On Sun, 2016-12-18 at 01:02 -0800, Himanshu Madhani wrote:
> diff --git a/drivers/scsi/qla2xxx/tcm_qla2xxx.c
> b/drivers/scsi/qla2xxx/tcm_qla2xxx.c
> index 6643f6f..9275f36 100644
> --- a/drivers/scsi/qla2xxx/tcm_qla2xxx.c
> +++ b/drivers/scsi/qla2xxx/tcm_qla2xxx.c
> @@ -567,6 +567,30 @@ static int tcm_qla2xxx_handle_tmr(struct
> qla_tgt_mgmt_cmd *mcmd, uint32_t lun,
> {
> struct qla_tgt_sess *sess = mcmd->sess;
> struct se_cmd *se_cmd = &mcmd->se_cmd;
> + struct se_session *se_sess = sess->se_sess;
> + bool found_lun = false;
> +
> + switch (tmr_func) {
> + case TMR_ABORT_TASK:
> + spin_lock(&se_sess->sess_cmd_lock);
> + list_for_each_entry(se_cmd, &se_sess->sess_cmd_list,
> se_cmd_list) {
> + struct qla_tgt_cmd *cmd =
> + container_of(se_cmd, struct qla_tgt_cmd,
> se_cmd);
> + struct abts_recv_from_24xx *abts =
> &mcmd->orig_iocb.abts;
> +
> + if (se_cmd->tag == abts->exchange_addr_to_abort) {
> + lun = cmd->unpacked_lun;
> + found_lun = true;
> + break;
> + }
> + }
> + spin_unlock(&se_sess->sess_cmd_lock);
> + if (!found_lun)
> + return -ENOBUFS;
> + break;
> + default:
> + break;
> + }
>
> return target_submit_tmr(se_cmd, sess->se_sess, NULL, lun, mcmd,
> tmr_func, GFP_ATOMIC, tag, TARGET_SCF_ACK_KREF);
Hello Himanshu,
Please consider removing the sess_cmd_list loop. Any lookups in
sess_cmd_list should be performed by the target core and not by a
target driver. Are you aware that core_tmr_abort_task() performs a very
similar lookup to the one above?
Bart.
