Re: [PATCH v7 5/5] loop: Add support for provision requests

From: Sarthak Kukreti
Date: Fri Oct 06 2023 - 21:30:11 EST

Next message: Sarthak Kukreti: "Re: [PATCH v7 0/5] Introduce provisioning primitives"
Previous message: Sarthak Kukreti: "[PATCH v8 4/5] dm: Add block provisioning support"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Mon, May 22, 2023 at 6:22 PM Darrick J. Wong <djwong@xxxxxxxxxx> wrote:
>
> On Mon, May 22, 2023 at 03:09:55PM -0700, Sarthak Kukreti wrote:
> > On Mon, May 22, 2023 at 9:37 AM Darrick J. Wong <djwong@xxxxxxxxxx> wrote:
> > >
> > > If someone calls fallocate(UNSHARE_RANGE) on a loop bdev, shouldn't
> > > there be a way to pass that through to the fallocate call to the backing
> > > file?
> > >
> > > --D
> > >
> >
> > Yeah, I think we could add a REQ_UNSHARE bit (similar to REQ_NOUNMAP) to pass down the intent to the backing file (and possibly beyond...).
> >
> > I took a stab at implementing it as a follow up patch so that there's
> > less review churn on the current series. If it looks good, I can add
> > it to the end of the series (or incorporate this into the existing
> > block and loop patches):
>
> It looks like a reasonable addition to the end of the series, assuming
> that filling holes in thinp devices is cheap but unsharing snapshot
> blocks is not.
>
> > From: Sarthak Kukreti <sarthakkukreti@xxxxxxxxxxxx>
> > Date: Mon, 22 May 2023 14:18:15 -0700
> > Subject: [PATCH] block: Pass unshare intent via REQ_OP_PROVISION
> >
> > Allow REQ_OP_PROVISION to pass in an extra REQ_UNSHARE bit to
> > annotate unshare requests to underlying layers. Layers that support
> > FALLOC_FL_UNSHARE will be able to use this as an indicator of which
> > fallocate() mode to use.
>
> > Signed-off-by: Sarthak Kukreti <sarthakkukreti@xxxxxxxxxxxx>
> > ---
> > block/blk-lib.c | 6 +++++-
> > block/fops.c | 6 +++++-
> > drivers/block/loop.c | 35 +++++++++++++++++++++++++++++------
> > include/linux/blk_types.h | 3 +++
> > include/linux/blkdev.h | 3 ++-
> > 5 files changed, 44 insertions(+), 9 deletions(-)
> >
> > diff --git a/block/blk-lib.c b/block/blk-lib.c
> > index 3cff5fb654f5..bea6f5a700b3 100644
> > --- a/block/blk-lib.c
> > +++ b/block/blk-lib.c
> > @@ -350,6 +350,7 @@ EXPORT_SYMBOL(blkdev_issue_secure_erase);
> > * @sector: start sector
> > * @nr_sects: number of sectors to provision
> > * @gfp_mask: memory allocation flags (for bio_alloc)
> > + * @flags: controls detailed behavior
> > *
> > * Description:
> > * Issues a provision request to the block device for the range of sectors.
> > @@ -357,7 +358,7 @@ EXPORT_SYMBOL(blkdev_issue_secure_erase);
> > * 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 nr_sects, gfp_t gfp, unsigned flags)
> > {
> > sector_t bs_mask = (bdev_logical_block_size(bdev) >> 9) - 1;
> > unsigned int max_sectors = bdev_max_provision_sectors(bdev);
> > @@ -380,6 +381,9 @@ int blkdev_issue_provision(struct block_device *bdev, sector_t sector,
> > bio->bi_iter.bi_sector = sector;
> > bio->bi_iter.bi_size = req_sects << SECTOR_SHIFT;
> >
> > + if (flags & BLKDEV_UNSHARE_RANGE)
>
> This is a provisioning flag, shouldn't this be ...
> BLKDEV_PROVISION_UNSHARE or something?
>
Done in v8, thanks!

> > + bio->bi_opf |= REQ_UNSHARE;
> > +
> > sector += req_sects;
> > nr_sects -= req_sects;
> > if (!nr_sects) {
> > diff --git a/block/fops.c b/block/fops.c
> > index be2e41f160bf..6848756f0557 100644
> > --- a/block/fops.c
> > +++ b/block/fops.c
> > @@ -659,7 +659,11 @@ static long blkdev_fallocate(struct file *file, int mode, loff_t start,
> > 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);
> > + len >> SECTOR_SHIFT, GFP_KERNEL,
> > + (mode &
> > + FALLOC_FL_UNSHARE_RANGE) ?
> > + BLKDEV_UNSHARE_RANGE :
> > + 0);
>
> You might want to do something about the six level indent here;
> Linus hates that.
>
Thanks for pointing it out, I switched it up a bit in v8 but it still
looks a bit weird to me.

Best
Sarthak

> --D
>
> > break;
> > case FALLOC_FL_ZERO_RANGE:
> > case FALLOC_FL_ZERO_RANGE | FALLOC_FL_KEEP_SIZE:
> > diff --git a/drivers/block/loop.c b/drivers/block/loop.c
> > index 7fe1a6629754..c844b145d666 100644
> > --- a/drivers/block/loop.c
> > +++ b/drivers/block/loop.c
> > @@ -306,6 +306,30 @@ static int lo_read_simple(struct loop_device *lo, struct request *rq,
> > return 0;
> > }
> >
> > +static bool validate_fallocate_mode(struct loop_device *lo, int mode)
> > +{
> > + bool ret = true;
> > +
> > + switch (mode) {
> > + case FALLOC_FL_PUNCH_HOLE:
> > + case FALLOC_FL_ZERO_RANGE:
> > + if (!bdev_max_discard_sectors(lo->lo_device))
> > + ret = false;
> > + break;
> > + case 0:
> > + case FALLOC_FL_UNSHARE_RANGE:
> > + if (!bdev_max_provision_sectors(lo->lo_device))
> > + ret = false;
> > + break;
> > +
> > + default:
> > + ret = false;
> > + }
> > +
> > + return ret;
> > +}
> > +
> > +
> > static int lo_fallocate(struct loop_device *lo, struct request *rq, loff_t pos,
> > int mode)
> > {
> > @@ -316,11 +340,7 @@ static int lo_fallocate(struct loop_device *lo, struct request *rq, loff_t pos,
> > struct file *file = lo->lo_backing_file;
> > int ret;
> >
> > - if (mode & (FALLOC_FL_PUNCH_HOLE | FALLOC_FL_ZERO_RANGE) &&
> > - !bdev_max_discard_sectors(lo->lo_device))
> > - return -EOPNOTSUPP;
> > -
> > - if (mode == 0 && !bdev_max_provision_sectors(lo->lo_device))
> > + if (!validate_fallocate_mode(lo, mode))
> > return -EOPNOTSUPP;
> >
> > mode |= FALLOC_FL_KEEP_SIZE;
> > @@ -493,7 +513,10 @@ static int do_req_filebacked(struct loop_device *lo, struct request *rq)
> > case REQ_OP_DISCARD:
> > return lo_fallocate(lo, rq, pos, FALLOC_FL_PUNCH_HOLE);
> > case REQ_OP_PROVISION:
> > - return lo_fallocate(lo, rq, pos, 0);
> > + return lo_fallocate(lo, rq, pos,
> > + (rq->cmd_flags & REQ_UNSHARE) ?
> > + FALLOC_FL_UNSHARE_RANGE :
> > + 0);
> > case REQ_OP_WRITE:
> > if (cmd->use_aio)
> > return lo_rw_aio(lo, cmd, pos, ITER_SOURCE);
> > diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
> > index b7bb0226fdee..1a536fd897cb 100644
> > --- a/include/linux/blk_types.h
> > +++ b/include/linux/blk_types.h
> > @@ -423,6 +423,8 @@ enum req_flag_bits {
> > */
> > /* for REQ_OP_WRITE_ZEROES: */
> > __REQ_NOUNMAP, /* do not free blocks when zeroing */
> > + /* for REQ_OP_PROVISION: */
> > + __REQ_UNSHARE, /* unshare blocks */
> >
> > __REQ_NR_BITS, /* stops here */
> > };
> > @@ -451,6 +453,7 @@ enum req_flag_bits {
> > #define REQ_FS_PRIVATE (__force blk_opf_t)(1ULL << __REQ_FS_PRIVATE)
> >
> > #define REQ_NOUNMAP (__force blk_opf_t)(1ULL << __REQ_NOUNMAP)
> > +#define REQ_UNSHARE (__force blk_opf_t)(1ULL << __REQ_UNSHARE)
> >
> > #define REQ_FAILFAST_MASK \
> > (REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT | REQ_FAILFAST_DRIVER)
> > diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
> > index 462ce586d46f..60c09b0d3fc9 100644
> > --- a/include/linux/blkdev.h
> > +++ b/include/linux/blkdev.h
> > @@ -1049,10 +1049,11 @@ 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);
> > + sector_t nr_sects, gfp_t gfp_mask, unsigned int flags);
> >
> > #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
> > #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
> > +#define BLKDEV_UNSHARE_RANGE (1 << 2) /* unshare range on provision */
> >
> > extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
> > sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
> > --
> > 2.39.2
> >

Next message: Sarthak Kukreti: "Re: [PATCH v7 0/5] Introduce provisioning primitives"
Previous message: Sarthak Kukreti: "[PATCH v8 4/5] dm: Add block provisioning support"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]