On Thu, 2012-01-19 at 13:39 -0800, Andy Grover wrote:
> Retval not very useful, and may even be harmful. Once submitted, fabrics
> should expect a sense error if anything goes wrong. All fabrics checking
> of this retval are useless or broken:
>
> fc checks it just to emit more debug output.
> ib_srpt trickles retval up, then it is ignored.
> qla2xxx trickles it up, which then causes a bug because the abort goto
> in qla_target.c thinks cmd hasn't been sent to target.
>
> Just returning nothing is best.
>
> Signed-off-by: Andy Grover <agrover@xxxxxxxxxx>
> ---
> drivers/infiniband/ulp/srpt/ib_srpt.c | 5 +----
> drivers/scsi/qla2xxx/tcm_qla2xxx.c | 7 ++-----
> drivers/target/target_core_transport.c | 5 ++---
> drivers/target/tcm_fc/tfc_cmd.c | 9 ++-------
> include/target/target_core_fabric.h | 2 +-
> 5 files changed, 8 insertions(+), 20 deletions(-)
>
<SNIP>
> diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
> index 15faca4..cfc4e02 100644
> --- a/drivers/target/target_core_transport.c
> +++ b/drivers/target/target_core_transport.c
> @@ -1633,7 +1633,7 @@ EXPORT_SYMBOL(transport_handle_cdb_direct);
> * This may only be called from process context, and also currently
> * assumes internal allocation of fabric payload buffer by target-core.
> **/
> -int target_submit_cmd(struct se_cmd *se_cmd, struct se_session *se_sess,
> +void target_submit_cmd(struct se_cmd *se_cmd, struct se_session *se_sess,
> unsigned char *cdb, unsigned char *sense, u32 unpacked_lun,
> u32 data_length, int task_attr, int data_dir, int flags)
> {
> @@ -1682,12 +1682,11 @@ int target_submit_cmd(struct se_cmd *se_cmd, struct se_session *se_sess,
> * when fabric has filled the incoming buffer.
> */
> transport_handle_cdb_direct(se_cmd);
> - return 0;
> + return;
>
> out_check_cond:
> transport_send_check_condition_and_sense(se_cmd,
> se_cmd->scsi_sense_reason, 0);
> - return 0;
> }
> EXPORT_SYMBOL(target_submit_cmd);
>
Your patch is fine here..
However my original usage of transport_send_check_condition_and_sense()
is wrong wrt target_submit_cmd() failures. The cmd_kref will have
already been taken by target_get_sess_cmd(), so this needs to be using
transport_generic_request_failure() around sense_check_condition to
handle proper TFO->check_stop_free() kref_puts + response queue_full
cases.
So I'm looking into addressing this breakage in lio-core now, and will
post a patch for 3.3-rc-fixes soon after I can verify the fix with
qla2xxx + ib_srpt failures.
--nab
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
