remove bi_phys_segments and related cleanups
Subsystems (5)
Christoph Hellwig
maintainerHi Jens,
this series removes the bi_phys_segments member in struct bio and cleans up various areas around it.
Christoph Hellwig
maintainerPATCH 1/6 block: initialize the write priority in blk_rq_bio_prep
The priority field also makes sense for passthrough requests, so initialize it in blk_rq_bio_prep.
--- block/blk-core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
block/blk-core.c +1 -1
@@ -680,7 +680,6 @@ void blk_init_request_from_bio(struct request *req, struct bio *bio) req->cmd_flags |=REQ_FAILFAST_MASK ; req->__sector = bio->bi_iter.bi_sector; - req->ioprio =bio_prio (bio); req->write_hint = bio->bi_write_hint;blk_rq_bio_prep (req->q, req, bio); }
@@ -1436,6 +1435,7 @@ void blk_rq_bio_prep(struct request_queue *q, struct request *rq, rq->__data_len = bio->bi_iter.bi_size; rq->bio = rq->biotail = bio; + rq->ioprio =bio_prio (bio);if (bio->bi_disk) rq->rq_disk = bio->bi_disk;
Signature · 1 line
2.20.1
Re: [PATCH 1/6] block: initialize the write priority in blk_rq_bio_prep
Quoted 23 lines · Christoph Hellwig
The priority field also makes sense for passthrough requests, so initialize it in
blk_rq_bio_prep.Signed-off-by Christoph Hellwig <[email protected]>Reviewed-by Chaitanya Kulkarni <[email protected]>--- block/blk-core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)block/blk-core.c +0
-1@@ -680,7 +680,6 @@ void blk_init_request_from_bio(struct request *req, struct bio *bio) req->cmd_flags |=REQ_FAILFAST_MASK ; req->__sector = bio->bi_iter.bi_sector; - req->ioprio =bio_prio (bio); req->write_hint = bio->bi_write_hint;blk_rq_bio_prep (req->q, req, bio); }
block/blk-core.c void blk_rq_bio_prep(struct request_queue *q, struct request *rq, · line 1441
Quoted 7 lines
block/blk-core.c +1
-0@@ -1436,6 +1435,7 @@ void blk_rq_bio_prep(struct request_queue *q, struct request *rq, rq->__data_len = bio->bi_iter.bi_size; rq->bio = rq->biotail = bio; + rq->ioprio =bio_prio (bio);if (bio->bi_disk) rq->rq_disk = bio->bi_disk;
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Teamlead Storage & Networking [email protected] +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: Felix Imendörffer, Mary Higgins, Sri Rasiah HRB 21284 (AG Nürnberg)
Re: [PATCH 1/6] block: initialize the write priority in blk_rq_bio_prep
In reply to Christoph HellwigIt looks good to me:
Re: [PATCH 1/6] block: initialize the write priority in blk_rq_bio_prep
In reply to Christoph HellwigLooks good to me"
Christoph Hellwig
maintainerPATCH 2/6 block: remove blk_init_request_from_bio
In reply to Christoph Hellwiglightnvm should have never used this function, as it is sending passthrough requests, so switch it to blk_rq_append_bio like all the other passthrough request users. Inline blk_init_request_from_bio into the only remaining caller.
--- block/blk-core.c | 11 ----------- block/blk-mq.c | 7 ++++++- drivers/nvme/host/lightnvm.c | 2 +- include/linux/blkdev.h | 1 - 4 files changed, 7 insertions(+), 14 deletions(-)
block/blk-core.c +0 -11
@@ -674,17 +674,6 @@ bool blk_attempt_plug_merge(struct request_queue *q, struct bio *bio,return false; } -void blk_init_request_from_bio (struct request *req,struct bio *bio) -{ -if (bio->bi_opf &REQ_RAHEAD ) - req->cmd_flags |=REQ_FAILFAST_MASK ; - - req->__sector = bio->bi_iter.bi_sector; - req->write_hint = bio->bi_write_hint; -blk_rq_bio_prep (req->q, req, bio); -} -EXPORT_SYMBOL_GPL (blk_init_request_from_bio); -static void handle_bad_sector (struct bio *bio,sector_t maxsector) {char b[BDEVNAME_SIZE ];
block/blk-mq.c +6 -1
@@ -1766,7 +1766,12 @@ void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule)static void blk_mq_bio_to_request (struct request *rq,struct bio *bio) { -blk_init_request_from_bio (rq, bio); +if (bio->bi_opf &REQ_RAHEAD ) + rq->cmd_flags |=REQ_FAILFAST_MASK ; + + rq->__sector = bio->bi_iter.bi_sector; + rq->write_hint = bio->bi_write_hint; +blk_rq_bio_prep (rq->q, rq, bio);blk_account_io_start (rq, true); }
drivers/nvme/host/lightnvm.c +1 -1
@@ -660,7 +660,7 @@ static struct request *nvme_nvm_alloc_request(struct request_queue *q, rq->cmd_flags &= ~REQ_FAILFAST_DRIVER ;if (rqd->bio) -blk_init_request_from_bio (rq, rqd->bio); +blk_rq_append_bio (rq, &rqd->bio);else rq->ioprio =IOPRIO_PRIO_VALUE (IOPRIO_CLASS_BE ,IOPRIO_NORM );
include/linux/blkdev.h +0 -1
@@ -828,7 +828,6 @@ extern void blk_unregister_queue(struct gendisk *disk);extern blk_qc_t generic_make_request (struct bio *bio);extern blk_qc_t direct_make_request (struct bio *bio);extern void blk_rq_init (struct request_queue *q,struct request *rq); -extern void blk_init_request_from_bio (struct request *req,struct bio *bio);extern void blk_put_request (struct request *);extern struct request *blk_get_request (struct request_queue *,unsigned int op,blk_mq_req_flags_t flags);
Signature · 1 line
2.20.1
Re: [PATCH 2/6] block: remove blk_init_request_from_bio
Quoted 5 lines · Christoph Hellwig
lightnvm should have never used this function, as it is sending passthrough requests, so switch it to
blk_rq_append_biolike all the other passthrough request users. Inlineblk_init_request_from_biointo the only remaining caller.
Quoted 8 lines
Signed-off-by Christoph Hellwig <[email protected]>--- block/blk-core.c | 11 ----------- block/blk-mq.c | 7 ++++++- drivers/nvme/host/lightnvm.c | 2 +- include/linux/blkdev.h | 1 - 4 files changed, 7 insertions(+), 14 deletions(-)
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Teamlead Storage & Networking [email protected] +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: Felix Imendörffer, Mary Higgins, Sri Rasiah HRB 21284 (AG Nürnberg)
Re: [PATCH 2/6] block: remove blk_init_request_from_bio
In reply to Christoph HellwigIt looks good to me.
Re: [PATCH 2/6] block: remove blk_init_request_from_bio
In reply to Christoph HellwigQuoted 81 lines · Christoph Hellwig
lightnvm should have never used this function, as it is sending passthrough requests, so switch it to
blk_rq_append_biolike all the other passthrough request users. Inlineblk_init_request_from_biointo the only remaining caller.Signed-off-by Christoph Hellwig <[email protected]>---
block/blk-core.c | 11 ----------- block/blk-mq.c | 7 ++++++- drivers/nvme/host/lightnvm.c | 2 +- include/linux/blkdev.h | 1 - 4 files changed, 7 insertions(+), 14 deletions(-)block/blk-core.c +0
-11@@ -674,17 +674,6 @@ bool blk_attempt_plug_merge(struct request_queue *q, struct bio *bio,return false; } -void blk_init_request_from_bio (struct request *req,struct bio *bio) -{ -if (bio->bi_opf &REQ_RAHEAD ) - req->cmd_flags |=REQ_FAILFAST_MASK ; - - req->__sector = bio->bi_iter.bi_sector; - req->write_hint = bio->bi_write_hint; -blk_rq_bio_prep (req->q, req, bio); -} -EXPORT_SYMBOL_GPL (blk_init_request_from_bio); -static void handle_bad_sector (struct bio *bio,sector_t maxsector) {char b[BDEVNAME_SIZE ];block/blk-mq.c +6
-1@@ -1766,7 +1766,12 @@ void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule)static void blk_mq_bio_to_request (struct request *rq,struct bio *bio) { -blk_init_request_from_bio (rq, bio); +if (bio->bi_opf &REQ_RAHEAD ) + rq->cmd_flags |=REQ_FAILFAST_MASK ; + + rq->__sector = bio->bi_iter.bi_sector; + rq->write_hint = bio->bi_write_hint; +blk_rq_bio_prep (rq->q, rq, bio);blk_account_io_start (rq, true); }drivers/nvme/host/lightnvm.c +1
-1@@ -660,7 +660,7 @@ static struct request *nvme_nvm_alloc_request(struct request_queue *q, rq->cmd_flags &= ~REQ_FAILFAST_DRIVER ;if (rqd->bio) -blk_init_request_from_bio (rq, rqd->bio); +blk_rq_append_bio (rq, &rqd->bio);else rq->ioprio =IOPRIO_PRIO_VALUE (IOPRIO_CLASS_BE ,IOPRIO_NORM );include/linux/blkdev.h +0
-2@@ -828,7 +828,6 @@ extern void blk_unregister_queue(struct gendisk *disk);extern blk_qc_t generic_make_request (struct bio *bio);extern blk_qc_t direct_make_request (struct bio *bio);extern void blk_rq_init (struct request_queue *q,struct request *rq); -extern void blk_init_request_from_bio (struct request *req,struct bio *bio);extern void blk_put_request (struct request *);extern struct request *blk_get_request (struct request_queue *,unsigned int op,blk_mq_req_flags_t flags); --
Quoted 1 line
2.20.1
Looks good.
Re: [PATCH 2/6] block: remove blk_init_request_from_bio
In reply to Christoph HellwigQuoted 72 lines · Christoph Hellwig
lightnvm should have never used this function, as it is sending passthrough requests, so switch it to
blk_rq_append_biolike all the other passthrough request users. Inlineblk_init_request_from_biointo the only remaining caller.Signed-off-by Christoph Hellwig <[email protected]>--- block/blk-core.c | 11 ----------- block/blk-mq.c | 7 ++++++- drivers/nvme/host/lightnvm.c | 2 +- include/linux/blkdev.h | 1 - 4 files changed, 7 insertions(+), 14 deletions(-)block/blk-core.c +0
-11@@ -674,17 +674,6 @@ bool blk_attempt_plug_merge(struct request_queue *q, struct bio *bio,return false; } -void blk_init_request_from_bio (struct request *req,struct bio *bio) -{ -if (bio->bi_opf &REQ_RAHEAD ) - req->cmd_flags |=REQ_FAILFAST_MASK ; - - req->__sector = bio->bi_iter.bi_sector; - req->write_hint = bio->bi_write_hint; -blk_rq_bio_prep (req->q, req, bio); -} -EXPORT_SYMBOL_GPL (blk_init_request_from_bio); -static void handle_bad_sector (struct bio *bio,sector_t maxsector) {char b[BDEVNAME_SIZE ];block/blk-mq.c +6
-1@@ -1766,7 +1766,12 @@ void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule)static void blk_mq_bio_to_request (struct request *rq,struct bio *bio) { -blk_init_request_from_bio (rq, bio); +if (bio->bi_opf &REQ_RAHEAD ) + rq->cmd_flags |=REQ_FAILFAST_MASK ; + + rq->__sector = bio->bi_iter.bi_sector; + rq->write_hint = bio->bi_write_hint; +blk_rq_bio_prep (rq->q, rq, bio);blk_account_io_start (rq, true); }drivers/nvme/host/lightnvm.c +1
-1@@ -660,7 +660,7 @@ static struct request *nvme_nvm_alloc_request(struct request_queue *q, rq->cmd_flags &= ~REQ_FAILFAST_DRIVER ;if (rqd->bio) -blk_init_request_from_bio (rq, rqd->bio); +blk_rq_append_bio (rq, &rqd->bio);else rq->ioprio =IOPRIO_PRIO_VALUE (IOPRIO_CLASS_BE ,IOPRIO_NORM );include/linux/blkdev.h +0
-0
include/linux/blkdev.h extern void blk_unregister_queue(struct gendisk *disk); · line 833
Quoted 7 lines
include/linux/blkdev.h +0
-1@@ -828,7 +828,6 @@ extern void blk_unregister_queue(struct gendisk *disk);extern blk_qc_t generic_make_request (struct bio *bio);extern blk_qc_t direct_make_request (struct bio *bio);extern void blk_rq_init (struct request_queue *q,struct request *rq); -extern void blk_init_request_from_bio (struct request *req,struct bio *bio);extern void blk_put_request (struct request *);extern struct request *blk_get_request (struct request_queue *,unsigned int op,blk_mq_req_flags_t flags);
Thanks Christoph.
Christoph Hellwig
maintainerPATCH 3/6 block: remove the bi_phys_segments field in struct bio
In reply to Christoph HellwigWe only need the number of segments in the blk-mq submission path. Remove the field from struct bio, and return it from a variant of blk_queue_split instead of that it can passed as an argument to those functions that need the value.
This also means we stop recounting segments except for cloning and partial segments.
To keep the number of arguments in this how path down remove pointless struct request_queue arguments from any of the functions that had it and grew a nr_segs argument.
--- Documentation/block/biodoc.txt | 1 - block/bfq-iosched.c | 5 ++- block/bio.c | 15 +------ block/blk-core.c | 32 +++++++-------- block/blk-map.c | 10 ++++- block/blk-merge.c | 75 ++++++++++++---------------------- block/blk-mq-sched.c | 26 +++++++----- block/blk-mq-sched.h | 10 +++-- block/blk-mq.c | 23 ++++++----- block/blk.h | 23 ++++++----- block/kyber-iosched.c | 5 ++- block/mq-deadline.c | 5 ++- drivers/md/raid5.c | 1 - include/linux/bio.h | 1 - include/linux/blk-mq.h | 2 +- include/linux/blk_types.h | 6 --- include/linux/blkdev.h | 1 - include/linux/elevator.h | 2 +- 18 files changed, 106 insertions(+), 137 deletions(-)
Documentation/block/biodoc.txt +0 -1
@@ -436,7 +436,6 @@ struct bio { struct bvec_iter bi_iter; /* current index into bio_vec array */ unsigned int bi_size; /* total size in bytes */ - unsigned short bi_phys_segments; /* segments after physaddr coalesce*/ unsigned short bi_hw_segments; /* segments after DMA remapping */ unsigned int bi_max; /* max bio_vecs we can hold used as index into pool */
block/bfq-iosched.c +3 -2
@@ -2027,7 +2027,8 @@ static void bfq_remove_request(struct request_queue *q, } -static bool bfq_bio_merge (struct blk_mq_hw_ctx *hctx,struct bio *bio) +static bool bfq_bio_merge (struct blk_mq_hw_ctx *hctx,struct bio *bio, +unsigned int nr_segs) {struct request_queue *q = hctx->queue;struct bfq_data *bfqd = q->elevator->elevator_data;
@@ -2050,7 +2051,7 @@ static bool bfq_bio_merge(struct blk_mq_hw_ctx *hctx, struct bio *bio) bfqd->bio_bfqq =NULL ; bfqd->bio_bic = bic; - ret =blk_mq_sched_try_merge (q, bio, &free); + ret =blk_mq_sched_try_merge (q, bio, nr_segs, &free);if (free)blk_mq_free_request (free);
block/bio.c +1 -14
@@ -558,14 +558,6 @@ void bio_put(struct bio *bio) }EXPORT_SYMBOL (bio_put); -int bio_phys_segments (struct request_queue *q,struct bio *bio) -{ -if (unlikely (!bio_flagged (bio,BIO_SEG_VALID ))) -blk_recount_segments (q, bio); - -return bio->bi_phys_segments; -} -/** * __bio_clone_fast - clone a bio that shares the original bio's biovec * @bio: destination bio
@@ -739,7 +731,7 @@ static int __bio_add_pc_page(struct request_queue *q, struct bio *bio,if (bio_full (bio))return 0 ; -if (bio->bi_phys_segments >=queue_max_segments (q)) +if (bio->bi_vcnt >=queue_max_segments (q))return 0 ; bvec = &bio->bi_io_vec[bio->bi_vcnt];
@@ -749,8 +741,6 @@ static int __bio_add_pc_page(struct request_queue *q, struct bio *bio, bio->bi_vcnt++; done: bio->bi_iter.bi_size += len; - bio->bi_phys_segments = bio->bi_vcnt; -bio_set_flag (bio,BIO_SEG_VALID );return len; }
@@ -1910,10 +1900,7 @@ void bio_trim(struct bio *bio, int offset, int size)if (offset ==0 && size == bio->bi_iter.bi_size)return ; -bio_clear_flag (bio,BIO_SEG_VALID ); -bio_advance (bio, offset <<9 ); - bio->bi_iter.bi_size = size;if (bio_integrity (bio))
block/blk-core.c +14 -18
@@ -537,15 +537,15 @@ void blk_put_request(struct request *req) }EXPORT_SYMBOL (blk_put_request); -bool bio_attempt_back_merge (struct request_queue *q,struct request *req, -struct bio *bio) +bool bio_attempt_back_merge (struct request *req,struct bio *bio, +unsigned int nr_segs) {const int ff = bio->bi_opf &REQ_FAILFAST_MASK ; -if (!ll_back_merge_fn (q, req, bio)) +if (!ll_back_merge_fn (req, bio, nr_segs))return false; -trace_block_bio_backmerge (q, req, bio); +trace_block_bio_backmerge (req->q, req, bio);if ((req->cmd_flags &REQ_FAILFAST_MASK ) != ff)blk_rq_set_mixed_merge (req);
@@ -558,15 +558,15 @@ bool bio_attempt_back_merge(struct request_queue *q, struct request *req,return true; } -bool bio_attempt_front_merge (struct request_queue *q,struct request *req, -struct bio *bio) +bool bio_attempt_front_merge (struct request *req,struct bio *bio, +unsigned int nr_segs) {const int ff = bio->bi_opf &REQ_FAILFAST_MASK ; -if (!ll_front_merge_fn (q, req, bio)) +if (!ll_front_merge_fn (req, bio, nr_segs))return false; -trace_block_bio_frontmerge (q, req, bio); +trace_block_bio_frontmerge (req->q, req, bio);if ((req->cmd_flags &REQ_FAILFAST_MASK ) != ff)blk_rq_set_mixed_merge (req);
@@ -608,6 +608,7 @@ bool bio_attempt_discard_merge(struct request_queue *q, struct request *req,* blk_attempt_plug_merge - try to merge with %current's plugged list * @q: request_queue new bio is being queued at * @bio: new bio being queued +* @nr_segs: number of segments in @bio * @same_queue_rq: pointer to &struct request that gets filled in when * another request associated with @q is found on the plug list * (optional, may be %NULL)
@@ -626,7 +627,7 @@ bool bio_attempt_discard_merge(struct request_queue *q, struct request *req,* Caller must ensure !blk_queue_nomerges(q) beforehand. */ bool blk_attempt_plug_merge (struct request_queue *q,struct bio *bio, -struct request **same_queue_rq) +unsigned int nr_segs,struct request **same_queue_rq) {struct blk_plug *plug;struct request *rq;
@@ -655,10 +656,10 @@ bool blk_attempt_plug_merge(struct request_queue *q, struct bio *bio,switch (blk_try_merge (rq, bio)) {case ELEVATOR_BACK_MERGE : - merged =bio_attempt_back_merge (q, rq, bio); + merged =bio_attempt_back_merge (rq, bio, nr_segs);break ;case ELEVATOR_FRONT_MERGE : - merged =bio_attempt_front_merge (q, rq, bio); + merged =bio_attempt_front_merge (rq, bio, nr_segs);break ;case ELEVATOR_DISCARD_MERGE : merged =bio_attempt_discard_merge (q, rq, bio);
@@ -1414,14 +1415,9 @@ bool blk_update_request(struct request *req, blk_status_t error, }EXPORT_SYMBOL_GPL (blk_update_request); -void blk_rq_bio_prep (struct request_queue *q,struct request *rq, -struct bio *bio) +void blk_rq_bio_prep (struct request *rq,struct bio *bio,unsigned int nr_segs) { -if (bio_has_data (bio)) - rq->nr_phys_segments =bio_phys_segments (q, bio); -else if (bio_op (bio) ==REQ_OP_DISCARD ) - rq->nr_phys_segments =1 ; - + rq->nr_phys_segments = nr_segs; rq->__data_len = bio->bi_iter.bi_size; rq->bio = rq->biotail = bio; rq->ioprio =bio_prio (bio);
block/blk-map.c +8 -2
@@ -18,13 +18,19 @@int blk_rq_append_bio (struct request *rq,struct bio **bio) {struct bio *orig_bio = *bio; +struct bvec_iter iter; +struct bio_vec bv; +unsigned int nr_segs =0 ;blk_queue_bounce (rq->q, bio); +bio_for_each_bvec (bv, *bio, iter) + nr_segs++; +if (!rq->bio) { -blk_rq_bio_prep (rq->q, rq, *bio); +blk_rq_bio_prep (rq, *bio, nr_segs); }else { -if (!ll_back_merge_fn (rq->q, rq, *bio)) { +if (!ll_back_merge_fn (rq, *bio, nr_segs)) {if (orig_bio != *bio) {bio_put (*bio); *bio = orig_bio;
block/blk-merge.c +27 -48
146 lines not on this page · raw message
block/blk-mq-sched.c +15 -11
@@ -224,7 +224,7 @@ void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx) }bool blk_mq_sched_try_merge (struct request_queue *q,struct bio *bio, -struct request **merged_request) +unsigned int nr_segs,struct request **merged_request) {struct request *rq;
@@ -232,7 +232,7 @@ bool blk_mq_sched_try_merge(struct request_queue *q, struct bio *bio,case ELEVATOR_BACK_MERGE :if (!blk_mq_sched_allow_merge (q, rq, bio))return false; -if (!bio_attempt_back_merge (q, rq, bio)) +if (!bio_attempt_back_merge (rq, bio, nr_segs))return false; *merged_request =attempt_back_merge (q, rq);if (!*merged_request)
@@ -241,7 +241,7 @@ bool blk_mq_sched_try_merge(struct request_queue *q, struct bio *bio,case ELEVATOR_FRONT_MERGE :if (!blk_mq_sched_allow_merge (q, rq, bio))return false; -if (!bio_attempt_front_merge (q, rq, bio)) +if (!bio_attempt_front_merge (rq, bio, nr_segs))return false; *merged_request =attempt_front_merge (q, rq);if (!*merged_request)
@@ -260,7 +260,7 @@ EXPORT_SYMBOL_GPL(blk_mq_sched_try_merge);* of them. */ bool blk_mq_bio_list_merge (struct request_queue *q,struct list_head *list, -struct bio *bio) +struct bio *bio,unsigned int nr_segs) {struct request *rq;int checked =8 ;
@@ -277,11 +277,13 @@ bool blk_mq_bio_list_merge(struct request_queue *q, struct list_head *list,switch (blk_try_merge (rq, bio)) {case ELEVATOR_BACK_MERGE :if (blk_mq_sched_allow_merge (q, rq, bio)) - merged =bio_attempt_back_merge (q, rq, bio); + merged =bio_attempt_back_merge (rq, bio, + nr_segs);break ;case ELEVATOR_FRONT_MERGE :if (blk_mq_sched_allow_merge (q, rq, bio)) - merged =bio_attempt_front_merge (q, rq, bio); + merged =bio_attempt_front_merge (rq, bio, + nr_segs);break ;case ELEVATOR_DISCARD_MERGE : merged =bio_attempt_discard_merge (q, rq, bio);
@@ -304,13 +306,14 @@ EXPORT_SYMBOL_GPL(blk_mq_bio_list_merge);*/ static bool blk_mq_attempt_merge (struct request_queue *q,struct blk_mq_hw_ctx *hctx, -struct blk_mq_ctx *ctx,struct bio *bio) +struct blk_mq_ctx *ctx,struct bio *bio, +unsigned int nr_segs) {enum hctx_type type = hctx->type;lockdep_assert_held (&ctx->lock); -if (blk_mq_bio_list_merge (q, &ctx->rq_lists[type], bio)) { +if (blk_mq_bio_list_merge (q, &ctx->rq_lists[type], bio, nr_segs)) { ctx->rq_merged++;return true; }
@@ -318,7 +321,8 @@ static bool blk_mq_attempt_merge(struct request_queue *q,return false; } -bool __blk_mq_sched_bio_merge (struct request_queue *q,struct bio *bio) +bool __blk_mq_sched_bio_merge (struct request_queue *q,struct bio *bio, +unsigned int nr_segs) {struct elevator_queue *e = q->elevator;struct blk_mq_ctx *ctx =blk_mq_get_ctx (q);
@@ -328,7 +332,7 @@ bool __blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio)if (e && e->type->ops.bio_merge) {blk_mq_put_ctx (ctx); -return e->type->ops.bio_merge (hctx, bio); +return e->type->ops.bio_merge (hctx, bio, nr_segs); } type = hctx->type;
@@ -336,7 +340,7 @@ bool __blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio) !list_empty_careful(&ctx->rq_lists[type])) {/* default per sw-queue merge */ spin_lock (&ctx->lock); - ret =blk_mq_attempt_merge (q, hctx, ctx, bio); + ret =blk_mq_attempt_merge (q, hctx, ctx, bio, nr_segs);spin_unlock (&ctx->lock); }
block/blk-mq-sched.h +6 -4
@@ -12,8 +12,9 @@ void blk_mq_sched_assign_ioc(struct request *rq);void blk_mq_sched_request_inserted (struct request *rq);bool blk_mq_sched_try_merge (struct request_queue *q,struct bio *bio, -struct request **merged_request); -bool __blk_mq_sched_bio_merge (struct request_queue *q,struct bio *bio); +unsigned int nr_segs,struct request **merged_request); +bool __blk_mq_sched_bio_merge (struct request_queue *q,struct bio *bio, +unsigned int nr_segs);bool blk_mq_sched_try_insert_merge (struct request_queue *q,struct request *rq);void blk_mq_sched_mark_restart_hctx (struct blk_mq_hw_ctx *hctx);void blk_mq_sched_restart (struct blk_mq_hw_ctx *hctx);
@@ -30,12 +31,13 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e);void blk_mq_exit_sched (struct request_queue *q,struct elevator_queue *e);static inline bool -blk_mq_sched_bio_merge (struct request_queue *q,struct bio *bio) +blk_mq_sched_bio_merge (struct request_queue *q,struct bio *bio, +unsigned int nr_segs) {if (blk_queue_nomerges (q) || !bio_mergeable (bio))return false; -return __blk_mq_sched_bio_merge (q, bio); +return __blk_mq_sched_bio_merge (q, bio, nr_segs); } static inline bool
block/blk-mq.c +12 -11
@@ -1764,14 +1764,15 @@ void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule) } } -static void blk_mq_bio_to_request (struct request *rq,struct bio *bio) +static void blk_mq_bio_to_request (struct request *rq,struct bio *bio, +unsigned int nr_segs) {if (bio->bi_opf &REQ_RAHEAD ) rq->cmd_flags |=REQ_FAILFAST_MASK ; rq->__sector = bio->bi_iter.bi_sector; rq->write_hint = bio->bi_write_hint; -blk_rq_bio_prep (rq->q, rq, bio); +blk_rq_bio_prep (rq, bio, nr_segs);blk_account_io_start (rq, true); }
@@ -1941,20 +1942,20 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)struct request *rq;struct blk_plug *plug;struct request *same_queue_rq =NULL ; +unsigned int nr_segs;blk_qc_t cookie;blk_queue_bounce (q, &bio); - -blk_queue_split (q, &bio); +__blk_queue_split (q, &bio, &nr_segs);if (!bio_integrity_prep (bio))return BLK_QC_T_NONE ;if (!is_flush_fua && !blk_queue_nomerges (q) && -blk_attempt_plug_merge (q, bio, &same_queue_rq)) +blk_attempt_plug_merge (q, bio, nr_segs, &same_queue_rq))return BLK_QC_T_NONE ; -if (blk_mq_sched_bio_merge (q, bio)) +if (blk_mq_sched_bio_merge (q, bio, nr_segs))return BLK_QC_T_NONE ;rq_qos_throttle (q, bio);
@@ -1977,7 +1978,7 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio) plug = current->plug;if (unlikely (is_flush_fua)) {blk_mq_put_ctx (data.ctx); -blk_mq_bio_to_request (rq, bio); +blk_mq_bio_to_request (rq, bio, nr_segs);/* bypass scheduler for flush rq */ blk_insert_flush (rq);
@@ -1991,7 +1992,7 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)struct request *last =NULL ;blk_mq_put_ctx (data.ctx); -blk_mq_bio_to_request (rq, bio); +blk_mq_bio_to_request (rq, bio, nr_segs);if (!request_count)trace_block_plug (q);
@@ -2006,7 +2007,7 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)blk_add_rq_to_plug (plug, rq); }else if (plug && !blk_queue_nomerges (q)) { -blk_mq_bio_to_request (rq, bio); +blk_mq_bio_to_request (rq, bio, nr_segs);/* * We do limited plugging. If the bio can be merged, do that.
@@ -2035,11 +2036,11 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio) }else if ((q->nr_hw_queues >1 && is_sync) || (!q->elevator && !data.hctx->dispatch_busy)) {blk_mq_put_ctx (data.ctx); -blk_mq_bio_to_request (rq, bio); +blk_mq_bio_to_request (rq, bio, nr_segs);blk_mq_try_issue_directly (data.hctx, rq, &cookie); }else {blk_mq_put_ctx (data.ctx); -blk_mq_bio_to_request (rq, bio); +blk_mq_bio_to_request (rq, bio, nr_segs);blk_mq_sched_insert_request (rq, false, true, true); }
block/blk.h +12 -11
@@ -50,8 +50,7 @@ struct blk_flush_queue *blk_alloc_flush_queue(struct request_queue *q, int node, int cmd_size, gfp_t flags);void blk_free_flush_queue (struct blk_flush_queue *q); -void blk_rq_bio_prep (struct request_queue *q,struct request *rq, -struct bio *bio); +void blk_rq_bio_prep (struct request *rq,struct bio *bio,unsigned int nr_segs);void blk_freeze_queue (struct request_queue *q);static inline void blk_queue_enter_live (struct request_queue *q)
@@ -153,14 +152,14 @@ static inline bool bio_integrity_endio(struct bio *bio)unsigned long blk_rq_timeout (unsigned long timeout);void blk_add_timer (struct request *req); -bool bio_attempt_front_merge (struct request_queue *q,struct request *req, -struct bio *bio); -bool bio_attempt_back_merge (struct request_queue *q,struct request *req, -struct bio *bio); +bool bio_attempt_front_merge (struct request *req,struct bio *bio, +unsigned int nr_segs); +bool bio_attempt_back_merge (struct request *req,struct bio *bio, +unsigned int nr_segs);bool bio_attempt_discard_merge (struct request_queue *q,struct request *req,struct bio *bio);bool blk_attempt_plug_merge (struct request_queue *q,struct bio *bio, -struct request **same_queue_rq); +unsigned int nr_segs,struct request **same_queue_rq);void blk_account_io_start (struct request *req,bool new_io);void blk_account_io_completion (struct request *req,unsigned int bytes);
@@ -194,10 +193,12 @@ static inline int blk_should_fake_timeout(struct request_queue *q) }#endif -int ll_back_merge_fn (struct request_queue *q,struct request *req, -struct bio *bio); -int ll_front_merge_fn (struct request_queue *q,struct request *req, -struct bio *bio); +void __blk_queue_split (struct request_queue *q,struct bio **bio, +unsigned int *nr_segs); +int ll_back_merge_fn (struct request *req,struct bio *bio, +unsigned int nr_segs); +int ll_front_merge_fn (struct request *req,struct bio *bio, +unsigned int nr_segs);struct request *attempt_back_merge (struct request_queue *q,struct request *rq);struct request *attempt_front_merge (struct request_queue *q,struct request *rq); int blk_attempt_req_merge(struct request_queue *q, struct request *rq,
block/kyber-iosched.c +3 -2
@@ -562,7 +562,8 @@ static void kyber_limit_depth(unsigned int op, struct blk_mq_alloc_data *data) } } -static bool kyber_bio_merge (struct blk_mq_hw_ctx *hctx,struct bio *bio) +static bool kyber_bio_merge (struct blk_mq_hw_ctx *hctx,struct bio *bio, +unsigned int nr_segs) {struct kyber_hctx_data *khd = hctx->sched_data;struct blk_mq_ctx *ctx =blk_mq_get_ctx (hctx->queue);
@@ -572,7 +573,7 @@ static bool kyber_bio_merge(struct blk_mq_hw_ctx *hctx, struct bio *bio)bool merged;spin_lock (&kcq->lock); - merged =blk_mq_bio_list_merge (hctx->queue, rq_list, bio); + merged =blk_mq_bio_list_merge (hctx->queue, rq_list, bio, nr_segs);spin_unlock (&kcq->lock);blk_mq_put_ctx (ctx);
block/mq-deadline.c +3 -2
@@ -469,7 +469,8 @@ static int dd_request_merge(struct request_queue *q, struct request **rq,return ELEVATOR_NO_MERGE ; } -static bool dd_bio_merge (struct blk_mq_hw_ctx *hctx,struct bio *bio) +static bool dd_bio_merge (struct blk_mq_hw_ctx *hctx,struct bio *bio, +unsigned int nr_segs) {struct request_queue *q = hctx->queue;struct deadline_data *dd = q->elevator->elevator_data;
@@ -477,7 +478,7 @@ static bool dd_bio_merge(struct blk_mq_hw_ctx *hctx, struct bio *bio)bool ret;spin_lock (&dd->lock); - ret =blk_mq_sched_try_merge (q, bio, &free); + ret =blk_mq_sched_try_merge (q, bio, nr_segs, &free);spin_unlock (&dd->lock);if (free)
drivers/md/raid5.c +0 -1
@@ -5251,7 +5251,6 @@ static int raid5_read_one_chunk(struct mddev *mddev, struct bio *raid_bio)rcu_read_unlock (); raid_bio->bi_next = (void *)rdev;bio_set_dev (align_bi, rdev->bdev); -bio_clear_flag (align_bi,BIO_SEG_VALID ); if (is_badblock(rdev, align_bi->bi_iter.bi_sector, bio_sectors(align_bi),
include/linux/bio.h +0 -1
@@ -408,7 +408,6 @@ static inline void bio_wouldblock_error(struct bio *bio) }struct request_queue ; -extern int bio_phys_segments (struct request_queue *,struct bio *);extern int submit_bio_wait (struct bio *bio);extern void bio_advance (struct bio *,unsigned );
include/linux/blk-mq.h +1 -1
@@ -306,7 +306,7 @@ void blk_mq_delay_kick_requeue_list(struct request_queue *q, unsigned long msecsbool blk_mq_complete_request (struct request *rq);void blk_mq_complete_request_sync (struct request *rq);bool blk_mq_bio_list_merge (struct request_queue *q,struct list_head *list, -struct bio *bio); +struct bio *bio,unsigned int nr_segs);bool blk_mq_queue_stopped (struct request_queue *q);void blk_mq_stop_hw_queue (struct blk_mq_hw_ctx *hctx);void blk_mq_start_hw_queue (struct blk_mq_hw_ctx *hctx);
include/linux/blk_types.h +0 -6
@@ -154,11 +154,6 @@ struct bio {blk_status_t bi_status;u8 bi_partno; -/* Number of segments in this BIO after -* physical address coalescing is performed. -*/ -unsigned int bi_phys_segments; -struct bvec_iter bi_iter;atomic_t __bi_remaining;
@@ -210,7 +205,6 @@ struct bio {*/ enum {BIO_NO_PAGE_REF ,/* don't put release vec pages */ -BIO_SEG_VALID ,/* bi_phys_segments valid */ BIO_CLONED ,/* doesn't own data */ BIO_BOUNCED ,/* bio is a bounce bio */ BIO_USER_MAPPED ,/* contains user pages */
include/linux/blkdev.h +0 -1
@@ -841,7 +841,6 @@ extern blk_status_t blk_insert_cloned_request(struct request_queue *q, struct request *rq);extern int blk_rq_append_bio (struct request *rq,struct bio **bio);extern void blk_queue_split (struct request_queue *,struct bio **); -extern void blk_recount_segments (struct request_queue *,struct bio *);extern int scsi_verify_blk_ioctl (struct block_device *,unsigned int );extern int scsi_cmd_blk_ioctl (struct block_device *,fmode_t ,unsigned int ,void __user *);
include/linux/elevator.h +1 -1
@@ -34,7 +34,7 @@ struct elevator_mq_ops {void (*depth_updated)(struct blk_mq_hw_ctx *);bool (*allow_merge)(struct request_queue *,struct request *,struct bio *); -bool (*bio_merge)(struct blk_mq_hw_ctx *,struct bio *); +bool (*bio_merge)(struct blk_mq_hw_ctx *,struct bio *,unsigned int );int (*request_merge)(struct request_queue *q,struct request **,struct bio *);void (*request_merged)(struct request_queue *,struct request *,enum elv_merge );void (*requests_merged)(struct request_queue *,struct request *,struct request *);
Signature · 1 line
2.20.1
Re: [PATCH 3/6] block: remove the bi_phys_segments field in struct bio
Quoted 25 lines · Christoph Hellwig
We only need the number of segments in the blk-mq submission path. Remove the field from struct bio, and return it from a variant of
blk_queue_splitinstead of that it can passed as an argument to those functions that need the value.This also means we stop recounting segments except for cloning and partial segments.
To keep the number of arguments in this how path down remove pointless struct
request_queuearguments from any of the functions that had it and grew anr_segsargument.Signed-off-by Christoph Hellwig <[email protected]>--- Documentation/block/biodoc.txt | 1 - block/bfq-iosched.c | 5 ++- block/bio.c | 15 +------ block/blk-core.c | 32 +++++++-------- block/blk-map.c | 10 ++++- block/blk-merge.c | 75 ++++++++++++---------------------- block/blk-mq-sched.c | 26 +++++++----- block/blk-mq-sched.h | 10 +++-- block/blk-mq.c | 23 ++++++----- block/blk.h | 23 ++++++----- block/kyber-iosched.c | 5 ++-
Quoted 9 lines
block/mq-deadline.c | 5 ++- drivers/md/raid5.c | 1 - include/linux/bio.h | 1 - include/linux/blk-mq.h | 2 +- include/linux/blk_types.h | 6 --- include/linux/blkdev.h | 1 - include/linux/elevator.h | 2 +- 18 files changed, 106 insertions(+), 137 deletions(-)
In general a very good idea, but:
block/blk-merge.c void blk_queue_split(struct request_queue *q, struct bio **bio) · line 312
Quoted 13 lines
block/blk-merge.c +7
-0@@ -304,6 +301,13 @@ void blk_queue_split(struct request_queue *q, struct bio **bio) *bio = split; } } + +void blk_queue_split (struct request_queue *q,struct bio **bio) +{ +unsigned int nr_segs; + +__blk_queue_split (q, bio, &nr_segs); +}EXPORT_SYMBOL (blk_queue_split);
That looks a bit weird, and I guess some or other compiler might complain here about nr_segs being unused.
Can't we modify __blk_queue_split() to accept a NULL argument here?
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Teamlead Storage & Networking [email protected] +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: Felix Imendörffer, Mary Higgins, Sri Rasiah HRB 21284 (AG Nürnberg)
Christoph Hellwig
maintainerRe: [PATCH 3/6] block: remove the bi_phys_segments field in struct bio
Quoted 11 lines · Hannes Reinecke
block/blk-merge.c void blk_queue_split(struct request_queue *q, struct bio **bio) · line 311
Quoted 8 lines
block/blk-merge.c +6
-0+void blk_queue_split (struct request_queue *q,struct bio **bio) +{ +unsigned int nr_segs; + +__blk_queue_split (q, bio, &nr_segs); +}EXPORT_SYMBOL (blk_queue_split);That looks a bit weird, and I guess some or other compiler might complain here about
nr_segsbeing unused.
Can't we modify__blk_queue_split()to accept a NULL argument here?
We could. But that would bloat the fast path for absolutely no reason. Passing a by reference output argument that is then ignored is a pretty common pattern.
Re: [PATCH 3/6] block: remove the bi_phys_segments field in struct bio
In reply to Hannes ReineckeQuoted 37 lines · Hannes Reinecke
Quoted 25 lines · Christoph Hellwig
We only need the number of segments in the blk-mq submission path. Remove the field from struct bio, and return it from a variant of
blk_queue_splitinstead of that it can passed as an argument to those functions that need the value.This also means we stop recounting segments except for cloning and partial segments.
To keep the number of arguments in this how path down remove pointless struct
request_queuearguments from any of the functions that had it and grew anr_segsargument.Signed-off-by Christoph Hellwig <[email protected]>--- Documentation/block/biodoc.txt | 1 - block/bfq-iosched.c | 5 ++- block/bio.c | 15 +------ block/blk-core.c | 32 +++++++-------- block/blk-map.c | 10 ++++- block/blk-merge.c | 75 ++++++++++++---------------------- block/blk-mq-sched.c | 26 +++++++----- block/blk-mq-sched.h | 10 +++-- block/blk-mq.c | 23 ++++++----- block/blk.h | 23 ++++++----- block/kyber-iosched.c | 5 ++-Quoted 9 lines
block/mq-deadline.c | 5 ++- drivers/md/raid5.c | 1 - include/linux/bio.h | 1 - include/linux/blk-mq.h | 2 +- include/linux/blk_types.h | 6 --- include/linux/blkdev.h | 1 - include/linux/elevator.h | 2 +- 18 files changed, 106 insertions(+), 137 deletions(-)In general a very good idea, but:
Quoted 16 lines
block/blk-merge.c void blk_queue_split(struct request_queue *q, struct bio **bio) · line 312
Quoted 13 lines
block/blk-merge.c +7
-0@@ -304,6 +301,13 @@ void blk_queue_split(struct request_queue *q, struct bio **bio) *bio = split; } } + +void blk_queue_split (struct request_queue *q,struct bio **bio) +{ +unsigned int nr_segs; + +__blk_queue_split (q, bio, &nr_segs); +}EXPORT_SYMBOL (blk_queue_split);That looks a bit weird, and I guess some or other compiler might complain here about
nr_segsbeing unused.
Can't we modify__blk_queue_split()to accept a NULL argument here?
How about __maybe_unused so the compiler doesn't complain?
Quoted 5 lines
Cheers,
Hannes
Re: [PATCH 3/6] block: remove the bi_phys_segments field in struct bio
In reply to Christoph HellwigQuoted 9 lines · Christoph Hellwig
We only need the number of segments in the blk-mq submission path. Remove the field from struct bio, and return it from a variant of
blk_queue_splitinstead of that it can passed as an argument to those functions that need the value.This also means we stop recounting segments except for cloning and partial segments.
To keep the number of arguments in this how path down remove
^^^
hot?I will have a further look at this patch series as soon as I have the time.
Bart.
Christoph Hellwig
maintainerPATCH 4/6 block: simplify blk_recalc_rq_segments
In reply to Christoph HellwigReturn the segement and let the callers assign them, which makes the code a littler more obvious. Also pass the request instead of q plus bio chain, allowing for the use of rq_for_each_bvec.
--- block/blk-core.c | 4 ++-- block/blk-merge.c | 21 ++++++--------------- block/blk.h | 2 +- 3 files changed, 9 insertions(+), 18 deletions(-)
block/blk-core.c +2 -2
@@ -1139,7 +1139,7 @@ static int blk_cloned_rq_check_limits(struct request_queue *q,* Recalculate it to check the request correctly on this queue's * limitation. */ -blk_recalc_rq_segments (rq); + rq->nr_phys_segments =blk_recalc_rq_segments (rq);if (rq->nr_phys_segments >queue_max_segments (q)) {printk (KERN_ERR "%s: over max segments limit. (%hu > %hu)\n" , __func__, rq->nr_phys_segments,queue_max_segments (q));
@@ -1408,7 +1408,7 @@ bool blk_update_request(struct request *req, blk_status_t error, }/* recalculate the number of segments */ -blk_recalc_rq_segments (req); + req->nr_phys_segments =blk_recalc_rq_segments (req); }return true;
block/blk-merge.c +6 -15
@@ -310,17 +310,16 @@ void blk_queue_split(struct request_queue *q, struct bio **bio) }EXPORT_SYMBOL (blk_queue_split); -static unsigned int __blk_recalc_rq_segments (struct request_queue *q, -struct bio *bio) +unsigned int blk_recalc_rq_segments (struct request *rq) {unsigned int nr_phys_segs =0 ; -struct bvec_iter iter; +struct req_iterator iter;struct bio_vec bv; -if (!bio) +if (!rq->bio)return 0 ; -switch (bio_op (bio)) { +switch (bio_op (rq->bio)) {case REQ_OP_DISCARD :case REQ_OP_SECURE_ERASE :case REQ_OP_WRITE_ZEROES :
@@ -329,19 +328,11 @@ static unsigned int __blk_recalc_rq_segments(struct request_queue *q,return 1 ; } -for_each_bio (bio) { -bio_for_each_bvec (bv, bio, iter) -bvec_split_segs (q, &bv, &nr_phys_segs,NULL ,UINT_MAX ); - } - +rq_for_each_bvec (bv, rq, iter) +bvec_split_segs (rq->q, &bv, &nr_phys_segs,NULL ,UINT_MAX );return nr_phys_segs; } -void blk_recalc_rq_segments (struct request *rq) -{ - rq->nr_phys_segments =__blk_recalc_rq_segments (rq->q, rq->bio); -} -static inline struct scatterlist *blk_next_sg (struct scatterlist **sg,struct scatterlist *sglist) {
block/blk.h +1 -1
@@ -203,7 +203,7 @@ struct request *attempt_back_merge(struct request_queue *q, struct request *rq);struct request *attempt_front_merge (struct request_queue *q,struct request *rq);int blk_attempt_req_merge (struct request_queue *q,struct request *rq,struct request *next); -void blk_recalc_rq_segments (struct request *rq); +unsigned int blk_recalc_rq_segments (struct request *rq);void blk_rq_set_mixed_merge (struct request *rq);bool blk_rq_merge_ok (struct request *rq,struct bio *bio);enum elv_merge blk_try_merge (struct request *rq,struct bio *bio);
Signature · 1 line
2.20.1
Re: [PATCH 4/6] block: simplify blk_recalc_rq_segments
Quoted 11 lines · Christoph Hellwig
Return the segement and let the callers assign them, which makes the code a littler more obvious. Also pass the request instead of q plus bio chain, allowing for the use of
rq_for_each_bvec.Signed-off-by Christoph Hellwig <[email protected]>--- block/blk-core.c | 4 ++-- block/blk-merge.c | 21 ++++++--------------- block/blk.h | 2 +- 3 files changed, 9 insertions(+), 18 deletions(-)Reviewed-by Hannes Reinecke <[email protected]>
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Teamlead Storage & Networking [email protected] +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: Felix Imendörffer, Mary Higgins, Sri Rasiah HRB 21284 (AG Nürnberg)
Christoph Hellwig
maintainerPATCH 5/6 block: untangle the end of blk_bio_segment_split
In reply to Christoph HellwigNow that we don't need to assign the front/back segment sizes, we can duplicating the segs assignment for the split vs no-split case and remove a whole chunk of boilerplate code.
--- block/blk-merge.c | 14 +++----------- 1 file changed, 3 insertions(+), 11 deletions(-)
block/blk-merge.c +3 -11
@@ -202,8 +202,6 @@ static struct bio *blk_bio_segment_split(struct request_queue *q,struct bio_vec bv, bvprv, *bvprvp =NULL ;struct bvec_iter iter;unsigned nsegs =0 , sectors =0 ; -bool do_split = true; -struct bio *new =NULL ;const unsigned max_sectors =get_max_io_size (q, bio);const unsigned max_segs =queue_max_segments (q);
@@ -245,17 +243,11 @@ static struct bio *blk_bio_segment_split(struct request_queue *q, } } - do_split = false; + *segs = nsegs; +return NULL ; split: *segs = nsegs; - -if (do_split) { - new =bio_split (bio, sectors,GFP_NOIO , bs); -if (new) - bio = new; - } - -return do_split ? new :NULL ; +return bio_split (bio, sectors,GFP_NOIO , bs); } void __blk_queue_split(struct request_queue *q, struct bio **bio,
Signature · 1 line
2.20.1
Re: [PATCH 5/6] block: untangle the end of blk_bio_segment_split
Quoted 9 lines · Christoph Hellwig
Now that we don't need to assign the front/back segment sizes, we can duplicating the segs assignment for the split vs no-split case and remove a whole chunk of boilerplate code.
Signed-off-by Christoph Hellwig <[email protected]>--- block/blk-merge.c | 14 +++----------- 1 file changed, 3 insertions(+), 11 deletions(-)
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Teamlead Storage & Networking [email protected] +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: Felix Imendörffer, Mary Higgins, Sri Rasiah HRB 21284 (AG Nürnberg)
Christoph Hellwig
maintainerPATCH 6/6 block: mark blk_rq_bio_prep as inline
In reply to Christoph HellwigThis function just has a few trivial assignments, has two callers with one of them being in the fastpath.
--- block/blk-core.c | 11 ----------- block/blk.h | 13 ++++++++++++- 2 files changed, 12 insertions(+), 12 deletions(-)
block/blk-core.c +0 -11
@@ -1415,17 +1415,6 @@ bool blk_update_request(struct request *req, blk_status_t error, }EXPORT_SYMBOL_GPL (blk_update_request); -void blk_rq_bio_prep (struct request *rq,struct bio *bio,unsigned int nr_segs) -{ - rq->nr_phys_segments = nr_segs; - rq->__data_len = bio->bi_iter.bi_size; - rq->bio = rq->biotail = bio; - rq->ioprio =bio_prio (bio); - -if (bio->bi_disk) - rq->rq_disk = bio->bi_disk; -} -#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE /** * rq_flush_dcache_pages - Helper function to flush all pages in a request
block/blk.h +12 -1
@@ -50,7 +50,6 @@ struct blk_flush_queue *blk_alloc_flush_queue(struct request_queue *q, int node, int cmd_size, gfp_t flags);void blk_free_flush_queue (struct blk_flush_queue *q); -void blk_rq_bio_prep (struct request *rq,struct bio *bio,unsigned int nr_segs);void blk_freeze_queue (struct request_queue *q);static inline void blk_queue_enter_live (struct request_queue *q)
@@ -99,6 +98,18 @@ static inline bool bvec_gap_to_prev(struct request_queue *q,return __bvec_gap_to_prev (q, bprv, offset); } +static inline void blk_rq_bio_prep (struct request *rq,struct bio *bio, +unsigned int nr_segs) +{ + rq->nr_phys_segments = nr_segs; + rq->__data_len = bio->bi_iter.bi_size; + rq->bio = rq->biotail = bio; + rq->ioprio =bio_prio (bio); + +if (bio->bi_disk) + rq->rq_disk = bio->bi_disk; +} +#ifdef CONFIG_BLK_DEV_INTEGRITY void blk_flush_integrity (void );bool __bio_integrity_endio (struct bio *);
Signature · 1 line
2.20.1
Re: [PATCH 6/6] block: mark blk_rq_bio_prep as inline
Quoted 10 lines · Christoph Hellwig
This function just has a few trivial assignments, has two callers with one of them being in the fastpath.
Signed-off-by Christoph Hellwig <[email protected]>Reviewed-by Chaitanya Kulkarni <[email protected]>--- block/blk-core.c | 11 ----------- block/blk.h | 13 ++++++++++++- 2 files changed, 12 insertions(+), 12 deletions(-)
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Teamlead Storage & Networking [email protected] +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: Felix Imendörffer, Mary Higgins, Sri Rasiah HRB 21284 (AG Nürnberg)
Christoph Hellwig
maintainerping?
Quoted 4 lines · Christoph Hellwig
Hi Jens,
this series removes the
bi_phys_segmentsmember in struct bio and cleans up various areas around it.
---end quoted text---
Jens Axboe
AppliedQuoted 4 lines · Christoph Hellwig
Hi Jens,
this series removes the
bi_phys_segmentsmember in struct bio and cleans up various areas around it.
Applied, thanks.
Signature · 1 line
Jens Axboe
Jens Axboe
maintainerQuoted 6 lines · Jens Axboe
Quoted 4 lines · Christoph Hellwig
Hi Jens,
this series removes the
bi_phys_segmentsmember in struct bio and cleans up various areas around it.>Applied, thanks.
Now that I'm back and can fully go through testing, I run into problems with this series. On one system, booting up makes it crash. I've managed to get it to boot, only for the first sync to make it crash. So seems likely related to a flush.
No console on that system, but that's OK since the trace itself isn't that interesting. We end up crashing in nvme_queue_rq(), here:
(gdb) l *nvme_queue_rq+0x2520xffffffff8148b292 is in nvme_queue_rq (./include/linux/blkdev.h:962).
957 */
958 static inline struct bio_vec req_bvec(struct request *rq)
959 {
960 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
961 return rq->special_vec;
962 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
963 }
with a NULL pointer dereference at 0x8.
Taking a look at the series, this is the offending commit:
commit 14ccb66b3f585b2bc21e7256c96090abed5a512c
Author: Christoph Hellwig <[email protected]> Date: Thu Jun 6 12:29:01 2019 +0200
block: remove the bi_phys_segments field in struct bio
with that as the HEAD, crash. Go back one commit, works fine.
Signature · 1 line
Jens Axboe
Christoph Hellwig
maintainerQuoted 3 lines · Jens Axboe
No console on that system, but that's OK since the trace itself isn't that interesting. We end up crashing in
nvme_queue_rq(), here:
Weird. Is that a nvme device with our without a volativle write cache? Can you send me your .config?
A message that is not in the archive (sent off-list or not yet received): [email protected]
Christoph Hellwig
maintainerQuoted 1 line · Jens Axboe
.config attached.
I couldn't get that to boot in my test systems even with mainline, but that seems to be do to systemd waiting for some crap not supported in the config.
But with my usual test config I've just completed a test run with KASAN enabled on a VWC=1 driver with no issues, so this keeps puzzling me.
Jens Axboe
maintainerQuoted 10 lines · Christoph Hellwig
Quoted 1 line · Jens Axboe
.config attached.
I couldn't get that to boot in my test systems even with mainline, but that seems to be do to systemd waiting for some crap not supported in the config.
But with my usual test config I've just completed a test run with KASAN enabled on a VWC=1 driver with no issues, so this keeps puzzling me.
Let me know what you want me to try. I can't reproduce it in qemu, but it's 100% on my laptop. My qemu drives also have VWC=1, so it's not (just) that.
Signature · 1 line
Jens Axboe
Christoph Hellwig
maintainerQuoted 11 lines · Jens Axboe
Quoted 7 lines
I couldn't get that to boot in my test systems even with mainline, but that seems to be do to systemd waiting for some crap not supported in the config.
But with my usual test config I've just completed a test run with KASAN enabled on a VWC=1 driver with no issues, so this keeps puzzling me.
Let me know what you want me to try. I can't reproduce it in qemu, but it's 100% on my laptop. My qemu drives also have VWC=1, so it's not (just) that.
I seriously have no idea unfortunately. It works fine for me both on qemu and on a real WD SN720 drive on my laptop. Just for curiosity you could try to pad the bio structure and see if bloating it to the old size makes any difference.
The other things that comes to mind is that when Johannes removed BIO_SEG_VALID there also were some unexplainable side effects, I'll look into seeing if there was any similarity.
Jens Axboe
maintainerQuoted 9 lines · Christoph Hellwig
Quoted 8 lines · Jens Axboe
Quoted 7 lines
I couldn't get that to boot in my test systems even with mainline, but that seems to be do to systemd waiting for some crap not supported in the config.
But with my usual test config I've just completed a test run with KASAN enabled on a VWC=1 driver with no issues, so this keeps puzzling me.
Quoted 8 lines
Quoted 3 lines
Let me know what you want me to try. I can't reproduce it in qemu, but it's 100% on my laptop. My qemu drives also have VWC=1, so it's not (just) that.
I seriously have no idea unfortunately. It works fine for me both on qemu and on a real WD SN720 drive on my laptop. Just for curiosity you could try to pad the bio structure and see if bloating it to the old size makes any difference.
No change with the padding, put it in the same place. Still insta crash before I get to login prompt, or right after I login and run sync.
Quoted 3 lines
The other things that comes to mind is that when Johannes removed
BIO_SEG_VALIDthere also were some unexplainable side effects, I'll look into seeing if there was any similarity.
Do you have a link?
I'll try and poke a bit here.
Signature · 1 line
Jens Axboe
Jens Axboe
maintainerQuoted 22 lines · Jens Axboe
Quoted 9 lines · Christoph Hellwig
Quoted 8 lines · Jens Axboe
Quoted 7 lines
I couldn't get that to boot in my test systems even with mainline, but that seems to be do to systemd waiting for some crap not supported in the config.
But with my usual test config I've just completed a test run with KASAN enabled on a VWC=1 driver with no issues, so this keeps puzzling me.
Quoted 8 lines
Quoted 3 lines
Let me know what you want me to try. I can't reproduce it in qemu, but it's 100% on my laptop. My qemu drives also have VWC=1, so it's not (just) that.
I seriously have no idea unfortunately. It works fine for me both on qemu and on a real WD SN720 drive on my laptop. Just for curiosity you could try to pad the bio structure and see if bloating it to the old size makes any difference.
No change with the padding, put it in the same place. Still insta crash before I get to login prompt, or right after I login and run sync.
Quoted 7 lines
Quoted 3 lines
The other things that comes to mind is that when Johannes removed
BIO_SEG_VALIDthere also were some unexplainable side effects, I'll look into seeing if there was any similarity.Do you have a link?
I'll try and poke a bit here.
The issue is a discard request, 4kb, with a bio but no bi_io_vec. The sync is a red herring, apart from the fact that it triggers the ext4 discards.
I'm guessing, when you tried to reproduce, that you didn't have discard enabled?
Signature · 1 line
Jens Axboe
Christoph Hellwig
maintainerQuoted 6 lines · Jens Axboe
The issue is a discard request, 4kb, with a bio but no
bi_io_vec. The sync is a red herring, apart from the fact that it triggers the ext4 discards.I'm guessing, when you tried to reproduce, that you didn't have discard enabled?
I didn't run with online discard, but I tried FITRIM (with XFS).
I'll try looking into ext4. I remember we had some weird one offs in that area for the different ways the physical segments are calculated, so this makes sense. Off to dinner now, but I'll look into it tomorrow.
Christoph Hellwig
maintainerQuoted 5 lines · Christoph Hellwig
I didn't run with online discard, but I tried FITRIM (with XFS).
I'll try looking into ext4. I remember we had some weird one offs in that area for the different ways the physical segments are calculated, so this makes sense. Off to dinner now, but I'll look into it tomorrow.
ext4 reproduced the bug instantly, but it was Write Zeroes, not Discard. Which explains the whole thing as ext4 is the only major user of Write Zeroes. A oatch will be on its way shortly.