80ksbr200 Integrated Device Technology, 80ksbr200 Datasheet - Page 41

• Manufacturer Part Number: 80ksbr200
• Description: Srio Serial Buffer Flow-control Device
• Manufacturer: Integrated Device Technology

„2005 Integrated Device Technology, Inc. All rights reserved. Advanced Datasheet for informational purposes only. Product specifications subject to change without notice.NOT AN OFFER FOR SALE The information presented herein is subject to a
Non-Disclosure Agreement (NDA) and is for planning purposes only. Nothing contained in this presentation, whether verbal or written, is intended as, or shall have the effect of, a sale or an offer for sale that creates a contractual power of acceptance.
IDT 80KSBR200
Notes
7.0 Error Management
Management Extensions Specification” or errors that do not.
Specification
7.1 sRIO Errors and Error Handling
upon notification off any of these errors is described minimally; for detail see Interrupt Generation. Reference RIO Intercon-
nect Specification Part 8 (Error Management Extensions Specifications) for more detail on specific errors described below.
7.1.1 Recoverable Errors
that RIO supports hardware detection of and a recovery mechanism for, as described in the RIO specification. In these
cases, the appropriate bit is set in the Port n Error Detect CSR. Only the packet containing the first detected recoverable
error that is enabled for error capture (by Port n Error Enable CSR) will be captured in the Port n Error Capture CSRs. No
interrupt is generated or actions required for a recoverable error. Recoverable errors are detected in the physical layer
only.
7.1.2 Notification Errors
and all logical/transport layer (LTL) errors captured). Because they are non-recoverable (and in some cases have caused
a packet to be dropped), notification by interrupt is available. However, because they are non-fatal, response to the inter-
rupt is not crucial to port performance; i.e., the port is still functional. When a notification error is detected, the appropriate
bit is set in the error-specific register, an interrupt is generated, and in some cases, the error is captured. The Degraded
Threshold error also causes the port to request training (parallel only) with the hope that port performance will improve. In
all cases, the RIO port continues operating. Notification errors are detected in both the physical and logical layer.
7.1.3 Fatal Errors
because it has received too many packet retries in a row. In the first case, GRIO will set the Output Failed-encountered bit
in the Port n Error and Status CSR; the RIO output hardware may or may not stop (based on Stop-on-Port-Failed-
Encounter-Enable and Drop-Packet-Enable bits). In the second case, RIO will set the Retry Counter Threshold Trigger
Exceeded bit in the Port n Implementation Error CSR; the RIO hardware will continue to operate. In both cases, an inter-
rupt is generated, and while the port will continue operating at least partially, a system-level fix (such as reset) is recom-
mended to clean up RIO’s internal queues and resume normal operation. Fatal errors are detected in the physical layer
only.
7.2 System Software Error Notification
system by means of interrupt pins if enabled, or a Maintenance port-write operation issued by SerB. For specifics on inter-
rupt mechanism, see
The SB handles errors in two ways. The errors are defined as either errors that fall under the “RapidIO Part 8: Error
The configuration registers associated with errors are found in the RapidIO Part 8: Error Management Extensions
This section is focused on errors and status information in addition to the minimum required by the RIO specification.
This section describes how the logical and physical layers will detect and react to RIO errors. The action of the SerB
RIO errors are classified under three categories:
These errors are non-fatal transmission errors (such as corrupt packet or control symbols, and general protocol errors)
These errors are non-recoverable non-fatal errors detected by RIO (such as Degraded Threshold, Port-Write received,
SerB detects two fatal errors:
In these cases, the port has failed because its Recoverable Error Rate has exceeded a predefined failed threshold or
System software is notified of logical, transport, and physical layer errors in two ways. An interrupt is issued to the local
Recoverable errors
Notification errors
Fatal errors
Exceeded failed threshold
Exceeded consecutive retry threshold
section
2.3, outlines the required registers for error management.
section
5, Doorbells and Interrupt of this datasheet. Maintenance port-write operations are sent to a
41 of 172
Advanced Datasheet*
March 19, 2007