提交 cc456b65 编写于 作者: C Christoph Hellwig

nvme-fabrics: allow internal passthrough command on deleting controllers

Without this we can't cleanly shut down.

Based on analysis an an earlier patch from Hannes Reinecke.

Fixes: bb06ec31 ("nvme: expand nvmf_check_if_ready checks")
Reported-by: NHannes Reinecke <hare@suse.de>
Tested-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Reviewed-by: NJames Smart <james.smart@broadcom.com>
上级 fe4a9791
......@@ -545,71 +545,54 @@ blk_status_t nvmf_check_if_ready(struct nvme_ctrl *ctrl, struct request *rq,
return BLK_STS_OK;
switch (ctrl->state) {
case NVME_CTRL_DELETING:
goto reject_io;
case NVME_CTRL_NEW:
case NVME_CTRL_CONNECTING:
case NVME_CTRL_DELETING:
/*
* This is the case of starting a new or deleting an association
* but connectivity was lost before it was fully created or torn
* down. We need to error the commands used to initialize the
* controller so the reconnect can go into a retry attempt. The
* commands should all be marked REQ_FAILFAST_DRIVER, which will
* hit the reject path below. Anything else will be queued while
* the state settles.
*/
if (!is_connected)
/*
* This is the case of starting a new
* association but connectivity was lost
* before it was fully created. We need to
* error the commands used to initialize the
* controller so the reconnect can go into a
* retry attempt. The commands should all be
* marked REQ_FAILFAST_DRIVER, which will hit
* the reject path below. Anything else will
* be queued while the state settles.
*/
goto reject_or_queue_io;
if ((queue_live &&
!(nvme_req(rq)->flags & NVME_REQ_USERCMD)) ||
(!queue_live && blk_rq_is_passthrough(rq) &&
cmd->common.opcode == nvme_fabrics_command &&
cmd->fabrics.fctype == nvme_fabrics_type_connect))
/*
* If queue is live, allow only commands that
* are internally generated pass through. These
* are commands on the admin queue to initialize
* the controller. This will reject any ioctl
* admin cmds received while initializing.
*
* If the queue is not live, allow only a
* connect command. This will reject any ioctl
* admin cmd as well as initialization commands
* if the controller reverted the queue to non-live.
*/
break;
/*
* If queue is live, allow only commands that are internally
* generated pass through. These are commands on the admin
* queue to initialize the controller. This will reject any
* ioctl admin cmds received while initializing.
*/
if (queue_live && !(nvme_req(rq)->flags & NVME_REQ_USERCMD))
return BLK_STS_OK;
/*
* fall-thru to the reject_or_queue_io clause
* If the queue is not live, allow only a connect command. This
* will reject any ioctl admin cmd as well as initialization
* commands if the controller reverted the queue to non-live.
*/
if (!queue_live && blk_rq_is_passthrough(rq) &&
cmd->common.opcode == nvme_fabrics_command &&
cmd->fabrics.fctype == nvme_fabrics_type_connect)
return BLK_STS_OK;
break;
/* these cases fall-thru
* case NVME_CTRL_LIVE:
* case NVME_CTRL_RESETTING:
*/
default:
break;
}
reject_or_queue_io:
/*
* Any other new io is something we're not in a state to send
* to the device. Default action is to busy it and retry it
* after the controller state is recovered. However, anything
* marked for failfast or nvme multipath is immediately failed.
* Note: commands used to initialize the controller will be
* marked for failfast.
* Any other new io is something we're not in a state to send to the
* device. Default action is to busy it and retry it after the
* controller state is recovered. However, anything marked for failfast
* or nvme multipath is immediately failed. Note: commands used to
* initialize the controller will be marked for failfast.
* Note: nvme cli/ioctl commands are marked for failfast.
*/
if (!blk_noretry_request(rq) && !(rq->cmd_flags & REQ_NVME_MPATH))
return BLK_STS_RESOURCE;
reject_io:
nvme_req(rq)->status = NVME_SC_ABORT_REQ;
return BLK_STS_IOERR;
}
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册