LSI SAS1064E and LSI SAS1068E Controllers Revisions B1, B2 and B3 may Cause a System Panic

Asset ID:	1-77-1168826.1
Update Date:	2014-02-13
Keywords:

Solution Type Sun Alert Sure

Solution 1168826.1 : LSI SAS1064E and LSI SAS1068E Controllers Revisions B1, B2 and B3 may Cause a System Panic

Applies to:

OpenSolaris Operating System
Sun SPARC Enterprise T5240 Server
Sun SPARC Sun OS
_________________________________________

Date of Workaround Release: 06-Aug-2010

Date of Resolved Release:11-Oct-2011
_________________________________________

Description

On Solaris 10 and OpenSolaris systems with revisions B1, B2 or B3 of Serial Attached SCSI controllers LSI SAS1064E or LSI SAS1068E installed, the mpt(7D) driver may trigger a panic when a PCI memory read crosses or ends on a 4-gigabyte boundary.

Note: LSI SAS1064E and LSI SAS1068E controllers are used both internally and externally as PCI-Express adapters.

Occurrence

This issue can occur on the following releases:

Systems with revision B1 of the Serial Attached SCSI controller LSI SAS1064E or LSI SAS1068E:

SPARC Platform

Solaris 10 without patch 147701-01

x86 Platform

Solaris 10 without patch 147702-01

Systems with revisions B2/B3 of the Serial Attached SCSI controllers LSI SAS1064E or LSI SAS1068E:

SPARC Platform

Solaris 10 without patch 143128-04
OpenSolaris based upon builds snv_01 through snv_128

x86 Platform

Solaris 10 without patch 143129-05
OpenSolaris based upon builds snv_01 through snv_134

Note 1: Solaris 8 and 9 are not impacted by this issue.

Note 2: This issue only occurs on systems with revision B1, B2 or B3 of Serial Attached SCSI controllers LSI SAS1064E or LSI SAS1068E. To determine if a system has these revision of the controllers installed, execute the following command:

    $ prtpicl -vc scsi-2 | egrep 'vendor|device-id|revision'
    :vendor-id     0x1000
    :device-id     0x58
    :revision-id   0x4
    :subsystem-vendor-id   0x1000

If the vendor-id is 0x1000 and device-id is 0x56 or 0x58 and revision-id is 0x2, 0x4 or 0x8 then the system is vulnerable to the issue.

0x2 represents B1 <- CR7026264 ==> Oracle BugID: 15701922
0x4 represents B2 <- CR6863127 ==> Oracle BugID: 15577476
0x8 represents B3 <- CR6863127 ==> Oracle BugID: 15577476

Symptoms

Below is a sample panic signature:

    panic[cpu4]/thread=2a100db3ca0: Fatal error has occurred in: PCIe fabric.(0x0)(0x41)

    000002a100e1bbc0 px:px_err_panic+1ac (1947400, 1359800, 41, 2a100e1bc70, 0, 0)
    %l0-3: 0000000000001001 0000000001947800 0000000000000000 0000000000000001
    %l4-7: 0000000000000000 0000000001875c00 0000000000000001 0000000000000000
    000002a100e1bcd0 px:px_err_fabric_intr+1b4 (30001d438c0, 0, 310000000000000, 1, 41, 310)
    %l0-3: 0000000000000000 0000000001947ad0 0000000001947800 0000000000000054
    %l4-7: 0000000001947ab8 0000000001947800 0000000001947ab0 0000000001947800
    000002a100e1be40 px:px_msiq_intr+1e4 (300014dc188, 30001d6ab18, 134c2b4, 0, 1, 3
    0004833e60)
    ...

Workaround

There is no work around for this issue.

This issue is addressed in the following releases:

Systems with revision B1 of the Serial Attached SCSI controllers LSI SAS1064E or LSI SAS1068E:

SPARC Platform

Solaris 10 with patch 147701-01 or later

x86 Platform

Solaris 10 without patch 147702-01 or later

Systems with revisions B2/B3 of the Serial Attached SCSI controllers LSI SAS1064E or LSI SAS1068E:

SPARC Platform

Solaris 10 with patch 143128-04 or later
OpenSolaris based upon builds snv_129 or later

x86 Platform

Solaris 10 with patch 143129-05 or later
OpenSolaris based upon builds snv_135 or later

Patches

<SUNPATCH:143128-04>
<SUNPATCH:143129-05>

Submitter: fred.herard@oracle.com
Responsible Engineer: fred.herard@oracle.com, vidhya.gopalan@oracle.com
BugID 7026264 addresses this issue for B1 and is being handled by vidhya.gopalan@oracle.com
Patches for this issue not expected to make update 10 so will not be available until October 2011.
Note: Serial Attached SCSI controllers LSI SAS1064E
and LSI SAS1068E are used as both internal controllers
and as external controllers as PCI-Express adapters.
Here are some examples:

Option SG-XPCIE4SAS-Z (LSI SAS164E)
Option SG-XPCIE8SAS-E-Z (LSI SAS1068E - Pandora)

As such, the only safe way to know if systems are affected
is to use the command given above.

These controllers are also commonly used in many systems
as listed in the Sun System Handbook.
As of 13 Aug, this list is:

SAS1064E

Workstations: Ultra 20 M2, 24, 40
Servers: Sun Fire X2100 M2, X2200 M2
and X4140, X4150, X4170, X4240, X4250,
X4270, X4275, X4440, X4450

SAS1068E is the 8 port 3gb SAS/SATA HBA

Sun Fire V215, V245, V445
Sun Fire X2100, X2100 M2, X2200 M2
Sun Fire X4100 M2, X4140, X4150, X4170
Sun Fire X4200, X4200 M2, X4240, X4250, X4270, X4275
Sun Fire X4440, X4450, X4540, X4600, X4600 M2, X4640
Sun Fire T1000, T2000
SPARC Enterprise M3000, M4000, M5000, M8000, M9000
SPARC Enterprise T5120, T5140, T5220, T5240, T5440

Storage J4200, J4400, J4500
Sun Storage 7310, 7410, F5100
StorageTek 2530

Netra T5220, T5440
Netra X4200 M2, X4250, X4450

History

06-Aug-2010: Date of Workaround release
18-Aug-2010: Updated Description statement with additional Note
04-Apr-2011: Added BugID. Updated Description, Occurrence, and Workaround sections
10-Oct-2011: Updated the Occurrence, and Workaround sections
07-Nov-2011: Updated Occurrence section
30-Mar-2012: Updated Occurrence section
30-Jul-2012: Maintenance update, no change in content
14-May-2013: Added bug reference to Oracle bugIDs in Occurrence section
13-Feb-2014: Maintenance/formatting update; no change in content

References

Attachments

This solution has no attachment