Hi Jinyoung,
can you work a bit on the commit log and especially the subject line?
I'd word this as something like:
"Subject: bio-integrity: create multi-page bvecs in bio_integrity_add_page()
Allow bio_integrity_add_page to create multi-page bvecs, just like
the bio payloads. This simplifies adding larger payloads, and fixes
support for non-tiny workloads with nvme, which stopped using scatterlist
for metadata a while ago"
It should probably also mentioned somewhere that you did an audit to
ensure all drivers and the core code is fine with these multi-page
segments. If it's not, this patch should only be added after that
has been made the case.
I think the extra arguments struct is a bit overcompliated, and mostly
due to me making the existing code to weird things in the low-level
helpers. With the "rationalize the flow in bio_add_page and friends"
series I just sent out, I think we can drop the previous patch and
simplify this one down to:
diff --git a/block/bio-integrity.c b/block/bio-integrity.c
index 4533eb49166109..85d70dc723f0ed 100644
--- a/block/bio-integrity.c
+++ b/block/bio-integrity.c
@@ -118,26 +118,44 @@ void bio_integrity_free(struct bio *bio)
* @len: number of bytes of integrity metadata in page
* @offset: start offset within page
*
- * Description: Attach a page containing integrity metadata to bio.
+ * Add a page containing integrity metadata to a bio while respecting
+ * the hardware max_sectors, max_segment and gap limitations.
*/
int bio_integrity_add_page(struct bio *bio, struct page *page,
unsigned int len, unsigned int offset)
{
+ struct request_queue *q = bdev_get_queue(bio->bi_bdev);
struct bio_integrity_payload *bip = bio_integrity(bio);
- if (bip->bip_vcnt >= bip->bip_max_vcnt) {
- printk(KERN_ERR "%s: bip_vec full\n", __func__);
+ if (((bip->bip_iter.bi_size + len) >> SECTOR_SHIFT) >
+ queue_max_hw_sectors(q))
return 0;
- }
- if (bip->bip_vcnt &&
- bvec_gap_to_prev(&bdev_get_queue(bio->bi_bdev)->limits,
- &bip->bip_vec[bip->bip_vcnt - 1], offset))
- return 0;
+ if (bip->bip_vcnt > 0) {
+ struct bio_vec *bv = &bip->bip_vec[bip->bip_vcnt - 1];
+ bool same_page = false;
+
+ if (bvec_try_merge_hw_page(q, bv, page, len, offset,
+ &same_page)) {
+ bip->bip_iter.bi_size += len;
+ return len;
+ }
+
+ if (bip->bip_vcnt >=
+ min(bip->bip_max_vcnt, queue_max_integrity_segments(q)))
+ return 0;
+
+ /*
+ * If the queue doesn't support SG gaps and adding this segment
+ * would create a gap, disallow it.
+ */
+ if (bvec_gap_to_prev(&q->limits, bv, offset))
+ return 0;
+ }
bvec_set_page(&bip->bip_vec[bip->bip_vcnt], page, len, offset);
bip->bip_vcnt++;
-
+ bip->bip_iter.bi_size += len;
return len;
}
EXPORT_SYMBOL(bio_integrity_add_page);
@@ -249,7 +267,6 @@ bool bio_integrity_prep(struct bio *bio)
}
bip->bip_flags |= BIP_BLOCK_INTEGRITY;
- bip->bip_iter.bi_size = len;
bip_set_seed(bip, bio->bi_iter.bi_sector);
if (bi->flags & BLK_INTEGRITY_IP_CHECKSUM)
diff --git a/block/bio.c b/block/bio.c
index 79e8aa600ddbe2..050b57e09ac362 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -934,7 +934,7 @@ static bool bvec_try_merge_page(struct bio_vec *bv, struct page *page,
* size limit. This is not for normal read/write bios, but for passthrough
* or Zone Append operations that we can't split.
*/
-static bool bvec_try_merge_hw_page(struct request_queue *q, struct bio_vec *bv,
+bool bvec_try_merge_hw_page(struct request_queue *q, struct bio_vec *bv,
struct page *page, unsigned len, unsigned offset,
bool *same_page)
{
diff --git a/block/blk.h b/block/blk.h
index 45547bcf111938..1e67f738b52191 100644
--- a/block/blk.h
+++ b/block/blk.h
@@ -486,4 +486,8 @@ static inline int req_ref_read(struct request *req)
return atomic_read(&req->ref);
}
+bool bvec_try_merge_hw_page(struct request_queue *q, struct bio_vec *bv,
+ struct page *page, unsigned len, unsigned offset,
+ bool *same_page);
+
#endif /* BLK_INTERNAL_H */
@@ -111,6 +111,23 @@ void bio_integrity_free(struct bio *bio)
bio->bi_opf &= ~REQ_INTEGRITY;
}
+/**
+ * bip_full - check if the bip is full
+ * @bip: bip to check
+ * @len: length of one segment to be added
+ *
+ * Return true if @bip is full and one segment with @len bytes can't be
+ * added to the bip, otherwise return false
+ */
+static inline bool bip_full(struct bio_integrity_payload *bip, unsigned int len)
+{
+ if (bip->bip_vcnt >= bip->bip_max_vcnt)
+ return true;
+ if (bip->bip_iter.bi_size > UINT_MAX - len)
+ return true;
+ return false;
+}
+
/**
* bio_integrity_add_page - Attach integrity metadata
* @bio: bio to update
@@ -118,25 +135,53 @@ void bio_integrity_free(struct bio *bio)
* @len: number of bytes of integrity metadata in page
* @offset: start offset within page
*
- * Description: Attach a page containing integrity metadata to bio.
+ * Add a page containing integrity metadata to a bio while respecting
+ * the hardware max_sectors, max_segment and gap limitations.
*/
int bio_integrity_add_page(struct bio *bio, struct page *page,
unsigned int len, unsigned int offset)
{
+ struct request_queue *q = bdev_get_queue(bio->bi_bdev);
struct bio_integrity_payload *bip = bio_integrity(bio);
- if (bip->bip_vcnt >= bip->bip_max_vcnt) {
+ if (((bip->bip_iter.bi_size + len) >> 9) > queue_max_hw_sectors(q))
+ return 0;
+
+ if (bip->bip_vcnt > 0) {
+ struct bio_vec *bv = &bip->bip_vec[bip->bip_vcnt - 1];
+ bool same_page = false;
+ struct page_merge_ctx pmc = {
+ .bv = bv,
+ .bi_vcnt = bip->bip_vcnt,
+ .bi_iter = &bip->bip_iter,
+ .page = page,
+ .len = len,
+ .offset = offset,
+ .same_page = &same_page,
+ };
+
+ if (bio_try_merge_hw_seg(q, &pmc))
+ return len;
+
+ /*
+ * If the queue doesn't support SG gaps and adding this segment
+ * would create a gap, disallow it.
+ */
+ if (bvec_gap_to_prev(&q->limits, bv, offset))
+ return 0;
+ }
+
+ if (bip_full(bip, len)) {
printk(KERN_ERR "%s: bip_vec full\n", __func__);
return 0;
}
- if (bip->bip_vcnt &&
- bvec_gap_to_prev(&bdev_get_queue(bio->bi_bdev)->limits,
- &bip->bip_vec[bip->bip_vcnt - 1], offset))
+ if (bip->bip_vcnt >= queue_max_integrity_segments(q))
return 0;
bvec_set_page(&bip->bip_vec[bip->bip_vcnt], page, len, offset);
bip->bip_vcnt++;
+ bip->bip_iter.bi_size += len;
return len;
}
@@ -249,7 +294,6 @@ bool bio_integrity_prep(struct bio *bio)
}
bip->bip_flags |= BIP_BLOCK_INTEGRITY;
- bip->bip_iter.bi_size = len;
bip_set_seed(bip, bio->bi_iter.bi_sector);
if (bi->flags & BLK_INTEGRITY_IP_CHECKSUM)
@@ -258,7 +302,6 @@ bool bio_integrity_prep(struct bio *bio)
/* Map it */
offset = offset_in_page(buf);
for (i = 0 ; i < nr_pages ; i++) {
- int ret;
bytes = PAGE_SIZE - offset;
if (len <= 0)
@@ -267,18 +310,13 @@ bool bio_integrity_prep(struct bio *bio)
if (bytes > len)
bytes = len;
- ret = bio_integrity_add_page(bio, virt_to_page(buf),
- bytes, offset);
-
- if (ret == 0) {
+ if (bio_integrity_add_page(bio, virt_to_page(buf),
+ bytes, offset) < bytes) {
printk(KERN_ERR "could not attach integrity payload\n");
status = BLK_STS_RESOURCE;
goto err_end_io;
}
- if (ret < bytes)
- break;
-
buf += bytes;
len -= bytes;
offset = 0;
@@ -926,22 +926,23 @@ static inline bool page_is_mergeable(const struct bio_vec *bv,
return (bv->bv_page + bv_end / PAGE_SIZE) == (page + off / PAGE_SIZE);
}
-static bool __bio_try_merge_page(struct bio *bio, struct page *page,
- unsigned int len, unsigned int off,
- bool *same_page)
+static bool __bio_try_merge_page(struct page_merge_ctx *pmc)
{
- struct bio_vec *bv = &bio->bi_io_vec[bio->bi_vcnt - 1];
+ struct bio_vec *bv = pmc->bv;
+ struct bvec_iter *bi_iter = pmc->bi_iter;
+ unsigned int len = pmc->len;
+ bool *same_page = pmc->same_page;
- if (!page_is_mergeable(bv, page, len, off, same_page))
+ if (!page_is_mergeable(bv, pmc->page, len, pmc->offset, same_page))
return false;
- if (bio->bi_iter.bi_size > UINT_MAX - len) {
+ if (bi_iter->bi_size > UINT_MAX - len) {
*same_page = false;
return false;
}
bv->bv_len += len;
- bio->bi_iter.bi_size += len;
+ bi_iter->bi_size += len;
return true;
}
@@ -966,13 +967,23 @@ static bool bio_try_merge_page(struct bio *bio, struct page *page,
unsigned int len, unsigned int off,
bool *same_page)
{
+ struct page_merge_ctx pmc;
+
if (WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED)))
return false;
if (!bio->bi_vcnt)
return false;
- return __bio_try_merge_page(bio, page, len, off, same_page);
+ pmc.bv = &bio->bi_io_vec[bio->bi_vcnt - 1];
+ pmc.bi_vcnt = bio->bi_vcnt;
+ pmc.bi_iter = &bio->bi_iter;
+ pmc.page = page;
+ pmc.len = len;
+ pmc.offset = off;
+ pmc.same_page = same_page;
+
+ return __bio_try_merge_page(&pmc);
}
/*
@@ -980,20 +991,19 @@ static bool bio_try_merge_page(struct bio *bio, struct page *page,
* size limit. This is not for normal read/write bios, but for passthrough
* or Zone Append operations that we can't split.
*/
-static bool bio_try_merge_hw_seg(struct request_queue *q, struct bio *bio,
- struct page *page, unsigned len,
- unsigned offset, bool *same_page)
+bool bio_try_merge_hw_seg(struct request_queue *q, struct page_merge_ctx *pmc)
{
- struct bio_vec *bv = &bio->bi_io_vec[bio->bi_vcnt - 1];
unsigned long mask = queue_segment_boundary(q);
+ struct bio_vec *bv = pmc->bv;
+ unsigned int len = pmc->len;
phys_addr_t addr1 = page_to_phys(bv->bv_page) + bv->bv_offset;
- phys_addr_t addr2 = page_to_phys(page) + offset + len - 1;
+ phys_addr_t addr2 = page_to_phys(pmc->page) + pmc->offset + len - 1;
if ((addr1 | mask) != (addr2 | mask))
return false;
if (bv->bv_len + len > queue_max_segment_size(q))
return false;
- return __bio_try_merge_page(bio, page, len, offset, same_page);
+ return __bio_try_merge_page(pmc);
}
/**
@@ -1013,8 +1023,6 @@ int bio_add_hw_page(struct request_queue *q, struct bio *bio,
struct page *page, unsigned int len, unsigned int offset,
unsigned int max_sectors, bool *same_page)
{
- struct bio_vec *bvec;
-
if (WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED)))
return 0;
@@ -1022,15 +1030,25 @@ int bio_add_hw_page(struct request_queue *q, struct bio *bio,
return 0;
if (bio->bi_vcnt > 0) {
- if (bio_try_merge_hw_seg(q, bio, page, len, offset, same_page))
+ struct bio_vec *bv = &bio->bi_io_vec[bio->bi_vcnt - 1];
+ struct page_merge_ctx pmc = {
+ .bv = bv,
+ .bi_vcnt = bio->bi_vcnt,
+ .bi_iter = &bio->bi_iter,
+ .page = page,
+ .len = len,
+ .offset = offset,
+ .same_page = same_page,
+ };
+
+ if (bio_try_merge_hw_seg(q, &pmc))
return len;
/*
* If the queue doesn't support SG gaps and adding this segment
* would create a gap, disallow it.
*/
- bvec = &bio->bi_io_vec[bio->bi_vcnt - 1];
- if (bvec_gap_to_prev(&q->limits, bvec, offset))
+ if (bvec_gap_to_prev(&q->limits, bv, offset))
return 0;
}
@@ -118,6 +118,19 @@ static inline bool bvec_gap_to_prev(const struct queue_limits *lim,
return __bvec_gap_to_prev(lim, bprv, offset);
}
+/* page merge context */
+struct page_merge_ctx {
+ struct bio_vec *bv; /* bvec where @page will be merged */
+ unsigned short bi_vcnt; /* how many bio_vec's */
+ struct bvec_iter *bi_iter; /* actual i/o information on device */
+ struct page *page; /* start page to add */
+ unsigned int len; /* length of the data to add */
+ unsigned int offset; /* offset of the data relative to @page */
+ bool *same_page; /* return if the segment has been merged inside the same page*/
+};
+
+bool bio_try_merge_hw_seg(struct request_queue *q, struct page_merge_ctx *pmc);
+
static inline bool rq_mergeable(struct request *rq)
{
if (blk_rq_is_passthrough(rq))