On Thu, 2012-02-16 at 06:26 +0800, Jim Barber wrote: > On 16/02/2012 5:30 AM, Nicholas A. Bellinger wrote: > > Hi Jim, > > > > Thanks for the quick turn-around.. > > > >> Feb 15 13:49:31 san kernel: [70231.165867] ABORT_TASK: Found referenced qla2xxx task_tag: 1160872 > >> Feb 15 13:49:31 san kernel: [70231.182553] qla2xxx 0000:0b:00.0: qla2x00_error_entry: error entry, type 0x12 status 0x8 > >> Feb 15 13:49:31 san kernel: [70231.436017] ABORT_TASK: Sending TMR_FUNCTION_COMPLETE for ref_tag: 1160872 > >> Feb 15 13:49:35 san kernel: [70234.529340] qla2xxx 0000:0b:00.0: qla2x00_error_entry: error entry, type 0x12 status 0x8 > > > > Strange.. So the ABORT_TASK is received, and immediately (in less than > > 1 sec) locates ref_tag: 1160872, and sends a TMR_FUNCITON_COMPLETION. > > > > Which would lead me to believe that ref_tag: 1160872 is *not* actually > > outstanding in the backend device, but is actually in the qla HW > > completion queue waiting to be pushed back onto the wire.. > > > > What I'm still not able to reproduce on my side with qla2xxx Linux > > clients is the bit: > > > > qla2xxx 0000:0b:00.0: qla2x00_error_entry: error entry, type 0x12 status 0x8 > > > > Which indicated an error entry for type CTIO_TYPE7 (likely the timed out > > completion I/O with ref_tag: 1160872). > > > > Just to make sure, can you verify which firmware that you're using with > > qla2xxx on the target side..? This value will appear in dmesg during > > 'modprobe qla2xxx' time like so: > > > > [ 3253.670910] scsi20 : qla2xxx > > [ 3253.674503] qla2xxx 0000:03:00.0: skipping scsi_scan_host() for non-initiator port > > [ 3253.682974] qla2xxx [0000:03:00.0]-00fa:20: QLogic Fibre Channed HBA Driver: 8.03.07.12-k. > > [ 3253.692186] qla2xxx [0000:03:00.0]-00fb:20: QLogic QLE2562 - PCI-Express Dual Channel 8Gb Fibre Channel HBA. > > [ 3253.703150] qla2xxx [0000:03:00.0]-00fc:20: ISP2532: PCIe (5.0GT/s x4) @ 0000:03:00.0 hdma+ host#=20 fw=5.06.01 (90d5). > > Hi Nicholas. > > Here are the results you're looking for. > I made sure the 3ware card had the latest firmware put on it only a matter of months ago (time flies but pretty sure within the past 6 months). > > [ 1.612442] scsi1 : qla2xxx > [ 1.612830] qla2xxx 0000:09:00.0: skipping scsi_scan_host() for non-initiator port > [ 1.612930] qla2xxx [0000:09:00.0]-00fa:1: QLogic Fibre Channed HBA Driver: 8.03.07.12-k. > [ 1.613024] qla2xxx [0000:09:00.0]-00fb:1: QLogic QLE2462 - PCI-Express Dual Channel 4Gb Fibre Channel HBA. > [ 1.613132] qla2xxx [0000:09:00.0]-00fc:1: ISP2432: PCIe (2.5GT/s x4) @ 0000:09:00.0 hdma+ host#=1 fw=4.04.05 (482). > Ok, 4.x.x is a fairly old firmware revision for target mode usage at this point. Please go ahead and grab the latest ql2400_fw.bin here, and put it into your /lib/firmware/ ftp://ftp.qlogic.com/outgoing/linux/firmware/ql2400_fw.bin Then please verify you're able to run with 5.06.xx firmware, and then retest again. Thanks! --nab P.S: In order for the fw to be picked up at boot time, you'll also need to rebuild initrd -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
