kern/90282: SCSI bus resets cause loss of ch device
Jason Bacon
bacon at smithers.neuro.mcw.edu
Mon Dec 12 07:40:09 PST 2005
>Number: 90282
>Category: kern
>Synopsis: SCSI bus resets cause loss of ch device
>Confidential: no
>Severity: serious
>Priority: low
>Responsible: freebsd-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Mon Dec 12 15:40:02 GMT 2005
>Closed-Date:
>Last-Modified:
>Originator: Jason Bacon
>Release: FreeBSD 5.4-RELEASE i386
>Organization:
Medical College of Wisconsin
>Environment:
FreeBSD gizmo.neuro.mcw.edu 5.4-RELEASE FreeBSD 5.4-RELEASE #0: Wed Nov 30 12:13:46 CST 2005 root at gizmo.neuro.mcw.edu:/usr/obj/usr/src/sys/gizmo i386
>Description:
When attempting to use the AIT3 tape drive in our Spectralogic
215 library, SCSI bus resets occur, which are followed by the
loss of /dev/ch0. The library works fine as long as the tape drive
isn't accessed - we can load and return tapes without problems.
The host adapter is a Tekram DC-390U3W. The transfer rate defaults
to 160mb/s:
sym0: <1010-33> port 0xe000-0xe0ff mem 0xf8802000-0xf8803fff,0xf8808000-0xf88083
ff irq 11 at device 9.0 on pci0
sym0: Symbios NVRAM, ID 7, Fast-80, LVD, parity checking
sym0: open drain IRQ line driver, using on-chip SRAM
sym0: using LOAD/STORE-based firmware.
sym0: handling phase mismatch from SCRIPTS.
sym0: SCAN FOR LUNS disabled for targets 2.
sym1: <1010-33> port 0xe400-0xe4ff mem 0xf8800000-0xf8801fff,0xf8804000-0xf88043
ff irq 11 at device 9.1 on pci0
sym1: Symbios NVRAM, ID 7, Fast-80, SE, parity checking
sym1: open drain IRQ line driver, using on-chip SRAM
sym1: using LOAD/STORE-based firmware.
sym1: handling phase mismatch from SCRIPTS.
.
.
.
sa0 at sym0 bus 0 target 2 lun 0
sa0: <SONY SDX-500C 01b1> Removable Sequential Access SCSI-2 device
sa0: 160.000MB/s transfers (80.000MHz, offset 62, 16bit)
Mounting root from ufs:/dev/ad0s1a
ch0 at sym0 bus 0 target 3 lun 0
ch0: <SPECTRA 215 1016> Removable Changer SCSI-2 device
ch0: 3.300MB/s transfers
ch0: 15 slots, 1 drive, 1 picker, 0 portals
>How-To-Repeat:
Load a tape, write for a while, then attempt to unload using
chio. After some period of write operation, a SCSI bus reset will
occur, and /dev/ch0 will be removed, making it impossible to return
the tape, and hence impossible to do multivolume backups.
>Fix:
Restarting the tape library and the operating system both appear to
be necessary to restore /dev/ch0.
WORKAROUND TO PREVENT THE PROBLEM:
Enter the Tekram SCSI BIOS, and reduce the transfer
rate on the library and tape drive to 40 mb/s. Since doing this,
I've been through several backups without incident. Other rates
may work as well, but I haven't tested them all. 40 is fast enough
to keep the drive streaming and slow enough to solve the bus reset
issue.
After adjusting the BIOS:
sa0 at sym0 bus 0 target 2 lun 0
sa0: <SONY SDX-500C 01b1> Removable Sequential Access SCSI-2 device
sa0: 40.000MB/s transfers (20.000MHz, offset 31, 16bit)
Mounting root from ufs:/dev/ad0s1a
ch0 at sym0 bus 0 target 3 lun 0
ch0: <SPECTRA 215 1016> Removable Changer SCSI-2 device
ch0: 3.300MB/s transfers
ch0: 15 slots, 1 drive, 1 picker, 0 portals
>Release-Note:
>Audit-Trail:
>Unformatted:
More information about the freebsd-bugs
mailing list