Hi,
This patch series adds a mechanism to pass through provision requests on
stacked thinly provisioned block devices.
The linux kernel provides several mechanisms to set up thinly provisioned
block storage abstractions (eg. dm-thin, loop devices over sparse files),
either directly as block devices or backing storage for filesystems. Currently,
short of writing data to either the device or filesystme, there is no way for
users to pre-allocate space for use in such storage setups. Consider the
following use-cases:
1) Suspend-to-disk and resume from a dm-thin device: In order to ensure that
the underlying thinpool metadata is not modified during the suspend
mechanism, the dm-thin device needs to be fully provisioned.
2) If a filesystem uses a loop device over a sparse file, fallocate() on the
filesystem will allocate blocks for files but the underlying sparse file
will remain intact.
3) Another example is virtual machine using a sparse file/dm-thin as a storage
device; by default, allocations within the VM boundaries will not affect
the host.
4) Several storage standards support mechanisms for thin provisioning on
real hardware devices. For example:
a. The NVMe spec 1.0b section 2.1.1 loosely talks about thin provisioning:
"When the THINP bit in the NSFEAT field of the Identify Namespace data
structure is set to ‘1’, the controller ... shall track the number of
allocated blocks in the Namespace Utilization field"
b. The SCSi Block Commands reference - 4 section references "Thin
provisioned logical units",
c. UFS 3.0 spec section 13.3.3 references "Thin provisioning".
In all the above situations, currently, the only way for pre-allocating space
is to issue writes (or use WRITE_ZEROES/WRITE_SAME). However, that does not
scale well with larger pre-allocation sizes.
This patchset introduces primitives to support block-level provisioning (note:
the term 'provisioning' is used to prevent overloading the term
'allocations/pre-allocations') requests across filesystems and block devices.
This allows fallocate() and file creation requests to reserve space across
stacked layers of block devices and filesystems. Currently, the patchset covers
a prototype on the device-mapper targets, loop device and ext4, but the same
mechanism can be extended to other filesystems/block devices as well as extended
for use with devices in 4 a-c.
Patch 1 introduces REQ_OP_PROVISION as a new request type.
The provision request acts like the inverse of a discard request; instead
of notifying lower layers that the block range will no longer be used, provision
acts as a request to lower layers to provision disk space for the given block
range. Real hardware storage devices will currently disable the provisioing
capability but for the standards listed in 4a.-c., REQ_OP_PROVISION can be
overloaded for use as the provisioing primitive for future devices.
Patch 2 implements REQ_OP_PROVISION handling for some of the device-mapper
targets. Device-mapper targets will usually mirror the support of underlying
devices. This patch also enables the use of fallocate in mode == 0 for block
devices.
Patch 3 wires up the loop device handling of REQ_OP_PROVISION and calls
fallocate() with mode 0 on the underlying file/block device.
Testing:
--------
- Tested on a VM running a 6.2 kernel.
- Preallocation of dm-thin devices:
As expected, avoiding the need to zero out thinly-provisioned block devices to
preallocate space speeds up the provisioning operation significantly:
The following was tested on a dm-thin device set up on top of a dm-thinp with
skip_block_zeroing=true.
A) Zeroout was measured using `fallocate -z ...`
B) Provision was measured using `fallocate -p ...`.
Size Time A B
512M real 1.093 0.034
user 0 0
sys 0.022 0.01
1G real 2.182 0.048
user 0 0.01
sys 0.022 0
2G real 4.344 0.082
user 0 0.01
sys 0.036 0
4G real 8.679 0.153
user 0 0.01
sys 0.073 0
8G real 17.777 0.318
user 0 0.01
sys 0.144 0
Changelog:
V3:
- Drop FALLOC_FL_PROVISION and use mode == 0 for provision requests.
- Drop fs-specific patches; will be sent out in a follow up series.
- Fix missing shared block handling for thin snapshots.
V2:
- Fix stacked limit handling.
- Enable provision request handling in dm-snapshot
- Don't call truncate_bdev_range if blkdev_fallocate() is called with
FALLOC_FL_PROVISION.
- Clarify semantics of FALLOC_FL_PROVISION and why it needs to be a separate flag
(as opposed to overloading mode == 0).
Sarthak Kukreti (3):
block: Introduce provisioning primitives
dm: Add support for block provisioning
loop: Add support for provision requests
block/blk-core.c | 5 ++
block/blk-lib.c | 53 ++++++++++++++++
block/blk-merge.c | 18 ++++++
block/blk-settings.c | 19 ++++++
block/blk-sysfs.c | 8 +++
block/bounce.c | 1 +
block/fops.c | 14 +++--
drivers/block/loop.c | 42 +++++++++++++
drivers/md/dm-crypt.c | 4 +-
drivers/md/dm-linear.c | 1 +
drivers/md/dm-snap.c | 7 +++
drivers/md/dm-table.c | 25 ++++++++
drivers/md/dm-thin.c | 110 +++++++++++++++++++++++++++++++---
drivers/md/dm.c | 4 ++
include/linux/bio.h | 6 +-
include/linux/blk_types.h | 5 +-
include/linux/blkdev.h | 16 +++++
include/linux/device-mapper.h | 11 ++++
18 files changed, 333 insertions(+), 16 deletions(-)
--
2.40.0.634.g4ca3ef3211-goog
Add support to dm devices for REQ_OP_PROVISION. The default mode
is to passthrough the request to the underlying device, if it
supports it. dm-thinpool uses the provision request to provision
blocks for a dm-thin device. dm-thinpool currently does not
pass through REQ_OP_PROVISION to underlying devices.
For shared blocks, provision requests will break sharing and copy the
contents of the entire block.
Signed-off-by: Sarthak Kukreti <[email protected]>
---
drivers/md/dm-crypt.c | 4 +-
drivers/md/dm-linear.c | 1 +
drivers/md/dm-snap.c | 7 +++
drivers/md/dm-table.c | 25 ++++++++
drivers/md/dm-thin.c | 110 +++++++++++++++++++++++++++++++---
drivers/md/dm.c | 4 ++
include/linux/device-mapper.h | 11 ++++
7 files changed, 153 insertions(+), 9 deletions(-)
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index 3ba53dc3cc3f..5c655bfd4686 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -3087,6 +3087,8 @@ static int crypt_ctr_optional(struct dm_target *ti, unsigned int argc, char **ar
if (ret)
return ret;
+ ti->num_provision_bios = 1;
+
while (opt_params--) {
opt_string = dm_shift_arg(&as);
if (!opt_string) {
@@ -3390,7 +3392,7 @@ static int crypt_map(struct dm_target *ti, struct bio *bio)
* - for REQ_OP_DISCARD caller must use flush if IO ordering matters
*/
if (unlikely(bio->bi_opf & REQ_PREFLUSH ||
- bio_op(bio) == REQ_OP_DISCARD)) {
+ bio_op(bio) == REQ_OP_DISCARD || bio_op(bio) == REQ_OP_PROVISION)) {
bio_set_dev(bio, cc->dev->bdev);
if (bio_sectors(bio))
bio->bi_iter.bi_sector = cc->start +
diff --git a/drivers/md/dm-linear.c b/drivers/md/dm-linear.c
index 3e622dcc9dbd..7843e548e850 100644
--- a/drivers/md/dm-linear.c
+++ b/drivers/md/dm-linear.c
@@ -62,6 +62,7 @@ static int linear_ctr(struct dm_target *ti, unsigned int argc, char **argv)
ti->num_discard_bios = 1;
ti->num_secure_erase_bios = 1;
ti->num_write_zeroes_bios = 1;
+ ti->num_provision_bios = 1;
ti->private = lc;
return 0;
diff --git a/drivers/md/dm-snap.c b/drivers/md/dm-snap.c
index f766c21408f1..f6b224a12000 100644
--- a/drivers/md/dm-snap.c
+++ b/drivers/md/dm-snap.c
@@ -1358,6 +1358,7 @@ static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
if (s->discard_zeroes_cow)
ti->num_discard_bios = (s->discard_passdown_origin ? 2 : 1);
ti->per_io_data_size = sizeof(struct dm_snap_tracked_chunk);
+ ti->num_provision_bios = 1;
/* Add snapshot to the list of snapshots for this origin */
/* Exceptions aren't triggered till snapshot_resume() is called */
@@ -2003,6 +2004,11 @@ static int snapshot_map(struct dm_target *ti, struct bio *bio)
/* If the block is already remapped - use that, else remap it */
e = dm_lookup_exception(&s->complete, chunk);
if (e) {
+ if (unlikely(bio_op(bio) == REQ_OP_PROVISION)) {
+ bio_endio(bio);
+ r = DM_MAPIO_SUBMITTED;
+ goto out_unlock;
+ }
remap_exception(s, e, bio, chunk);
if (unlikely(bio_op(bio) == REQ_OP_DISCARD) &&
io_overlaps_chunk(s, bio)) {
@@ -2413,6 +2419,7 @@ static void snapshot_io_hints(struct dm_target *ti, struct queue_limits *limits)
/* All discards are split on chunk_size boundary */
limits->discard_granularity = snap->store->chunk_size;
limits->max_discard_sectors = snap->store->chunk_size;
+ limits->max_provision_sectors = snap->store->chunk_size;
up_read(&_origins_lock);
}
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 2055a758541d..5985343384a7 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -1850,6 +1850,26 @@ static bool dm_table_supports_write_zeroes(struct dm_table *t)
return true;
}
+static int device_provision_capable(struct dm_target *ti, struct dm_dev *dev,
+ sector_t start, sector_t len, void *data)
+{
+ return !bdev_max_provision_sectors(dev->bdev);
+}
+
+static bool dm_table_supports_provision(struct dm_table *t)
+{
+ for (unsigned int i = 0; i < t->num_targets; i++) {
+ struct dm_target *ti = dm_table_get_target(t, i);
+
+ if (ti->provision_supported ||
+ (ti->type->iterate_devices &&
+ ti->type->iterate_devices(ti, device_provision_capable, NULL)))
+ return true;
+ }
+
+ return false;
+}
+
static int device_not_nowait_capable(struct dm_target *ti, struct dm_dev *dev,
sector_t start, sector_t len, void *data)
{
@@ -1983,6 +2003,11 @@ int dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
if (!dm_table_supports_write_zeroes(t))
q->limits.max_write_zeroes_sectors = 0;
+ if (dm_table_supports_provision(t))
+ blk_queue_max_provision_sectors(q, UINT_MAX >> 9);
+ else
+ q->limits.max_provision_sectors = 0;
+
dm_table_verify_integrity(t);
/*
diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index 13d4677baafd..b08b7ae617be 100644
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -909,7 +909,8 @@ static void __inc_remap_and_issue_cell(void *context,
struct bio *bio;
while ((bio = bio_list_pop(&cell->bios))) {
- if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD)
+ if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD ||
+ bio_op(bio) == REQ_OP_PROVISION)
bio_list_add(&info->defer_bios, bio);
else {
inc_all_io_entry(info->tc->pool, bio);
@@ -1013,6 +1014,15 @@ static void process_prepared_mapping(struct dm_thin_new_mapping *m)
goto out;
}
+ /*
+ * For provision requests, once the prepared block has been inserted
+ * into the mapping btree, return.
+ */
+ if (bio && bio_op(bio) == REQ_OP_PROVISION) {
+ bio_endio(bio);
+ return;
+ }
+
/*
* Release any bios held while the block was being provisioned.
* If we are processing a write bio that completely covers the block,
@@ -1241,7 +1251,7 @@ static int io_overlaps_block(struct pool *pool, struct bio *bio)
static int io_overwrites_block(struct pool *pool, struct bio *bio)
{
- return (bio_data_dir(bio) == WRITE) &&
+ return (bio_data_dir(bio) == WRITE) && bio_op(bio) != REQ_OP_PROVISION &&
io_overlaps_block(pool, bio);
}
@@ -1334,10 +1344,11 @@ static void schedule_copy(struct thin_c *tc, dm_block_t virt_block,
/*
* IO to pool_dev remaps to the pool target's data_dev.
*
- * If the whole block of data is being overwritten, we can issue the
- * bio immediately. Otherwise we use kcopyd to clone the data first.
+ * If the whole block of data is being overwritten and if this is not a
+ * provision request, we can issue the bio immediately.
+ * Otherwise we use kcopyd to clone the data first.
*/
- if (io_overwrites_block(pool, bio))
+ if (io_overwrites_block(pool, bio) && bio_op(bio) != REQ_OP_PROVISION)
remap_and_issue_overwrite(tc, bio, data_dest, m);
else {
struct dm_io_region from, to;
@@ -1356,7 +1367,8 @@ static void schedule_copy(struct thin_c *tc, dm_block_t virt_block,
/*
* Do we need to zero a tail region?
*/
- if (len < pool->sectors_per_block && pool->pf.zero_new_blocks) {
+ if (len < pool->sectors_per_block && pool->pf.zero_new_blocks &&
+ bio_op(bio) != REQ_OP_PROVISION) {
atomic_inc(&m->prepare_actions);
ll_zero(tc, m,
data_dest * pool->sectors_per_block + len,
@@ -1390,6 +1402,10 @@ static void schedule_zero(struct thin_c *tc, dm_block_t virt_block,
m->data_block = data_block;
m->cell = cell;
+ /* Provision requests are chained on the original bio. */
+ if (bio && bio_op(bio) == REQ_OP_PROVISION)
+ m->bio = bio;
+
/*
* If the whole block of data is being overwritten or we are not
* zeroing pre-existing data, we can issue the bio immediately.
@@ -1865,7 +1881,8 @@ static void process_shared_bio(struct thin_c *tc, struct bio *bio,
if (bio_data_dir(bio) == WRITE && bio->bi_iter.bi_size) {
break_sharing(tc, bio, block, &key, lookup_result, data_cell);
- cell_defer_no_holder(tc, virt_cell);
+ if (bio_op(bio) != REQ_OP_PROVISION)
+ cell_defer_no_holder(tc, virt_cell);
} else {
struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
@@ -1982,6 +1999,73 @@ static void process_cell(struct thin_c *tc, struct dm_bio_prison_cell *cell)
}
}
+static void process_provision_cell(struct thin_c *tc, struct dm_bio_prison_cell *cell)
+{
+ int r;
+ struct pool *pool = tc->pool;
+ struct bio *bio = cell->holder;
+ dm_block_t begin, end;
+ struct dm_thin_lookup_result lookup_result;
+
+ if (tc->requeue_mode) {
+ cell_requeue(pool, cell);
+ return;
+ }
+
+ get_bio_block_range(tc, bio, &begin, &end);
+
+ while (begin != end) {
+ r = ensure_next_mapping(pool);
+ if (r)
+ /* we did our best */
+ return;
+
+ r = dm_thin_find_block(tc->td, begin, 1, &lookup_result);
+ switch (r) {
+ case 0:
+ if (lookup_result.shared)
+ process_shared_bio(tc, bio, begin,
+ &lookup_result, cell);
+ begin++;
+ break;
+ case -ENODATA:
+ bio_inc_remaining(bio);
+ provision_block(tc, bio, begin, cell);
+ begin++;
+ break;
+ default:
+ DMERR_LIMIT(
+ "%s: dm_thin_find_block() failed: error = %d",
+ __func__, r);
+ cell_defer_no_holder(tc, cell);
+ bio_io_error(bio);
+ begin++;
+ break;
+ }
+ }
+ bio_endio(bio);
+ cell_defer_no_holder(tc, cell);
+}
+
+static void process_provision_bio(struct thin_c *tc, struct bio *bio)
+{
+ dm_block_t begin, end;
+ struct dm_cell_key virt_key;
+ struct dm_bio_prison_cell *virt_cell;
+
+ get_bio_block_range(tc, bio, &begin, &end);
+ if (begin == end) {
+ bio_endio(bio);
+ return;
+ }
+
+ build_key(tc->td, VIRTUAL, begin, end, &virt_key);
+ if (bio_detain(tc->pool, &virt_key, bio, &virt_cell))
+ return;
+
+ process_provision_cell(tc, virt_cell);
+}
+
static void process_bio(struct thin_c *tc, struct bio *bio)
{
struct pool *pool = tc->pool;
@@ -2202,6 +2286,8 @@ static void process_thin_deferred_bios(struct thin_c *tc)
if (bio_op(bio) == REQ_OP_DISCARD)
pool->process_discard(tc, bio);
+ else if (bio_op(bio) == REQ_OP_PROVISION)
+ process_provision_bio(tc, bio);
else
pool->process_bio(tc, bio);
@@ -2723,7 +2809,8 @@ static int thin_bio_map(struct dm_target *ti, struct bio *bio)
return DM_MAPIO_SUBMITTED;
}
- if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD) {
+ if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD ||
+ bio_op(bio) == REQ_OP_PROVISION) {
thin_defer_bio_with_throttle(tc, bio);
return DM_MAPIO_SUBMITTED;
}
@@ -3370,6 +3457,8 @@ static int pool_ctr(struct dm_target *ti, unsigned int argc, char **argv)
pt->adjusted_pf = pt->requested_pf = pf;
ti->num_flush_bios = 1;
ti->limit_swap_bios = true;
+ ti->num_provision_bios = 1;
+ ti->provision_supported = true;
/*
* Only need to enable discards if the pool should pass
@@ -4068,6 +4157,7 @@ static void pool_io_hints(struct dm_target *ti, struct queue_limits *limits)
blk_limits_io_opt(limits, pool->sectors_per_block << SECTOR_SHIFT);
}
+
/*
* pt->adjusted_pf is a staging area for the actual features to use.
* They get transferred to the live pool in bind_control_target()
@@ -4261,6 +4351,9 @@ static int thin_ctr(struct dm_target *ti, unsigned int argc, char **argv)
ti->num_discard_bios = 1;
}
+ ti->num_provision_bios = 1;
+ ti->provision_supported = true;
+
mutex_unlock(&dm_thin_pool_table.mutex);
spin_lock_irq(&tc->pool->lock);
@@ -4475,6 +4568,7 @@ static void thin_io_hints(struct dm_target *ti, struct queue_limits *limits)
limits->discard_granularity = pool->sectors_per_block << SECTOR_SHIFT;
limits->max_discard_sectors = 2048 * 1024 * 16; /* 16G */
+ limits->max_provision_sectors = 2048 * 1024 * 16; /* 16G */
}
static struct target_type thin_target = {
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index dfde0088147a..d8f1803062b7 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -1593,6 +1593,7 @@ static bool is_abnormal_io(struct bio *bio)
case REQ_OP_DISCARD:
case REQ_OP_SECURE_ERASE:
case REQ_OP_WRITE_ZEROES:
+ case REQ_OP_PROVISION:
return true;
default:
break;
@@ -1617,6 +1618,9 @@ static blk_status_t __process_abnormal_io(struct clone_info *ci,
case REQ_OP_WRITE_ZEROES:
num_bios = ti->num_write_zeroes_bios;
break;
+ case REQ_OP_PROVISION:
+ num_bios = ti->num_provision_bios;
+ break;
default:
break;
}
diff --git a/include/linux/device-mapper.h b/include/linux/device-mapper.h
index 7975483816e4..e9f687521ae6 100644
--- a/include/linux/device-mapper.h
+++ b/include/linux/device-mapper.h
@@ -334,6 +334,12 @@ struct dm_target {
*/
unsigned int num_write_zeroes_bios;
+ /*
+ * The number of PROVISION bios that will be submitted to the target.
+ * The bio number can be accessed with dm_bio_get_target_bio_nr.
+ */
+ unsigned int num_provision_bios;
+
/*
* The minimum number of extra bytes allocated in each io for the
* target to use.
@@ -358,6 +364,11 @@ struct dm_target {
*/
bool discards_supported:1;
+ /* Set if this target needs to receive provision requests regardless of
+ * whether or not its underlying devices have support.
+ */
+ bool provision_supported:1;
+
/*
* Set if we need to limit the number of in-flight bios when swapping.
*/
--
2.40.0.634.g4ca3ef3211-goog
Hi,
This patch series is revision 4 of introducing a new mechanism to pass through provision requests on stacked thinly provisioned storage devices. See [1] for original cover letter.
[1] https://lore.kernel.org/lkml/[email protected]/T/#md4f20113c2242755747ae069f84be720a6751012
Changelog:
V4:
- Fix UNSHARE and KEEP_SIZE handling in blkdev_fallocate.
- Split dm-thin support into a separate patch.
- Remove ranged provision request handling and adjust io hints to handle provisioning one block at a time.
- Add missing provision_supported for dm targets.
V3:
- Drop FALLOC_FL_PROVISION and use mode == 0 for provision requests.
- Drop fs-specific patches; will be sent out in a follow up series.
- Fix missing shared block handling for thin snapshots.
V2:
- Fix stacked limit handling.
- Enable provision request handling in dm-snapshot
- Don't call truncate_bdev_range if blkdev_fallocate() is called with
FALLOC_FL_PROVISION.
- Clarify semantics of FALLOC_FL_PROVISION and why it needs to be a separate flag
(as opposed to overloading mode == 0).
Sarthak Kukreti (4):
block: Introduce provisioning primitives
dm: Add block provisioning support
dm-thin: Add REQ_OP_PROVISION support
loop: Add support for provision requests
block/blk-core.c | 5 +++
block/blk-lib.c | 53 +++++++++++++++++++++++++
block/blk-merge.c | 18 +++++++++
block/blk-settings.c | 19 +++++++++
block/blk-sysfs.c | 8 ++++
block/bounce.c | 1 +
block/fops.c | 25 +++++++++---
drivers/block/loop.c | 42 ++++++++++++++++++++
drivers/md/dm-crypt.c | 5 ++-
drivers/md/dm-linear.c | 2 +
drivers/md/dm-snap.c | 8 ++++
drivers/md/dm-table.c | 23 +++++++++++
drivers/md/dm-thin.c | 73 ++++++++++++++++++++++++++++++++---
drivers/md/dm.c | 6 +++
include/linux/bio.h | 6 ++-
include/linux/blk_types.h | 5 ++-
include/linux/blkdev.h | 16 ++++++++
include/linux/device-mapper.h | 17 ++++++++
18 files changed, 317 insertions(+), 15 deletions(-)
--
2.40.0.634.g4ca3ef3211-goog
Introduce block request REQ_OP_PROVISION. The intent of this request
is to request underlying storage to preallocate disk space for the given
block range. Block devices that support this capability will export
a provision limit within their request queues.
This patch also adds the capability to call fallocate() in mode 0
on block devices, which will send REQ_OP_PROVISION to the block
device for the specified range,
Signed-off-by: Sarthak Kukreti <[email protected]>
---
block/blk-core.c | 5 ++++
block/blk-lib.c | 53 +++++++++++++++++++++++++++++++++++++++
block/blk-merge.c | 18 +++++++++++++
block/blk-settings.c | 19 ++++++++++++++
block/blk-sysfs.c | 8 ++++++
block/bounce.c | 1 +
block/fops.c | 25 +++++++++++++-----
include/linux/bio.h | 6 +++--
include/linux/blk_types.h | 5 +++-
include/linux/blkdev.h | 16 ++++++++++++
10 files changed, 147 insertions(+), 9 deletions(-)
diff --git a/block/blk-core.c b/block/blk-core.c
index 42926e6cb83c..4a2342ba3a8b 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -123,6 +123,7 @@ static const char *const blk_op_name[] = {
REQ_OP_NAME(WRITE_ZEROES),
REQ_OP_NAME(DRV_IN),
REQ_OP_NAME(DRV_OUT),
+ REQ_OP_NAME(PROVISION)
};
#undef REQ_OP_NAME
@@ -798,6 +799,10 @@ void submit_bio_noacct(struct bio *bio)
if (!q->limits.max_write_zeroes_sectors)
goto not_supported;
break;
+ case REQ_OP_PROVISION:
+ if (!q->limits.max_provision_sectors)
+ goto not_supported;
+ break;
default:
break;
}
diff --git a/block/blk-lib.c b/block/blk-lib.c
index e59c3069e835..647b6451660b 100644
--- a/block/blk-lib.c
+++ b/block/blk-lib.c
@@ -343,3 +343,56 @@ int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector,
return ret;
}
EXPORT_SYMBOL(blkdev_issue_secure_erase);
+
+/**
+ * blkdev_issue_provision - provision a block range
+ * @bdev: blockdev to write
+ * @sector: start sector
+ * @nr_sects: number of sectors to provision
+ * @gfp_mask: memory allocation flags (for bio_alloc)
+ *
+ * Description:
+ * Issues a provision request to the block device for the range of sectors.
+ * For thinly provisioned block devices, this acts as a signal for the
+ * underlying storage pool to allocate space for this block range.
+ */
+int blkdev_issue_provision(struct block_device *bdev, sector_t sector,
+ sector_t nr_sects, gfp_t gfp)
+{
+ sector_t bs_mask = (bdev_logical_block_size(bdev) >> 9) - 1;
+ unsigned int max_sectors = bdev_max_provision_sectors(bdev);
+ struct bio *bio = NULL;
+ struct blk_plug plug;
+ int ret = 0;
+
+ if (max_sectors == 0)
+ return -EOPNOTSUPP;
+ if ((sector | nr_sects) & bs_mask)
+ return -EINVAL;
+ if (bdev_read_only(bdev))
+ return -EPERM;
+
+ blk_start_plug(&plug);
+ for (;;) {
+ unsigned int req_sects = min_t(sector_t, nr_sects, max_sectors);
+
+ bio = blk_next_bio(bio, bdev, 0, REQ_OP_PROVISION, gfp);
+ bio->bi_iter.bi_sector = sector;
+ bio->bi_iter.bi_size = req_sects << SECTOR_SHIFT;
+
+ sector += req_sects;
+ nr_sects -= req_sects;
+ if (!nr_sects) {
+ ret = submit_bio_wait(bio);
+ if (ret == -EOPNOTSUPP)
+ ret = 0;
+ bio_put(bio);
+ break;
+ }
+ cond_resched();
+ }
+ blk_finish_plug(&plug);
+
+ return ret;
+}
+EXPORT_SYMBOL(blkdev_issue_provision);
diff --git a/block/blk-merge.c b/block/blk-merge.c
index 6460abdb2426..a3ffebb97a1d 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -158,6 +158,21 @@ static struct bio *bio_split_write_zeroes(struct bio *bio,
return bio_split(bio, lim->max_write_zeroes_sectors, GFP_NOIO, bs);
}
+static struct bio *bio_split_provision(struct bio *bio,
+ const struct queue_limits *lim,
+ unsigned int *nsegs, struct bio_set *bs)
+{
+ *nsegs = 0;
+
+ if (!lim->max_provision_sectors)
+ return NULL;
+
+ if (bio_sectors(bio) <= lim->max_provision_sectors)
+ return NULL;
+
+ return bio_split(bio, lim->max_provision_sectors, GFP_NOIO, bs);
+}
+
/*
* Return the maximum number of sectors from the start of a bio that may be
* submitted as a single request to a block device. If enough sectors remain,
@@ -366,6 +381,9 @@ struct bio *__bio_split_to_limits(struct bio *bio,
case REQ_OP_WRITE_ZEROES:
split = bio_split_write_zeroes(bio, lim, nr_segs, bs);
break;
+ case REQ_OP_PROVISION:
+ split = bio_split_provision(bio, lim, nr_segs, bs);
+ break;
default:
split = bio_split_rw(bio, lim, nr_segs, bs,
get_max_io_size(bio, lim) << SECTOR_SHIFT);
diff --git a/block/blk-settings.c b/block/blk-settings.c
index 896b4654ab00..d303e6614c36 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -59,6 +59,7 @@ void blk_set_default_limits(struct queue_limits *lim)
lim->zoned = BLK_ZONED_NONE;
lim->zone_write_granularity = 0;
lim->dma_alignment = 511;
+ lim->max_provision_sectors = 0;
}
/**
@@ -82,6 +83,7 @@ void blk_set_stacking_limits(struct queue_limits *lim)
lim->max_dev_sectors = UINT_MAX;
lim->max_write_zeroes_sectors = UINT_MAX;
lim->max_zone_append_sectors = UINT_MAX;
+ lim->max_provision_sectors = UINT_MAX;
}
EXPORT_SYMBOL(blk_set_stacking_limits);
@@ -208,6 +210,20 @@ void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
}
EXPORT_SYMBOL(blk_queue_max_write_zeroes_sectors);
+/**
+ * blk_queue_max_provision_sectors - set max sectors for a single provision
+ *
+ * @q: the request queue for the device
+ * @max_provision_sectors: maximum number of sectors to provision per command
+ **/
+
+void blk_queue_max_provision_sectors(struct request_queue *q,
+ unsigned int max_provision_sectors)
+{
+ q->limits.max_provision_sectors = max_provision_sectors;
+}
+EXPORT_SYMBOL(blk_queue_max_provision_sectors);
+
/**
* blk_queue_max_zone_append_sectors - set max sectors for a single zone append
* @q: the request queue for the device
@@ -578,6 +594,9 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
t->max_segment_size = min_not_zero(t->max_segment_size,
b->max_segment_size);
+ t->max_provision_sectors = min_not_zero(t->max_provision_sectors,
+ b->max_provision_sectors);
+
t->misaligned |= b->misaligned;
alignment = queue_limit_alignment_offset(b, start);
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index f1fce1c7fa44..202aa78f933e 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -132,6 +132,12 @@ static ssize_t queue_max_discard_segments_show(struct request_queue *q,
return queue_var_show(queue_max_discard_segments(q), page);
}
+static ssize_t queue_max_provision_sectors_show(struct request_queue *q,
+ char *page)
+{
+ return queue_var_show(queue_max_provision_sectors(q), (page));
+}
+
static ssize_t queue_max_integrity_segments_show(struct request_queue *q, char *page)
{
return queue_var_show(q->limits.max_integrity_segments, page);
@@ -599,6 +605,7 @@ QUEUE_RO_ENTRY(queue_io_min, "minimum_io_size");
QUEUE_RO_ENTRY(queue_io_opt, "optimal_io_size");
QUEUE_RO_ENTRY(queue_max_discard_segments, "max_discard_segments");
+QUEUE_RO_ENTRY(queue_max_provision_sectors, "max_provision_sectors");
QUEUE_RO_ENTRY(queue_discard_granularity, "discard_granularity");
QUEUE_RO_ENTRY(queue_discard_max_hw, "discard_max_hw_bytes");
QUEUE_RW_ENTRY(queue_discard_max, "discard_max_bytes");
@@ -648,6 +655,7 @@ static struct attribute *queue_attrs[] = {
&queue_max_sectors_entry.attr,
&queue_max_segments_entry.attr,
&queue_max_discard_segments_entry.attr,
+ &queue_max_provision_sectors_entry.attr,
&queue_max_integrity_segments_entry.attr,
&queue_max_segment_size_entry.attr,
&elv_iosched_entry.attr,
diff --git a/block/bounce.c b/block/bounce.c
index 7cfcb242f9a1..ab9d8723ae64 100644
--- a/block/bounce.c
+++ b/block/bounce.c
@@ -176,6 +176,7 @@ static struct bio *bounce_clone_bio(struct bio *bio_src)
case REQ_OP_DISCARD:
case REQ_OP_SECURE_ERASE:
case REQ_OP_WRITE_ZEROES:
+ case REQ_OP_PROVISION:
break;
default:
bio_for_each_segment(bv, bio_src, iter)
diff --git a/block/fops.c b/block/fops.c
index d2e6be4e3d1c..e1775269654a 100644
--- a/block/fops.c
+++ b/block/fops.c
@@ -611,9 +611,13 @@ static ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to)
return ret;
}
+#define BLKDEV_FALLOC_FL_TRUNCATE \
+ (FALLOC_FL_PUNCH_HOLE | FALLOC_FL_ZERO_RANGE | \
+ FALLOC_FL_NO_HIDE_STALE)
+
#define BLKDEV_FALLOC_FL_SUPPORTED \
- (FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE | \
- FALLOC_FL_ZERO_RANGE | FALLOC_FL_NO_HIDE_STALE)
+ (BLKDEV_FALLOC_FL_TRUNCATE | FALLOC_FL_KEEP_SIZE | \
+ FALLOC_FL_UNSHARE_RANGE)
static long blkdev_fallocate(struct file *file, int mode, loff_t start,
loff_t len)
@@ -625,7 +629,7 @@ static long blkdev_fallocate(struct file *file, int mode, loff_t start,
int error;
/* Fail if we don't recognize the flags. */
- if (mode & ~BLKDEV_FALLOC_FL_SUPPORTED)
+ if (mode != 0 && mode & ~BLKDEV_FALLOC_FL_SUPPORTED)
return -EOPNOTSUPP;
/* Don't go off the end of the device. */
@@ -649,11 +653,20 @@ static long blkdev_fallocate(struct file *file, int mode, loff_t start,
filemap_invalidate_lock(inode->i_mapping);
/* Invalidate the page cache, including dirty pages. */
- error = truncate_bdev_range(bdev, file->f_mode, start, end);
- if (error)
- goto fail;
+ if (mode & BLKDEV_FALLOC_FL_TRUNCATE) {
+ error = truncate_bdev_range(bdev, file->f_mode, start, end);
+ if (error)
+ goto fail;
+ }
switch (mode) {
+ case 0:
+ case FALLOC_FL_UNSHARE_RANGE:
+ case FALLOC_FL_KEEP_SIZE:
+ case FALLOC_FL_UNSHARE_RANGE | FALLOC_FL_KEEP_SIZE:
+ error = blkdev_issue_provision(bdev, start >> SECTOR_SHIFT,
+ len >> SECTOR_SHIFT, GFP_KERNEL);
+ break;
case FALLOC_FL_ZERO_RANGE:
case FALLOC_FL_ZERO_RANGE | FALLOC_FL_KEEP_SIZE:
error = blkdev_issue_zeroout(bdev, start >> SECTOR_SHIFT,
diff --git a/include/linux/bio.h b/include/linux/bio.h
index d766be7152e1..9820b3b039f2 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -57,7 +57,8 @@ static inline bool bio_has_data(struct bio *bio)
bio->bi_iter.bi_size &&
bio_op(bio) != REQ_OP_DISCARD &&
bio_op(bio) != REQ_OP_SECURE_ERASE &&
- bio_op(bio) != REQ_OP_WRITE_ZEROES)
+ bio_op(bio) != REQ_OP_WRITE_ZEROES &&
+ bio_op(bio) != REQ_OP_PROVISION)
return true;
return false;
@@ -67,7 +68,8 @@ static inline bool bio_no_advance_iter(const struct bio *bio)
{
return bio_op(bio) == REQ_OP_DISCARD ||
bio_op(bio) == REQ_OP_SECURE_ERASE ||
- bio_op(bio) == REQ_OP_WRITE_ZEROES;
+ bio_op(bio) == REQ_OP_WRITE_ZEROES ||
+ bio_op(bio) == REQ_OP_PROVISION;
}
static inline void *bio_data(struct bio *bio)
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index 99be590f952f..27bdf88f541c 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -385,7 +385,10 @@ enum req_op {
REQ_OP_DRV_IN = (__force blk_opf_t)34,
REQ_OP_DRV_OUT = (__force blk_opf_t)35,
- REQ_OP_LAST = (__force blk_opf_t)36,
+ /* request device to provision block */
+ REQ_OP_PROVISION = (__force blk_opf_t)37,
+
+ REQ_OP_LAST = (__force blk_opf_t)38,
};
enum req_flag_bits {
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 941304f17492..239e2f418b6e 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -303,6 +303,7 @@ struct queue_limits {
unsigned int discard_granularity;
unsigned int discard_alignment;
unsigned int zone_write_granularity;
+ unsigned int max_provision_sectors;
unsigned short max_segments;
unsigned short max_integrity_segments;
@@ -921,6 +922,8 @@ extern void blk_queue_max_discard_sectors(struct request_queue *q,
unsigned int max_discard_sectors);
extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
unsigned int max_write_same_sectors);
+extern void blk_queue_max_provision_sectors(struct request_queue *q,
+ unsigned int max_provision_sectors);
extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
extern void blk_queue_max_zone_append_sectors(struct request_queue *q,
unsigned int max_zone_append_sectors);
@@ -1060,6 +1063,9 @@ int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector,
sector_t nr_sects, gfp_t gfp);
+extern int blkdev_issue_provision(struct block_device *bdev, sector_t sector,
+ sector_t nr_sects, gfp_t gfp_mask);
+
#define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
#define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
@@ -1139,6 +1145,11 @@ static inline unsigned short queue_max_discard_segments(const struct request_que
return q->limits.max_discard_segments;
}
+static inline unsigned short queue_max_provision_sectors(const struct request_queue *q)
+{
+ return q->limits.max_provision_sectors;
+}
+
static inline unsigned int queue_max_segment_size(const struct request_queue *q)
{
return q->limits.max_segment_size;
@@ -1281,6 +1292,11 @@ static inline bool bdev_nowait(struct block_device *bdev)
return test_bit(QUEUE_FLAG_NOWAIT, &bdev_get_queue(bdev)->queue_flags);
}
+static inline unsigned int bdev_max_provision_sectors(struct block_device *bdev)
+{
+ return bdev_get_queue(bdev)->limits.max_provision_sectors;
+}
+
static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
{
return blk_queue_zoned_model(bdev_get_queue(bdev));
--
2.40.0.634.g4ca3ef3211-goog
dm-thinpool uses the provision request to provision
blocks for a dm-thin device. dm-thinpool currently does not
pass through REQ_OP_PROVISION to underlying devices.
For shared blocks, provision requests will break sharing and copy the
contents of the entire block. Additionally, if 'skip_block_zeroing'
is not set, dm-thin will opt to zero out the entire range as a part
of provisioning.
Signed-off-by: Sarthak Kukreti <[email protected]>
---
drivers/md/dm-thin.c | 73 +++++++++++++++++++++++++++++++++++++++++---
1 file changed, 68 insertions(+), 5 deletions(-)
diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index 2b13c949bd72..58d633f5c928 100644
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -274,6 +274,7 @@ struct pool {
process_bio_fn process_bio;
process_bio_fn process_discard;
+ process_bio_fn process_provision;
process_cell_fn process_cell;
process_cell_fn process_discard_cell;
@@ -913,7 +914,8 @@ static void __inc_remap_and_issue_cell(void *context,
struct bio *bio;
while ((bio = bio_list_pop(&cell->bios))) {
- if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD)
+ if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD ||
+ bio_op(bio) == REQ_OP_PROVISION)
bio_list_add(&info->defer_bios, bio);
else {
inc_all_io_entry(info->tc->pool, bio);
@@ -1245,8 +1247,8 @@ static int io_overlaps_block(struct pool *pool, struct bio *bio)
static int io_overwrites_block(struct pool *pool, struct bio *bio)
{
- return (bio_data_dir(bio) == WRITE) &&
- io_overlaps_block(pool, bio);
+ return (bio_data_dir(bio) == WRITE) && io_overlaps_block(pool, bio) &&
+ bio_op(bio) != REQ_OP_PROVISION;
}
static void save_and_set_endio(struct bio *bio, bio_end_io_t **save,
@@ -1891,7 +1893,8 @@ static void process_shared_bio(struct thin_c *tc, struct bio *bio,
if (bio_data_dir(bio) == WRITE && bio->bi_iter.bi_size) {
break_sharing(tc, bio, block, &key, lookup_result, data_cell);
- cell_defer_no_holder(tc, virt_cell);
+ if (bio_op(bio) != REQ_OP_PROVISION)
+ cell_defer_no_holder(tc, virt_cell);
} else {
struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
@@ -1953,6 +1956,51 @@ static void provision_block(struct thin_c *tc, struct bio *bio, dm_block_t block
}
}
+static void process_provision_bio(struct thin_c *tc, struct bio *bio)
+{
+ int r;
+ struct pool *pool = tc->pool;
+ dm_block_t block = get_bio_block(tc, bio);
+ struct dm_bio_prison_cell *cell;
+ struct dm_cell_key key;
+ struct dm_thin_lookup_result lookup_result;
+
+ /*
+ * If cell is already occupied, then the block is already
+ * being provisioned so we have nothing further to do here.
+ */
+ build_virtual_key(tc->td, block, &key);
+ if (bio_detain(pool, &key, bio, &cell))
+ return;
+
+ if (tc->requeue_mode) {
+ cell_requeue(pool, cell);
+ return;
+ }
+
+ r = dm_thin_find_block(tc->td, block, 1, &lookup_result);
+ switch (r) {
+ case 0:
+ if (lookup_result.shared) {
+ process_shared_bio(tc, bio, block, &lookup_result, cell);
+ } else {
+ bio_endio(bio);
+ cell_defer_no_holder(tc, cell);
+ }
+ break;
+ case -ENODATA:
+ provision_block(tc, bio, block, cell);
+ break;
+
+ default:
+ DMERR_LIMIT("%s: dm_thin_find_block() failed: error = %d",
+ __func__, r);
+ cell_defer_no_holder(tc, cell);
+ bio_io_error(bio);
+ break;
+ }
+}
+
static void process_cell(struct thin_c *tc, struct dm_bio_prison_cell *cell)
{
int r;
@@ -2228,6 +2276,8 @@ static void process_thin_deferred_bios(struct thin_c *tc)
if (bio_op(bio) == REQ_OP_DISCARD)
pool->process_discard(tc, bio);
+ else if (bio_op(bio) == REQ_OP_PROVISION)
+ pool->process_provision(tc, bio);
else
pool->process_bio(tc, bio);
@@ -2579,6 +2629,7 @@ static void set_pool_mode(struct pool *pool, enum pool_mode new_mode)
dm_pool_metadata_read_only(pool->pmd);
pool->process_bio = process_bio_fail;
pool->process_discard = process_bio_fail;
+ pool->process_provision = process_bio_fail;
pool->process_cell = process_cell_fail;
pool->process_discard_cell = process_cell_fail;
pool->process_prepared_mapping = process_prepared_mapping_fail;
@@ -2592,6 +2643,7 @@ static void set_pool_mode(struct pool *pool, enum pool_mode new_mode)
dm_pool_metadata_read_only(pool->pmd);
pool->process_bio = process_bio_read_only;
pool->process_discard = process_bio_success;
+ pool->process_provision = process_bio_fail;
pool->process_cell = process_cell_read_only;
pool->process_discard_cell = process_cell_success;
pool->process_prepared_mapping = process_prepared_mapping_fail;
@@ -2612,6 +2664,7 @@ static void set_pool_mode(struct pool *pool, enum pool_mode new_mode)
pool->out_of_data_space = true;
pool->process_bio = process_bio_read_only;
pool->process_discard = process_discard_bio;
+ pool->process_provision = process_bio_fail;
pool->process_cell = process_cell_read_only;
pool->process_prepared_mapping = process_prepared_mapping;
set_discard_callbacks(pool);
@@ -2628,6 +2681,7 @@ static void set_pool_mode(struct pool *pool, enum pool_mode new_mode)
dm_pool_metadata_read_write(pool->pmd);
pool->process_bio = process_bio;
pool->process_discard = process_discard_bio;
+ pool->process_provision = process_provision_bio;
pool->process_cell = process_cell;
pool->process_prepared_mapping = process_prepared_mapping;
set_discard_callbacks(pool);
@@ -2749,7 +2803,8 @@ static int thin_bio_map(struct dm_target *ti, struct bio *bio)
return DM_MAPIO_SUBMITTED;
}
- if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD) {
+ if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD ||
+ bio_op(bio) == REQ_OP_PROVISION) {
thin_defer_bio_with_throttle(tc, bio);
return DM_MAPIO_SUBMITTED;
}
@@ -3396,6 +3451,9 @@ static int pool_ctr(struct dm_target *ti, unsigned int argc, char **argv)
pt->adjusted_pf = pt->requested_pf = pf;
ti->num_flush_bios = 1;
ti->limit_swap_bios = true;
+ ti->num_provision_bios = 1;
+ ti->provision_supported = true;
+ ti->max_provision_granularity = true;
/*
* Only need to enable discards if the pool should pass
@@ -4288,6 +4346,9 @@ static int thin_ctr(struct dm_target *ti, unsigned int argc, char **argv)
ti->max_discard_granularity = true;
}
+ ti->num_provision_bios = 1;
+ ti->provision_supported = true;
+
mutex_unlock(&dm_thin_pool_table.mutex);
spin_lock_irq(&tc->pool->lock);
@@ -4502,6 +4563,8 @@ static void thin_io_hints(struct dm_target *ti, struct queue_limits *limits)
limits->discard_granularity = pool->sectors_per_block << SECTOR_SHIFT;
limits->max_discard_sectors = pool->sectors_per_block * BIO_PRISON_MAX_RANGE;
+
+ limits->max_provision_sectors = pool->sectors_per_block;
}
static struct target_type thin_target = {
--
2.40.0.634.g4ca3ef3211-goog
Add support for provision requests to loopback devices.
Loop devices will configure provision support based on
whether the underlying block device/file can support
the provision request and upon receiving a provision bio,
will map it to the backing device/storage. For loop devices
over files, a REQ_OP_PROVISION request will translate to
an fallocate mode 0 call on the backing file.
Signed-off-by: Sarthak Kukreti <[email protected]>
---
drivers/block/loop.c | 42 ++++++++++++++++++++++++++++++++++++++++++
1 file changed, 42 insertions(+)
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index bc31bb7072a2..13c4b4f8b9c1 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -327,6 +327,24 @@ static int lo_fallocate(struct loop_device *lo, struct request *rq, loff_t pos,
return ret;
}
+static int lo_req_provision(struct loop_device *lo, struct request *rq, loff_t pos)
+{
+ struct file *file = lo->lo_backing_file;
+ struct request_queue *q = lo->lo_queue;
+ int ret;
+
+ if (!q->limits.max_provision_sectors) {
+ ret = -EOPNOTSUPP;
+ goto out;
+ }
+
+ ret = file->f_op->fallocate(file, 0, pos, blk_rq_bytes(rq));
+ if (unlikely(ret && ret != -EINVAL && ret != -EOPNOTSUPP))
+ ret = -EIO;
+ out:
+ return ret;
+}
+
static int lo_req_flush(struct loop_device *lo, struct request *rq)
{
int ret = vfs_fsync(lo->lo_backing_file, 0);
@@ -488,6 +506,8 @@ static int do_req_filebacked(struct loop_device *lo, struct request *rq)
FALLOC_FL_PUNCH_HOLE);
case REQ_OP_DISCARD:
return lo_fallocate(lo, rq, pos, FALLOC_FL_PUNCH_HOLE);
+ case REQ_OP_PROVISION:
+ return lo_req_provision(lo, rq, pos);
case REQ_OP_WRITE:
if (cmd->use_aio)
return lo_rw_aio(lo, cmd, pos, ITER_SOURCE);
@@ -754,6 +774,25 @@ static void loop_sysfs_exit(struct loop_device *lo)
&loop_attribute_group);
}
+static void loop_config_provision(struct loop_device *lo)
+{
+ struct file *file = lo->lo_backing_file;
+ struct inode *inode = file->f_mapping->host;
+
+ /*
+ * If the backing device is a block device, mirror its provisioning
+ * capability.
+ */
+ if (S_ISBLK(inode->i_mode)) {
+ blk_queue_max_provision_sectors(lo->lo_queue,
+ bdev_max_provision_sectors(I_BDEV(inode)));
+ } else if (file->f_op->fallocate) {
+ blk_queue_max_provision_sectors(lo->lo_queue, UINT_MAX >> 9);
+ } else {
+ blk_queue_max_provision_sectors(lo->lo_queue, 0);
+ }
+}
+
static void loop_config_discard(struct loop_device *lo)
{
struct file *file = lo->lo_backing_file;
@@ -1092,6 +1131,7 @@ static int loop_configure(struct loop_device *lo, fmode_t mode,
blk_queue_io_min(lo->lo_queue, bsize);
loop_config_discard(lo);
+ loop_config_provision(lo);
loop_update_rotational(lo);
loop_update_dio(lo);
loop_sysfs_init(lo);
@@ -1304,6 +1344,7 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info)
}
loop_config_discard(lo);
+ loop_config_provision(lo);
/* update dio if lo_offset or transfer is changed */
__loop_update_dio(lo, lo->use_dio);
@@ -1830,6 +1871,7 @@ static blk_status_t loop_queue_rq(struct blk_mq_hw_ctx *hctx,
case REQ_OP_FLUSH:
case REQ_OP_DISCARD:
case REQ_OP_WRITE_ZEROES:
+ case REQ_OP_PROVISION:
cmd->use_aio = false;
break;
default:
--
2.40.0.634.g4ca3ef3211-goog
Add block provisioning support for device-mapper targets.
dm-crypt, dm-snap and dm-linear will, by default, passthrough
REQ_OP_PROVISION requests to the underlying device, if
supported.
Signed-off-by: Sarthak Kukreti <[email protected]>
---
drivers/md/dm-crypt.c | 5 ++++-
drivers/md/dm-linear.c | 2 ++
drivers/md/dm-snap.c | 8 ++++++++
drivers/md/dm-table.c | 23 +++++++++++++++++++++++
drivers/md/dm.c | 6 ++++++
include/linux/device-mapper.h | 17 +++++++++++++++++
6 files changed, 60 insertions(+), 1 deletion(-)
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index 8b47b913ee83..aa8072d6d7bf 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -3336,6 +3336,9 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
cc->tag_pool_max_sectors <<= cc->sector_shift;
}
+ ti->num_provision_bios = 1;
+ ti->provision_supported = true;
+
ret = -ENOMEM;
cc->io_queue = alloc_workqueue("kcryptd_io/%s", WQ_MEM_RECLAIM, 1, devname);
if (!cc->io_queue) {
@@ -3390,7 +3393,7 @@ static int crypt_map(struct dm_target *ti, struct bio *bio)
* - for REQ_OP_DISCARD caller must use flush if IO ordering matters
*/
if (unlikely(bio->bi_opf & REQ_PREFLUSH ||
- bio_op(bio) == REQ_OP_DISCARD)) {
+ bio_op(bio) == REQ_OP_DISCARD || bio_op(bio) == REQ_OP_PROVISION)) {
bio_set_dev(bio, cc->dev->bdev);
if (bio_sectors(bio))
bio->bi_iter.bi_sector = cc->start +
diff --git a/drivers/md/dm-linear.c b/drivers/md/dm-linear.c
index f4448d520ee9..66e50f5b0665 100644
--- a/drivers/md/dm-linear.c
+++ b/drivers/md/dm-linear.c
@@ -62,6 +62,8 @@ static int linear_ctr(struct dm_target *ti, unsigned int argc, char **argv)
ti->num_discard_bios = 1;
ti->num_secure_erase_bios = 1;
ti->num_write_zeroes_bios = 1;
+ ti->num_provision_bios = 1;
+ ti->provision_supported = true;
ti->private = lc;
return 0;
diff --git a/drivers/md/dm-snap.c b/drivers/md/dm-snap.c
index 9c49f53760d0..07927bb1e711 100644
--- a/drivers/md/dm-snap.c
+++ b/drivers/md/dm-snap.c
@@ -1358,6 +1358,8 @@ static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
if (s->discard_zeroes_cow)
ti->num_discard_bios = (s->discard_passdown_origin ? 2 : 1);
ti->per_io_data_size = sizeof(struct dm_snap_tracked_chunk);
+ ti->num_provision_bios = 1;
+ ti->provision_supported = true;
/* Add snapshot to the list of snapshots for this origin */
/* Exceptions aren't triggered till snapshot_resume() is called */
@@ -2003,6 +2005,11 @@ static int snapshot_map(struct dm_target *ti, struct bio *bio)
/* If the block is already remapped - use that, else remap it */
e = dm_lookup_exception(&s->complete, chunk);
if (e) {
+ if (unlikely(bio_op(bio) == REQ_OP_PROVISION)) {
+ bio_endio(bio);
+ r = DM_MAPIO_SUBMITTED;
+ goto out_unlock;
+ }
remap_exception(s, e, bio, chunk);
if (unlikely(bio_op(bio) == REQ_OP_DISCARD) &&
io_overlaps_chunk(s, bio)) {
@@ -2413,6 +2420,7 @@ static void snapshot_io_hints(struct dm_target *ti, struct queue_limits *limits)
/* All discards are split on chunk_size boundary */
limits->discard_granularity = snap->store->chunk_size;
limits->max_discard_sectors = snap->store->chunk_size;
+ limits->max_provision_sectors = snap->store->chunk_size;
up_read(&_origins_lock);
}
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 119db5e01080..9301f050529f 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -1854,6 +1854,26 @@ static bool dm_table_supports_write_zeroes(struct dm_table *t)
return true;
}
+static int device_provision_capable(struct dm_target *ti, struct dm_dev *dev,
+ sector_t start, sector_t len, void *data)
+{
+ return !bdev_max_provision_sectors(dev->bdev);
+}
+
+static bool dm_table_supports_provision(struct dm_table *t)
+{
+ for (unsigned int i = 0; i < t->num_targets; i++) {
+ struct dm_target *ti = dm_table_get_target(t, i);
+
+ if (ti->provision_supported ||
+ (ti->type->iterate_devices &&
+ ti->type->iterate_devices(ti, device_provision_capable, NULL)))
+ return true;
+ }
+
+ return false;
+}
+
static int device_not_nowait_capable(struct dm_target *ti, struct dm_dev *dev,
sector_t start, sector_t len, void *data)
{
@@ -1987,6 +2007,9 @@ int dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
if (!dm_table_supports_write_zeroes(t))
q->limits.max_write_zeroes_sectors = 0;
+ if (!dm_table_supports_provision(t))
+ q->limits.max_provision_sectors = 0;
+
dm_table_verify_integrity(t);
/*
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 3b694ba3a106..9b94121b8d38 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -1609,6 +1609,7 @@ static bool is_abnormal_io(struct bio *bio)
case REQ_OP_DISCARD:
case REQ_OP_SECURE_ERASE:
case REQ_OP_WRITE_ZEROES:
+ case REQ_OP_PROVISION:
return true;
default:
break;
@@ -1641,6 +1642,11 @@ static blk_status_t __process_abnormal_io(struct clone_info *ci,
if (ti->max_write_zeroes_granularity)
max_granularity = limits->max_write_zeroes_sectors;
break;
+ case REQ_OP_PROVISION:
+ num_bios = ti->num_provision_bios;
+ if (ti->max_provision_granularity)
+ max_granularity = limits->max_provision_sectors;
+ break;
default:
break;
}
diff --git a/include/linux/device-mapper.h b/include/linux/device-mapper.h
index a52d2b9a6846..9981378457d2 100644
--- a/include/linux/device-mapper.h
+++ b/include/linux/device-mapper.h
@@ -334,6 +334,12 @@ struct dm_target {
*/
unsigned int num_write_zeroes_bios;
+ /*
+ * The number of PROVISION bios that will be submitted to the target.
+ * The bio number can be accessed with dm_bio_get_target_bio_nr.
+ */
+ unsigned int num_provision_bios;
+
/*
* The minimum number of extra bytes allocated in each io for the
* target to use.
@@ -358,6 +364,11 @@ struct dm_target {
*/
bool discards_supported:1;
+ /* Set if this target needs to receive provision requests regardless of
+ * whether or not its underlying devices have support.
+ */
+ bool provision_supported:1;
+
/*
* Set if this target requires that discards be split on
* 'max_discard_sectors' boundaries.
@@ -376,6 +387,12 @@ struct dm_target {
*/
bool max_write_zeroes_granularity:1;
+ /*
+ * Set if this target requires that provisions be split on
+ * 'max_provision_sectors' boundaries.
+ */
+ bool max_provision_granularity:1;
+
/*
* Set if we need to limit the number of in-flight bios when swapping.
*/
--
2.40.0.634.g4ca3ef3211-goog
On Fri, Apr 14, 2023 at 2:58 PM Mike Snitzer <[email protected]> wrote:
>
> On Thu, Apr 13 2023 at 8:02P -0400,
> Sarthak Kukreti <[email protected]> wrote:
>
> > Add support to dm devices for REQ_OP_PROVISION. The default mode
> > is to passthrough the request to the underlying device, if it
> > supports it. dm-thinpool uses the provision request to provision
> > blocks for a dm-thin device. dm-thinpool currently does not
> > pass through REQ_OP_PROVISION to underlying devices.
> >
> > For shared blocks, provision requests will break sharing and copy the
> > contents of the entire block.
> >
> > Signed-off-by: Sarthak Kukreti <[email protected]>
> > ---
> > drivers/md/dm-crypt.c | 4 +-
> > drivers/md/dm-linear.c | 1 +
> > drivers/md/dm-snap.c | 7 +++
>
> Have you tested REQ_OP_PROVISION with these targets? Just want to
> make sure you have an explicit need (and vested interest) for them
> passing through REQ_OP_PROVISION.
>
Yes. I have a vested interest in dm-linear and dm-crypt; I kept
dm-snap support mostly for consistency with thin snapshots.
> > diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
> > index 2055a758541d..5985343384a7 100644
> > --- a/drivers/md/dm-table.c
> > +++ b/drivers/md/dm-table.c
> > @@ -1850,6 +1850,26 @@ static bool dm_table_supports_write_zeroes(struct dm_table *t)
> > return true;
> > }
> >
> > +static int device_provision_capable(struct dm_target *ti, struct dm_dev *dev,
> > + sector_t start, sector_t len, void *data)
> > +{
> > + return !bdev_max_provision_sectors(dev->bdev);
> > +}
> > +
> > +static bool dm_table_supports_provision(struct dm_table *t)
> > +{
> > + for (unsigned int i = 0; i < t->num_targets; i++) {
> > + struct dm_target *ti = dm_table_get_target(t, i);
> > +
> > + if (ti->provision_supported ||
> > + (ti->type->iterate_devices &&
> > + ti->type->iterate_devices(ti, device_provision_capable, NULL)))
> > + return true;
> > + }
> > +
> > + return false;
> > +}
> > +
> > static int device_not_nowait_capable(struct dm_target *ti, struct dm_dev *dev,
> > sector_t start, sector_t len, void *data)
> > {
> > @@ -1983,6 +2003,11 @@ int dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
> > if (!dm_table_supports_write_zeroes(t))
> > q->limits.max_write_zeroes_sectors = 0;
> >
> > + if (dm_table_supports_provision(t))
> > + blk_queue_max_provision_sectors(q, UINT_MAX >> 9);
>
> This doesn't seem correct in that it'll override whatever
> max_provision_sectors was set by a target (like thinp).
>
> I think you only need the if (!dm_table_supports_provision)) case:
>
Done
> > + else
> > + q->limits.max_provision_sectors = 0;
> > +
> > dm_table_verify_integrity(t);
> >
> > /*
> > diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
> > index 13d4677baafd..b08b7ae617be 100644
> > --- a/drivers/md/dm-thin.c
> > +++ b/drivers/md/dm-thin.c
>
> I think it'll make the most sense to split out the dm-thin.c changes
> in a separate patch.
>
Separated the dm-thin changes into a separate patch that follows this one in v4.
> > @@ -909,7 +909,8 @@ static void __inc_remap_and_issue_cell(void *context,
> > struct bio *bio;
> >
> > while ((bio = bio_list_pop(&cell->bios))) {
> > - if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD)
> > + if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD ||
> > + bio_op(bio) == REQ_OP_PROVISION)
> > bio_list_add(&info->defer_bios, bio);
> > else {
> > inc_all_io_entry(info->tc->pool, bio);
> > @@ -1013,6 +1014,15 @@ static void process_prepared_mapping(struct dm_thin_new_mapping *m)
> > goto out;
> > }
> >
> > + /*
> > + * For provision requests, once the prepared block has been inserted
> > + * into the mapping btree, return.
> > + */
> > + if (bio && bio_op(bio) == REQ_OP_PROVISION) {
> > + bio_endio(bio);
> > + return;
> > + }
> > +
> > /*
> > * Release any bios held while the block was being provisioned.
> > * If we are processing a write bio that completely covers the block,
> > @@ -1241,7 +1251,7 @@ static int io_overlaps_block(struct pool *pool, struct bio *bio)
> >
> > static int io_overwrites_block(struct pool *pool, struct bio *bio)
> > {
> > - return (bio_data_dir(bio) == WRITE) &&
> > + return (bio_data_dir(bio) == WRITE) && bio_op(bio) != REQ_OP_PROVISION &&
> > io_overlaps_block(pool, bio);
> > }
> >
> > @@ -1334,10 +1344,11 @@ static void schedule_copy(struct thin_c *tc, dm_block_t virt_block,
> > /*
> > * IO to pool_dev remaps to the pool target's data_dev.
> > *
> > - * If the whole block of data is being overwritten, we can issue the
> > - * bio immediately. Otherwise we use kcopyd to clone the data first.
> > + * If the whole block of data is being overwritten and if this is not a
> > + * provision request, we can issue the bio immediately.
> > + * Otherwise we use kcopyd to clone the data first.
> > */
> > - if (io_overwrites_block(pool, bio))
> > + if (io_overwrites_block(pool, bio) && bio_op(bio) != REQ_OP_PROVISION)
> > remap_and_issue_overwrite(tc, bio, data_dest, m);
> > else {
> > struct dm_io_region from, to;
> > @@ -1356,7 +1367,8 @@ static void schedule_copy(struct thin_c *tc, dm_block_t virt_block,
> > /*
> > * Do we need to zero a tail region?
> > */
> > - if (len < pool->sectors_per_block && pool->pf.zero_new_blocks) {
> > + if (len < pool->sectors_per_block && pool->pf.zero_new_blocks &&
> > + bio_op(bio) != REQ_OP_PROVISION) {
> > atomic_inc(&m->prepare_actions);
> > ll_zero(tc, m,
> > data_dest * pool->sectors_per_block + len,
> > @@ -1390,6 +1402,10 @@ static void schedule_zero(struct thin_c *tc, dm_block_t virt_block,
> > m->data_block = data_block;
> > m->cell = cell;
> >
> > + /* Provision requests are chained on the original bio. */
> > + if (bio && bio_op(bio) == REQ_OP_PROVISION)
> > + m->bio = bio;
> > +
> > /*
> > * If the whole block of data is being overwritten or we are not
> > * zeroing pre-existing data, we can issue the bio immediately.
> > @@ -1865,7 +1881,8 @@ static void process_shared_bio(struct thin_c *tc, struct bio *bio,
> >
> > if (bio_data_dir(bio) == WRITE && bio->bi_iter.bi_size) {
> > break_sharing(tc, bio, block, &key, lookup_result, data_cell);
> > - cell_defer_no_holder(tc, virt_cell);
> > + if (bio_op(bio) != REQ_OP_PROVISION)
> > + cell_defer_no_holder(tc, virt_cell);
> > } else {
> > struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
> >
>
> Not confident in the above changes given the request that we only
> handle REQ_OP_PROVISION one thinp block at a time. So I'll gloss over
> them for now.
>
Yeah, the majority of this got removed in v4. I added a check in
io_overwrites_block() to return false for all provision requests.
> > @@ -1982,6 +1999,73 @@ static void process_cell(struct thin_c *tc, struct dm_bio_prison_cell *cell)
> > }
> > }
> >
> > +static void process_provision_cell(struct thin_c *tc, struct dm_bio_prison_cell *cell)
> > +{
> > + int r;
> > + struct pool *pool = tc->pool;
> > + struct bio *bio = cell->holder;
> > + dm_block_t begin, end;
> > + struct dm_thin_lookup_result lookup_result;
> > +
> > + if (tc->requeue_mode) {
> > + cell_requeue(pool, cell);
> > + return;
> > + }
> > +
> > + get_bio_block_range(tc, bio, &begin, &end);
> > +
> > + while (begin != end) {
> > + r = ensure_next_mapping(pool);
> > + if (r)
> > + /* we did our best */
> > + return;
> > +
> > + r = dm_thin_find_block(tc->td, begin, 1, &lookup_result);
> > + switch (r) {
> > + case 0:
> > + if (lookup_result.shared)
> > + process_shared_bio(tc, bio, begin,
> > + &lookup_result, cell);
> > + begin++;
> > + break;
> > + case -ENODATA:
> > + bio_inc_remaining(bio);
> > + provision_block(tc, bio, begin, cell);
> > + begin++;
> > + break;
> > + default:
> > + DMERR_LIMIT(
> > + "%s: dm_thin_find_block() failed: error = %d",
> > + __func__, r);
> > + cell_defer_no_holder(tc, cell);
> > + bio_io_error(bio);
> > + begin++;
> > + break;
> > + }
> > + }
> > + bio_endio(bio);
> > + cell_defer_no_holder(tc, cell);
> > +}
> > +
> > +static void process_provision_bio(struct thin_c *tc, struct bio *bio)
> > +{
> > + dm_block_t begin, end;
> > + struct dm_cell_key virt_key;
> > + struct dm_bio_prison_cell *virt_cell;
> > +
> > + get_bio_block_range(tc, bio, &begin, &end);
> > + if (begin == end) {
> > + bio_endio(bio);
> > + return;
> > + }
>
> Like Joe mentioned, this pattern was fine for discards because they
> are advisory/optional. But we need to make sure we don't truncate
> REQ_OP_PROVISION -- so we need to round up if we partially bleed into
> the blocks to the left or right.
>
> BUT ranged REQ_OP_PROVISION support is for later, this can be dealt
> with more simply in that each REQ_OP_PROVISION will be handled a block
> at a time initially. SO you'll want to honor _all_ REQ_OP_PROVISION,
> never returning early.
>
Thanks. The next patch in the series has the simplified version. It
had a lot in common with process_bio() so there was a possibility for
merging the two code paths, but I opted to keep it like this to make
ranged handling and passdown support easier to implement.
> > +
> > + build_key(tc->td, VIRTUAL, begin, end, &virt_key);
> > + if (bio_detain(tc->pool, &virt_key, bio, &virt_cell))
> > + return;
> > +
> > + process_provision_cell(tc, virt_cell);
> > +}
> > +
> > static void process_bio(struct thin_c *tc, struct bio *bio)
> > {
> > struct pool *pool = tc->pool;
> > @@ -2202,6 +2286,8 @@ static void process_thin_deferred_bios(struct thin_c *tc)
> >
> > if (bio_op(bio) == REQ_OP_DISCARD)
> > pool->process_discard(tc, bio);
> > + else if (bio_op(bio) == REQ_OP_PROVISION)
> > + process_provision_bio(tc, bio);
>
> This should be pool->process_provision() (or ->process_provision_bio
> if you like). Point is, you need to be switching these methods
> if/when the pool_mode transitions in set_pool_mode().
>
Done
> > else
> > pool->process_bio(tc, bio);
> >
> > @@ -2723,7 +2809,8 @@ static int thin_bio_map(struct dm_target *ti, struct bio *bio)
> > return DM_MAPIO_SUBMITTED;
> > }
> >
> > - if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD) {
> > + if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD ||
> > + bio_op(bio) == REQ_OP_PROVISION) {
> > thin_defer_bio_with_throttle(tc, bio);
> > return DM_MAPIO_SUBMITTED;
> > }
> > @@ -3370,6 +3457,8 @@ static int pool_ctr(struct dm_target *ti, unsigned int argc, char **argv)
> > pt->adjusted_pf = pt->requested_pf = pf;
> > ti->num_flush_bios = 1;
> > ti->limit_swap_bios = true;
> > + ti->num_provision_bios = 1;
> > + ti->provision_supported = true;
> >
> > /*
> > * Only need to enable discards if the pool should pass
> > @@ -4068,6 +4157,7 @@ static void pool_io_hints(struct dm_target *ti, struct queue_limits *limits)
> > blk_limits_io_opt(limits, pool->sectors_per_block << SECTOR_SHIFT);
> > }
> >
> > +
>
> Please fix this extra whitespace damage.
>
Done
> > /*
> > * pt->adjusted_pf is a staging area for the actual features to use.
> > * They get transferred to the live pool in bind_control_target()
> > @@ -4261,6 +4351,9 @@ static int thin_ctr(struct dm_target *ti, unsigned int argc, char **argv)
> > ti->num_discard_bios = 1;
> > }
> >
> > + ti->num_provision_bios = 1;
> > + ti->provision_supported = true;
> > +
> > mutex_unlock(&dm_thin_pool_table.mutex);
> >
> > spin_lock_irq(&tc->pool->lock);
> > @@ -4475,6 +4568,7 @@ static void thin_io_hints(struct dm_target *ti, struct queue_limits *limits)
> >
> > limits->discard_granularity = pool->sectors_per_block << SECTOR_SHIFT;
> > limits->max_discard_sectors = 2048 * 1024 * 16; /* 16G */
> > + limits->max_provision_sectors = 2048 * 1024 * 16; /* 16G */
>
> Building on my previous reply, with suggested update to
> dm.c:__process_abnormal_io(), once you rebase on dm-6.4's dm-thin.c
> you'll want to instead:
>
> limits->max_provision_sectors = pool->sectors_per_block << SECTOR_SHIFT;
>
> And you'll want to drop any of the above code that deals with handling
> bio-prison range locking and processing of REQ_OP_PROVISION for
> multiple thinp blocks at once.
>
> Simple REQ_OP_PROVISION processing one thinp block at a time first and
> then we can worry about handling REQ_OP_PROVISION that span blocks
> later.
>
Thanks, done in v4.
> > static struct target_type thin_target = {
> > diff --git a/drivers/md/dm.c b/drivers/md/dm.c
> > index dfde0088147a..d8f1803062b7 100644
> > --- a/drivers/md/dm.c
> > +++ b/drivers/md/dm.c
> > @@ -1593,6 +1593,7 @@ static bool is_abnormal_io(struct bio *bio)
> > case REQ_OP_DISCARD:
> > case REQ_OP_SECURE_ERASE:
> > case REQ_OP_WRITE_ZEROES:
> > + case REQ_OP_PROVISION:
> > return true;
> > default:
> > break;
> > @@ -1617,6 +1618,9 @@ static blk_status_t __process_abnormal_io(struct clone_info *ci,
> > case REQ_OP_WRITE_ZEROES:
> > num_bios = ti->num_write_zeroes_bios;
> > break;
> > + case REQ_OP_PROVISION:
> > + num_bios = ti->num_provision_bios;
> > + break;
> > default:
> > break;
> > }
>
> Please be sure to include my suggested __process_abnormal_io change
> from my previous reply.
>
Done.
> > diff --git a/include/linux/device-mapper.h b/include/linux/device-mapper.h
> > index 7975483816e4..e9f687521ae6 100644
> > --- a/include/linux/device-mapper.h
> > +++ b/include/linux/device-mapper.h
> > @@ -334,6 +334,12 @@ struct dm_target {
> > */
> > unsigned int num_write_zeroes_bios;
> >
> > + /*
> > + * The number of PROVISION bios that will be submitted to the target.
> > + * The bio number can be accessed with dm_bio_get_target_bio_nr.
> > + */
> > + unsigned int num_provision_bios;
> > +
> > /*
> > * The minimum number of extra bytes allocated in each io for the
> > * target to use.
> > @@ -358,6 +364,11 @@ struct dm_target {
> > */
> > bool discards_supported:1;
> >
> > + /* Set if this target needs to receive provision requests regardless of
> > + * whether or not its underlying devices have support.
> > + */
> > + bool provision_supported:1;
> > +
> > /*
> > * Set if we need to limit the number of in-flight bios when swapping.
> > */
>
> You'll need to add max_provision_granularity bool too (as implied by
> the dm.c:__process_abnormal_io() change suggested in my first reply to
> this patch).
>
> I'm happy to wait for you to consume the v3 feedback we've provided so
> you can create a v4. I'm thinking I can base dm-thin.c's WRITE_ZEROES
> support ontop of your REQ_OP_PROVISION v4 changes -- they should be
> complementary.
>
Done. Thanks for the review and guidance!
Best
Sarthak
> Thanks,
> Mike
On 4/18/23 15:12, Sarthak Kukreti wrote:
> /* Fail if we don't recognize the flags. */
> - if (mode & ~BLKDEV_FALLOC_FL_SUPPORTED)
> + if (mode != 0 && mode & ~BLKDEV_FALLOC_FL_SUPPORTED)
> return -EOPNOTSUPP;
Is this change necessary? Doesn't (mode & ~BLKDEV_FALLOC_FL_SUPPORTED)
!= 0 imply that mode != 0?
Thanks,
Bart.
On Wed, Apr 19, 2023 at 4:21 PM Dave Chinner <[email protected]> wrote:
>
> On Wed, Apr 19, 2023 at 10:26:02AM -0700, Darrick J. Wong wrote:
> > On Wed, Apr 19, 2023 at 12:17:34PM -0400, Mike Snitzer wrote:
> > > (And obviously needs fixing independent of this patchset)
> > >
> > > Shouldn't mkfs first check that the underlying storage supports
> > > REQ_OP_PROVISION by verifying
> > > /sys/block/<dev>/queue/provision_max_bytes exists and is not 0?
> > > (Just saying, we need to add new features more defensively.. you just
> > > made the case based on this scenario's implications alone)
> >
> > Not for fallocate -- for regular files, there's no way to check if the
> > filesystem actually supports the operation requested other than to try
> > it and see if it succeeds. We probably should've defined a DRY_RUN flag
> > for that purpose back when it was introduced.
>
> That ignores the fact that fallocate() was never intended to
> guarantee it will work in all contexts - it's an advisory interface
> at it's most basic level. If the call succeeds, then great, it does
> what is says on the box, but if it fails then it should have no
> visible effect on user data at all and the application still needs
> to perform whatever modification it needed done in some other way.
>
> IOWs, calling it one a block device without first checking if the
> block device supports that fallocate operation is exactly how it is
> supposed to be used. If the kernel can't handle this gracefully
> without corrupting data, then that's a kernel bug and not an
> application problem.
>
> > For fallocate calls to block devices, yes, the program can check the
> > queue limits in sysfs if fstat says the supplied path is a block device,
> > but I don't know that most programs are that thorough. The fallocate(1)
> > CLI program does not check.
>
> Right. fallocate() was intended to just do the right thing without
> the application having to jump thrown an unknown number of hoops to
> determine if fallocate() can be used or not in the context it is
> executing in. The kernel implementation is supposed to abstract all that
> context-dependent behaviour away from the application; all the
> application has to do is implement the fallocate() fast path and a
> single generic "do the right thing the slow way" fallback when the
> fallocate() method it called is not supported...
>
I added a separate commit[1] to fix this so that we only
truncate_bdev_range() iff we are in a supported de-allocate mode call.
Subsequently, the REQ_OP_PROVISION patch is a bit simpler when rebased
on top.
[1] https://www.spinics.net/lists/kernel/msg4765688.html
Best
Sarthak
> -Dave.
> --
> Dave Chinner
> [email protected]