block: ublk: enable zoned storage support
Commit Message
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
This allows implementation of zoned storage devices in user space. An
example user space implementation based on ubdsrv is available [1].
[1] https://github.com/metaspace/ubdsrv/commit/14a2b708f74f70cfecb076d92e680dc718cc1f6d
Signed-off-by: Andreas Hindborg <a.hindborg@samsung.com>
---
drivers/block/ublk_drv.c | 145 ++++++++++++++++++++++++++++++++--
include/uapi/linux/ublk_cmd.h | 18 +++++
2 files changed, 157 insertions(+), 6 deletions(-)
Comments
Hello Andreas,
On Fri, Feb 24, 2023 at 01:59:50PM +0100, Andreas Hindborg wrote:
> 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
>
> This allows implementation of zoned storage devices in user space. An
> example user space implementation based on ubdsrv is available [1].
>
> [1] https://github.com/metaspace/ubdsrv/commit/14a2b708f74f70cfecb076d92e680dc718cc1f6d
>
> Signed-off-by: Andreas Hindborg <a.hindborg@samsung.com>
> ---
Did you try to build this patch with CONFIG_BLK_DEV_ZONED disabled?
I got the following build errors when building it on top of vanilla v6.2
when CONFIG_BLK_DEV_ZONED is disabled:
drivers/block/ublk_drv.c: In function ‘ublk_dev_param_basic_apply’:
drivers/block/ublk_drv.c:221:28: error: ‘struct gendisk’ has no member named ‘nr_zones’
221 | ub->ub_disk->nr_zones = p->dev_sectors / p->chunk_sectors;
| ^~
drivers/block/ublk_drv.c: In function ‘ublk_dev_param_zoned_apply’:
drivers/block/ublk_drv.c:244:17: error: implicit declaration of function ‘disk_set_max_active_zones’; did you mean ‘bdev_max_active_zones’? [-Werror=implicit-function-declaration]
244 | disk_set_max_active_zones(ub->ub_disk, p->max_active_zones);
| ^~~~~~~~~~~~~~~~~~~~~~~~~
| bdev_max_active_zones
drivers/block/ublk_drv.c:245:17: error: implicit declaration of function ‘disk_set_max_open_zones’; did you mean ‘bdev_max_open_zones’? [-Werror=implicit-function-declaration]
245 | disk_set_max_open_zones(ub->ub_disk, p->max_open_zones);
> drivers/block/ublk_drv.c | 145 ++++++++++++++++++++++++++++++++--
> include/uapi/linux/ublk_cmd.h | 18 +++++
> 2 files changed, 157 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
> index 6368b56eacf1..9741b4b1647b 100644
> --- a/drivers/block/ublk_drv.c
> +++ b/drivers/block/ublk_drv.c
> @@ -20,6 +20,7 @@
> #include <linux/major.h>
> #include <linux/wait.h>
> #include <linux/blkdev.h>
> +#include <linux/blkzoned.h>
> #include <linux/init.h>
> #include <linux/swap.h>
> #include <linux/slab.h>
> @@ -51,10 +52,12 @@
> | UBLK_F_URING_CMD_COMP_IN_TASK \
> | UBLK_F_NEED_GET_DATA \
> | UBLK_F_USER_RECOVERY \
> - | UBLK_F_USER_RECOVERY_REISSUE)
> + | UBLK_F_USER_RECOVERY_REISSUE \
> + | UBLK_F_ZONED)
>
> /* All UBLK_PARAM_TYPE_* should be included here */
> -#define UBLK_PARAM_TYPE_ALL (UBLK_PARAM_TYPE_BASIC | UBLK_PARAM_TYPE_DISCARD)
> +#define UBLK_PARAM_TYPE_ALL (UBLK_PARAM_TYPE_BASIC | UBLK_PARAM_TYPE_DISCARD \
> + | UBLK_PARAM_TYPE_ZONED)
>
> struct ublk_rq_data {
> struct llist_node node;
> @@ -212,6 +215,11 @@ 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);
> +
> + if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
> + ub->dev_info.flags & UBLK_F_ZONED && p->chunk_sectors) {
> + ub->ub_disk->nr_zones = p->dev_sectors / p->chunk_sectors;
> + }
Kernel coding style says to not unnecessarily use braces where a single
statement will do:
https://www.kernel.org/doc/html/latest/process/coding-style.html#placing-braces-and-spaces
> }
>
> static void ublk_dev_param_discard_apply(struct ublk_device *ub)
> @@ -227,6 +235,19 @@ static void ublk_dev_param_discard_apply(struct ublk_device *ub)
> blk_queue_max_discard_segments(q, p->max_discard_segments);
> }
>
> +static void ublk_dev_param_zoned_apply(struct ublk_device *ub)
> +{
> + const struct ublk_param_zoned *p = &ub->params.zoned;
> +
> + if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
> + ub->dev_info.flags & UBLK_F_ZONED) {
> + disk_set_max_active_zones(ub->ub_disk, p->max_active_zones);
> + disk_set_max_open_zones(ub->ub_disk, p->max_open_zones);
> + /* We do not support zone append yet */
> + //blk_queue_max_zone_append_sectors(q, zone_size);
Remove commented out code.
It can be added once that feature is added.
> + }
> +}
> +
> static int ublk_validate_params(const struct ublk_device *ub)
> {
> /* basic param is the only one which must be set */
> @@ -268,6 +289,9 @@ static int ublk_apply_params(struct ublk_device *ub)
> if (ub->params.types & UBLK_PARAM_TYPE_DISCARD)
> ublk_dev_param_discard_apply(ub);
>
> + if (ub->params.types & UBLK_PARAM_TYPE_ZONED)
> + ublk_dev_param_zoned_apply(ub);
> +
> return 0;
> }
>
> @@ -361,9 +385,18 @@ static void ublk_free_disk(struct gendisk *disk)
> put_device(&ub->cdev_dev);
> }
>
> +#if IS_ENABLED(CONFIG_BLK_DEV_ZONED)
Here you use "#if IS_ENABLED(CONFIG_BLK_DEV_ZONED)"
yet further down you use
"#ifdef CONFIG_BLK_DEV_ZONED".
IS_ENABLED() is usually used when testing for a Kconfig within
a normal C conditional, e.g. if (IS_ENABLED()), where possible,
in order to avoid ugly C preprocessor conditionals completely.
However, if you have to use C preprocessor conditionals, then
#ifdef CONFIG_SOMETHING
is the most common way to conditionally include something.
> +static int ublk_report_zones(struct gendisk *disk, sector_t sector,
> + unsigned int nr_zones, report_zones_cb cb,
> + void *data);
> +#endif
> +
> static const struct block_device_operations ub_fops = {
> - .owner = THIS_MODULE,
> - .free_disk = ublk_free_disk,
> + .owner = THIS_MODULE,
> + .free_disk = ublk_free_disk,
> +#if IS_ENABLED(CONFIG_BLK_DEV_ZONED)
> + .report_zones = ublk_report_zones,
> +#endif
> };
>
> #define UBLK_MAX_PIN_PAGES 32
> @@ -499,7 +532,7 @@ static int ublk_unmap_io(const struct ublk_queue *ubq,
> {
> const unsigned int rq_bytes = blk_rq_bytes(req);
>
> - if (req_op(req) == REQ_OP_READ && ublk_rq_has_data(req)) {
> + if ((req_op(req) == REQ_OP_READ || req_op(req) == REQ_OP_DRV_IN) && ublk_rq_has_data(req)) {
> struct ublk_map_data data = {
> .ubq = ubq,
> .rq = req,
> @@ -566,6 +599,26 @@ 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;
> +#ifdef CONFIG_BLK_DEV_ZONED
> + 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 = UBLK_IO_OP_DRV_IN;
> + break;
> + case REQ_OP_ZONE_APPEND:
> + /* We do not support zone append yet */
> + fallthrough;
> +#endif
> default:
> return BLK_STS_IOERR;
> }
> @@ -612,7 +665,8 @@ static void ublk_complete_rq(struct request *req)
> *
> * Both the two needn't unmap.
> */
> - if (req_op(req) != REQ_OP_READ && req_op(req) != REQ_OP_WRITE) {
> + if (req_op(req) != REQ_OP_READ && req_op(req) != REQ_OP_WRITE &&
> + req_op(req) != REQ_OP_DRV_IN) {
> blk_mq_end_request(req, BLK_STS_OK);
> return;
> }
> @@ -1493,6 +1547,73 @@ static struct ublk_device *ublk_get_device_from_id(int idx)
> return ub;
> }
>
> +#ifdef CONFIG_BLK_DEV_ZONED
> +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;
> + unsigned int zone_size;
> + unsigned int first_zone;
> + int ret = 0;
> +
> + ub = disk->private_data;
> +
> + if (!(ub->dev_info.flags & UBLK_F_ZONED))
> + return -EINVAL;
> +
> + zone_size = disk->queue->limits.chunk_sectors;
> + first_zone = sector >> ilog2(zone_size);
> +
> + nr_zones = min(ub->ub_disk->nr_zones - first_zone, nr_zones);
> +
> + for (unsigned int i = 0; i < nr_zones; i++) {
> +
No need for empty line here.
> + struct request *req;
> + blk_status_t status;
> + struct blk_zone info;
Since this is allocated on the stack, does it need memset()?
null_blk seems to memset blk_zone before the loop, but not inside the loop:
https://github.com/torvalds/linux/blob/v6.2/drivers/block/null_blk/zoned.c#L206
So I'm a bit confused...
If e.g. cb() is blkdev_copy_zone_to_user()
https://github.com/torvalds/linux/blob/v6.2/block/blk-zoned.c#L318
It will copy whatever the kernel allocated.
So I guess that since you can't guarantee that the ublk user space driver
will fill in all fields in the zone when receiving a REQ_OP_DRV_IN,
I assume that you do need to do a memset() in order to not potentially
leak kernel space data.
> +
> + req = blk_mq_alloc_request(disk->queue, REQ_OP_DRV_IN, 0);
> +
> + if (IS_ERR(req)) {
> + ret = PTR_ERR(req);
> + goto out;
> + }
> +
> + req->__sector = sector;
> +
> + ret = blk_rq_map_kern(disk->queue, req, &info, sizeof(info),
> + GFP_KERNEL);
> +
> + if (ret)
> + goto out;
> +
> + status = blk_execute_rq(req, 0);
> + ret = blk_status_to_errno(status);
> + if (ret)
> + goto out;
> +
> + blk_mq_free_request(req);
> +
> + ret = cb(&info, i, data);
> + if (ret)
> + goto out;
> +
> + /* A zero length zone means don't ask for more zones */
> + if (!info.len) {
> + nr_zones = i;
> + break;
> + }
> +
> + sector += zone_size;
> + }
> + ret = nr_zones;
> +
> + out:
> + return ret;
> +}
> +#endif
> +
> static int ublk_ctrl_start_dev(struct io_uring_cmd *cmd)
> {
> struct ublksrv_ctrl_cmd *header = (struct ublksrv_ctrl_cmd *)cmd->cmd;
> @@ -1535,6 +1656,15 @@ static int ublk_ctrl_start_dev(struct io_uring_cmd *cmd)
> if (ret)
> goto out_put_disk;
>
> + if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
> + ub->dev_info.flags & UBLK_F_ZONED) {
> + disk_set_zoned(disk, BLK_ZONED_HM);
> + blk_queue_required_elevator_features(disk->queue, ELEVATOR_F_ZBD_SEQ_WRITE);
> + ret = blk_revalidate_disk_zones(disk, NULL);
> + if (ret)
> + goto out_put_disk;
> + }
> +
> get_device(&ub->cdev_dev);
> ret = add_disk(disk);
> if (ret) {
> @@ -1673,6 +1803,9 @@ static int ublk_ctrl_add_dev(struct io_uring_cmd *cmd)
> if (!IS_BUILTIN(CONFIG_BLK_DEV_UBLK))
> ub->dev_info.flags |= UBLK_F_URING_CMD_COMP_IN_TASK;
>
> + if (!IS_ENABLED(CONFIG_BLK_DEV_ZONED))
> + ub->dev_info.flags &= ~UBLK_F_ZONED;
> +
> /* We are not ready to support zero copy */
> ub->dev_info.flags &= ~UBLK_F_SUPPORT_ZERO_COPY;
>
> diff --git a/include/uapi/linux/ublk_cmd.h b/include/uapi/linux/ublk_cmd.h
> index 8f88e3a29998..074b97821575 100644
> --- a/include/uapi/linux/ublk_cmd.h
> +++ b/include/uapi/linux/ublk_cmd.h
> @@ -78,6 +78,10 @@
> #define UBLK_F_USER_RECOVERY (1UL << 3)
>
> #define UBLK_F_USER_RECOVERY_REISSUE (1UL << 4)
> +/*
> + * Enable zoned device support
> + */
> +#define UBLK_F_ZONED (1ULL << 5)
>
> /* device state */
> #define UBLK_S_DEV_DEAD 0
> @@ -129,6 +133,12 @@ struct ublksrv_ctrl_dev_info {
> #define UBLK_IO_OP_DISCARD 3
> #define UBLK_IO_OP_WRITE_SAME 4
> #define UBLK_IO_OP_WRITE_ZEROES 5
> +#define UBLK_IO_OP_ZONE_OPEN 10
> +#define UBLK_IO_OP_ZONE_CLOSE 11
> +#define UBLK_IO_OP_ZONE_FINISH 12
> +#define UBLK_IO_OP_ZONE_APPEND 13
> +#define UBLK_IO_OP_ZONE_RESET 15
> +#define UBLK_IO_OP_DRV_IN 34
>
> #define UBLK_IO_F_FAILFAST_DEV (1U << 8)
> #define UBLK_IO_F_FAILFAST_TRANSPORT (1U << 9)
> @@ -214,6 +224,12 @@ struct ublk_param_discard {
> __u16 reserved0;
> };
>
> +struct ublk_param_zoned {
> + __u64 max_open_zones;
> + __u64 max_active_zones;
> + __u64 max_append_size;
> +};
> +
> struct ublk_params {
> /*
> * Total length of parameters, userspace has to set 'len' for both
> @@ -224,10 +240,12 @@ struct ublk_params {
> __u32 len;
> #define UBLK_PARAM_TYPE_BASIC (1 << 0)
> #define UBLK_PARAM_TYPE_DISCARD (1 << 1)
> +#define UBLK_PARAM_TYPE_ZONED (1 << 2)
> __u32 types; /* types of parameter included */
>
> struct ublk_param_basic basic;
> struct ublk_param_discard discard;
> + struct ublk_param_zoned zoned;
> };
>
> #endif
> --
> 2.39.2
>
On Fri, Feb 24, 2023 at 04:53:52PM +0100, Niklas Cassel wrote:
> Hello Andreas,
>
> On Fri, Feb 24, 2023 at 01:59:50PM +0100, Andreas Hindborg wrote:
> > 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
> >
> > This allows implementation of zoned storage devices in user space. An
> > example user space implementation based on ubdsrv is available [1].
> >
> > [1] https://github.com/metaspace/ubdsrv/commit/14a2b708f74f70cfecb076d92e680dc718cc1f6d
> >
> > Signed-off-by: Andreas Hindborg <a.hindborg@samsung.com>
> > ---
>
> Did you try to build this patch with CONFIG_BLK_DEV_ZONED disabled?
>
> I got the following build errors when building it on top of vanilla v6.2
> when CONFIG_BLK_DEV_ZONED is disabled:
>
> drivers/block/ublk_drv.c: In function ‘ublk_dev_param_basic_apply’:
> drivers/block/ublk_drv.c:221:28: error: ‘struct gendisk’ has no member named ‘nr_zones’
> 221 | ub->ub_disk->nr_zones = p->dev_sectors / p->chunk_sectors;
> | ^~
> drivers/block/ublk_drv.c: In function ‘ublk_dev_param_zoned_apply’:
> drivers/block/ublk_drv.c:244:17: error: implicit declaration of function ‘disk_set_max_active_zones’; did you mean ‘bdev_max_active_zones’? [-Werror=implicit-function-declaration]
> 244 | disk_set_max_active_zones(ub->ub_disk, p->max_active_zones);
> | ^~~~~~~~~~~~~~~~~~~~~~~~~
> | bdev_max_active_zones
> drivers/block/ublk_drv.c:245:17: error: implicit declaration of function ‘disk_set_max_open_zones’; did you mean ‘bdev_max_open_zones’? [-Werror=implicit-function-declaration]
> 245 | disk_set_max_open_zones(ub->ub_disk, p->max_open_zones);
>
The problem here is probably that blkdev.h does not have dummy functions for
disk_set_max_active_zones() and disk_set_max_open_zones()
in the:
#else /* CONFIG_BLK_DEV_ZONED */
case.
I do see dummy functions in blkdev.h for disk_nr_zones(), disk_zone_is_seq()
and disk_zone_no() when CONFIG_BLK_DEV_ZONED is not set.
Kind regards,
Niklas
Hi Andreas,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on v6.2]
[cannot apply to axboe-block/for-next linus/master next-20230224]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Andreas-Hindborg/block-ublk-enable-zoned-storage-support/20230224-210205
patch link: https://lore.kernel.org/r/20230224125950.214779-1-nmi%40metaspace.dk
patch subject: [PATCH] block: ublk: enable zoned storage support
config: ia64-buildonly-randconfig-r006-20230222 (https://download.01.org/0day-ci/archive/20230225/202302250222.XOrw9j6z-lkp@intel.com/config)
compiler: ia64-linux-gcc (GCC) 12.1.0
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# https://github.com/intel-lab-lkp/linux/commit/6d088fc1b115a63e8888b12fa47aabd45be97460
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Andreas-Hindborg/block-ublk-enable-zoned-storage-support/20230224-210205
git checkout 6d088fc1b115a63e8888b12fa47aabd45be97460
# save the config file
mkdir build_dir && cp config build_dir/.config
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross W=1 O=build_dir ARCH=ia64 olddefconfig
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross W=1 O=build_dir ARCH=ia64 SHELL=/bin/bash drivers/
If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <lkp@intel.com>
| Link: https://lore.kernel.org/oe-kbuild-all/202302250222.XOrw9j6z-lkp@intel.com/
All errors (new ones prefixed by >>):
drivers/block/ublk_drv.c: In function 'ublk_dev_param_basic_apply':
>> drivers/block/ublk_drv.c:221:28: error: 'struct gendisk' has no member named 'nr_zones'
221 | ub->ub_disk->nr_zones = p->dev_sectors / p->chunk_sectors;
| ^~
drivers/block/ublk_drv.c: In function 'ublk_dev_param_zoned_apply':
>> drivers/block/ublk_drv.c:244:17: error: implicit declaration of function 'disk_set_max_active_zones'; did you mean 'bdev_max_active_zones'? [-Werror=implicit-function-declaration]
244 | disk_set_max_active_zones(ub->ub_disk, p->max_active_zones);
| ^~~~~~~~~~~~~~~~~~~~~~~~~
| bdev_max_active_zones
>> drivers/block/ublk_drv.c:245:17: error: implicit declaration of function 'disk_set_max_open_zones'; did you mean 'bdev_max_open_zones'? [-Werror=implicit-function-declaration]
245 | disk_set_max_open_zones(ub->ub_disk, p->max_open_zones);
| ^~~~~~~~~~~~~~~~~~~~~~~
| bdev_max_open_zones
drivers/block/ublk_drv.c: In function 'ublk_ctrl_start_dev':
>> drivers/block/ublk_drv.c:1663:23: error: implicit declaration of function 'blk_revalidate_disk_zones' [-Werror=implicit-function-declaration]
1663 | ret = blk_revalidate_disk_zones(disk, NULL);
| ^~~~~~~~~~~~~~~~~~~~~~~~~
cc1: some warnings being treated as errors
vim +221 drivers/block/ublk_drv.c
192
193 static void ublk_dev_param_basic_apply(struct ublk_device *ub)
194 {
195 struct request_queue *q = ub->ub_disk->queue;
196 const struct ublk_param_basic *p = &ub->params.basic;
197
198 blk_queue_logical_block_size(q, 1 << p->logical_bs_shift);
199 blk_queue_physical_block_size(q, 1 << p->physical_bs_shift);
200 blk_queue_io_min(q, 1 << p->io_min_shift);
201 blk_queue_io_opt(q, 1 << p->io_opt_shift);
202
203 blk_queue_write_cache(q, p->attrs & UBLK_ATTR_VOLATILE_CACHE,
204 p->attrs & UBLK_ATTR_FUA);
205 if (p->attrs & UBLK_ATTR_ROTATIONAL)
206 blk_queue_flag_clear(QUEUE_FLAG_NONROT, q);
207 else
208 blk_queue_flag_set(QUEUE_FLAG_NONROT, q);
209
210 blk_queue_max_hw_sectors(q, p->max_sectors);
211 blk_queue_chunk_sectors(q, p->chunk_sectors);
212 blk_queue_virt_boundary(q, p->virt_boundary_mask);
213
214 if (p->attrs & UBLK_ATTR_READ_ONLY)
215 set_disk_ro(ub->ub_disk, true);
216
217 set_capacity(ub->ub_disk, p->dev_sectors);
218
219 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
220 ub->dev_info.flags & UBLK_F_ZONED && p->chunk_sectors) {
> 221 ub->ub_disk->nr_zones = p->dev_sectors / p->chunk_sectors;
222 }
223 }
224
225 static void ublk_dev_param_discard_apply(struct ublk_device *ub)
226 {
227 struct request_queue *q = ub->ub_disk->queue;
228 const struct ublk_param_discard *p = &ub->params.discard;
229
230 q->limits.discard_alignment = p->discard_alignment;
231 q->limits.discard_granularity = p->discard_granularity;
232 blk_queue_max_discard_sectors(q, p->max_discard_sectors);
233 blk_queue_max_write_zeroes_sectors(q,
234 p->max_write_zeroes_sectors);
235 blk_queue_max_discard_segments(q, p->max_discard_segments);
236 }
237
238 static void ublk_dev_param_zoned_apply(struct ublk_device *ub)
239 {
240 const struct ublk_param_zoned *p = &ub->params.zoned;
241
242 if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
243 ub->dev_info.flags & UBLK_F_ZONED) {
> 244 disk_set_max_active_zones(ub->ub_disk, p->max_active_zones);
> 245 disk_set_max_open_zones(ub->ub_disk, p->max_open_zones);
246 /* We do not support zone append yet */
247 //blk_queue_max_zone_append_sectors(q, zone_size);
248 }
249 }
250
Niklas Cassel <Niklas.Cassel@wdc.com> writes:
> On Fri, Feb 24, 2023 at 04:53:52PM +0100, Niklas Cassel wrote:
>> Hello Andreas,
>>
>> On Fri, Feb 24, 2023 at 01:59:50PM +0100, Andreas Hindborg wrote:
>> > 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
>> >
>> > This allows implementation of zoned storage devices in user space. An
>> > example user space implementation based on ubdsrv is available [1].
>> >
>> > [1] https://github.com/metaspace/ubdsrv/commit/14a2b708f74f70cfecb076d92e680dc718cc1f6d
>> >
>> > Signed-off-by: Andreas Hindborg <a.hindborg@samsung.com>
>> > ---
>>
>> Did you try to build this patch with CONFIG_BLK_DEV_ZONED disabled?
>>
>> I got the following build errors when building it on top of vanilla v6.2
>> when CONFIG_BLK_DEV_ZONED is disabled:
>>
>> drivers/block/ublk_drv.c: In function ‘ublk_dev_param_basic_apply’:
>> drivers/block/ublk_drv.c:221:28: error: ‘struct gendisk’ has no member named ‘nr_zones’
>> 221 | ub->ub_disk->nr_zones = p->dev_sectors / p->chunk_sectors;
>> | ^~
>> drivers/block/ublk_drv.c: In function ‘ublk_dev_param_zoned_apply’:
>> drivers/block/ublk_drv.c:244:17: error: implicit declaration of function ‘disk_set_max_active_zones’; did you mean ‘bdev_max_active_zones’? [-Werror=implicit-function-declaration]
>> 244 | disk_set_max_active_zones(ub->ub_disk, p->max_active_zones);
>> | ^~~~~~~~~~~~~~~~~~~~~~~~~
>> | bdev_max_active_zones
>> drivers/block/ublk_drv.c:245:17: error: implicit declaration of function ‘disk_set_max_open_zones’; did you mean ‘bdev_max_open_zones’? [-Werror=implicit-function-declaration]
>> 245 | disk_set_max_open_zones(ub->ub_disk, p->max_open_zones);
>>
>
> The problem here is probably that blkdev.h does not have dummy functions for
> disk_set_max_active_zones() and disk_set_max_open_zones()
>
> in the:
> #else /* CONFIG_BLK_DEV_ZONED */
> case.
>
> I do see dummy functions in blkdev.h for disk_nr_zones(), disk_zone_is_seq()
> and disk_zone_no() when CONFIG_BLK_DEV_ZONED is not set.
Thanks for the comments Niklas :)
I completely forgot to test without CONFIG_BLK_DEV_ZONED 🤦 What is the
preferred resolution here? Conditionally remove the lines with #ifdef
rather than use if(IS_ENABLED(...))? Dummy functions would resolve lines
244 and 245, but would not help with assignment of nr_zones at 221 as
that will not exist when CONFIG_BLK_DEV_ZONED is disabled.
BR Andreas
Hi Andreas,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on v6.2]
[cannot apply to axboe-block/for-next linus/master next-20230224]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Andreas-Hindborg/block-ublk-enable-zoned-storage-support/20230224-210205
patch link: https://lore.kernel.org/r/20230224125950.214779-1-nmi%40metaspace.dk
patch subject: [PATCH] block: ublk: enable zoned storage support
config: mips-allyesconfig (https://download.01.org/0day-ci/archive/20230225/202302250349.1eKfsTvO-lkp@intel.com/config)
compiler: mips-linux-gcc (GCC) 12.1.0
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# https://github.com/intel-lab-lkp/linux/commit/6d088fc1b115a63e8888b12fa47aabd45be97460
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Andreas-Hindborg/block-ublk-enable-zoned-storage-support/20230224-210205
git checkout 6d088fc1b115a63e8888b12fa47aabd45be97460
# save the config file
mkdir build_dir && cp config build_dir/.config
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross W=1 O=build_dir ARCH=mips olddefconfig
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross W=1 O=build_dir ARCH=mips SHELL=/bin/bash
If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <lkp@intel.com>
| Link: https://lore.kernel.org/oe-kbuild-all/202302250349.1eKfsTvO-lkp@intel.com/
All errors (new ones prefixed by >>):
arch/mips/kernel/head.o: in function `kernel_entry':
(.ref.text+0xac): relocation truncated to fit: R_MIPS_26 against `start_kernel'
init/main.o: in function `set_reset_devices':
main.c:(.init.text+0x20): relocation truncated to fit: R_MIPS_26 against `_mcount'
main.c:(.init.text+0x30): relocation truncated to fit: R_MIPS_26 against `__sanitizer_cov_trace_pc'
init/main.o: in function `debug_kernel':
main.c:(.init.text+0xa4): relocation truncated to fit: R_MIPS_26 against `_mcount'
main.c:(.init.text+0xb4): relocation truncated to fit: R_MIPS_26 against `__sanitizer_cov_trace_pc'
init/main.o: in function `quiet_kernel':
main.c:(.init.text+0x128): relocation truncated to fit: R_MIPS_26 against `_mcount'
main.c:(.init.text+0x138): relocation truncated to fit: R_MIPS_26 against `__sanitizer_cov_trace_pc'
init/main.o: in function `warn_bootconfig':
main.c:(.init.text+0x1ac): relocation truncated to fit: R_MIPS_26 against `_mcount'
main.c:(.init.text+0x1bc): relocation truncated to fit: R_MIPS_26 against `__sanitizer_cov_trace_pc'
init/main.o: in function `init_setup':
main.c:(.init.text+0x234): relocation truncated to fit: R_MIPS_26 against `_mcount'
main.c:(.init.text+0x254): additional relocation overflows omitted from the output
mips-linux-ld: drivers/block/ublk_drv.o: in function `ublk_dev_param_basic_apply':
>> ublk_drv.c:(.text.ublk_dev_param_basic_apply+0x2d8): undefined reference to `__udivdi3'
@@ -20,6 +20,7 @@
#include <linux/major.h>
#include <linux/wait.h>
#include <linux/blkdev.h>
+#include <linux/blkzoned.h>
#include <linux/init.h>
#include <linux/swap.h>
#include <linux/slab.h>
@@ -51,10 +52,12 @@
| UBLK_F_URING_CMD_COMP_IN_TASK \
| UBLK_F_NEED_GET_DATA \
| UBLK_F_USER_RECOVERY \
- | UBLK_F_USER_RECOVERY_REISSUE)
+ | UBLK_F_USER_RECOVERY_REISSUE \
+ | UBLK_F_ZONED)
/* All UBLK_PARAM_TYPE_* should be included here */
-#define UBLK_PARAM_TYPE_ALL (UBLK_PARAM_TYPE_BASIC | UBLK_PARAM_TYPE_DISCARD)
+#define UBLK_PARAM_TYPE_ALL (UBLK_PARAM_TYPE_BASIC | UBLK_PARAM_TYPE_DISCARD \
+ | UBLK_PARAM_TYPE_ZONED)
struct ublk_rq_data {
struct llist_node node;
@@ -212,6 +215,11 @@ 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);
+
+ if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
+ ub->dev_info.flags & UBLK_F_ZONED && p->chunk_sectors) {
+ ub->ub_disk->nr_zones = p->dev_sectors / p->chunk_sectors;
+ }
}
static void ublk_dev_param_discard_apply(struct ublk_device *ub)
@@ -227,6 +235,19 @@ static void ublk_dev_param_discard_apply(struct ublk_device *ub)
blk_queue_max_discard_segments(q, p->max_discard_segments);
}
+static void ublk_dev_param_zoned_apply(struct ublk_device *ub)
+{
+ const struct ublk_param_zoned *p = &ub->params.zoned;
+
+ if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
+ ub->dev_info.flags & UBLK_F_ZONED) {
+ disk_set_max_active_zones(ub->ub_disk, p->max_active_zones);
+ disk_set_max_open_zones(ub->ub_disk, p->max_open_zones);
+ /* We do not support zone append yet */
+ //blk_queue_max_zone_append_sectors(q, zone_size);
+ }
+}
+
static int ublk_validate_params(const struct ublk_device *ub)
{
/* basic param is the only one which must be set */
@@ -268,6 +289,9 @@ static int ublk_apply_params(struct ublk_device *ub)
if (ub->params.types & UBLK_PARAM_TYPE_DISCARD)
ublk_dev_param_discard_apply(ub);
+ if (ub->params.types & UBLK_PARAM_TYPE_ZONED)
+ ublk_dev_param_zoned_apply(ub);
+
return 0;
}
@@ -361,9 +385,18 @@ static void ublk_free_disk(struct gendisk *disk)
put_device(&ub->cdev_dev);
}
+#if IS_ENABLED(CONFIG_BLK_DEV_ZONED)
+static int ublk_report_zones(struct gendisk *disk, sector_t sector,
+ unsigned int nr_zones, report_zones_cb cb,
+ void *data);
+#endif
+
static const struct block_device_operations ub_fops = {
- .owner = THIS_MODULE,
- .free_disk = ublk_free_disk,
+ .owner = THIS_MODULE,
+ .free_disk = ublk_free_disk,
+#if IS_ENABLED(CONFIG_BLK_DEV_ZONED)
+ .report_zones = ublk_report_zones,
+#endif
};
#define UBLK_MAX_PIN_PAGES 32
@@ -499,7 +532,7 @@ static int ublk_unmap_io(const struct ublk_queue *ubq,
{
const unsigned int rq_bytes = blk_rq_bytes(req);
- if (req_op(req) == REQ_OP_READ && ublk_rq_has_data(req)) {
+ if ((req_op(req) == REQ_OP_READ || req_op(req) == REQ_OP_DRV_IN) && ublk_rq_has_data(req)) {
struct ublk_map_data data = {
.ubq = ubq,
.rq = req,
@@ -566,6 +599,26 @@ 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;
+#ifdef CONFIG_BLK_DEV_ZONED
+ 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 = UBLK_IO_OP_DRV_IN;
+ break;
+ case REQ_OP_ZONE_APPEND:
+ /* We do not support zone append yet */
+ fallthrough;
+#endif
default:
return BLK_STS_IOERR;
}
@@ -612,7 +665,8 @@ static void ublk_complete_rq(struct request *req)
*
* Both the two needn't unmap.
*/
- if (req_op(req) != REQ_OP_READ && req_op(req) != REQ_OP_WRITE) {
+ if (req_op(req) != REQ_OP_READ && req_op(req) != REQ_OP_WRITE &&
+ req_op(req) != REQ_OP_DRV_IN) {
blk_mq_end_request(req, BLK_STS_OK);
return;
}
@@ -1493,6 +1547,73 @@ static struct ublk_device *ublk_get_device_from_id(int idx)
return ub;
}
+#ifdef CONFIG_BLK_DEV_ZONED
+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;
+ unsigned int zone_size;
+ unsigned int first_zone;
+ int ret = 0;
+
+ ub = disk->private_data;
+
+ if (!(ub->dev_info.flags & UBLK_F_ZONED))
+ return -EINVAL;
+
+ zone_size = disk->queue->limits.chunk_sectors;
+ first_zone = sector >> ilog2(zone_size);
+
+ nr_zones = min(ub->ub_disk->nr_zones - first_zone, nr_zones);
+
+ for (unsigned int i = 0; i < nr_zones; i++) {
+
+ struct request *req;
+ blk_status_t status;
+ struct blk_zone info;
+
+ req = blk_mq_alloc_request(disk->queue, REQ_OP_DRV_IN, 0);
+
+ if (IS_ERR(req)) {
+ ret = PTR_ERR(req);
+ goto out;
+ }
+
+ req->__sector = sector;
+
+ ret = blk_rq_map_kern(disk->queue, req, &info, sizeof(info),
+ GFP_KERNEL);
+
+ if (ret)
+ goto out;
+
+ status = blk_execute_rq(req, 0);
+ ret = blk_status_to_errno(status);
+ if (ret)
+ goto out;
+
+ blk_mq_free_request(req);
+
+ ret = cb(&info, i, data);
+ if (ret)
+ goto out;
+
+ /* A zero length zone means don't ask for more zones */
+ if (!info.len) {
+ nr_zones = i;
+ break;
+ }
+
+ sector += zone_size;
+ }
+ ret = nr_zones;
+
+ out:
+ return ret;
+}
+#endif
+
static int ublk_ctrl_start_dev(struct io_uring_cmd *cmd)
{
struct ublksrv_ctrl_cmd *header = (struct ublksrv_ctrl_cmd *)cmd->cmd;
@@ -1535,6 +1656,15 @@ static int ublk_ctrl_start_dev(struct io_uring_cmd *cmd)
if (ret)
goto out_put_disk;
+ if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
+ ub->dev_info.flags & UBLK_F_ZONED) {
+ disk_set_zoned(disk, BLK_ZONED_HM);
+ blk_queue_required_elevator_features(disk->queue, ELEVATOR_F_ZBD_SEQ_WRITE);
+ ret = blk_revalidate_disk_zones(disk, NULL);
+ if (ret)
+ goto out_put_disk;
+ }
+
get_device(&ub->cdev_dev);
ret = add_disk(disk);
if (ret) {
@@ -1673,6 +1803,9 @@ static int ublk_ctrl_add_dev(struct io_uring_cmd *cmd)
if (!IS_BUILTIN(CONFIG_BLK_DEV_UBLK))
ub->dev_info.flags |= UBLK_F_URING_CMD_COMP_IN_TASK;
+ if (!IS_ENABLED(CONFIG_BLK_DEV_ZONED))
+ ub->dev_info.flags &= ~UBLK_F_ZONED;
+
/* We are not ready to support zero copy */
ub->dev_info.flags &= ~UBLK_F_SUPPORT_ZERO_COPY;
@@ -78,6 +78,10 @@
#define UBLK_F_USER_RECOVERY (1UL << 3)
#define UBLK_F_USER_RECOVERY_REISSUE (1UL << 4)
+/*
+ * Enable zoned device support
+ */
+#define UBLK_F_ZONED (1ULL << 5)
/* device state */
#define UBLK_S_DEV_DEAD 0
@@ -129,6 +133,12 @@ struct ublksrv_ctrl_dev_info {
#define UBLK_IO_OP_DISCARD 3
#define UBLK_IO_OP_WRITE_SAME 4
#define UBLK_IO_OP_WRITE_ZEROES 5
+#define UBLK_IO_OP_ZONE_OPEN 10
+#define UBLK_IO_OP_ZONE_CLOSE 11
+#define UBLK_IO_OP_ZONE_FINISH 12
+#define UBLK_IO_OP_ZONE_APPEND 13
+#define UBLK_IO_OP_ZONE_RESET 15
+#define UBLK_IO_OP_DRV_IN 34
#define UBLK_IO_F_FAILFAST_DEV (1U << 8)
#define UBLK_IO_F_FAILFAST_TRANSPORT (1U << 9)
@@ -214,6 +224,12 @@ struct ublk_param_discard {
__u16 reserved0;
};
+struct ublk_param_zoned {
+ __u64 max_open_zones;
+ __u64 max_active_zones;
+ __u64 max_append_size;
+};
+
struct ublk_params {
/*
* Total length of parameters, userspace has to set 'len' for both
@@ -224,10 +240,12 @@ struct ublk_params {
__u32 len;
#define UBLK_PARAM_TYPE_BASIC (1 << 0)
#define UBLK_PARAM_TYPE_DISCARD (1 << 1)
+#define UBLK_PARAM_TYPE_ZONED (1 << 2)
__u32 types; /* types of parameter included */
struct ublk_param_basic basic;
struct ublk_param_discard discard;
+ struct ublk_param_zoned zoned;
};
#endif