| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <87a5wad4qt.fsf@metaspace.dk>
Date: Wed, 05 Jul 2023 10:29:50 +0200
From: "Andreas Hindborg (Samsung)" <nmi@...aspace.dk>
To: Damien Le Moal <dlemoal@...nel.org>
Cc: Ming Lei <ming.lei@...hat.com>,
open list <linux-kernel@...r.kernel.org>,
Matias Bjorling <Matias.Bjorling@....com>,
Hans Holmberg <Hans.Holmberg@....com>,
Jens Axboe <axboe@...nel.dk>,
Minwoo Im <minwoo.im.dev@...il.com>,
Aravind Ramesh <Aravind.Ramesh@....com>, gost.dev@...sung.com,
"open list:BLOCK LAYER" <linux-block@...r.kernel.org>,
Christoph Hellwig <hch@...radead.org>
Subject: Re: [PATCH v5 5/5] ublk: enable zoned storage support
Damien Le Moal <dlemoal@...nel.org> writes:
> On 7/5/23 01:52, Andreas Hindborg wrote:
>> From: Andreas Hindborg <a.hindborg@...sung.com>
>>
>> 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.
>>
>> Signed-off-by: Andreas Hindborg <a.hindborg@...sung.com>
>> ---
>> MAINTAINERS | 1 +
>> drivers/block/Kconfig | 5 +
>> drivers/block/Makefile | 1 +
>> drivers/block/ublk-zoned.c | 225 ++++++++++++++++++++++++++++++++++
>> drivers/block/ublk.c | 92 +++++++++++---
>> drivers/block/ublk.h | 72 +++++++++++
>> include/uapi/linux/ublk_cmd.h | 28 +++++
>> 7 files changed, 410 insertions(+), 14 deletions(-)
>> create mode 100644 drivers/block/ublk-zoned.c
>>
>> diff --git a/MAINTAINERS b/MAINTAINERS
>> index 1f193cd43958..8277f3e3e8e9 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -21553,6 +21553,7 @@ M: Ming Lei <ming.lei@...hat.com>
>> L: linux-block@...r.kernel.org
>> S: Maintained
>> F: Documentation/block/ublk.rst
>> +F: drivers/block/ublk-zoned.c
>> F: drivers/block/ublk.c
>> F: drivers/block/ublk.h
>> F: include/uapi/linux/ublk_cmd.h
>> diff --git a/drivers/block/Kconfig b/drivers/block/Kconfig
>> index 5b9d4aaebb81..2d76e9693c2d 100644
>> --- a/drivers/block/Kconfig
>> +++ b/drivers/block/Kconfig
>> @@ -373,6 +373,7 @@ config BLK_DEV_RBD
>> config BLK_DEV_UBLK
>> tristate "Userspace block driver (Experimental)"
>> select IO_URING
>> + select BLK_DEV_UBLK_ZONED if BLK_DEV_ZONED
>> help
>> io_uring based userspace block driver. Together with ublk server, ublk
>> has been working well, but interface with userspace or command data
>> @@ -402,6 +403,10 @@ config BLKDEV_UBLK_LEGACY_OPCODES
>> suggested to enable N if your application(ublk server) switches to
>> ioctl command encoding.
>>
>> +config BLK_DEV_UBLK_ZONED
>> + bool
>> + depends on BLK_DEV_UBLK && BLK_DEV_ZONED
>
> I do not think this line is needed.
Seems you are right 👍
>
>> +
>> source "drivers/block/rnbd/Kconfig"
>>
>> endif # BLK_DEV
>> diff --git a/drivers/block/Makefile b/drivers/block/Makefile
>> index 38f2229623a8..6c700936d427 100644
>> --- a/drivers/block/Makefile
>> +++ b/drivers/block/Makefile
>> @@ -39,5 +39,6 @@ obj-$(CONFIG_BLK_DEV_NULL_BLK) += null_blk/
>>
>> obj-$(CONFIG_BLK_DEV_UBLK) += ublk_drv.o
>> ublk_drv-$(CONFIG_BLK_DEV_UBLK) += ublk.o
>> +ublk_drv-$(CONFIG_BLK_DEV_UBLK_ZONED) += ublk-zoned.o
>>
>> swim_mod-y := swim.o swim_asm.o
>> diff --git a/drivers/block/ublk-zoned.c b/drivers/block/ublk-zoned.c
>> new file mode 100644
>> index 000000000000..6ddde2ac63e3
>> --- /dev/null
>> +++ b/drivers/block/ublk-zoned.c
>> @@ -0,0 +1,225 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * Copyright 2023 Andreas Hindborg <a.hindborg@...sung.com>
>> + */
>> +#include <linux/blkzoned.h>
>> +#include <linux/ublk_cmd.h>
>> +#include <linux/vmalloc.h>
>> +
>> +#include "ublk.h"
>> +
>> +
>
> extra blank line not needed.
👍
>
>> +static int get_nr_zones(const struct ublk_device *ub)
>> +{
>> + const struct ublk_param_basic *p = &ub->params.basic;
>> +
>> + if (p->chunk_sectors)
>> + return p->dev_sectors / p->chunk_sectors;
>
> This one is likely causing the build bot warning about __udivdi3 being undefined.
>
>> + else
>
> No need for the else.
>
>> + return 0;
>> +}
>
> Overall, I would reverse this:
>
> static int get_nr_zones(const struct ublk_device *ub)
> {
> const struct ublk_param_basic *p = &ub->params.basic;
>
> if (!p->chunk_sectors)
> return 0;
>
> return p->dev_sectors >> ilog2( p->chunk_sectors);
> }
>
> And given that for a non-zoned drive chunk_sectors should be 0 as well (unless
> it is used by regular ublk), this function will return 0 zones for regular
> devies, and it does not depend on CONFIG_BLK_DEV_ZONED.
Ok for the edit. Not sure about moving to ublk.c since it is actually
static linkage without a prototype in the header.
>
>> +
>> +int ublk_set_nr_zones(struct ublk_device *ub)
>> +{
>> + const struct ublk_param_basic *p = &ub->params.basic;
>> +
>> + if (ublk_dev_is_zoned(ub) && !p->chunk_sectors)
>> + return -EINVAL;
>
> This is a little odd. Why would p->chunk_sectors be 0 if ublk_dev_is_zoned() is
> true ?
It is a "defensive programming" check. It should not happen because we
validate this when parameters are set from user space and we should not
be able to get here if parameters are not set.
I think it enhances clarity and intention while helping mitigate
potential problems from future refactoring and edits. It's also not in a
hot path.
But I can remove it 👍
>
>> +
>> + if (ublk_dev_is_zoned(ub)) {
>> + ub->ub_disk->nr_zones = get_nr_zones(ub);
>> + if (!ub->ub_disk->nr_zones)
>> + return -EINVAL;
>> + }
>
> This is the only hunk you should need.
>
>> +
>> + return 0;
>> +}
>> +
>> +void ublk_disk_set_zoned(struct ublk_device *ub)
>> +{
>> + if (ublk_dev_is_zoned(ub)) {
>> + disk_set_zoned(ub->ub_disk, BLK_ZONED_HM);
>> + blk_queue_required_elevator_features(ub->ub_disk->queue,
>> + ELEVATOR_F_ZBD_SEQ_WRITE);
>
> zone size (chunk_sectors) ? max open and max active zones ? max append sectors ?
> It would be nice to have all that set in the same place.
Yes, that makes sense.
>
>> + }
>> +}
>> +
>> +void ublk_zoned_commit_completion(const struct ublksrv_io_cmd *ub_cmd,
>> + struct ublk_io *io, struct request *req)
>> +{
>> + if (io->flags & UBLK_IO_FLAG_ZONE_APPEND)
>> + req->__sector = ub_cmd->addr;
>> + io->flags &= ~UBLK_IO_FLAG_ZONE_APPEND;
>
> Clearing the flag can be inside the if. But this helper is not needed at all.
> Nothing in it depends on CONFIG_BLK_DEV_ZONED so you can open code it in
> ublk_commit_completion(). For the !CONFIG_BLK_DEV_ZONED case, you will never
> have a zoned device, so you will never see UBLK_IO_FLAG_ZONE_APPEND being set.
I put it behind the config check to not add the instructions to this
path in case they are not required. I'm fine with open coding it as well 👍
>
>> +}
>> +
>> +int ublk_revalidate_disk_zones(struct ublk_device *ub)
>> +{
>> + int ret = 0;
>> +
>> + if (ublk_dev_is_zoned(ub))
>> + ret = blk_revalidate_disk_zones(ub->ub_disk, NULL);
>> +
>> + return ret;
>
> Variable ret is not necessary.
Right 👍
>
>> +}
>> +
>> +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) && !(ublk_dev_params_zoned(ub)))
>
> extra parenthesis around ublk_dev_params_zoned(ub) are not needed.
Thanks 👍
>
>> + return -EINVAL;
>> +
>> + if (!ublk_dev_is_zoned(ub) && ublk_dev_params_zoned(ub))
>
> This one could be an || condition with the previous if. But overall, cannot this
> be simplified to:
>
> if ((!ublk_dev_is_zoned(ub)) && (!ublk_dev_params_zoned(ub)))
> return -EINVAL;
I don't think that is the same. The EINVAL should be hit when the xor of
the two checks is true. I considered doing that but decided the one I
submitted is more clear. I also considered combining the checks in one
conditional statement, but I liked two statements better. Compiler
should be cleaver enough to optimize.
>
> ?
>
> Not sure as this is not clear. If ublk_dev_params_zoned() is false, I do not see
> how ublk_dev_is_zoned() can be true...
This checks for user error. User might set UBLK_F_ZONED during
UBLK_CMD_ADD_DEV command but might fail to set zoned parameters
(UBLK_PARAM_TYPE_ZONED) during UBLK_CMD_SET_PARAMS command. Or the other
way around. Setting zoned parameters on a non zoned ublk device should
also result in an error.
>
>> + return -EINVAL;
>> +
>> + if (!ublk_dev_params_zoned(ub))
>> + return 0;
>
> Shouldn't this be first ? That would simplify the previous ifs...
No, because we want to check that the user did not set zoned parameters
for a regular device.
>
>> +
>> + if (!p->max_zone_append_sectors)
>> + return -EINVAL;
>> +
>> + nr_zones = get_nr_zones(ub);
>> +
>> + if (p->max_active_zones > nr_zones)
>> + return -EINVAL;
>> +
>> + if (p->max_open_zones > nr_zones)
>> + return -EINVAL;
>
> nr_zones = get_nr_zones(ub);
> if (!nr_zones ||
> p->max_active_zones > nr_zones ||
> p->max_open_zones > nr_zones)
> return -EINVAL;
I think this makes the code less readable, and compiler should be smart
enough to do this for us anyway. If you insist I will change 😏
>
>> +
>> + return 0;
>> +}
>> +
>> +int ublk_dev_param_zoned_apply(struct ublk_device *ub)
>> +{
>> + const struct ublk_param_zoned *p = &ub->params.zoned;
>> +
>> + if (ublk_dev_is_zoned(ub) && !(ublk_dev_params_zoned(ub)))
>> + return -EINVAL;
>> +
>> + if (!ublk_dev_is_zoned(ub) && ublk_dev_params_zoned(ub))
>> + return -EINVAL;
>> +
>> + if (!ublk_dev_params_zoned(ub))
>> + return 0;
>
> How many times do you need to check these ? these are checked in
> ublk_dev_param_zoned_validate() already. And given the name, I would assume that
> the parameters were validated before being applied.
I'll remove the checks.
>
>> +
>> + 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);
>> +
>
> Shouldn't this all be in ublk_disk_set_zoned() ? Or even better, merge these 2
> functions together.
Yes, you are right.
>
>> + return 0;
>> +}
>> +
>> +/* Based on virtblk_alloc_report_buffer */
>> +static void *ublk_alloc_report_buffer(struct ublk_device *ublk,
>> + unsigned int nr_zones,
>> + unsigned int zone_sectors, size_t *buflen)
>> +{
>> + struct request_queue *q = ublk->ub_disk->queue;
>> + size_t bufsize;
>> + void *buf;
>> +
>> + nr_zones = min_t(unsigned int, nr_zones,
>> + get_capacity(ublk->ub_disk) >> ilog2(zone_sectors));
>
> ublk->ub_disk->nr_zones
Thanks 👍
>
> But ublk_report_zones() does the capping of nr_zones already. So it should not
> be needed at all here.
I would leave it in so that potential future callers of this function
get no surprises, but I get the sense that you don't prefer that kind of
checks. Hard no to leaving it in or OK to have it?
>
>> +
>> + 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);
>> +
>> + 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;
>> +}
>> +
>> +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;
>> +
>> + if (!ublk_dev_is_zoned(ub))
>> + return -EOPNOTSUPP;
>> +
>> + nr_zones = min_t(unsigned int, ub->ub_disk->nr_zones - first_zone,
>> + nr_zones);
>> +
>> + buffer = ublk_alloc_report_buffer(ub, nr_zones, zone_size_sectors,
>> + &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);
>
> Nit: splitting the line after "=" is the more usual way.
Alright. I am using clang-format and for some reason it likes the other
way. I wonder if there is a config option for that behavior.
>
>> + 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;
>> +
>> + 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;
>> +
>> + ret = cb(zone, i, data);
>> + if (ret)
>> + goto out;
>> +
>> + done_zones++;
>> + sector += zone_size_sectors;
>> +
>> + /* A zero length zone means don't ask for more zones */
>> + if (!zone->len) {
>> + ret = done_zones;
>> + goto out;
>
> You should only need to "break" here. The reason is that the report call may
> have been short, but that does not mean that the overall report request from the
> user has been served yet.
Yes.
>
>> + }
>> + }
>> + }
>> +
>> + ret = done_zones;
>> +
>> +out:
>> + kvfree(buffer);
>> + return ret;
>> +}
>> diff --git a/drivers/block/ublk.c b/drivers/block/ublk.c
>> index 0b1ec102aaae..c97a8f14f8a9 100644
>> --- a/drivers/block/ublk.c
>> +++ b/drivers/block/ublk.c
>> @@ -57,12 +57,13 @@
>> | 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_uring_cmd_pdu {
>> struct ublk_queue *ubq;
>> @@ -137,7 +138,7 @@ 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;
>> @@ -162,6 +163,8 @@ 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 ublk_set_nr_zones(ub);
>
> This is odd. Why is this not called only under a "if (ublk_dev_is_zoned(ub)" ?
The call is a noop if not ublk_dev_is_zoned(). I will hoist the check to here to
enhance clarity.
>
>> }
>>
>> static void ublk_dev_param_discard_apply(struct ublk_device *ub)
>> @@ -191,6 +194,9 @@ 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;
>>
>> @@ -209,20 +215,24 @@ static int ublk_validate_params(const struct ublk_device *ub)
>> if (ub->params.types & UBLK_PARAM_TYPE_DEVT)
>> return -EINVAL;
>>
>> - return 0;
>> + return ublk_dev_param_zoned_validate(ub);
>
> Same here.
In this case we need to make the call no matter what, in order to do
some sanity checks on the parameters. It is an error to set zoned
parameters when the ublk device does not have the zoned flag.
>
>> }
>>
>> 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);
>>
>> - return 0;
>> + return ublk_dev_param_zoned_apply(ub);
>
> Same.
Right.
>
>> }
>>
>> static inline bool ublk_support_user_copy(const struct ublk_queue *ubq)
>> @@ -392,6 +402,7 @@ 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
>> @@ -506,7 +517,8 @@ 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,
>> @@ -590,6 +602,7 @@ 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);
>> u32 ublk_op;
>>
>> switch (req_op(req)) {
>> @@ -608,6 +621,37 @@ 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 zone append or reset_all yet */
>> + fallthrough;
>
> Wrong comment. Append support is above it.
Thanks
>
>> default:
>> return BLK_STS_IOERR;
>> }
>> @@ -661,7 +705,8 @@ static inline 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)
>> goto exit;
>>
>> /* for READ request, writing data in iod->addr to rq buffers */
>> @@ -1025,6 +1070,8 @@ static void ublk_commit_completion(struct ublk_device *ub,
>> /* find the io request and complete */
>> req = blk_mq_tag_to_rq(ub->tag_set.tags[qid], tag);
>>
>> + ublk_zoned_commit_completion(ub_cmd, io, req);
>> +
>> if (req && likely(!blk_should_fake_timeout(req->q)))
>> ublk_put_req_ref(ubq, req);
>> }
>> @@ -1324,7 +1371,8 @@ static int __ublk_ch_uring_cmd(struct io_uring_cmd *cmd,
>> ^ (_IOC_NR(cmd_op) == UBLK_IO_NEED_GET_DATA))
>> goto out;
>>
>> - if (ublk_support_user_copy(ubq) && ub_cmd->addr) {
>> + if (ublk_support_user_copy(ubq) &&
>> + !(io->flags & UBLK_IO_FLAG_ZONE_APPEND) && ub_cmd->addr) {
>> ret = -EINVAL;
>> goto out;
>> }
>> @@ -1447,11 +1495,14 @@ static inline bool ublk_check_ubuf_dir(const struct request *req,
>> int ubuf_dir)
>> {
>> /* copy ubuf to request pages */
>> - if (req_op(req) == REQ_OP_READ && ubuf_dir == ITER_SOURCE)
>> + if ((req_op(req) == REQ_OP_READ || req_op(req) == REQ_OP_DRV_IN) &&
>> + ubuf_dir == ITER_SOURCE)
>> return true;
>>
>> /* copy request pages to ubuf */
>> - if (req_op(req) == REQ_OP_WRITE && ubuf_dir == ITER_DEST)
>> + if ((req_op(req) == REQ_OP_WRITE ||
>> + req_op(req) == REQ_OP_ZONE_APPEND) &&
>> + ubuf_dir == ITER_DEST)
>> return true;
>>
>> return false;
>> @@ -1780,6 +1831,8 @@ static int ublk_ctrl_start_dev(struct ublk_device *ub, struct io_uring_cmd *cmd)
>> ub->dev_info.ublksrv_pid = ublksrv_pid;
>> ub->ub_disk = disk;
>>
>> + ublk_disk_set_zoned(ub);
>> +
>
> This really needs to go inside ublk_apply_params() I think.
Yes
>
>> ret = ublk_apply_params(ub);
>> if (ret)
>> goto out_put_disk;
>> @@ -1788,8 +1841,12 @@ static int ublk_ctrl_start_dev(struct ublk_device *ub, struct io_uring_cmd *cmd)
>> if (ub->nr_privileged_daemon != ub->nr_queues_ready)
>> set_bit(GD_SUPPRESS_PART_SCAN, &disk->state);
>>
>> - get_device(&ub->cdev_dev);
>> ub->dev_info.state = UBLK_S_DEV_LIVE;
>> + ret = ublk_revalidate_disk_zones(ub);
>> + if (ret)
>> + goto out_put_disk;
>> +
>> + get_device(&ub->cdev_dev);
>> ret = add_disk(disk);
>> if (ret) {
>> /*
>> @@ -1950,6 +2007,13 @@ static int ublk_ctrl_add_dev(struct io_uring_cmd *cmd)
>> if (ublk_dev_is_user_copy(ub))
>> ub->dev_info.flags &= ~UBLK_F_NEED_GET_DATA;
>>
>> + /* Zoned storage support requires user copy feature */
>> + if (ublk_dev_is_zoned(ub) &&
>> + (!IS_ENABLED(CONFIG_BLK_DEV_ZONED) || !ublk_dev_is_user_copy(ub))) {
>
> CONFIG_BLK_DEV_UBLK_ZONED ?
Yes, although it the same in this translation unit because of kconfig.
>
>> + ret = -EINVAL;
>> + goto out_free_dev_number;
>> + }
>> +
>> /* We are not ready to support zero copy */
>> ub->dev_info.flags &= ~UBLK_F_SUPPORT_ZERO_COPY;
>>
>> diff --git a/drivers/block/ublk.h b/drivers/block/ublk.h
>> index fcbcc6b02aa0..b78ab43cea82 100644
>> --- a/drivers/block/ublk.h
>> +++ b/drivers/block/ublk.h
>> @@ -45,6 +45,10 @@
>> */
>> #define UBLK_IO_FLAG_NEED_GET_DATA 0x08
>>
>> +/*
>> + * Set when IO is Zone Append
>> + */
>> +#define UBLK_IO_FLAG_ZONE_APPEND 0x10
>>
>> struct ublk_device {
>> struct gendisk *ub_disk;
>> @@ -89,6 +93,9 @@ struct ublk_rq_data {
>> struct llist_node node;
>>
>> struct kref ref;
>> + __u32 operation;
>> + __u64 sector;
>> + __u32 nr_sectors;
>> };
>>
>> struct ublk_io {
>> @@ -100,9 +107,74 @@ struct ublk_io {
>> struct io_uring_cmd *cmd;
>> };
>>
>> +static inline int ublk_dev_params_zoned(const struct ublk_device *ub)
>> +{
>> + return ub->params.types & UBLK_PARAM_TYPE_ZONED;
>> +}
>> +
>> +static inline bool ublk_dev_is_zoned(const struct ublk_device *ub)
>> +{
>> + return ub->dev_info.flags & UBLK_F_ZONED;
>> +}
>> +
>> static inline bool ublk_dev_is_user_copy(const struct ublk_device *ub)
>> {
>> return ub->dev_info.flags & UBLK_F_USER_COPY;
>> }
>>
>> +#ifndef CONFIG_BLK_DEV_ZONED
>
> CONFIG_BLK_DEV_UBLK_ZONED
Thanks
>
>> +
>> +static inline int ublk_set_nr_zones(struct ublk_device *ub)
>> +{
>> + return 0;
>> +}
>> +
>> +static inline void ublk_disk_set_zoned(struct ublk_device *ub)
>> +{
>> +}
>> +
>> +static inline int ublk_dev_param_zoned_validate(const struct ublk_device *ub)
>> +{
>> + if (ublk_dev_params_zoned(ub))
>> + return -EINVAL;
>> + return 0;
>> +}
>> +
>> +static inline int ublk_dev_param_zoned_apply(struct ublk_device *ub)
>> +{
>> + if (ublk_dev_params_zoned(ub))
>> + return -EINVAL;
>> + return 0;
>> +}
>> +
>> +static inline void
>> +ublk_zoned_commit_completion(const struct ublksrv_io_cmd *ub_cmd,
>> + struct ublk_io *io, struct request *req)
>> +{
>> +}
>> +
>> +static inline int ublk_revalidate_disk_zones(struct ublk_device *ub)
>> +{
>> + return 0;
>> +}
>> +
>> +#define ublk_report_zones (NULL)
>> +
>> +#else
>> +
>> +int ublk_set_nr_zones(struct ublk_device *ub);
>> +void ublk_disk_set_zoned(struct ublk_device *ub);
>> +int ublk_dev_param_zoned_validate(const struct ublk_device *ub);
>> +int ublk_dev_param_zoned_apply(struct ublk_device *ub);
>> +void ublk_zoned_commit_completion(const struct ublksrv_io_cmd *ub_cmd,
>> + struct ublk_io *io, struct request *req);
>> +int ublk_revalidate_disk_zones(struct ublk_device *ub);
>> +int ublk_report_zones(struct gendisk *disk, sector_t sector,
>> + unsigned int nr_zones, report_zones_cb cb,
>> + void *data);
>> +
>> +#endif
>> +
>> +
>> +
>> #endif
>> diff --git a/include/uapi/linux/ublk_cmd.h b/include/uapi/linux/ublk_cmd.h
>> index a32810c8ef2b..13b76e665aed 100644
>> --- a/include/uapi/linux/ublk_cmd.h
>> +++ b/include/uapi/linux/ublk_cmd.h
>> @@ -176,6 +176,12 @@
>> /* Copy between request and user buffer by pread()/pwrite() */
>> #define UBLK_F_USER_COPY (1UL << 7)
>>
>> +/*
>> + * User space sets this flag when setting up the device to request zoned storage support. Kernel may
>> + * deny the request by returning an error.
>> + */
>> +#define UBLK_F_ZONED (1ULL << 8)
>> +
>> /* device state */
>> #define UBLK_S_DEV_DEAD 0
>> #define UBLK_S_DEV_LIVE 1
>> @@ -235,7 +241,20 @@ 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_START (32)
>> +/* Construct a zone report. The report request is carried in `struct ublksrv_io_desc`. The
>> + * `start_sector` field must be the first sector of a zone and shall indicate the first zone of the
>> + * report. The `nr_sectors` shall indicate how many zones should be reported (divide by zone size to
>> + * get number of zones in the report) and must be an integer multiple of the zone size. The report
>> + * shall be delivered as a `struct blk_zone` array. To report fewer zones than requested, zero the
>> + * last entry of the returned array.
>> + */
>
> Long lines and first line is not "/*"
I swear I read somewhere 100 character line limit, but I see quite
clearly it says 80 in the style guide. I wonder why checkpatch.pl did
not complain.
>
>> +#define UBLK_IO_OP_REPORT_ZONES (__UBLK_IO_OP_DRV_IN_START)
>> #define __UBLK_IO_OP_DRV_IN_END (96)
>> #define __UBLK_IO_OP_DRV_OUT_START (__UBLK_IO_OP_DRV_IN_END)
>> #define __UBLK_IO_OP_DRV_OUT_END (160)
>> @@ -335,6 +354,13 @@ struct ublk_param_devt {
>> __u32 disk_minor;
>> };
>>
>> +struct ublk_param_zoned {
>> + __u32 max_open_zones;
>> + __u32 max_active_zones;
>> + __u32 max_zone_append_sectors;
>> + __u8 reserved[20];
>> +};
>> +
>> struct ublk_params {
>> /*
>> * Total length of parameters, userspace has to set 'len' for both
>> @@ -346,11 +372,13 @@ struct ublk_params {
>> #define UBLK_PARAM_TYPE_BASIC (1 << 0)
>> #define UBLK_PARAM_TYPE_DISCARD (1 << 1)
>> #define UBLK_PARAM_TYPE_DEVT (1 << 2)
>> +#define UBLK_PARAM_TYPE_ZONED (1 << 3)
>> __u32 types; /* types of parameter included */
>>
>> struct ublk_param_basic basic;
>> struct ublk_param_discard discard;
>> struct ublk_param_devt devt;
>> + struct ublk_param_zoned zoned;
>> };
>>
>> #endif
Powered by blists - more mailing lists