With recovery feature enabled, if ublk chardev is ready to be released
and quiesce_work has been scheduled, we:
(1) reinit all ubqs, including:
(a) put the task_struct and reset ->ubq_daemon to NULL.
(b) reset all ublk_io.
(2) reset ub->mm to NULL.
Then ublk chardev is released and new process can open it.
RECOVER_DEV is introduced as a new ctrl-cmd for recovery feature.
After the chardev is opened and all ubqs are ready, user should send
RECOVER_DEV to:
(1) wait until all new ubq_daemons getting ready.
(2) update ublksrv_pid
(3) unquiesce the request queue and expect incoming ublk_queue_rq()
(4) convert ub's state to UBLK_S_DEV_LIVE
(5) reschedule monitor_work
Signed-off-by: ZiyangZhang <ZiyangZhang@xxxxxxxxxxxxxxxxx>
---
drivers/block/ublk_drv.c | 67 +++++++++++++++++++++++++++++++++++
include/uapi/linux/ublk_cmd.h | 1 +
2 files changed, 68 insertions(+)
diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
index 3977869d2bc4..ac8bf497567f 100644
--- a/drivers/block/ublk_drv.c
+++ b/drivers/block/ublk_drv.c
@@ -898,10 +898,40 @@ static int ublk_ch_open(struct inode *inode, struct file *filp)
return 0;
}
+static void ublk_queue_reinit(struct ublk_device *ub, struct ublk_queue *ubq)
+{
+ int i;
+
+ /* old daemon is PF_EXITING, put it now */
+ if (ubq->ubq_daemon) {
+ put_task_struct(ubq->ubq_daemon);
+ /* We have to reset it to NULL, otherwise ub won't accept new FETCH_REQ */
+ ubq->ubq_daemon = NULL;
+ }
+
+ for (i = 0; i < ubq->q_depth; i++) {
+ struct ublk_io *io = &ubq->ios[i];
+
+ /* forget everything now and be ready for new FETCH_REQ */
+ io->flags = 0;
+ io->cmd = NULL;
+ io->addr = 0;
+ }
+ ubq->nr_io_ready = 0;
+}
+
static int ublk_ch_release(struct inode *inode, struct file *filp)
{
struct ublk_device *ub = filp->private_data;
+ int i;
+ pr_devel("%s: reinit queues for dev id %d.\n", __func__, ub->dev_info.dev_id);
+ for (i = 0; i < ub->dev_info.nr_hw_queues; i++)
+ ublk_queue_reinit(ub, ublk_get_queue(ub, i));
+ /* set to NULL, otherwise new ubq_daemon cannot mmap the io_cmd_buf */
+ ub->mm = NULL;
+ ub->nr_queues_ready = 0;
+ init_completion(&ub->completion);
clear_bit(UB_STATE_OPEN, &ub->state);
return 0;
}
@@ -1873,6 +1903,40 @@ static int ublk_ctrl_set_params(struct io_uring_cmd *cmd)
return ret;
}
+static int ublk_ctrl_recover_dev(struct io_uring_cmd *cmd)
+{
+ struct ublksrv_ctrl_cmd *header = (struct ublksrv_ctrl_cmd *)cmd->cmd;
+ int ublksrv_pid = (int)header->data[0];
+ struct ublk_device *ub;
+ int ret = -EINVAL;
+
+ ub = ublk_get_device_from_id(header->dev_id);
+ if (!ub)
+ return ret;
+
+ /* wait until new ubq_daemon sending all FETCH_REQ */
+ wait_for_completion_interruptible(&ub->completion);
+
+ mutex_lock(&ub->mutex);
+ if (!ublk_can_use_recovery(ub))
+ goto out_unlock;
+
+ if (ub->dev_info.state != UBLK_S_DEV_QUIESCED) {
+ ret = -EBUSY;
+ goto out_unlock;
+ }
+ ub->dev_info.ublksrv_pid = ublksrv_pid;
+ blk_mq_unquiesce_queue(ub->ub_disk->queue);
+ blk_mq_kick_requeue_list(ub->ub_disk->queue);
+ ub->dev_info.state = UBLK_S_DEV_LIVE;
+ schedule_delayed_work(&ub->monitor_work, UBLK_DAEMON_MONITOR_PERIOD);
+ ret = 0;
+ out_unlock:
+ mutex_unlock(&ub->mutex);
+ ublk_put_device(ub);
+ return ret;
+}
+
static int ublk_ctrl_uring_cmd(struct io_uring_cmd *cmd,
unsigned int issue_flags)
{
@@ -1914,6 +1978,9 @@ static int ublk_ctrl_uring_cmd(struct io_uring_cmd *cmd,
case UBLK_CMD_SET_PARAMS:
ret = ublk_ctrl_set_params(cmd);
break;
+ case UBLK_CMD_RECOVER_DEV:
+ ret = ublk_ctrl_recover_dev(cmd);
+ break;
default:
break;
}
diff --git a/include/uapi/linux/ublk_cmd.h b/include/uapi/linux/ublk_cmd.h
index 332370628757..deb674f1fbc9 100644
--- a/include/uapi/linux/ublk_cmd.h
+++ b/include/uapi/linux/ublk_cmd.h
@@ -17,6 +17,7 @@
#define UBLK_CMD_STOP_DEV 0x07
#define UBLK_CMD_SET_PARAMS 0x08
#define UBLK_CMD_GET_PARAMS 0x09
+#define UBLK_CMD_RECOVER_DEV 0x10
/*
* IO commands, issued by ublk server, and handled by ublk driver.
--
2.27.0
