Error Log Analysis
System Machine Check (SCB: 660)
A system machine check is a system- or processor-detected error that occurred as a result
of an "off-chip" request to the system. The following conditions cause PALcode to build
the 660/670 machine check logout frame and invoke the 660 error handler.
•
CIA-detected noncorrectable ECC error
•
CIA-detected command/address parity error
•
CIA-detected nonexistent memory reference
•
PCI bus parity error
•
PCI command/address parity error
•
PCI master abort
•
PCI target abort
•
Scatter/gather invalid translations table entry
•
I/O timeout
•
System overtemperature (event)
•
Unexpected IPL 23 interrupt request
Processor-Detected Correctable Errors (SCB: 630)
Processor-detected correctable errors are generic Alpha 21164A microprocessor
correctable errors. The error-handling code should be common across all platforms that
use the Alpha 21164A. The following conditions cause PALcode to invoke the operating
system 630 error handler:
•
Single-bit I-stream ECC error
•
Single-bit D-stream ECC error
System-Detected Nonfatal Error (SCB: 620)
These nonfatal errors are
following conditions cause the PALcode to invoke the operating system 620 error handler:
•
DMA read errors
•
High temperature warning (event)
•
Fan warning/failure (event)—This event is fatal and non-correctable, though reported
as nonfatal. The OS then shuts the system down.
5–4 DIGITAL Server 3300/3300R
Server 3300/3300R-specific correctable errors. The
DIGITAL