block/dm: add bio_rewind to improve DM requeue
Subsystems (2)
Mike Snitzer
maintainerv2 cover block/dm: add bio_rewind to improve DM requeue
Hi,
This v2 is the by-product of my having picked up v1's patch 2+3 and folding them (into patch 1 of this v2 series). So patch 1 of this v2 is settled and will be merged into 5.20.
Patches 2 and 3 are the result of me having gone over the v1 code. I folded in the copy-n-paste bugfix that Eric Biggers kindly pointed out in v1. I also updated patch headers and code comments for clarity. And I also renamed some variables, tweaked some style knits (subjective but whatever).
This code is available in linux-dm.git's 'dm-5.20' branch:
https://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm.git/log/?h=dm-5.20
Which is based on Jens's for-5.20/block
I'll obviously rebase 'dm-5.20' ontop of Jens's branch if/when he picks up patch 2 in this series.
This code is passing all my tests so far but testing is ongoing.
All comments welcome.
ps. I was told Kent had replied to the v1 thread while I was working on this v2, I kept focus on finishing this v2. I'll check my email in a bit.
Ming Lei (3):
dm: improve BLK_STS_DM_REQUEUE and BLK_STS_AGAIN handling
block: add bio_rewind() API
dm: add two stage requeue mechanism
block/bio-integrity.c | 19 +++++ block/bio.c | 20 +++++ block/blk-crypto-internal.h | 7 ++ block/blk-crypto.c | 25 ++++++ drivers/md/dm-core.h | 11 ++- drivers/md/dm.c | 189 ++++++++++++++++++++++++++++++++++---------- include/linux/bio.h | 21 +++++ include/linux/bvec.h | 33 ++++++++ 8 files changed, 281 insertions(+), 44 deletions(-)
Signature · 1 line
2.15.0
Mike Snitzer
maintainerPATCH v2 1/3 dm: improve BLK_STS_DM_REQUEUE and BLK_STS_AGAIN handling
From: Ming Lei <[email protected]>
If either BLK_STS_DM_REQUEUE or BLK_STS_AGAIN is returned for POLLED io, we requeue the original bio into deferred list and kick md->wq to re-submit it to block layer.
Improve the handling in the following way:
1) Factor out dm_handle_requeue() for handling dm_io requeue.
2) Unify handling for BLK_STS_DM_REQUEUE and BLK_STS_AGAIN: clear REQ_POLLED for BLK_STS_DM_REQUEUE too, for the sake of simplicity, given BLK_STS_DM_REQUEUE is very unusual.
3) Queue md->wq explicitly in dm_handle_requeue(), so requeue handling becomes more robust.
--- drivers/md/dm.c | 70 ++++++++++++++++++++++++++++++++++++--------------------- 1 file changed, 45 insertions(+), 25 deletions(-)
drivers/md/dm.c +45 -25
@@ -880,22 +880,41 @@ static int __noflush_suspending(struct mapped_device *md)return test_bit (DMF_NOFLUSH_SUSPENDING , &md->flags); } -static void dm_io_complete (struct dm_io *io) +/* +* Return true if the dm_io's original bio is requeued. +* io->status is updated with error if requeue disallowed. +*/ +static bool dm_handle_requeue (struct dm_io *io) { -blk_status_t io_error; -struct mapped_device *md = io->md;struct bio *bio = io->split_bio ? io->split_bio : io->orig_bio; +bool handle_requeue = (io->status ==BLK_STS_DM_REQUEUE ); +bool handle_polled_eagain = ((io->status ==BLK_STS_AGAIN ) && + (bio->bi_opf &REQ_POLLED )); +struct mapped_device *md = io->md; +bool requeued = false; -if (io->status ==BLK_STS_DM_REQUEUE ) { +if (handle_requeue || handle_polled_eagain) {unsigned long flags; + +if (bio->bi_opf &REQ_POLLED ) { +/* +* Upper layer won't help us poll split bio +* (io->orig_bio may only reflect a subset of the +* pre-split original) so clear REQ_POLLED. +*/ +bio_clear_polled (bio); + } +/* -* Target requested pushing back the I/O. +* Target requested pushing back the I/O or +* polled IO hit BLK_STS_AGAIN. */ spin_lock_irqsave (&md->deferred_lock, flags); -if (__noflush_suspending (md) && - !WARN_ON_ONCE (dm_is_zone_write (md, bio))) { -/* NOTE early return due to BLK_STS_DM_REQUEUE below */ +if ((__noflush_suspending (md) && + !WARN_ON_ONCE (dm_is_zone_write (md, bio))) || + handle_polled_eagain) {bio_list_add_head (&md->deferred, bio); + requeued = true; }else {/* * noflush suspend was interrupted or this is
@@ -906,6 +925,21 @@ static void dm_io_complete(struct dm_io *io)spin_unlock_irqrestore (&md->deferred_lock, flags); } +if (requeued) +queue_work (md->wq, &md->work); + +return requeued; +} + +static void dm_io_complete (struct dm_io *io) +{ +struct bio *bio = io->split_bio ? io->split_bio : io->orig_bio; +struct mapped_device *md = io->md; +blk_status_t io_error; +bool requeued; + + requeued =dm_handle_requeue (io); + io_error = io->status;if (dm_io_flagged (io,DM_IO_ACCOUNTED ))dm_end_io_acct (io);
@@ -925,23 +959,9 @@ static void dm_io_complete(struct dm_io *io)if (unlikely (wq_has_sleeper (&md->wait)))wake_up (&md->wait); -if (io_error ==BLK_STS_DM_REQUEUE || io_error ==BLK_STS_AGAIN ) { -if (bio->bi_opf &REQ_POLLED ) { -/* -* Upper layer won't help us poll split bio (io->orig_bio -* may only reflect a subset of the pre-split original) -* so clear REQ_POLLED in case of requeue. -*/ -bio_clear_polled (bio); -if (io_error ==BLK_STS_AGAIN ) { -/* io_uring doesn't handle BLK_STS_AGAIN (yet) */ -queue_io (md, bio); -return ; - } - } -if (io_error ==BLK_STS_DM_REQUEUE ) -return ; - } +/* Return early if the original bio was requeued */ +if (requeued) +return ;if (bio_is_flush_with_data (bio)) {/*
Signature · 1 line
2.15.0
Mike Snitzer
maintainerPATCH v2 2/3 block: add bio_rewind() API
In reply to Mike SnitzerFrom: Ming Lei <[email protected]>
Commit 7759eb23fd98 ("block: remove bio_rewind_iter()") removes the similar API because the following reasons:
```
It is pointed that bio_rewind_iter() is one very bad API[1]:
1) bio size may not be restored after rewinding
2) it causes some bogus change, such as 5151842b9d8732 (block: reset
bi_iter.bi_done after splitting bio)
3) rewinding really makes things complicated wrt. bio splitting
4) unnecessary updating of .bi_done in fast path
[1] https://marc.info/?t=153549924200005&r=1&w=2
So this patch takes Kent's suggestion to restore one bio into its original
state via saving bio iterator(struct bvec_iter) in bio_integrity_prep(),
given now bio_rewind_iter() is only used by bio integrity code.
```However, saving off a copy of the 32 bytes bio->bi_iter in case rewind needed isn't efficient because it bloats per-bio-data for what is an unlikely case. That suggestion also ignores the need to restore crypto and integrity info.
Add bio_rewind() API for a specific use-case that is much more narrow than the previous more generic rewind code that was reverted:
1) most bios have a fixed end sector since bio split is done from front of the bio, if driver just records how many sectors between current bio's start sector and the original bio's end sector, the original position can be restored. Keeping the original bio's end sector fixed is a _hard_ requirement for this bio_rewind() interface!
2) if a bio's end sector won't change (usually bio_trim() isn't called, or in the case of DM it preserves original bio), user can restore original position by the storing sector offset from the current ->bi_iter.bi_sector to bio's end sector; together with saving bio size, only 8 bytes is needed to restore to original bio.
3) DM's requeue use case: when BLK_STS_DM_REQUEUE happens, DM core needs to restore to an "original bio" which represents the current dm_io to be requeued (which may be a subset of the original bio). By storing the sector offset from the original bio's end sector and dm_io's size, bio_rewind() can restore such original bio. See commit 7dd76d1feec7 ("dm: improve bio splitting and associated IO accounting") for more details on how DM does this. Leveraging this, allows DM core to shift the need for bio cloning from bio-split time (during IO submission) to the less likely BLK_STS_DM_REQUEUE handling (after IO completes with that error).
4) Unlike the original rewind API, bio_rewind() doesn't add .bi_done to bvec_iter and there is no effect on the fast path.
Implement bio_wind() by factoring out clear helpers that it calls:bio_integrity_rewind, bio_crypt_rewind and bio_rewind_iter.
--- block/bio-integrity.c | 19 +++++++++++++++++++ block/bio.c | 20 ++++++++++++++++++++ block/blk-crypto-internal.h | 7 +++++++ block/blk-crypto.c | 25 +++++++++++++++++++++++++ include/linux/bio.h | 21 +++++++++++++++++++++ include/linux/bvec.h | 33 +++++++++++++++++++++++++++++++++ 6 files changed, 125 insertions(+)
block/bio-integrity.c +19 -0
@@ -378,6 +378,25 @@ void bio_integrity_advance(struct bio *bio, unsigned int bytes_done)bvec_iter_advance (bip->bip_vec, &bip->bip_iter, bytes); } +/** +* bio_integrity_rewind - Rewind integrity vector +* @bio: bio whose integrity vector to update +* @bytes_done: number of data bytes to rewind +* +* Description: This function calculates how many integrity bytes the +* number of completed data bytes correspond to and rewind the +* integrity vector accordingly. +*/ +void bio_integrity_rewind (struct bio *bio,unsigned int bytes_done) +{ +struct bio_integrity_payload *bip =bio_integrity (bio); +struct blk_integrity *bi =blk_get_integrity (bio->bi_bdev->bd_disk); +unsigned bytes =bio_integrity_bytes (bi, bytes_done >>9 ); + + bip->bip_iter.bi_sector -=bio_integrity_intervals (bi, bytes_done >>9 ); +bvec_iter_rewind (bip->bip_vec, &bip->bip_iter, bytes); +} +/** * bio_integrity_trim - Trim integrity vector * @bio: bio whose integrity vector to update
block/bio.c +20 -0
@@ -1348,6 +1348,26 @@ void __bio_advance(struct bio *bio, unsigned bytes) }EXPORT_SYMBOL (__bio_advance); +/** +* bio_rewind - update ->bi_iter of @bio by rewinding @bytes. +* @bio: bio to rewind +* @bytes: how many bytes to rewind +* +* WARNING: +* Caller must ensure that @bio has a fixed end sector, to allow +* rewinding from end of bio and restoring its original position. +* Caller is also responsibile for restoring bio's size. +*/ +void bio_rewind (struct bio *bio,unsigned bytes) +{ +if (bio_integrity (bio)) +bio_integrity_rewind (bio, bytes); + +bio_crypt_rewind (bio, bytes); +bio_rewind_iter (bio, &bio->bi_iter, bytes); +} +EXPORT_SYMBOL (bio_rewind); +void bio_copy_data_iter (struct bio *dst,struct bvec_iter *dst_iter,struct bio *src,struct bvec_iter *src_iter) {
block/blk-crypto-internal.h +7 -0
@@ -114,6 +114,13 @@ static inline void bio_crypt_advance(struct bio *bio, unsigned int bytes)__bio_crypt_advance (bio, bytes); } +void __bio_crypt_rewind (struct bio *bio,unsigned int bytes); +static inline void bio_crypt_rewind (struct bio *bio,unsigned int bytes) +{ +if (bio_has_crypt_ctx (bio)) +__bio_crypt_rewind (bio, bytes); +} +void __bio_crypt_free_ctx (struct bio *bio);static inline void bio_crypt_free_ctx (struct bio *bio) {
block/blk-crypto.c +25 -0
@@ -134,6 +134,23 @@ void bio_crypt_dun_increment(u64 dun[BLK_CRYPTO_DUN_ARRAY_SIZE], } } +/* Decrements @dun by @dec, treating @dun as a multi-limb integer. */ +void bio_crypt_dun_decrement (u64 dun[BLK_CRYPTO_DUN_ARRAY_SIZE ], +unsigned int dec) +{ +int i; + +for (i =0 ; dec && i <BLK_CRYPTO_DUN_ARRAY_SIZE ; i++) { +u64 prev = dun[i]; + + dun[i] -= dec; +if (dun[i] > prev) + dec =1 ; +else + dec =0 ; + } +} +void __bio_crypt_advance (struct bio *bio,unsigned int bytes) {struct bio_crypt_ctx *bc = bio->bi_crypt_context;
@@ -142,6 +159,14 @@ void __bio_crypt_advance(struct bio *bio, unsigned int bytes) bytes >> bc->bc_key->data_unit_size_bits); } +void __bio_crypt_rewind (struct bio *bio,unsigned int bytes) +{ +struct bio_crypt_ctx *bc = bio->bi_crypt_context; + +bio_crypt_dun_decrement (bc->bc_dun, + bytes >> bc->bc_key->data_unit_size_bits); +} +/* * Returns true if @bc->bc_dun plus @bytes converted to data units is equal to * @next_dun, treating the DUNs as multi-limb integers.
include/linux/bio.h +21 -0
@@ -105,6 +105,19 @@ static inline void bio_advance_iter(const struct bio *bio,/* TODO: It is reasonable to complete bio with error here. */ } +static inline void bio_rewind_iter (const struct bio *bio, +struct bvec_iter *iter,unsigned int bytes) +{ + iter->bi_sector -= bytes >>9 ; + +/* No advance means no rewind */ +if (bio_no_advance_iter (bio)) + iter->bi_size += bytes; +else +bvec_iter_rewind (bio->bi_io_vec, iter, bytes); +/* TODO: It is reasonable to complete bio with error here. */ +} +/* @bytes should be less or equal to bvec[i->bi_idx].bv_len */ static inline void bio_advance_iter_single(const struct bio *bio, struct bvec_iter *iter,
@@ -119,6 +132,7 @@ static inline void bio_advance_iter_single(const struct bio *bio, }void __bio_advance (struct bio *,unsigned bytes); +void bio_rewind (struct bio *,unsigned bytes);/** * bio_advance - increment/complete a bio by some number of bytes
@@ -699,6 +713,7 @@ extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unextern int bio_integrity_add_page (struct bio *,struct page *,unsigned int ,unsigned int );extern bool bio_integrity_prep (struct bio *);extern void bio_integrity_advance (struct bio *,unsigned int ); +extern void bio_integrity_rewind (struct bio *,unsigned int );extern void bio_integrity_trim (struct bio *);extern int bio_integrity_clone (struct bio *,struct bio *,gfp_t );extern int bioset_integrity_create (struct bio_set *,int );
@@ -739,6 +754,12 @@ static inline void bio_integrity_advance(struct bio *bio,return ; } +static inline void bio_integrity_rewind (struct bio *bio, +unsigned int bytes_done) +{ +return ; +} +static inline void bio_integrity_trim (struct bio *bio) {return ;
include/linux/bvec.h +33 -0
@@ -122,6 +122,39 @@ static inline bool bvec_iter_advance(const struct bio_vec *bv,return true; } +static inline bool bvec_iter_rewind (const struct bio_vec *bv, +struct bvec_iter *iter, +unsigned int bytes) +{ +int idx; + + iter->bi_size += bytes; +if (bytes <= iter->bi_bvec_done) { + iter->bi_bvec_done -= bytes; +return true; + } + + bytes -= iter->bi_bvec_done; + idx = iter->bi_idx -1 ; + +while (idx >=0 && bytes && bytes > bv[idx].bv_len) { + bytes -= bv[idx].bv_len; + idx--; + } + +if (WARN_ONCE (idx <0 && bytes, +"Attempted to rewind iter beyond bvec's boundaries\n" )) { + iter->bi_size -= bytes; + iter->bi_bvec_done =0 ; + iter->bi_idx =0 ; +return false; + } + + iter->bi_idx = idx; + iter->bi_bvec_done = bv[idx].bv_len - bytes; +return true; +} +/* * A simpler version of bvec_iter_advance(), @bytes should not span * across multiple bvec entries, i.e. bytes <= bv[i->bi_idx].bv_len
Signature · 1 line
2.15.0
Mike Snitzer
maintainerPATCH v2 3/3 dm: add two stage requeue mechanism
In reply to Mike SnitzerFrom: Ming Lei <[email protected]>
Commit 61b6e2e5321d ("dm: fix BLK_STS_DM_REQUEUE handling when dm_io represents split bio") reverted DM core's bio splitting back to using bio_split()+bio_chain() because it was found that otherwise DM's BLK_STS_DM_REQUEUE would trigger a live-lock waiting for bio completion that would never occur.
Restore using bio_trim()+bio_inc_remaining(), like was done in commit 7dd76d1feec7 ("dm: improve bio splitting and associated IO accounting"), but this time with proper handling for the above scenario that is covered in more detail in the commit header for 61b6e2e5321d.
Solve this issue by adding a two staged dm_io requeue mechanism that uses the new bio_rewind():
1) requeue the dm_io into the requeue_list added to struct mapped_device, and schedule it via new added requeue work. This workqueue just clones the dm_io->orig_bio (which DM saves and ensures its end sector isn't modified). Using the sectors and sectors_offset members of the dm_io that are recorded relative to the end of orig_bio: bio_rewind()+bio_trim() are then used to make that cloned bio reflect the subset of the original bio that is represented by the dm_io that is being requeued.
2) the 2nd stage requeue is same with original requeue, but io->orig_bio points to new cloned bio (which matches the requeued dm_io as described above).
This allows DM core to shift the need for bio cloning from bio-split time (during IO submission) to the less likely BLK_STS_DM_REQUEUE handling (after IO completes with that error).
--- drivers/md/dm-core.h | 11 ++++- drivers/md/dm.c | 133 +++++++++++++++++++++++++++++++++++++++++---------- 2 files changed, 118 insertions(+), 26 deletions(-)
drivers/md/dm-core.h +10 -1
@@ -22,6 +22,8 @@#define DM_RESERVED_MAX_IOS 1024 +struct dm_io ; +struct dm_kobject_holder {struct kobject kobj;struct completion completion;
@@ -91,6 +93,14 @@ struct mapped_device {spinlock_t deferred_lock;struct bio_list deferred; +/* +* requeue work context is needed for cloning one new bio +* to represent the dm_io to be requeued, since each +* dm_io may point to the original bio from FS. +*/ +struct work_struct requeue_work; +struct dm_io *requeue_list; +void *interface_ptr;/*
@@ -275,7 +285,6 @@ struct dm_io {atomic_t io_count;struct mapped_device *md; -struct bio *split_bio;/* The three fields represent mapped part of original bio */ struct bio *orig_bio;unsigned int sector_offset;/* offset to end of orig_bio */
drivers/md/dm.c +108 -25
198 lines not on this page · raw message
Signature · 1 line
2.15.0