On 11/13/19 5:42 AM, Roman Bolshakov wrote:
On Mon, Nov 11, 2019 at 07:57:52PM -0800, Bart Van Assche wrote:
- target_get_sess_cmd(&cmd->se_cmd, true);
+ if (target_get_sess_cmd(&cmd->se_cmd, true) < 0)
+ return iscsit_add_reject_cmd(cmd,
+ ISCSI_REASON_BOOKMARK_NO_RESOURCES, buf);
Sending the reject reason implies initiator can resend original PDU,
according to https://tools.ietf.org/html/rfc7143#section-11.17.1
Is it intended, i.e. do we want the retry?
Hi Roman,
The two new iscsit_add_reject_cmd() can only be triggered if the
initiator sends one or more iSCSI PDUs after the logout PDU. I think
that's a violation of the iSCSI protocol. Anyway, how about changing the
reject reason into ISCSI_REASON_PROTOCOL_ERROR?
Thanks,
Bart.
