ecc(ADM)

ecc, eccd -- memory Error Correction Code (ECC) facility

Syntax

/etc/ecc

Description

The memory Error Correction Code (ECC) utilities periodically check RAM for single and double-bit errors to increase data integrity. This feature is specific to Corollary smp RAM used in Corollary and Corollary-compatible systems.

The ECC daemon: eccd

The ECC daemon, or background program, scans the smp RAM checking for single bit errors. Single bit errors themselves are harmless and are automatically corrected by hardware. However, if an additional bit is corrupted at the same location, a double bit error occurs and the system panics.

The ECC daemon helps avoid double bit errors by informing the system administrator of existing single bit errors. Errors are reported via the system console and /usr/adm/messages. The system administrator should periodically check /usr/adm/messages for any single bit error notifications and use the ecc utility to map the affected 4KB page out of memory.

The script that controls this process, /etc/idrc.d/ecc, is built manually using mkdev eccd. The /etc/idrc.d/ecc script is thereafter invoked automatically when the system enters multiuser mode.

The ecc utility

The administrator should check periodically for memory ECC errors. The ECC errors are displayed on the console and stored in /usr/adm/messages. The messages appear as follows:

   found a single-bit error
   board=n bad_addr=xxxx

where n is the board number and xxxx is the address of the error.

ECC errors are mapped and stored in a bad page table using the ecc utility. The main ecc menu appears as follows:

1. Print Current Bad Memory Page Table 2. Add Entries to Current Bad Memory Page Table 3. Delete Entries from Current Bad Memory Page Table

Enter your choice or `q' to quit:

Option 1 prints the current bad memory page table. Option 2 is used to add new entries to the table. Option 3 can be used to delete entires from the bad page table when memory boards or individual SIMMs have been replaced.

You must reboot the system after modifying the bad page table for it to take effect.

Limitations

This utility works only with Corollary smp RAM and compatibles.

Files

/dev/ecc: ECC device
/etc/idrc.d/ecc: ECC daemon startup script
/etc/idsd.d/ecc: ECC daemon shutdown script
/etc/eccdata: bad page table
/etc/eccdpid: daemon process ID

Standards conformance

ecc is not part of any currently supported standard; it is an extension of AT&T System V provided by The Santa Cruz Operation, Inc.