Re: [PATCH v8 2/2] ublk: enable zoned storage support

From: Ming Lei
Date: Wed Jul 12 2023 - 11:33:36 EST


On Tue, Jul 11, 2023 at 09:23:53AM +0200, Andreas Hindborg wrote:
> From: Andreas Hindborg <a.hindborg@xxxxxxxxxxx>
>
> Add zoned storage support to ublk: report_zones and operations:
> - REQ_OP_ZONE_OPEN
> - REQ_OP_ZONE_CLOSE
> - REQ_OP_ZONE_FINISH
> - REQ_OP_ZONE_RESET
> - REQ_OP_ZONE_APPEND
>
> The zone append feature uses the `addr` field of `struct ublksrv_io_cmd` to
> communicate ALBA back to the kernel. Therefore ublk must be used with the
> user copy feature (UBLK_F_USER_COPY) for zoned storage support to be
> available. Without this feature, ublk will not allow zoned storage support.

Without one userspace implementation, it is hard to verify the kernel
change, so I write one zoned target example based on libublk-rs by
taking one day.

https://github.com/ming1/libublk-rs.git [1]

It is one zoned-ramdisk example, and should be easy to
add persistence zoned. All zoned commands are covered, except for
REQ_OP_ZONE_RESET_ALL.

How to create ublk-zoned:

cargo run --example zoned -- add -1 1024 #1024MB

So far it passes all blktests/zbd tests for external zoned device.

'mkfs.btrfs -O zoned' & mount & git clone linux_kernel in this btrfs
works.

So this patchset gets verified now, so far so good.

[1] not publish to crates.io yet, still need some cleanup

>
> Signed-off-by: Andreas Hindborg <a.hindborg@xxxxxxxxxxx>
> ---
> drivers/block/ublk_drv.c | 314 ++++++++++++++++++++++++++++++++--
> include/uapi/linux/ublk_cmd.h | 36 +++-
> 2 files changed, 331 insertions(+), 19 deletions(-)
>
> diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
> index 8d271901efac..a57be961b708 100644
> --- a/drivers/block/ublk_drv.c
> +++ b/drivers/block/ublk_drv.c
> @@ -56,16 +56,21 @@
> | UBLK_F_USER_RECOVERY_REISSUE \
> | UBLK_F_UNPRIVILEGED_DEV \
> | UBLK_F_CMD_IOCTL_ENCODE \
> - | UBLK_F_USER_COPY)
> + | UBLK_F_USER_COPY \
> + | UBLK_F_ZONED)
>
> /* All UBLK_PARAM_TYPE_* should be included here */
> -#define UBLK_PARAM_TYPE_ALL (UBLK_PARAM_TYPE_BASIC | \
> - UBLK_PARAM_TYPE_DISCARD | UBLK_PARAM_TYPE_DEVT)
> +#define UBLK_PARAM_TYPE_ALL \
> + (UBLK_PARAM_TYPE_BASIC | UBLK_PARAM_TYPE_DISCARD | \
> + UBLK_PARAM_TYPE_DEVT | UBLK_PARAM_TYPE_ZONED)
>
> struct ublk_rq_data {
> struct llist_node node;
>
> struct kref ref;
> + __u64 sector;
> + __u32 operation;
> + __u32 nr_sectors;
> };
>
> struct ublk_uring_cmd_pdu {
> @@ -110,6 +115,11 @@ struct ublk_uring_cmd_pdu {
> */
> #define UBLK_IO_FLAG_NEED_GET_DATA 0x08
>
> +/*
> + * Set when IO is Zone Append
> + */
> +#define UBLK_IO_FLAG_ZONE_APPEND 0x10
> +
> struct ublk_io {
> /* userspace buffer address from io cmd */
> __u64 addr;
> @@ -190,6 +200,198 @@ static inline bool ublk_dev_is_user_copy(const struct ublk_device *ub)
> return ub->dev_info.flags & UBLK_F_USER_COPY;
> }
>
> +static inline bool ublk_dev_is_zoned(const struct ublk_device *ub)
> +{
> + return ub->dev_info.flags & UBLK_F_ZONED;
> +}
> +
> +static inline bool ublk_queue_is_zoned(struct ublk_queue *ubq)
> +{
> + return ubq->flags & UBLK_F_ZONED;
> +}
> +
> +#ifdef CONFIG_BLK_DEV_ZONED
> +
> +static int ublk_get_nr_zones(const struct ublk_device *ub)
> +{
> + const struct ublk_param_basic *p = &ub->params.basic;
> +
> + /* Zone size is a power of 2 */
> + return p->dev_sectors >> ilog2(p->chunk_sectors);
> +}
> +
> +static int ublk_revalidate_disk_zones(struct ublk_device *ub)
> +{
> + return blk_revalidate_disk_zones(ub->ub_disk, NULL);
> +}
> +
> +static int ublk_dev_param_zoned_validate(const struct ublk_device *ub)
> +{
> + const struct ublk_param_zoned *p = &ub->params.zoned;
> + int nr_zones;
> +
> + if (!ublk_dev_is_zoned(ub))
> + return -EINVAL;
> +
> + if (!p->max_zone_append_sectors)
> + return -EINVAL;
> +
> + nr_zones = ublk_get_nr_zones(ub);
> +
> + if (p->max_active_zones > nr_zones)
> + return -EINVAL;
> +
> + if (p->max_open_zones > nr_zones)
> + return -EINVAL;
> +
> + return 0;
> +}
> +
> +static int ublk_dev_param_zoned_apply(struct ublk_device *ub)
> +{
> + const struct ublk_param_zoned *p = &ub->params.zoned;
> +
> + disk_set_zoned(ub->ub_disk, BLK_ZONED_HM);
> + blk_queue_required_elevator_features(ub->ub_disk->queue,
> + ELEVATOR_F_ZBD_SEQ_WRITE);
> + disk_set_max_active_zones(ub->ub_disk, p->max_active_zones);
> + disk_set_max_open_zones(ub->ub_disk, p->max_open_zones);
> + blk_queue_max_zone_append_sectors(ub->ub_disk->queue, p->max_zone_append_sectors);
> +
> + ub->ub_disk->nr_zones = ublk_get_nr_zones(ub);
> +
> + return 0;
> +}
> +
> +/* Based on virtblk_alloc_report_buffer */
> +static void *ublk_alloc_report_buffer(struct ublk_device *ublk,
> + unsigned int nr_zones, size_t *buflen)
> +{
> + struct request_queue *q = ublk->ub_disk->queue;
> + size_t bufsize;
> + void *buf;
> +
> + nr_zones = min_t(unsigned int, nr_zones,
> + ublk->ub_disk->nr_zones);
> +
> + bufsize = nr_zones * sizeof(struct blk_zone);
> + bufsize =
> + min_t(size_t, bufsize, queue_max_hw_sectors(q) << SECTOR_SHIFT);
> + bufsize = min_t(size_t, bufsize, queue_max_segments(q) << PAGE_SHIFT);

'bufsize' shouldn't related with queue_max_segments() given we don't
enable segment limit yet.

> +
> + while (bufsize >= sizeof(struct blk_zone)) {
> + buf = __vmalloc(bufsize, GFP_KERNEL | __GFP_NORETRY);
> + if (buf) {
> + *buflen = bufsize;
> + return buf;
> + }
> + bufsize >>= 1;
> + }
> +
> + *buflen = 0;
> + return NULL;
> +}
> +
> +static int ublk_report_zones(struct gendisk *disk, sector_t sector,
> + unsigned int nr_zones, report_zones_cb cb, void *data)
> +{
> + struct ublk_device *ub = disk->private_data;
> + unsigned int zone_size_sectors = disk->queue->limits.chunk_sectors;
> + unsigned int first_zone = sector >> ilog2(zone_size_sectors);
> + unsigned int done_zones = 0;
> + unsigned int max_zones_per_request;
> + int ret;
> + struct blk_zone *buffer;
> + size_t buffer_length;
> +
> + nr_zones = min_t(unsigned int, ub->ub_disk->nr_zones - first_zone,
> + nr_zones);
> +
> + buffer = ublk_alloc_report_buffer(ub, nr_zones, &buffer_length);
> + if (!buffer)
> + return -ENOMEM;
> +
> + max_zones_per_request = buffer_length / sizeof(struct blk_zone);
> +
> + while (done_zones < nr_zones) {
> + unsigned int remaining_zones = nr_zones - done_zones;
> + unsigned int zones_in_request =
> + min_t(unsigned int, remaining_zones, max_zones_per_request);
> + struct request *req;
> + struct ublk_rq_data *pdu;
> + blk_status_t status;
> +
> + memset(buffer, 0, buffer_length);
> +
> + req = blk_mq_alloc_request(disk->queue, REQ_OP_DRV_IN, 0);
> + if (IS_ERR(req)) {
> + ret = PTR_ERR(req);
> + goto out;
> + }
> +
> + pdu = blk_mq_rq_to_pdu(req);
> + pdu->operation = UBLK_IO_OP_REPORT_ZONES;
> + pdu->sector = sector;
> + pdu->nr_sectors = remaining_zones * zone_size_sectors;

pdu->nr_sectors will be set to ublksrv_io_desc->nr_sectors as part of
UAPI, I feel it is hard to understand to pass 'remaining_zones * zone_size_sectors',
probably it is fine to re-define as buffer size or requested
nr_blk_zone, or other ideas?

> +
> + ret = blk_rq_map_kern(disk->queue, req, buffer, buffer_length,
> + GFP_KERNEL);
> + if (ret) {
> + blk_mq_free_request(req);
> + goto out;
> + }
> +
> + status = blk_execute_rq(req, 0);
> + ret = blk_status_to_errno(status);
> + blk_mq_free_request(req);
> + if (ret)
> + goto out;
> +
> + for (unsigned int i = 0; i < zones_in_request; i++) {
> + struct blk_zone *zone = buffer + i;
> +
> + /* A zero length zone means no more zones in this response */
> + if (!zone->len)
> + break;
> +
> + ret = cb(zone, i, data);
> + if (ret)
> + goto out;
> +
> + done_zones++;
> + sector += zone_size_sectors;
> +
> + }
> + }
> +
> + ret = done_zones;
> +
> +out:
> + kvfree(buffer);
> + return ret;
> +}
> +
> +#else
> +
> +#define ublk_report_zones (NULL)
> +
> +static int ublk_dev_param_zoned_validate(const struct ublk_device *ub)
> +{
> + return -EOPNOTSUPP;
> +}
> +
> +static int ublk_dev_param_zoned_apply(struct ublk_device *ub)
> +{
> + return -EOPNOTSUPP;
> +}
> +
> +static int ublk_revalidate_disk_zones(struct ublk_device *ub)
> +{
> + return 0;
> +}
> +
> +#endif
> +
> static inline void __ublk_complete_rq(struct request *req);
> static void ublk_complete_rq(struct kref *ref);
>
> @@ -232,7 +434,7 @@ static inline unsigned ublk_pos_to_tag(loff_t pos)
> UBLK_TAG_BITS_MASK;
> }
>
> -static void ublk_dev_param_basic_apply(struct ublk_device *ub)
> +static int ublk_dev_param_basic_apply(struct ublk_device *ub)
> {
> struct request_queue *q = ub->ub_disk->queue;
> const struct ublk_param_basic *p = &ub->params.basic;
> @@ -257,6 +459,8 @@ static void ublk_dev_param_basic_apply(struct ublk_device *ub)
> set_disk_ro(ub->ub_disk, true);
>
> set_capacity(ub->ub_disk, p->dev_sectors);
> +
> + return 0;
> }
>
> static void ublk_dev_param_discard_apply(struct ublk_device *ub)
> @@ -286,6 +490,9 @@ static int ublk_validate_params(const struct ublk_device *ub)
>
> if (p->max_sectors > (ub->dev_info.max_io_buf_bytes >> 9))
> return -EINVAL;
> +
> + if (ublk_dev_is_zoned(ub) && !p->chunk_sectors)
> + return -EINVAL;
> } else
> return -EINVAL;
>
> @@ -304,19 +511,31 @@ static int ublk_validate_params(const struct ublk_device *ub)
> if (ub->params.types & UBLK_PARAM_TYPE_DEVT)
> return -EINVAL;
>
> + if (ub->params.types & UBLK_PARAM_TYPE_ZONED)
> + return ublk_dev_param_zoned_validate(ub);
> + else if (ublk_dev_is_zoned(ub))
> + return -EINVAL;
> +
> return 0;
> }
>
> static int ublk_apply_params(struct ublk_device *ub)
> {
> + int ret;
> +
> if (!(ub->params.types & UBLK_PARAM_TYPE_BASIC))
> return -EINVAL;
>
> - ublk_dev_param_basic_apply(ub);
> + ret = ublk_dev_param_basic_apply(ub);
> + if (ret)
> + return ret;
>
> if (ub->params.types & UBLK_PARAM_TYPE_DISCARD)
> ublk_dev_param_discard_apply(ub);
>
> + if (ub->params.types & UBLK_PARAM_TYPE_ZONED)
> + return ublk_dev_param_zoned_apply(ub);
> +
> return 0;
> }
>
> @@ -487,6 +706,7 @@ static const struct block_device_operations ub_fops = {
> .owner = THIS_MODULE,
> .open = ublk_open,
> .free_disk = ublk_free_disk,
> + .report_zones = ublk_report_zones,
> };
>
> #define UBLK_MAX_PIN_PAGES 32
> @@ -601,7 +821,8 @@ static inline bool ublk_need_map_req(const struct request *req)
>
> static inline bool ublk_need_unmap_req(const struct request *req)
> {
> - return ublk_rq_has_data(req) && req_op(req) == REQ_OP_READ;
> + return ublk_rq_has_data(req) &&
> + (req_op(req) == REQ_OP_READ || req_op(req) == REQ_OP_DRV_IN);
> }
>
> static int ublk_map_io(const struct ublk_queue *ubq, const struct request *req,
> @@ -685,8 +906,14 @@ static blk_status_t ublk_setup_iod(struct ublk_queue *ubq, struct request *req)
> {
> struct ublksrv_io_desc *iod = ublk_get_iod(ubq, req->tag);
> struct ublk_io *io = &ubq->ios[req->tag];
> + struct ublk_rq_data *pdu = blk_mq_rq_to_pdu(req);
> + enum req_op op = req_op(req);
> u32 ublk_op;
>
> + if (!ublk_queue_is_zoned(ubq) &&
> + (op_is_zone_mgmt(op) || op == REQ_OP_ZONE_APPEND))
> + return -EIO;
> +
> switch (req_op(req)) {
> case REQ_OP_READ:
> ublk_op = UBLK_IO_OP_READ;
> @@ -703,6 +930,37 @@ static blk_status_t ublk_setup_iod(struct ublk_queue *ubq, struct request *req)
> case REQ_OP_WRITE_ZEROES:
> ublk_op = UBLK_IO_OP_WRITE_ZEROES;
> break;
> + case REQ_OP_ZONE_OPEN:
> + ublk_op = UBLK_IO_OP_ZONE_OPEN;
> + break;
> + case REQ_OP_ZONE_CLOSE:
> + ublk_op = UBLK_IO_OP_ZONE_CLOSE;
> + break;
> + case REQ_OP_ZONE_FINISH:
> + ublk_op = UBLK_IO_OP_ZONE_FINISH;
> + break;
> + case REQ_OP_ZONE_RESET:
> + ublk_op = UBLK_IO_OP_ZONE_RESET;
> + break;
> + case REQ_OP_DRV_IN:
> + ublk_op = pdu->operation;
> + switch (ublk_op) {
> + case UBLK_IO_OP_REPORT_ZONES:
> + iod->op_flags = ublk_op | ublk_req_build_flags(req);
> + iod->nr_sectors = pdu->nr_sectors;
> + iod->start_sector = pdu->sector;
> + return BLK_STS_OK;
> + default:
> + return BLK_STS_IOERR;
> + }
> + case REQ_OP_ZONE_APPEND:
> + ublk_op = UBLK_IO_OP_ZONE_APPEND;
> + io->flags |= UBLK_IO_FLAG_ZONE_APPEND;
> + break;
> + case REQ_OP_ZONE_RESET_ALL:
> + case REQ_OP_DRV_OUT:
> + /* We do not support reset_all and drv_out */
> + return BLK_STS_NOTSUPP;

I'd suggest to move all zoned handling & the above check into one helper,
such as:

default:
if (ublk_queue_is_zoned())
return ublk_zoned_setup_iod();

Then zoned check can be avoided for other generic OPs, and zoned patch
conflict could be avoided usually.

Anyway, this version looks much better, and I think we are close to
go.

Thanks,
Ming