On Mon, 2016-12-19 at 07:59 -0800, h...@infradead.org wrote: > On Mon, Dec 19, 2016 at 03:33:27PM +0000, Bart Van Assche wrote: > > 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? > > This was my first reaction as well, but it seems like qla2xxx hardware > doesn't pass up the LUN for an abort request. If that's really the > case (which seems really odd to me) we'll need this loop. If there is a > way to get the lun out of the hardware it would be preferable to make > use of that passed up lun.
Hello Christoph, The SCSI Architecture Manual (SAM-6) specifies that the SCSI transport protocol defines whether the scope of the ABORT TASK task management function is I_T_L or I_T. In the Fibre Channel Protocol for SCSI (FCP) document I read that for FC ABORT TASK corresponds to the ABTS-LS frame. As far as I know no LUN information is present in the FC ABTS frame. I think this means that target_submit_tmr() should be modified such that it supports "LUN not specified". Bart.
