kern/55960: IO to PATA drives on Promise SATA150 TX2plus (PDC20375) causes kernel lock-up

[Jesper Monsted]

>Number:         55960
>Category:       kern
>Synopsis:       IO to PATA drives on Promise SATA150 TX2plus (PDC20375) causes kernel lock-up
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Mon Aug 25 09:00:18 PDT 2003
>Closed-Date:
>Last-Modified:
>Originator:     Jesper Monsted
>Release:        FreeBSD 5.1-RELEASE-p2 i386
>Organization:
none
>Environment:
System: FreeBSD  5.1-RELEASE-p2 FreebSD 5.1-RELEASE-p2 #2: Sun Aug 24 22:45:33 CEST 2003    root at chef:/usr/obj/usr/src/sys/CHEF  i386

P3-1000, 512 MB RAM, Adaptec 2940UW SCSI controller, Promise Ultra133 TX2 (2xATA133), Promise SATA150 TX2plus (2xSATA, 1xATA133) controllers, one SCSI, five ATA, two SATA drives, one ATA CDrom.

# atacontrol list
ATA channel 0:
    Master:  ad0 <IC35L040AVER07-0/ER4OA44A> ATA/ATAPI rev 5
    Slave:   ad1 <IC35L040AVER07-0/ER4OA46A> ATA/ATAPI rev 5
ATA channel 1:
    Master: acd0 <PLEXTOR CD-R PX-W4012A/1.00> ATA/ATAPI rev 0
    Slave:       no device present
ATA channel 2:
    Master:  ad4 <ST3160023AS/3.05> ATA/ATAPI rev 6
    Slave:       no device present
ATA channel 3:
    Master:  ad6 <ST3160023AS/3.05> ATA/ATAPI rev 6
    Slave:       no device present
ATA channel 4:
    Master:  ad8 <Maxtor 4G160J8/GAK819K0> ATA/ATAPI rev 6
    Slave:       no device present
ATA channel 5:
    Master: ad10 <Maxtor 6Y160P0/YAR41VW0> ATA/ATAPI rev 7
    Slave:       no device present
ATA channel 6:
    Master: ad12 <Maxtor 4G160J8/GAK819K0> ATA/ATAPI rev 6
    Slave:       no device present

>Description:
Any command generating IO on the drives connected to the ATA133 channel of the SATA150 controller triggers a lock-up. There is no output on the terminal and the only option is a hard reset. First, it seemed to be a problem with vinum, but later study found it to be reproducable with 'dd if=/dev/ad8s1e of=/dev/null bs=1m' and similar. 
Similar commands running on the SATA drives work flawlessly.

The drive in question is known good, since it was just taken off another controller and moved to the SATA one.

The problem was seen on both 5.1-RELEASE and CURRENT.

4.x was not tested since it didn't support the SATA controller.

The controller has been replaced with the same result.

>How-To-Repeat:
dd if=/dev/ad8s1e of=/dev/null bs=1m
Most traffic to a vinum array containing drive on the failing channel.

>Fix:

None known.

>Release-Note:
>Audit-Trail:
>Unformatted: