Bug 201157 - [cam] mpt stuck in loop attempting to abort
Summary: [cam] mpt stuck in loop attempting to abort
Status: New
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: 10.1-STABLE
Hardware: Any Any
: --- Affects Only Me
Assignee: freebsd-bugs (Nobody)
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2015-06-28 07:21 UTC by G. Paul Ziemba
Modified: 2015-06-28 07:21 UTC (History)
0 users

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description G. Paul Ziemba 2015-06-28 07:21:48 UTC 
% uname -vm
FreeBSD 10.1-STABLE #1 r280974M: Thu Apr  2 23:07:57 PDT 2015     root@hairball:/usr/obj/usr/src/sys/GPZ-280974  amd64

% pciconf -lv mpt0
mpt0@pci0:16:8:0:       class=0x010000 card=0x50c01000 chip=0x00301000 rev=0xc1 hdr=0x00
    vendor     = 'LSI Logic / Symbios Logic'
    device     = '53c1030 PCI-X Fusion-MPT Dual Ultra320 SCSI'
    class      = mass storage
    subclass   = SCSI

DMESG output:
    mpt0: <LSILogic 1030 Ultra4 Adapter> port 0xa000-0xa0ff mem 0xf3720000-0xf373ffff,0xf3700000-0xf371ffff irq 16 at device 8.0 on pci16
    mpt0: MPI Version=1.2.14.0

Attached via SCSI is a Dell Powervault 122T (LTO2) tape library which contains:
    sa1: <IBM ULTRIUM-TD2 53Y3>
    ch1: <DELL PV-122T K17r>

After powering on the tape library and waiting two minutes, I ran "camcontrol rescan 1" which got stuck. The following syslog messages appeared:

Jun 27 23:37:01 hairball kernel: sa1 at mpt0 bus 0 scbus1 target 4 lun 0
Jun 27 23:37:01 hairball kernel: sa1: <IBM ULTRIUM-TD2 53Y3> Removable Sequential Access SCSI-3 device
Jun 27 23:37:01 hairball kernel: sa1: Serial Number 1110289184
Jun 27 23:37:01 hairball kernel: sa1: 160.000MB/s transfers
Jun 27 23:37:01 hairball kernel: (80.000MHz DT, offset 31, 16bit)
Jun 27 23:37:02 hairball kernel: (probe0:mpt0:0:9:0): INQUIRY. CDB: 12 00 00 00 24 00 
Jun 27 23:37:02 hairball kernel: (probe0:mpt0:0:9:0): CAM status: Target Bus Phase Sequence Failure
Jun 27 23:37:02 hairball kernel: (probe0:mpt0:0:9:0): Retrying command
Jun 27 23:37:07 hairball kernel: mpt0: request 0xfffffe0000fe1148:1120 timed out for ccb 0xfffff80186f8f800 (req->ccb 0xfffff80186f8f800)
Jun 27 23:37:07 hairball kernel: mpt0: attempting to abort req 0xfffffe0000fe1148:1120 function 0
Jun 27 23:37:07 hairball kernel: mpt0: abort of req 0xfffffe0000fe1148:1120 completed

The last two lines (attempting to abort; abort completed) repeated endlessly. After 30 minutes, I powered off the the tape library. The camcontrol command completed, and the final syslog messages were:

Jun 28 00:07:17 hairball kernel: mpt0: attempting to abort req 0xfffffe0000fe2020:1145 function 0
Jun 28 00:07:17 hairball kernel: mpt0: completing timedout/aborted req 0xfffffe0000fe2020:1145
Jun 28 00:07:17 hairball kernel: mpt0: abort of req 0xfffffe0000fe2020:0 completed

This behavior is inconsistent. Sometimes everything works as expected; other times the system behaves as described above.