diff mbox series

[06/16] RDMA/srpt: Fix handling of command / TMF submission failure

Message ID 20190125183458.220477-7-bvanassche@acm.org (mailing list archive)
State Accepted
Headers show
Series SCSI target patches for kernel v5.1 | expand

Commit Message

Bart Van Assche Jan. 25, 2019, 6:34 p.m. UTC
If submitting an SRP IU to the target core fails, send the SCSI
response "BUSY" to the initiator instead of not sending any
response.

Reviewed-by: Hannes Reinecke <hare@suse.com>
Cc: Doug Ledford <dledford@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/infiniband/ulp/srpt/ib_srpt.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

Comments

Mike Christie Feb. 2, 2019, 12:15 a.m. UTC | #1
On 01/25/2019 12:34 PM, Bart Van Assche wrote:
> If submitting an SRP IU to the target core fails, send the SCSI
> response "BUSY" to the initiator instead of not sending any
> response.
> 
> Reviewed-by: Hannes Reinecke <hare@suse.com>
> Cc: Doug Ledford <dledford@redhat.com>
> Cc: Jason Gunthorpe <jgg@ziepe.ca>
> Cc: Nicholas Bellinger <nab@linux-iscsi.org>
> Cc: Mike Christie <mchristi@redhat.com>
> Cc: Christoph Hellwig <hch@lst.de>
> Signed-off-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  drivers/infiniband/ulp/srpt/ib_srpt.c | 9 ++++-----
>  1 file changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c
> index 8cee8c6c6be9..4fc901d1c0c1 100644
> --- a/drivers/infiniband/ulp/srpt/ib_srpt.c
> +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c
> @@ -1505,7 +1505,7 @@ static void srpt_handle_cmd(struct srpt_rdma_ch *ch,
>  			pr_err("0x%llx: parsing SRP descriptor table failed.\n",
>  			       srp_cmd->tag);
>  		}
> -		goto release_ioctx;
> +		goto busy;
>  	}
>  
>  	rc = target_submit_cmd_map_sgls(cmd, ch->sess, srp_cmd->cdb,
> @@ -1516,13 +1516,12 @@ static void srpt_handle_cmd(struct srpt_rdma_ch *ch,
>  	if (rc != 0) {
>  		pr_debug("target_submit_cmd() returned %d for tag %#llx\n", rc,
>  			 srp_cmd->tag);
> -		goto release_ioctx;
> +		goto busy;
>  	}
>  	return;
>  
> -release_ioctx:
> -	send_ioctx->state = SRPT_STATE_DONE;
> -	srpt_release_cmd(cmd);
> +busy:
> +	target_send_busy(cmd);
>  }
>  
>  static int srp_tmr_to_tcm(int fn)
> 

Can you not do transport_generic_request_failure(cmd, TCM_LUN_BUSY),
because some of the cmd's bits are not yet set?
Bart Van Assche Feb. 2, 2019, 12:53 a.m. UTC | #2
On Fri, 2019-02-01 at 18:15 -0600, Mike Christie wrote:
> On 01/25/2019 12:34 PM, Bart Van Assche wrote:
> > If submitting an SRP IU to the target core fails, send the SCSI
> > response "BUSY" to the initiator instead of not sending any
> > response.
> > 
> > Reviewed-by: Hannes Reinecke <hare@suse.com>
> > Cc: Doug Ledford <dledford@redhat.com>
> > Cc: Jason Gunthorpe <jgg@ziepe.ca>
> > Cc: Nicholas Bellinger <nab@linux-iscsi.org>
> > Cc: Mike Christie <mchristi@redhat.com>
> > Cc: Christoph Hellwig <hch@lst.de>
> > Signed-off-by: Bart Van Assche <bvanassche@acm.org>
> > ---
> >  drivers/infiniband/ulp/srpt/ib_srpt.c | 9 ++++-----
> >  1 file changed, 4 insertions(+), 5 deletions(-)
> > 
> > diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c
> > index 8cee8c6c6be9..4fc901d1c0c1 100644
> > --- a/drivers/infiniband/ulp/srpt/ib_srpt.c
> > +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c
> > @@ -1505,7 +1505,7 @@ static void srpt_handle_cmd(struct srpt_rdma_ch *ch,
> >  			pr_err("0x%llx: parsing SRP descriptor table failed.\n",
> >  			       srp_cmd->tag);
> >  		}
> > -		goto release_ioctx;
> > +		goto busy;
> >  	}
> >  
> >  	rc = target_submit_cmd_map_sgls(cmd, ch->sess, srp_cmd->cdb,
> > @@ -1516,13 +1516,12 @@ static void srpt_handle_cmd(struct srpt_rdma_ch *ch,
> >  	if (rc != 0) {
> >  		pr_debug("target_submit_cmd() returned %d for tag %#llx\n", rc,
> >  			 srp_cmd->tag);
> > -		goto release_ioctx;
> > +		goto busy;
> >  	}
> >  	return;
> >  
> > -release_ioctx:
> > -	send_ioctx->state = SRPT_STATE_DONE;
> > -	srpt_release_cmd(cmd);
> > +busy:
> > +	target_send_busy(cmd);
> >  }
> >  
> >  static int srp_tmr_to_tcm(int fn)
> > 
> 
> Can you not do transport_generic_request_failure(cmd, TCM_LUN_BUSY),
> because some of the cmd's bits are not yet set?

Hi Mike,

That's correct. I think it is only safe to call transport_generic_request_failure()
after a command has been submitted to the target core. target_send_busy() is called
if submitting a command to the target core failed. Hence the following comment above
target_send_busy():

"Note: Only call this function if target_submit_cmd*() failed."

Bart.
diff mbox series

Patch

diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c
index 8cee8c6c6be9..4fc901d1c0c1 100644
--- a/drivers/infiniband/ulp/srpt/ib_srpt.c
+++ b/drivers/infiniband/ulp/srpt/ib_srpt.c
@@ -1505,7 +1505,7 @@  static void srpt_handle_cmd(struct srpt_rdma_ch *ch,
 			pr_err("0x%llx: parsing SRP descriptor table failed.\n",
 			       srp_cmd->tag);
 		}
-		goto release_ioctx;
+		goto busy;
 	}
 
 	rc = target_submit_cmd_map_sgls(cmd, ch->sess, srp_cmd->cdb,
@@ -1516,13 +1516,12 @@  static void srpt_handle_cmd(struct srpt_rdma_ch *ch,
 	if (rc != 0) {
 		pr_debug("target_submit_cmd() returned %d for tag %#llx\n", rc,
 			 srp_cmd->tag);
-		goto release_ioctx;
+		goto busy;
 	}
 	return;
 
-release_ioctx:
-	send_ioctx->state = SRPT_STATE_DONE;
-	srpt_release_cmd(cmd);
+busy:
+	target_send_busy(cmd);
 }
 
 static int srp_tmr_to_tcm(int fn)