bio: bio containing bip to be unmapped and freed
> + *
> + * Description: Used to unmap and free the user mapped integrity portion of a
> + * bio. Submitter attaching the user integrity buffer is responsible for
> + * unmapping and freeing it during completion.
> + */
> +void bio_integrity_unmap_free_user(struct bio *bio)
> +{
> + struct bio_integrity_payload *bip = bio_integrity(bio);
> + struct bio_set *bs = bio->bi_pool;
> +
> + WARN_ON(!(bip->bip_flags & BIP_INTEGRITY_USER));
> + bio_integrity_unmap_user(bip);
> + __bio_integrity_free(bs, bip);
> + bio->bi_integrity = NULL;
> + bio->bi_opf &= ~REQ_INTEGRITY;
> +}
> +EXPORT_SYMBOL(bio_integrity_unmap_free_user);
Should this be a
if (WARN_ON_ONCE(!(bip->bip_flags & BIP_INTEGRITY_USER)))
return;
?
--
Jens Axboe
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] block: unmap and free user mapped integrity via submitter
2024-05-10 13:45 ` Jens Axboe
@ 2024-05-10 18:42 ` Anuj gupta
0 siblings, 0 replies; 5+ messages in thread
From: Anuj gupta @ 2024-05-10 18:42 UTC (permalink / raw)
To: Jens Axboe
Cc: Anuj Gupta, hch, kbusch, linux-nvme, linux-block, martin.petersen,
Kanchan Joshi
On Fri, May 10, 2024 at 7:15 PM Jens Axboe <axboe@kernel.dk> wrote:
>
> On 5/10/24 3:44 AM, Anuj Gupta wrote:
> > +/**
> > + * bio_integrity_unmap_free_user - Unmap and free bio user integrity payload
> > + * @bio: bio containing bip to be unmapped and freed
> > + *
> > + * Description: Used to unmap and free the user mapped integrity portion of a
> > + * bio. Submitter attaching the user integrity buffer is responsible for
> > + * unmapping and freeing it during completion.
> > + */
> > +void bio_integrity_unmap_free_user(struct bio *bio)
> > +{
> > + struct bio_integrity_payload *bip = bio_integrity(bio);
> > + struct bio_set *bs = bio->bi_pool;
> > +
> > + WARN_ON(!(bip->bip_flags & BIP_INTEGRITY_USER));
> > + bio_integrity_unmap_user(bip);
> > + __bio_integrity_free(bs, bip);
> > + bio->bi_integrity = NULL;
> > + bio->bi_opf &= ~REQ_INTEGRITY;
> > +}
> > +EXPORT_SYMBOL(bio_integrity_unmap_free_user);
>
> Should this be a
>
> if (WARN_ON_ONCE(!(bip->bip_flags & BIP_INTEGRITY_USER)))
> return;
>
> ?
Right, I will fix this in the next version.
>
> --
> Jens Axboe
>
--
Anuj Gupta
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] block: unmap and free user mapped integrity via submitter
2024-05-10 9:44 ` [PATCH] block: unmap and free user mapped integrity via submitter Anuj Gupta
2024-05-10 13:45 ` Jens Axboe
@ 2024-05-10 18:56 ` Keith Busch
2024-05-11 11:22 ` Anuj gupta
1 sibling, 1 reply; 5+ messages in thread
From: Keith Busch @ 2024-05-10 18:56 UTC (permalink / raw)
To: Anuj Gupta
Cc: axboe, hch, linux-nvme, linux-block, martin.petersen,
Kanchan Joshi
On Fri, May 10, 2024 at 03:14:29PM +0530, Anuj Gupta wrote:
> The user mapped intergity is copied back and unpinned by
> bio_integrity_free which is a low-level routine. Do it via the submitter
> rather than doing it in the low-level block layer code, to split the
> submitter side from the consumer side of the bio.
Thanks, this looks pretty good.
> out_unmap:
> - if (bio)
> + if (bio) {
> + if (bio_integrity(bio))
> + bio_integrity_unmap_free_user(bio);
> blk_rq_unmap_user(bio);
> + }
Since we're adding more cleanup responsibilities on the caller, and this
pattern is repeated 4 times, I think a little helper function is
warranted: 'nvme_unmap_bio(struct bio *bio)', or something like that.
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] block: unmap and free user mapped integrity via submitter
2024-05-10 18:56 ` Keith Busch
@ 2024-05-11 11:22 ` Anuj gupta
0 siblings, 0 replies; 5+ messages in thread
From: Anuj gupta @ 2024-05-11 11:22 UTC (permalink / raw)
To: Keith Busch
Cc: Anuj Gupta, axboe, hch, linux-nvme, linux-block, martin.petersen,
Kanchan Joshi
On Sat, May 11, 2024 at 12:31 AM Keith Busch <kbusch@kernel.org> wrote:
>
> On Fri, May 10, 2024 at 03:14:29PM +0530, Anuj Gupta wrote:
> > The user mapped intergity is copied back and unpinned by
> > bio_integrity_free which is a low-level routine. Do it via the submitter
> > rather than doing it in the low-level block layer code, to split the
> > submitter side from the consumer side of the bio.
>
> Thanks, this looks pretty good.
>
> > out_unmap:
> > - if (bio)
> > + if (bio) {
> > + if (bio_integrity(bio))
> > + bio_integrity_unmap_free_user(bio);
> > blk_rq_unmap_user(bio);
> > + }
>
> Since we're adding more cleanup responsibilities on the caller, and this
> pattern is repeated 4 times, I think a little helper function is
> warranted: 'nvme_unmap_bio(struct bio *bio)', or something like that.
Makes sense, I will add this in the next version.
Thanks,
Anuj Gupta
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2024-05-11 11:23 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
[not found] <CGME20240510095142epcas5p4fde65328020139931417f83ccedbce90@epcas5p4.samsung.com>
2024-05-10 9:44 ` [PATCH] block: unmap and free user mapped integrity via submitter Anuj Gupta
2024-05-10 13:45 ` Jens Axboe
2024-05-10 18:42 ` Anuj gupta
2024-05-10 18:56 ` Keith Busch
2024-05-11 11:22 ` Anuj gupta
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox