Re: [PATCH V4 16/27] ublk: add new feature UBLK_F_BATCH_IO

[Ming Lei <ming.lei@redhat.com>]

On Mon, Dec 01, 2025 at 01:16:04PM -0800, Caleb Sander Mateos wrote:
> On Thu, Nov 20, 2025 at 6:00 PM Ming Lei <ming.lei@redhat.com> wrote:
> >
> > Add new feature UBLK_F_BATCH_IO which replaces the following two
> > per-io commands:
> >
> >         - UBLK_U_IO_FETCH_REQ
> >
> >         - UBLK_U_IO_COMMIT_AND_FETCH_REQ
> >
> > with three per-queue batch io uring_cmd:
> >
> >         - UBLK_U_IO_PREP_IO_CMDS
> >
> >         - UBLK_U_IO_COMMIT_IO_CMDS
> >
> >         - UBLK_U_IO_FETCH_IO_CMDS
> >
> > Then ublk can deliver batch io commands to ublk server in single
> > multishort uring_cmd, also allows to prepare & commit multiple
> > commands in batch style via single uring_cmd, communication cost is
> > reduced a lot.
> >
> > This feature also doesn't limit task context any more for all supported
> > commands, so any allowed uring_cmd can be issued in any task context.
> > ublk server implementation becomes much easier.
> >
> > Meantime load balance becomes much easier to support with this feature.
> > The command `UBLK_U_IO_FETCH_IO_CMDS` can be issued from multiple task
> > contexts, so each task can adjust this command's buffer length or number
> > of inflight commands for controlling how much load is handled by current
> > task.
> >
> > Later, priority parameter will be added to command `UBLK_U_IO_FETCH_IO_CMDS`
> > for improving load balance support.
> >
> > UBLK_U_IO_GET_DATA isn't supported in batch io yet, but it may be
> 
> UBLK_U_IO_NEED_GET_DATA?

Yeah.

> 
> > enabled in future via its batch pair.
> >
> > Signed-off-by: Ming Lei <ming.lei@redhat.com>
> > ---
> >  drivers/block/ublk_drv.c      | 58 ++++++++++++++++++++++++++++++++---
> >  include/uapi/linux/ublk_cmd.h | 16 ++++++++++
> >  2 files changed, 69 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
> > index 849199771f86..90cd1863bc83 100644
> > --- a/drivers/block/ublk_drv.c
> > +++ b/drivers/block/ublk_drv.c
> > @@ -74,7 +74,8 @@
> >                 | UBLK_F_AUTO_BUF_REG \
> >                 | UBLK_F_QUIESCE \
> >                 | UBLK_F_PER_IO_DAEMON \
> > -               | UBLK_F_BUF_REG_OFF_DAEMON)
> > +               | UBLK_F_BUF_REG_OFF_DAEMON \
> > +               | UBLK_F_BATCH_IO)
> >
> >  #define UBLK_F_ALL_RECOVERY_FLAGS (UBLK_F_USER_RECOVERY \
> >                 | UBLK_F_USER_RECOVERY_REISSUE \
> > @@ -320,12 +321,12 @@ static void ublk_batch_dispatch(struct ublk_queue *ubq,
> >
> >  static inline bool ublk_dev_support_batch_io(const struct ublk_device *ub)
> >  {
> > -       return false;
> > +       return ub->dev_info.flags & UBLK_F_BATCH_IO;
> >  }
> >
> >  static inline bool ublk_support_batch_io(const struct ublk_queue *ubq)
> >  {
> > -       return false;
> > +       return ubq->flags & UBLK_F_BATCH_IO;
> >  }
> >
> >  static inline void ublk_io_lock(struct ublk_io *io)
> > @@ -3450,6 +3451,41 @@ static int ublk_validate_batch_fetch_cmd(struct ublk_batch_io_data *data,
> >         return 0;
> >  }
> >
> > +static int ublk_handle_non_batch_cmd(struct io_uring_cmd *cmd,
> > +                                    unsigned int issue_flags)
> > +{
> > +       const struct ublksrv_io_cmd *ub_cmd = io_uring_sqe_cmd(cmd->sqe);
> > +       struct ublk_device *ub = cmd->file->private_data;
> > +       unsigned tag = READ_ONCE(ub_cmd->tag);
> > +       unsigned q_id = READ_ONCE(ub_cmd->q_id);
> > +       unsigned index = READ_ONCE(ub_cmd->addr);
> > +       struct ublk_queue *ubq;
> > +       struct ublk_io *io;
> > +       int ret = -EINVAL;
> 
> I think it would be clearer to just return -EINVAL instead of adding
> this variable, but up to you
> 
> > +
> > +       if (!ub)
> > +               return ret;
> 
> How is this case possible?

Will remove the check.

> 
> > +
> > +       if (q_id >= ub->dev_info.nr_hw_queues)
> > +               return ret;
> > +
> > +       ubq = ublk_get_queue(ub, q_id);
> > +       if (tag >= ubq->q_depth)
> 
> Can avoid the likely cache miss here by using ub->dev_info.queue_depth
> instead, analogous to ublk_ch_uring_cmd_local()

OK.

> 
> > +               return ret;
> > +
> > +       io = &ubq->ios[tag];
> > +
> > +       switch (cmd->cmd_op) {
> > +       case UBLK_U_IO_REGISTER_IO_BUF:
> > +               return ublk_register_io_buf(cmd, ub, q_id, tag, io, index,
> > +                               issue_flags);
> > +       case UBLK_U_IO_UNREGISTER_IO_BUF:
> > +               return ublk_unregister_io_buf(cmd, ub, index, issue_flags);
> > +       default:
> > +               return -EOPNOTSUPP;
> > +       }
> > +}
> > +
> >  static int ublk_ch_batch_io_uring_cmd(struct io_uring_cmd *cmd,
> >                                        unsigned int issue_flags)
> >  {
> > @@ -3497,7 +3533,8 @@ static int ublk_ch_batch_io_uring_cmd(struct io_uring_cmd *cmd,
> >                 ret = ublk_handle_batch_fetch_cmd(&data);
> >                 break;
> >         default:
> > -               ret = -EOPNOTSUPP;
> > +               ret = ublk_handle_non_batch_cmd(cmd, issue_flags);
> 
> We should probably skip the if (data.header.q_id >=
> ub->dev_info.nr_hw_queues) check for a non-batch command?

It is true only for UBLK_IO_UNREGISTER_IO_BUF.

> 
> > +               break;
> >         }
> >  out:
> >         return ret;
> > @@ -4163,9 +4200,13 @@ static int ublk_ctrl_add_dev(const struct ublksrv_ctrl_cmd *header)
> >
> >         ub->dev_info.flags |= UBLK_F_CMD_IOCTL_ENCODE |
> >                 UBLK_F_URING_CMD_COMP_IN_TASK |
> > -               UBLK_F_PER_IO_DAEMON |
> > +               (ublk_dev_support_batch_io(ub) ? 0 : UBLK_F_PER_IO_DAEMON) |
> 
> Seems redundant with the logic below to clear UBLK_F_PER_IO_DAEMON if
> (ublk_dev_support_batch_io(ub))?

Good catch.

> 
> >                 UBLK_F_BUF_REG_OFF_DAEMON;
> >
> > +       /* So far, UBLK_F_PER_IO_DAEMON won't be exposed for BATCH_IO */
> > +       if (ublk_dev_support_batch_io(ub))
> > +               ub->dev_info.flags &= ~UBLK_F_PER_IO_DAEMON;
> > +
> >         /* GET_DATA isn't needed any more with USER_COPY or ZERO COPY */
> >         if (ub->dev_info.flags & (UBLK_F_USER_COPY | UBLK_F_SUPPORT_ZERO_COPY |
> >                                 UBLK_F_AUTO_BUF_REG))
> > @@ -4518,6 +4559,13 @@ static int ublk_wait_for_idle_io(struct ublk_device *ub,
> >         unsigned int elapsed = 0;
> >         int ret;
> >
> > +       /*
> > +        * For UBLK_F_BATCH_IO ublk server can get notified with existing
> > +        * or new fetch command, so needn't wait any more
> > +        */
> > +       if (ublk_dev_support_batch_io(ub))
> > +               return 0;
> > +
> >         while (elapsed < timeout_ms && !signal_pending(current)) {
> >                 unsigned int queues_cancelable = 0;
> >                 int i;
> > diff --git a/include/uapi/linux/ublk_cmd.h b/include/uapi/linux/ublk_cmd.h
> > index cd894c1d188e..5e8b1211b7f4 100644
> > --- a/include/uapi/linux/ublk_cmd.h
> > +++ b/include/uapi/linux/ublk_cmd.h
> > @@ -335,6 +335,22 @@
> >   */
> >  #define UBLK_F_BUF_REG_OFF_DAEMON (1ULL << 14)
> >
> > +
> > +/*
> > + * Support the following commands for delivering & committing io command
> > + * in batch.
> > + *
> > + *     - UBLK_U_IO_PREP_IO_CMDS
> > + *     - UBLK_U_IO_COMMIT_IO_CMDS
> > + *     - UBLK_U_IO_FETCH_IO_CMDS
> > + *     - UBLK_U_IO_REGISTER_IO_BUF
> > + *     - UBLK_U_IO_UNREGISTER_IO_BUF
> 
> Seems like it might make sense to provided batched versions of
> UBLK_U_IO_REGISTER_IO_BUF and UBLK_U_IO_UNREGISTER_IO_BUF. That could
> be done in the future, I guess, but it might simplify
> ublk_ch_batch_io_uring_cmd() to only have to handle struct
> ublk_batch_io.

Agree, and it can be added in future.




Thanks,
Ming