On Fri, Mar 31, 2023 at 5:26 AM Brian Foster <bfoster@xxxxxxxxxx> wrote:
>
> On Thu, Mar 30, 2023 at 05:30:22PM -0700, Sarthak Kukreti wrote:
> > On Thu, Jan 5, 2023 at 6:42 AM Brian Foster <bfoster@xxxxxxxxxx> wrote:
> > >
> > > On Thu, Dec 29, 2022 at 12:12:47AM -0800, Sarthak Kukreti wrote:
> > > > Add support to dm devices for REQ_OP_PROVISION. The default mode
> > > > is to pass through the request and dm-thin will utilize it to provision
> > > > blocks.
> > > >
> > > > Signed-off-by: Sarthak Kukreti <sarthakkukreti@xxxxxxxxxxxx>
> > > > ---
> > > > 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 | 90 ++++++++++++++++++++++++++++++++++-
> > > > drivers/md/dm.c | 4 ++
> > > > include/linux/device-mapper.h | 11 +++++
> > > > 7 files changed, 139 insertions(+), 3 deletions(-)
> > > >
> > > ...
> > > > diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
> > > > index 64cfcf46881d..ab3f1abfabaf 100644
> > > > --- a/drivers/md/dm-thin.c
> > > > +++ b/drivers/md/dm-thin.c
> > > ...
> > > > @@ -1980,6 +1992,70 @@ 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);
> > >
> > > Hi Sarthak,
> > >
> > > I think we discussed this before.. but remind me if/how we wanted to
> > > handle the case if the thin blocks are shared..? Would a provision op
> > > carry enough information to distinguish an FALLOC_FL_UNSHARE_RANGE
> > > request from upper layers to conditionally provision in that case?
> > >
> > I think that should depend on how the filesystem implements unsharing:
> > assuming that we use provision on first allocation, unsharing on xfs
> > should result in xfs calling REQ_OP_PROVISION on the newly allocated
> > blocks first. But for ext4, we'd fail UNSHARE_RANGE unless provision
> > (instead of noprovision, provision_on_alloc), in which case, we'd send
> > REQ_OP_PROVISION.
> >
>
> I think my question was unclear... It doesn't necessarily have much to
> do with the filesystem or associated provision policy. Since dm-thin can
> share blocks internally via snapshots, do you intend to support
> FL_UNSHARE_RANGE via blkdev_fallocate() and REQ_OP_PROVISION?
>
> If so, then presumably this wants an UNSHARE request flag to pair with
> REQ_OP_PROVISION. Also, the dm-thin code above needs to check whether an
> existing block it finds is shared and basically do whatever COW breaking
> is necessary during the PROVISION request.
>
> If not, why? And what is expected behavior if blkdev_fallocate() is
> called with FL_UNSHARE_RANGE?
>
I think the handling of REQ_OP_PROVISION by each snapshot target is
kind-of implicit:
- snapshot-origin: do nothing
- snapshot: send REQ_OP_PROVISION to the COW device
- snapshot-merge: send REQ_OP_PROVISION to the origin.
>From the thinpool's perspective, REQ_OP_PROVISION reuses the
provision_block() primitive to break sharing (there's a bug in the
below code, as you pointed out: case 0 should also call
provision_block() if the lookup result shows that this is a shared
block).
So, I think the provision op would carry enough information to
conditionally provision and copy the block. Are there other cases
where UNSHARE_RANGE would be useful?
Best
Sarthak
> Brian
>
> > Best
> > Sarthak
> >
> >
> > Sarthak
> >
> > > Brian
> > >
> > > > + switch (r) {
> > > > + case 0:
> > > > + 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;
> > > > @@ -2200,6 +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)
> > > > + process_provision_bio(tc, bio);
> > > > else
> > > > pool->process_bio(tc, bio);
> > > >
> > > > @@ -2716,7 +2794,8 @@ 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;
> > > > }
> > > > @@ -3355,6 +3434,8 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv)
> > > > pt->low_water_blocks = low_water_blocks;
> > > > pt->adjusted_pf = pt->requested_pf = pf;
> > > > ti->num_flush_bios = 1;
> > > > + ti->num_provision_bios = 1;
> > > > + ti->provision_supported = true;
> > > >
> > > > /*
> > > > * Only need to enable discards if the pool should pass
> > > > @@ -4053,6 +4134,7 @@ 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()
> > > > @@ -4243,6 +4325,9 @@ static int thin_ctr(struct dm_target *ti, unsigned 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);
> > > > @@ -4457,6 +4542,7 @@ 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 e1ea3a7bd9d9..4d19bae9da4a 100644
> > > > --- a/drivers/md/dm.c
> > > > +++ b/drivers/md/dm.c
> > > > @@ -1587,6 +1587,7 @@ 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;
> > > > @@ -1611,6 +1612,9 @@ 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 04c6acf7faaa..b4d97d5d75b8 100644
> > > > --- a/include/linux/device-mapper.h
> > > > +++ b/include/linux/device-mapper.h
> > > > @@ -333,6 +333,12 @@ struct dm_target {
> > > > */
> > > > unsigned 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 num_provision_bios;
> > > > +
> > > > /*
> > > > * The minimum number of extra bytes allocated in each io for the
> > > > * target to use.
> > > > @@ -357,6 +363,11 @@ 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.
> > > > */
> > > > --
> > > > 2.37.3
> > > >
> > >
> >
>
