Sun Fire X4600 M2 Server
 
Sun Fire X2100 Server
 
Sun Fire X4100 M2 Server
 
Sun Blade X8400 Server Module
 
Sun Blade X6220 Server Module
 
Sun Ultra 20 M2 Workstation
 
Sun Fire X4640 Server
 
Sun Blade X6420 Server Module
 
Sun Fire X4100 Server
 
Sun Fire X2100 M2 Server
 
Sun Java Workstation W2100z
 
Sun Blade X6440 Server Module
 
Sun Blade X8420 Server Module
 
Sun Fire X4240 Server
 
Sun Fire X4600 Server
 
Sun Fire V20z Server
 
Sun Fire X4200 Server
 
Sun Blade 6000 System
 
Sun Fire V40z Server
 
Sun Fire X2200 M2 Server
 
Sun Netra X4200 M2 Server
 
Sun Fire X4500 Server
 
Sun Blade 8000 System
 
Sun Ultra 20 Workstation
 
Sun Blade X8440 Server Module
 
Sun Fire X4540 Server
 
Sun Ultra 40 M2 Workstation
 
Solaris x64/x86 Operating System
 
Sun Java Workstation W1100z
 
Sun Fire X4440 Server
 
Sun Blade X6240 Server Module
 

GCS>Sun Microsystems>Servers>x64 Servers
 

Applies to:

Goal

When an unrecoverable Machine Check Errors is detected, a Machine Check Exception is generated.

When such condition occurs, Solaris[TM] Operating System for x86 Platforms can panic with a Trap of type 0x12 (Machine check exception). When the Solaris{TM} Operating System (OS) receives a Machine check exception, it shows MCE warning messages just before the panic.

Examples of the warning:

WARNING: MCE: Bank 0: error code 15:addr = cf53c000, model errcode = 0
WARNING: MCE: Bank 2: error code 152:addr = e748, model errcode = 2
WARNING: MCE: Bank 4: error code 0xf0f, mserrcode = 0x7

The meaning of MCE messages depends on processors. This document explains how to understand MCE messages on AMD Opteron based systems.

Solution

An MCE message has 3 or 4 values; bank, error code, address (not always exist) and model error code. All values are displayed as hexadecimal numbers regardless of existence of the prefix "0x".

WARNING: MCE: Bank 0: error code 15:addr = cf53c000, model errcode = 0

- Bank (THIS HAS NOTHING TO DO WITH MEMORY BANK)

Opteron processors have five error reporting banks associated with specific hardware blocks.

• Data Cache unit(DC) - Includes the cache structures that hold data and tags, the data TLBs, and the data cache probing logic.
• Instruction cache unit(IC) - Includes the instruction cache structures that hold instructions and tags, the instruction TLBs, and the instruction cache probing logic.
• Bus unit(BU) - Includes the system bus interface to the Northbridge and the level 2 cache.
• Load/store unit(LS) - Includes logic used to manage loads and stores.
• Northbridge unit(NB) - Includes the Northbridge and DRAM controller.

These banks correspond to Bank 0-4 in order.

- Error code

Error code field shows 16bit MCA (Machine Check Architecture) Error Code contained in Machine Check Status Registers. The MCA Error Code has the following format.

      Error Value     Error Type        Description
-------------------  -------------- ----------------------------------
0000 0000 0001 TTLL  TLB errors     Errors in the Gart TLB cache
0000 0001 RRRR TTLL  Memory errors  Errors in the cache hierarchy
0000 1PPT RRRR IILL  Bus Errors     General bus errors including errors
in the HyperTransport link or DRAM

 Transaction Type Bits (TT)
00 Instruction
01 Data
10 Generic
11 reserved

 Cache Level Bits (LL)
00 Level 0
01 Level 1
10 Level 2
11 Generig

 Participation Processor Bits (PP)
00 Local node originated the request
01 Local node responded to the request
10 Local node observed error as 3rd party
11 Generic

 Time-out Bit (T)
0 Request did not time out
1 Request timed out

 Memory Transaction Type Bits (RRRR)
0000 Generic error
0001 Generic read
0010 Generic write
0011 Data read
0100 Data write
0101 Instruction fetch
0110 Prefetch
0111 Evict
1000 Snoop

 Memory or I/O Bits (II)
00 Memory access
01 reserved
10 I/O access
11 Generic

- Address

Address field shows the address where the Machine check exception occurs.

- Model error code

Model error code field shows 4bit Extended Error Code contained in Machine Check Status Registers.

The meaning is:

 DC/IC
0000 TLB parity error in physical array
0001 TLB parity error in virtual array (multi-match error)

 BU
0000 Bus or cache data array error
0010 Cache tag array error

 LS Reserved

 NB
0000 ECC error
0001 CRC error
0010 Sync error
0011 Master Abort
0100 Target Abort
0101 GART error
0110 RMW error
0111 Watchdog error
1000 ChipKill ECC error

Detailed information to decode the message can be obtained from AMD's document

"BIOS and Kernel Developer's Guide for AMD Athlon 64 and AMD Opteron Processors"

Decode examples:

  WARNING: MCE: Bank 0: error code 15:addr = cf53c000, model errcode = 0

  TLB error at L1 cache detected by Data Cache Unit.

  WARNING: MCE: Bank 2: error code 152:addr = e748, model errcode = 2

  Tag parity error during an instruction fetch at L2 cache detected by Bus Unit.

  WARNING: MCE: Bank 4: error code 0xf0f, mserrcode = 0x7

  Watchdog error detected by Northbridge.

References

BIOS and Kernel Developer's Guide for AMD Athlon 64 and AMD Opteron Processors

AMD64 Architecture Programmer's Manual Volume 2: System Programming

Other MCE Related InfoDocs

Technical Instruction - KERNEL: Understanding Trap Type 18 (0x12) Panics on x86 platform
Technical Instruction - Understanding and Decoding Machine Check Errors on Sun Fire[TM] V20z/Sun Fire[TM] V40z running Red Hat OS

This solution has no attachment