From 7988613b0e5b2638caf6cd493cc78e9595eba19c Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Sat, 23 Nov 2013 17:19:00 -0800 Subject: block: Convert bio_for_each_segment() to bvec_iter More prep work for immutable biovecs - with immutable bvecs drivers won't be able to use the biovec directly, they'll need to use helpers that take into account bio->bi_iter.bi_bvec_done. This updates callers for the new usage without changing the implementation yet. Signed-off-by: Kent Overstreet Cc: Jens Axboe Cc: Geert Uytterhoeven Cc: Benjamin Herrenschmidt Cc: Paul Mackerras Cc: "Ed L. Cashin" Cc: Nick Piggin Cc: Lars Ellenberg Cc: Jiri Kosina Cc: Paul Clements Cc: Jim Paris Cc: Geoff Levand Cc: Yehuda Sadeh Cc: Sage Weil Cc: Alex Elder Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris Cc: Philip Kelleher Cc: Konrad Rzeszutek Wilk Cc: Jeremy Fitzhardinge Cc: Neil Brown Cc: Martin Schwidefsky Cc: Heiko Carstens Cc: linux390@de.ibm.com Cc: Nagalakshmi Nandigama Cc: Sreekanth Reddy Cc: support@lsi.com Cc: "James E.J. Bottomley" Cc: Greg Kroah-Hartman Cc: Alexander Viro Cc: Steven Whitehouse Cc: Herton Ronaldo Krzesinski Cc: Tejun Heo Cc: Andrew Morton Cc: Guo Chao Cc: Asai Thambi S P Cc: Selvan Mani Cc: Sam Bradshaw Cc: Matthew Wilcox Cc: Keith Busch Cc: Stephen Hemminger Cc: Quoc-Son Anh Cc: Sebastian Ott Cc: Nitin Gupta Cc: Minchan Kim Cc: Jerome Marchand Cc: Seth Jennings Cc: "Martin K. Petersen" Cc: Mike Snitzer Cc: Vivek Goyal Cc: "Darrick J. Wong" Cc: Chris Metcalf Cc: Jan Kara Cc: linux-m68k@lists.linux-m68k.org Cc: linuxppc-dev@lists.ozlabs.org Cc: drbd-user@lists.linbit.com Cc: nbd-general@lists.sourceforge.net Cc: cbe-oss-dev@lists.ozlabs.org Cc: xen-devel@lists.xensource.com Cc: virtualization@lists.linux-foundation.org Cc: linux-raid@vger.kernel.org Cc: linux-s390@vger.kernel.org Cc: DL-MPTFusionLinux@lsi.com Cc: linux-scsi@vger.kernel.org Cc: devel@driverdev.osuosl.org Cc: linux-fsdevel@vger.kernel.org Cc: cluster-devel@redhat.com Cc: linux-mm@kvack.org Acked-by: Geoff Levand --- include/linux/blkdev.h | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'include/linux/blkdev.h') diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 1b135d49b279..337b92a54658 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -735,7 +735,7 @@ struct rq_map_data { }; struct req_iterator { - int i; + struct bvec_iter iter; struct bio *bio; }; @@ -748,10 +748,11 @@ struct req_iterator { #define rq_for_each_segment(bvl, _rq, _iter) \ __rq_for_each_bio(_iter.bio, _rq) \ - bio_for_each_segment(bvl, _iter.bio, _iter.i) + bio_for_each_segment(bvl, _iter.bio, _iter.iter) #define rq_iter_last(rq, _iter) \ - (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1) + (_iter.bio->bi_next == NULL && \ + bio_iter_last(_iter.bio, _iter.iter)) #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform" -- cgit v1.2.3-59-g8ed1b From 4550dd6c6b062fc5e5b647296d55da22616123c3 Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Wed, 7 Aug 2013 14:26:21 -0700 Subject: block: Immutable bio vecs This adds a mechanism by which we can advance a bio by an arbitrary number of bytes without modifying the biovec: bio->bi_iter.bi_bvec_done indicates the number of bytes completed in the current bvec. Various driver code still needs to be updated to not refer to the bvec directly before we can use this for interesting things, like efficient bio splitting. Signed-off-by: Kent Overstreet Cc: Jens Axboe Cc: Lars Ellenberg Cc: Paul Clements Cc: drbd-user@lists.linbit.com Cc: nbd-general@lists.sourceforge.net --- Documentation/block/biovecs.txt | 111 ++++++++++++++++++++++++++++++++++++++++ drivers/block/drbd/drbd_main.c | 4 +- drivers/block/nbd.c | 2 +- fs/bio.c | 27 +--------- include/linux/bio.h | 81 ++++++++++++++++++++++++++--- include/linux/blk_types.h | 3 ++ include/linux/blkdev.h | 4 +- 7 files changed, 194 insertions(+), 38 deletions(-) create mode 100644 Documentation/block/biovecs.txt (limited to 'include/linux/blkdev.h') diff --git a/Documentation/block/biovecs.txt b/Documentation/block/biovecs.txt new file mode 100644 index 000000000000..74a32ad52f53 --- /dev/null +++ b/Documentation/block/biovecs.txt @@ -0,0 +1,111 @@ + +Immutable biovecs and biovec iterators: +======================================= + +Kent Overstreet + +As of 3.13, biovecs should never be modified after a bio has been submitted. +Instead, we have a new struct bvec_iter which represents a range of a biovec - +the iterator will be modified as the bio is completed, not the biovec. + +More specifically, old code that needed to partially complete a bio would +update bi_sector and bi_size, and advance bi_idx to the next biovec. If it +ended up partway through a biovec, it would increment bv_offset and decrement +bv_len by the number of bytes completed in that biovec. + +In the new scheme of things, everything that must be mutated in order to +partially complete a bio is segregated into struct bvec_iter: bi_sector, +bi_size and bi_idx have been moved there; and instead of modifying bv_offset +and bv_len, struct bvec_iter has bi_bvec_done, which represents the number of +bytes completed in the current bvec. + +There are a bunch of new helper macros for hiding the gory details - in +particular, presenting the illusion of partially completed biovecs so that +normal code doesn't have to deal with bi_bvec_done. + + * Driver code should no longer refer to biovecs directly; we now have + bio_iovec() and bio_iovec_iter() macros that return literal struct biovecs, + constructed from the raw biovecs but taking into account bi_bvec_done and + bi_size. + + bio_for_each_segment() has been updated to take a bvec_iter argument + instead of an integer (that corresponded to bi_idx); for a lot of code the + conversion just required changing the types of the arguments to + bio_for_each_segment(). + + * Advancing a bvec_iter is done with bio_advance_iter(); bio_advance() is a + wrapper around bio_advance_iter() that operates on bio->bi_iter, and also + advances the bio integrity's iter if present. + + There is a lower level advance function - bvec_iter_advance() - which takes + a pointer to a biovec, not a bio; this is used by the bio integrity code. + +What's all this get us? +======================= + +Having a real iterator, and making biovecs immutable, has a number of +advantages: + + * Before, iterating over bios was very awkward when you weren't processing + exactly one bvec at a time - for example, bio_copy_data() in fs/bio.c, + which copies the contents of one bio into another. Because the biovecs + wouldn't necessarily be the same size, the old code was tricky convoluted - + it had to walk two different bios at the same time, keeping both bi_idx and + and offset into the current biovec for each. + + The new code is much more straightforward - have a look. This sort of + pattern comes up in a lot of places; a lot of drivers were essentially open + coding bvec iterators before, and having common implementation considerably + simplifies a lot of code. + + * Before, any code that might need to use the biovec after the bio had been + completed (perhaps to copy the data somewhere else, or perhaps to resubmit + it somewhere else if there was an error) had to save the entire bvec array + - again, this was being done in a fair number of places. + + * Biovecs can be shared between multiple bios - a bvec iter can represent an + arbitrary range of an existing biovec, both starting and ending midway + through biovecs. This is what enables efficient splitting of arbitrary + bios. Note that this means we _only_ use bi_size to determine when we've + reached the end of a bio, not bi_vcnt - and the bio_iovec() macro takes + bi_size into account when constructing biovecs. + + * Splitting bios is now much simpler. The old bio_split() didn't even work on + bios with more than a single bvec! Now, we can efficiently split arbitrary + size bios - because the new bio can share the old bio's biovec. + + Care must be taken to ensure the biovec isn't freed while the split bio is + still using it, in case the original bio completes first, though. Using + bio_chain() when splitting bios helps with this. + + * Submitting partially completed bios is now perfectly fine - this comes up + occasionally in stacking block drivers and various code (e.g. md and + bcache) had some ugly workarounds for this. + + It used to be the case that submitting a partially completed bio would work + fine to _most_ devices, but since accessing the raw bvec array was the + norm, not all drivers would respect bi_idx and those would break. Now, + since all drivers _must_ go through the bvec iterator - and have been + audited to make sure they are - submitting partially completed bios is + perfectly fine. + +Other implications: +=================== + + * Almost all usage of bi_idx is now incorrect and has been removed; instead, + where previously you would have used bi_idx you'd now use a bvec_iter, + probably passing it to one of the helper macros. + + I.e. instead of using bio_iovec_idx() (or bio->bi_iovec[bio->bi_idx]), you + now use bio_iter_iovec(), which takes a bvec_iter and returns a + literal struct bio_vec - constructed on the fly from the raw biovec but + taking into account bi_bvec_done (and bi_size). + + * bi_vcnt can't be trusted or relied upon by driver code - i.e. anything that + doesn't actually own the bio. The reason is twofold: firstly, it's not + actually needed for iterating over the bio anymore - we only use bi_size. + Secondly, when cloning a bio and reusing (a portion of) the original bio's + biovec, in order to calculate bi_vcnt for the new bio we'd have to iterate + over all the biovecs in the new bio - which is silly as it's not needed. + + So, don't use bi_vcnt anymore. diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index f4e5440aba05..929468e1512a 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -1546,7 +1546,7 @@ static int _drbd_send_bio(struct drbd_conf *mdev, struct bio *bio) err = _drbd_no_send_page(mdev, bvec.bv_page, bvec.bv_offset, bvec.bv_len, - bio_iter_last(bio, iter) + bio_iter_last(bvec, iter) ? 0 : MSG_MORE); if (err) return err; @@ -1565,7 +1565,7 @@ static int _drbd_send_zc_bio(struct drbd_conf *mdev, struct bio *bio) err = _drbd_send_page(mdev, bvec.bv_page, bvec.bv_offset, bvec.bv_len, - bio_iter_last(bio, iter) ? 0 : MSG_MORE); + bio_iter_last(bvec, iter) ? 0 : MSG_MORE); if (err) return err; } diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c index aa362f493216..55298db36b2d 100644 --- a/drivers/block/nbd.c +++ b/drivers/block/nbd.c @@ -278,7 +278,7 @@ static int nbd_send_req(struct nbd_device *nbd, struct request *req) */ rq_for_each_segment(bvec, req, iter) { flags = 0; - if (!rq_iter_last(req, iter)) + if (!rq_iter_last(bvec, iter)) flags = MSG_MORE; dprintk(DBG_TX, "%s: request %p: sending %d bytes data\n", nbd->disk->disk_name, req, bvec.bv_len); diff --git a/fs/bio.c b/fs/bio.c index 8b7f14a95503..07b4b7afa695 100644 --- a/fs/bio.c +++ b/fs/bio.c @@ -532,13 +532,11 @@ void __bio_clone(struct bio *bio, struct bio *bio_src) * most users will be overriding ->bi_bdev with a new target, * so we don't set nor calculate new physical/hw segment counts here */ - bio->bi_iter.bi_sector = bio_src->bi_iter.bi_sector; bio->bi_bdev = bio_src->bi_bdev; bio->bi_flags |= 1 << BIO_CLONED; bio->bi_rw = bio_src->bi_rw; bio->bi_vcnt = bio_src->bi_vcnt; - bio->bi_iter.bi_size = bio_src->bi_iter.bi_size; - bio->bi_iter.bi_idx = bio_src->bi_iter.bi_idx; + bio->bi_iter = bio_src->bi_iter; } EXPORT_SYMBOL(__bio_clone); @@ -808,28 +806,7 @@ void bio_advance(struct bio *bio, unsigned bytes) if (bio_integrity(bio)) bio_integrity_advance(bio, bytes); - bio->bi_iter.bi_sector += bytes >> 9; - bio->bi_iter.bi_size -= bytes; - - if (bio->bi_rw & BIO_NO_ADVANCE_ITER_MASK) - return; - - while (bytes) { - if (unlikely(bio->bi_iter.bi_idx >= bio->bi_vcnt)) { - WARN_ONCE(1, "bio idx %d >= vcnt %d\n", - bio->bi_iter.bi_idx, bio->bi_vcnt); - break; - } - - if (bytes >= bio_iovec(bio).bv_len) { - bytes -= bio_iovec(bio).bv_len; - bio->bi_iter.bi_idx++; - } else { - bio_iovec(bio).bv_len -= bytes; - bio_iovec(bio).bv_offset += bytes; - bytes = 0; - } - } + bio_advance_iter(bio, &bio->bi_iter, bytes); } EXPORT_SYMBOL(bio_advance); diff --git a/include/linux/bio.h b/include/linux/bio.h index c16adb5f69f8..04e592e74c92 100644 --- a/include/linux/bio.h +++ b/include/linux/bio.h @@ -64,11 +64,38 @@ #define bio_iovec_idx(bio, idx) (&((bio)->bi_io_vec[(idx)])) #define __bio_iovec(bio) bio_iovec_idx((bio), (bio)->bi_iter.bi_idx) -#define bio_iter_iovec(bio, iter) ((bio)->bi_io_vec[(iter).bi_idx]) +#define __bvec_iter_bvec(bvec, iter) (&(bvec)[(iter).bi_idx]) -#define bio_page(bio) (bio_iovec((bio)).bv_page) -#define bio_offset(bio) (bio_iovec((bio)).bv_offset) -#define bio_iovec(bio) (*__bio_iovec(bio)) +#define bvec_iter_page(bvec, iter) \ + (__bvec_iter_bvec((bvec), (iter))->bv_page) + +#define bvec_iter_len(bvec, iter) \ + min((iter).bi_size, \ + __bvec_iter_bvec((bvec), (iter))->bv_len - (iter).bi_bvec_done) + +#define bvec_iter_offset(bvec, iter) \ + (__bvec_iter_bvec((bvec), (iter))->bv_offset + (iter).bi_bvec_done) + +#define bvec_iter_bvec(bvec, iter) \ +((struct bio_vec) { \ + .bv_page = bvec_iter_page((bvec), (iter)), \ + .bv_len = bvec_iter_len((bvec), (iter)), \ + .bv_offset = bvec_iter_offset((bvec), (iter)), \ +}) + +#define bio_iter_iovec(bio, iter) \ + bvec_iter_bvec((bio)->bi_io_vec, (iter)) + +#define bio_iter_page(bio, iter) \ + bvec_iter_page((bio)->bi_io_vec, (iter)) +#define bio_iter_len(bio, iter) \ + bvec_iter_len((bio)->bi_io_vec, (iter)) +#define bio_iter_offset(bio, iter) \ + bvec_iter_offset((bio)->bi_io_vec, (iter)) + +#define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter) +#define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter) +#define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter) #define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_iter.bi_idx) #define bio_sectors(bio) ((bio)->bi_iter.bi_size >> 9) @@ -145,16 +172,54 @@ static inline void *bio_data(struct bio *bio) bvl = bio_iovec_idx((bio), (i)), i < (bio)->bi_vcnt; \ i++) +static inline void bvec_iter_advance(struct bio_vec *bv, struct bvec_iter *iter, + unsigned bytes) +{ + WARN_ONCE(bytes > iter->bi_size, + "Attempted to advance past end of bvec iter\n"); + + while (bytes) { + unsigned len = min(bytes, bvec_iter_len(bv, *iter)); + + bytes -= len; + iter->bi_size -= len; + iter->bi_bvec_done += len; + + if (iter->bi_bvec_done == __bvec_iter_bvec(bv, *iter)->bv_len) { + iter->bi_bvec_done = 0; + iter->bi_idx++; + } + } +} + +#define for_each_bvec(bvl, bio_vec, iter, start) \ + for ((iter) = start; \ + (bvl) = bvec_iter_bvec((bio_vec), (iter)), \ + (iter).bi_size; \ + bvec_iter_advance((bio_vec), &(iter), (bvl).bv_len)) + + +static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter, + unsigned bytes) +{ + iter->bi_sector += bytes >> 9; + + if (bio->bi_rw & BIO_NO_ADVANCE_ITER_MASK) + iter->bi_size -= bytes; + else + bvec_iter_advance(bio->bi_io_vec, iter, bytes); +} + #define __bio_for_each_segment(bvl, bio, iter, start) \ for (iter = (start); \ - bvl = bio_iter_iovec((bio), (iter)), \ - (iter).bi_idx < (bio)->bi_vcnt; \ - (iter).bi_idx++) + (iter).bi_size && \ + ((bvl = bio_iter_iovec((bio), (iter))), 1); \ + bio_advance_iter((bio), &(iter), (bvl).bv_len)) #define bio_for_each_segment(bvl, bio, iter) \ __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter) -#define bio_iter_last(bio, iter) ((iter).bi_idx == (bio)->bi_vcnt - 1) +#define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len) /* * get a reference to a bio, so it won't disappear. the intended use is diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h index 29b5b84d8a29..d369f8f6af79 100644 --- a/include/linux/blk_types.h +++ b/include/linux/blk_types.h @@ -34,6 +34,9 @@ struct bvec_iter { unsigned int bi_size; /* residual I/O count */ unsigned int bi_idx; /* current index into bvl_vec */ + + unsigned int bi_bvec_done; /* number of bytes completed in + current bvec */ }; /* diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 337b92a54658..02cb6f0ea71d 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -750,9 +750,9 @@ struct req_iterator { __rq_for_each_bio(_iter.bio, _rq) \ bio_for_each_segment(bvl, _iter.bio, _iter.iter) -#define rq_iter_last(rq, _iter) \ +#define rq_iter_last(bvec, _iter) \ (_iter.bio->bi_next == NULL && \ - bio_iter_last(_iter.bio, _iter.iter)) + bio_iter_last(bvec, _iter.iter)) #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform" -- cgit v1.2.3-59-g8ed1b From c78afc6261b09f74abff8c0719b80692a4959768 Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Thu, 11 Jul 2013 22:39:53 -0700 Subject: bcache/md: Use raid stripe size Now that we've got code for raid5/6 stripe awareness, bcache just needs to know about the stripes and when writing partial stripes is expensive - we probably don't want to enable this optimization for raid1 or 10, even though they have stripes. So add a flag to queue_limits. Signed-off-by: Kent Overstreet --- block/blk-settings.c | 4 ++++ drivers/md/bcache/super.c | 6 ++++++ drivers/md/raid5.c | 1 + include/linux/blkdev.h | 1 + 4 files changed, 12 insertions(+) (limited to 'include/linux/blkdev.h') diff --git a/block/blk-settings.c b/block/blk-settings.c index 05e826793e4e..5d21239bc859 100644 --- a/block/blk-settings.c +++ b/block/blk-settings.c @@ -592,6 +592,10 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b, ret = -1; } + t->raid_partial_stripes_expensive = + max(t->raid_partial_stripes_expensive, + b->raid_partial_stripes_expensive); + /* Find lowest common alignment_offset */ t->alignment_offset = lcm(t->alignment_offset, alignment) & (max(t->physical_block_size, t->io_min) - 1); diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c index 63ebef78df4a..e363efcf2b76 100644 --- a/drivers/md/bcache/super.c +++ b/drivers/md/bcache/super.c @@ -1134,6 +1134,12 @@ static int cached_dev_init(struct cached_dev *dc, unsigned block_size) hlist_add_head(&io->hash, dc->io_hash + RECENT_IO); } + dc->disk.stripe_size = q->limits.io_opt >> 9; + + if (dc->disk.stripe_size) + dc->partial_stripes_expensive = + q->limits.raid_partial_stripes_expensive; + ret = bcache_device_init(&dc->disk, block_size, dc->bdev->bd_part->nr_sects - dc->sb.data_offset); if (ret) diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index eea63372e4d3..1cfb22c025b6 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -6101,6 +6101,7 @@ static int run(struct mddev *mddev) blk_queue_io_min(mddev->queue, chunk_size); blk_queue_io_opt(mddev->queue, chunk_size * (conf->raid_disks - conf->max_degraded)); + mddev->queue->limits.raid_partial_stripes_expensive = 1; /* * We can only discard a whole stripe. It doesn't make sense to * discard data disk but write parity disk diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 02cb6f0ea71d..0375654adb28 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -291,6 +291,7 @@ struct queue_limits { unsigned char discard_misaligned; unsigned char cluster; unsigned char discard_zeroes_data; + unsigned char raid_partial_stripes_expensive; }; struct request_queue { -- cgit v1.2.3-59-g8ed1b From 6897fc22ea01b562b55c6168592bcbd3ee62b006 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 30 Jan 2014 15:45:47 -0800 Subject: kernel: use lockless list for smp_call_function_single Make smp_call_function_single and friends more efficient by using a lockless list. Signed-off-by: Christoph Hellwig Reviewed-by: Jan Kara Cc: Jens Axboe Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/blkdev.h | 5 +---- include/linux/smp.h | 6 +++++- kernel/smp.c | 51 +++++++++++++------------------------------------- 3 files changed, 19 insertions(+), 43 deletions(-) (limited to 'include/linux/blkdev.h') diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 0375654adb28..8678c4322b44 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -95,10 +95,7 @@ enum rq_cmd_type_bits { * as well! */ struct request { - union { - struct list_head queuelist; - struct llist_node ll_list; - }; + struct list_head queuelist; union { struct call_single_data csd; struct work_struct mq_flush_data; diff --git a/include/linux/smp.h b/include/linux/smp.h index 5da22ee42e16..3834f43f9993 100644 --- a/include/linux/smp.h +++ b/include/linux/smp.h @@ -11,12 +11,16 @@ #include #include #include +#include extern void cpu_idle(void); typedef void (*smp_call_func_t)(void *info); struct call_single_data { - struct list_head list; + union { + struct list_head list; + struct llist_node llist; + }; smp_call_func_t func; void *info; u16 flags; diff --git a/kernel/smp.c b/kernel/smp.c index bd9f94028838..4ad913e7c253 100644 --- a/kernel/smp.c +++ b/kernel/smp.c @@ -28,12 +28,7 @@ struct call_function_data { static DEFINE_PER_CPU_SHARED_ALIGNED(struct call_function_data, cfd_data); -struct call_single_queue { - struct list_head list; - raw_spinlock_t lock; -}; - -static DEFINE_PER_CPU_SHARED_ALIGNED(struct call_single_queue, call_single_queue); +static DEFINE_PER_CPU_SHARED_ALIGNED(struct llist_head, call_single_queue); static int hotplug_cfd(struct notifier_block *nfb, unsigned long action, void *hcpu) @@ -85,12 +80,8 @@ void __init call_function_init(void) void *cpu = (void *)(long)smp_processor_id(); int i; - for_each_possible_cpu(i) { - struct call_single_queue *q = &per_cpu(call_single_queue, i); - - raw_spin_lock_init(&q->lock); - INIT_LIST_HEAD(&q->list); - } + for_each_possible_cpu(i) + init_llist_head(&per_cpu(call_single_queue, i)); hotplug_cfd(&hotplug_cfd_notifier, CPU_UP_PREPARE, cpu); register_cpu_notifier(&hotplug_cfd_notifier); @@ -141,18 +132,9 @@ static void csd_unlock(struct call_single_data *csd) */ static void generic_exec_single(int cpu, struct call_single_data *csd, int wait) { - struct call_single_queue *dst = &per_cpu(call_single_queue, cpu); - unsigned long flags; - int ipi; - if (wait) csd->flags |= CSD_FLAG_WAIT; - raw_spin_lock_irqsave(&dst->lock, flags); - ipi = list_empty(&dst->list); - list_add_tail(&csd->list, &dst->list); - raw_spin_unlock_irqrestore(&dst->lock, flags); - /* * The list addition should be visible before sending the IPI * handler locks the list to pull the entry off it because of @@ -164,7 +146,7 @@ static void generic_exec_single(int cpu, struct call_single_data *csd, int wait) * locking and barrier primitives. Generic code isn't really * equipped to do the right thing... */ - if (ipi) + if (llist_add(&csd->llist, &per_cpu(call_single_queue, cpu))) arch_send_call_function_single_ipi(cpu); if (wait) @@ -177,27 +159,26 @@ static void generic_exec_single(int cpu, struct call_single_data *csd, int wait) */ void generic_smp_call_function_single_interrupt(void) { - struct call_single_queue *q = &__get_cpu_var(call_single_queue); - LIST_HEAD(list); + struct llist_node *entry, *next; /* * Shouldn't receive this interrupt on a cpu that is not yet online. */ WARN_ON_ONCE(!cpu_online(smp_processor_id())); - raw_spin_lock(&q->lock); - list_replace_init(&q->list, &list); - raw_spin_unlock(&q->lock); + entry = llist_del_all(&__get_cpu_var(call_single_queue)); + entry = llist_reverse_order(entry); - while (!list_empty(&list)) { + while (entry) { struct call_single_data *csd; - csd = list_entry(list.next, struct call_single_data, list); - list_del(&csd->list); + next = entry->next; + csd = llist_entry(entry, struct call_single_data, llist); csd->func(csd->info); - csd_unlock(csd); + + entry = next; } } @@ -411,17 +392,11 @@ void smp_call_function_many(const struct cpumask *mask, for_each_cpu(cpu, cfd->cpumask) { struct call_single_data *csd = per_cpu_ptr(cfd->csd, cpu); - struct call_single_queue *dst = - &per_cpu(call_single_queue, cpu); - unsigned long flags; csd_lock(csd); csd->func = func; csd->info = info; - - raw_spin_lock_irqsave(&dst->lock, flags); - list_add_tail(&csd->list, &dst->list); - raw_spin_unlock_irqrestore(&dst->lock, flags); + llist_add(&csd->llist, &per_cpu(call_single_queue, cpu)); } /* Send a message to all CPUs in the map */ -- cgit v1.2.3-59-g8ed1b From 18741986a4b1dc4b1f171634c4191abc3b0fa023 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 10 Feb 2014 09:29:00 -0700 Subject: blk-mq: rework flush sequencing logic Witch to using a preallocated flush_rq for blk-mq similar to what's done with the old request path. This allows us to set up the request properly with a tag from the actually allowed range and ->rq_disk as needed by some drivers. To make life easier we also switch to dynamic allocation of ->flush_rq for the old path. This effectively reverts most of "blk-mq: fix for flush deadlock" and "blk-mq: Don't reserve a tag for flush request" Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- block/blk-core.c | 15 +++++-- block/blk-flush.c | 105 ++++++++++++++++++------------------------------- block/blk-mq.c | 54 +++++++++---------------- block/blk-mq.h | 1 + block/blk-sysfs.c | 2 + include/linux/blk-mq.h | 5 +-- include/linux/blkdev.h | 11 ++---- 7 files changed, 76 insertions(+), 117 deletions(-) (limited to 'include/linux/blkdev.h') diff --git a/block/blk-core.c b/block/blk-core.c index 06636f3ad424..853f92749202 100644 --- a/block/blk-core.c +++ b/block/blk-core.c @@ -693,11 +693,20 @@ blk_init_queue_node(request_fn_proc *rfn, spinlock_t *lock, int node_id) if (!uninit_q) return NULL; + uninit_q->flush_rq = kzalloc(sizeof(struct request), GFP_KERNEL); + if (!uninit_q->flush_rq) + goto out_cleanup_queue; + q = blk_init_allocated_queue(uninit_q, rfn, lock); if (!q) - blk_cleanup_queue(uninit_q); - + goto out_free_flush_rq; return q; + +out_free_flush_rq: + kfree(uninit_q->flush_rq); +out_cleanup_queue: + blk_cleanup_queue(uninit_q); + return NULL; } EXPORT_SYMBOL(blk_init_queue_node); @@ -1127,7 +1136,7 @@ static struct request *blk_old_get_request(struct request_queue *q, int rw, struct request *blk_get_request(struct request_queue *q, int rw, gfp_t gfp_mask) { if (q->mq_ops) - return blk_mq_alloc_request(q, rw, gfp_mask, false); + return blk_mq_alloc_request(q, rw, gfp_mask); else return blk_old_get_request(q, rw, gfp_mask); } diff --git a/block/blk-flush.c b/block/blk-flush.c index 9143e85226c7..66e2b697f5db 100644 --- a/block/blk-flush.c +++ b/block/blk-flush.c @@ -130,20 +130,26 @@ static void blk_flush_restore_request(struct request *rq) blk_clear_rq_complete(rq); } -static void mq_flush_data_run(struct work_struct *work) +static void mq_flush_run(struct work_struct *work) { struct request *rq; - rq = container_of(work, struct request, mq_flush_data); + rq = container_of(work, struct request, mq_flush_work); memset(&rq->csd, 0, sizeof(rq->csd)); blk_mq_run_request(rq, true, false); } -static void blk_mq_flush_data_insert(struct request *rq) +static bool blk_flush_queue_rq(struct request *rq) { - INIT_WORK(&rq->mq_flush_data, mq_flush_data_run); - kblockd_schedule_work(rq->q, &rq->mq_flush_data); + if (rq->q->mq_ops) { + INIT_WORK(&rq->mq_flush_work, mq_flush_run); + kblockd_schedule_work(rq->q, &rq->mq_flush_work); + return false; + } else { + list_add_tail(&rq->queuelist, &rq->q->queue_head); + return true; + } } /** @@ -187,12 +193,7 @@ static bool blk_flush_complete_seq(struct request *rq, unsigned int seq, case REQ_FSEQ_DATA: list_move_tail(&rq->flush.list, &q->flush_data_in_flight); - if (q->mq_ops) - blk_mq_flush_data_insert(rq); - else { - list_add(&rq->queuelist, &q->queue_head); - queued = true; - } + queued = blk_flush_queue_rq(rq); break; case REQ_FSEQ_DONE: @@ -216,9 +217,6 @@ static bool blk_flush_complete_seq(struct request *rq, unsigned int seq, } kicked = blk_kick_flush(q); - /* blk_mq_run_flush will run queue */ - if (q->mq_ops) - return queued; return kicked | queued; } @@ -230,10 +228,9 @@ static void flush_end_io(struct request *flush_rq, int error) struct request *rq, *n; unsigned long flags = 0; - if (q->mq_ops) { - blk_mq_free_request(flush_rq); + if (q->mq_ops) spin_lock_irqsave(&q->mq_flush_lock, flags); - } + running = &q->flush_queue[q->flush_running_idx]; BUG_ON(q->flush_pending_idx == q->flush_running_idx); @@ -263,48 +260,14 @@ static void flush_end_io(struct request *flush_rq, int error) * kblockd. */ if (queued || q->flush_queue_delayed) { - if (!q->mq_ops) - blk_run_queue_async(q); - else - /* - * This can be optimized to only run queues with requests - * queued if necessary. - */ - blk_mq_run_queues(q, true); + WARN_ON(q->mq_ops); + blk_run_queue_async(q); } q->flush_queue_delayed = 0; if (q->mq_ops) spin_unlock_irqrestore(&q->mq_flush_lock, flags); } -static void mq_flush_work(struct work_struct *work) -{ - struct request_queue *q; - struct request *rq; - - q = container_of(work, struct request_queue, mq_flush_work); - - rq = blk_mq_alloc_request(q, WRITE_FLUSH|REQ_FLUSH_SEQ, - __GFP_WAIT|GFP_ATOMIC, false); - rq->cmd_type = REQ_TYPE_FS; - rq->end_io = flush_end_io; - - blk_mq_run_request(rq, true, false); -} - -/* - * We can't directly use q->flush_rq, because it doesn't have tag and is not in - * hctx->rqs[]. so we must allocate a new request, since we can't sleep here, - * so offload the work to workqueue. - * - * Note: we assume a flush request finished in any hardware queue will flush - * the whole disk cache. - */ -static void mq_run_flush(struct request_queue *q) -{ - kblockd_schedule_work(q, &q->mq_flush_work); -} - /** * blk_kick_flush - consider issuing flush request * @q: request_queue being kicked @@ -339,19 +302,31 @@ static bool blk_kick_flush(struct request_queue *q) * different from running_idx, which means flush is in flight. */ q->flush_pending_idx ^= 1; + if (q->mq_ops) { - mq_run_flush(q); - return true; + struct blk_mq_ctx *ctx = first_rq->mq_ctx; + struct blk_mq_hw_ctx *hctx = q->mq_ops->map_queue(q, ctx->cpu); + + blk_mq_rq_init(hctx, q->flush_rq); + q->flush_rq->mq_ctx = ctx; + + /* + * Reuse the tag value from the fist waiting request, + * with blk-mq the tag is generated during request + * allocation and drivers can rely on it being inside + * the range they asked for. + */ + q->flush_rq->tag = first_rq->tag; + } else { + blk_rq_init(q, q->flush_rq); } - blk_rq_init(q, &q->flush_rq); - q->flush_rq.cmd_type = REQ_TYPE_FS; - q->flush_rq.cmd_flags = WRITE_FLUSH | REQ_FLUSH_SEQ; - q->flush_rq.rq_disk = first_rq->rq_disk; - q->flush_rq.end_io = flush_end_io; + q->flush_rq->cmd_type = REQ_TYPE_FS; + q->flush_rq->cmd_flags = WRITE_FLUSH | REQ_FLUSH_SEQ; + q->flush_rq->rq_disk = first_rq->rq_disk; + q->flush_rq->end_io = flush_end_io; - list_add_tail(&q->flush_rq.queuelist, &q->queue_head); - return true; + return blk_flush_queue_rq(q->flush_rq); } static void flush_data_end_io(struct request *rq, int error) @@ -407,11 +382,8 @@ void blk_insert_flush(struct request *rq) /* * @policy now records what operations need to be done. Adjust * REQ_FLUSH and FUA for the driver. - * We keep REQ_FLUSH for mq to track flush requests. For !FUA, - * we never dispatch the request directly. */ - if (rq->cmd_flags & REQ_FUA) - rq->cmd_flags &= ~REQ_FLUSH; + rq->cmd_flags &= ~REQ_FLUSH; if (!(fflags & REQ_FUA)) rq->cmd_flags &= ~REQ_FUA; @@ -560,5 +532,4 @@ EXPORT_SYMBOL(blkdev_issue_flush); void blk_mq_init_flush(struct request_queue *q) { spin_lock_init(&q->mq_flush_lock); - INIT_WORK(&q->mq_flush_work, mq_flush_work); } diff --git a/block/blk-mq.c b/block/blk-mq.c index 14c8f35946e1..a59b0565e940 100644 --- a/block/blk-mq.c +++ b/block/blk-mq.c @@ -194,27 +194,9 @@ static void blk_mq_rq_ctx_init(struct request_queue *q, struct blk_mq_ctx *ctx, } static struct request *__blk_mq_alloc_request(struct blk_mq_hw_ctx *hctx, - gfp_t gfp, bool reserved, - int rw) + gfp_t gfp, bool reserved) { - struct request *req; - bool is_flush = false; - /* - * flush need allocate a request, leave at least one request for - * non-flush IO to avoid deadlock - */ - if ((rw & REQ_FLUSH) && !(rw & REQ_FLUSH_SEQ)) { - if (atomic_inc_return(&hctx->pending_flush) >= - hctx->queue_depth - hctx->reserved_tags - 1) { - atomic_dec(&hctx->pending_flush); - return NULL; - } - is_flush = true; - } - req = blk_mq_alloc_rq(hctx, gfp, reserved); - if (!req && is_flush) - atomic_dec(&hctx->pending_flush); - return req; + return blk_mq_alloc_rq(hctx, gfp, reserved); } static struct request *blk_mq_alloc_request_pinned(struct request_queue *q, @@ -227,7 +209,7 @@ static struct request *blk_mq_alloc_request_pinned(struct request_queue *q, struct blk_mq_ctx *ctx = blk_mq_get_ctx(q); struct blk_mq_hw_ctx *hctx = q->mq_ops->map_queue(q, ctx->cpu); - rq = __blk_mq_alloc_request(hctx, gfp & ~__GFP_WAIT, reserved, rw); + rq = __blk_mq_alloc_request(hctx, gfp & ~__GFP_WAIT, reserved); if (rq) { blk_mq_rq_ctx_init(q, ctx, rq, rw); break; @@ -244,15 +226,14 @@ static struct request *blk_mq_alloc_request_pinned(struct request_queue *q, return rq; } -struct request *blk_mq_alloc_request(struct request_queue *q, int rw, - gfp_t gfp, bool reserved) +struct request *blk_mq_alloc_request(struct request_queue *q, int rw, gfp_t gfp) { struct request *rq; if (blk_mq_queue_enter(q)) return NULL; - rq = blk_mq_alloc_request_pinned(q, rw, gfp, reserved); + rq = blk_mq_alloc_request_pinned(q, rw, gfp, false); if (rq) blk_mq_put_ctx(rq->mq_ctx); return rq; @@ -276,7 +257,7 @@ EXPORT_SYMBOL(blk_mq_alloc_reserved_request); /* * Re-init and set pdu, if we have it */ -static void blk_mq_rq_init(struct blk_mq_hw_ctx *hctx, struct request *rq) +void blk_mq_rq_init(struct blk_mq_hw_ctx *hctx, struct request *rq) { blk_rq_init(hctx->queue, rq); @@ -290,9 +271,6 @@ static void __blk_mq_free_request(struct blk_mq_hw_ctx *hctx, const int tag = rq->tag; struct request_queue *q = rq->q; - if ((rq->cmd_flags & REQ_FLUSH) && !(rq->cmd_flags & REQ_FLUSH_SEQ)) - atomic_dec(&hctx->pending_flush); - blk_mq_rq_init(hctx, rq); blk_mq_put_tag(hctx->tags, tag); @@ -946,14 +924,14 @@ static void blk_mq_make_request(struct request_queue *q, struct bio *bio) hctx = q->mq_ops->map_queue(q, ctx->cpu); trace_block_getrq(q, bio, rw); - rq = __blk_mq_alloc_request(hctx, GFP_ATOMIC, false, bio->bi_rw); + rq = __blk_mq_alloc_request(hctx, GFP_ATOMIC, false); if (likely(rq)) - blk_mq_rq_ctx_init(q, ctx, rq, bio->bi_rw); + blk_mq_rq_ctx_init(q, ctx, rq, rw); else { blk_mq_put_ctx(ctx); trace_block_sleeprq(q, bio, rw); - rq = blk_mq_alloc_request_pinned(q, bio->bi_rw, - __GFP_WAIT|GFP_ATOMIC, false); + rq = blk_mq_alloc_request_pinned(q, rw, __GFP_WAIT|GFP_ATOMIC, + false); ctx = rq->mq_ctx; hctx = q->mq_ops->map_queue(q, ctx->cpu); } @@ -1230,9 +1208,7 @@ static int blk_mq_init_hw_queues(struct request_queue *q, hctx->queue_num = i; hctx->flags = reg->flags; hctx->queue_depth = reg->queue_depth; - hctx->reserved_tags = reg->reserved_tags; hctx->cmd_size = reg->cmd_size; - atomic_set(&hctx->pending_flush, 0); blk_mq_init_cpu_notifier(&hctx->cpu_notifier, blk_mq_hctx_notify, hctx); @@ -1412,9 +1388,14 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_reg *reg, blk_mq_init_flush(q); blk_mq_init_cpu_queues(q, reg->nr_hw_queues); - if (blk_mq_init_hw_queues(q, reg, driver_data)) + q->flush_rq = kzalloc(round_up(sizeof(struct request) + reg->cmd_size, + cache_line_size()), GFP_KERNEL); + if (!q->flush_rq) goto err_hw; + if (blk_mq_init_hw_queues(q, reg, driver_data)) + goto err_flush_rq; + blk_mq_map_swqueue(q); mutex_lock(&all_q_mutex); @@ -1422,6 +1403,9 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_reg *reg, mutex_unlock(&all_q_mutex); return q; + +err_flush_rq: + kfree(q->flush_rq); err_hw: kfree(q->mq_map); err_map: diff --git a/block/blk-mq.h b/block/blk-mq.h index f29b645f0e1c..ed0035cd458e 100644 --- a/block/blk-mq.h +++ b/block/blk-mq.h @@ -28,6 +28,7 @@ void blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async); void blk_mq_init_flush(struct request_queue *q); void blk_mq_drain_queue(struct request_queue *q); void blk_mq_free_queue(struct request_queue *q); +void blk_mq_rq_init(struct blk_mq_hw_ctx *hctx, struct request *rq); /* * CPU hotplug helpers diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c index 8095c4a21fc0..7500f876dae4 100644 --- a/block/blk-sysfs.c +++ b/block/blk-sysfs.c @@ -549,6 +549,8 @@ static void blk_release_queue(struct kobject *kobj) if (q->mq_ops) blk_mq_free_queue(q); + kfree(q->flush_rq); + blk_trace_shutdown(q); bdi_destroy(&q->backing_dev_info); diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h index 468be242db90..18ba8a627f46 100644 --- a/include/linux/blk-mq.h +++ b/include/linux/blk-mq.h @@ -36,15 +36,12 @@ struct blk_mq_hw_ctx { struct list_head page_list; struct blk_mq_tags *tags; - atomic_t pending_flush; - unsigned long queued; unsigned long run; #define BLK_MQ_MAX_DISPATCH_ORDER 10 unsigned long dispatched[BLK_MQ_MAX_DISPATCH_ORDER]; unsigned int queue_depth; - unsigned int reserved_tags; unsigned int numa_node; unsigned int cmd_size; /* per-request extra data */ @@ -129,7 +126,7 @@ void blk_mq_insert_request(struct request_queue *, struct request *, void blk_mq_run_queues(struct request_queue *q, bool async); void blk_mq_free_request(struct request *rq); bool blk_mq_can_queue(struct blk_mq_hw_ctx *); -struct request *blk_mq_alloc_request(struct request_queue *q, int rw, gfp_t gfp, bool reserved); +struct request *blk_mq_alloc_request(struct request_queue *q, int rw, gfp_t gfp); struct request *blk_mq_alloc_reserved_request(struct request_queue *q, int rw, gfp_t gfp); struct request *blk_mq_rq_from_tag(struct request_queue *q, unsigned int tag); diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 0375654adb28..b2d25ecbcbc1 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -101,7 +101,7 @@ struct request { }; union { struct call_single_data csd; - struct work_struct mq_flush_data; + struct work_struct mq_flush_work; }; struct request_queue *q; @@ -451,13 +451,8 @@ struct request_queue { unsigned long flush_pending_since; struct list_head flush_queue[2]; struct list_head flush_data_in_flight; - union { - struct request flush_rq; - struct { - spinlock_t mq_flush_lock; - struct work_struct mq_flush_work; - }; - }; + struct request *flush_rq; + spinlock_t mq_flush_lock; struct mutex sysfs_lock; -- cgit v1.2.3-59-g8ed1b