Table 1064. Nar Error Types - STMicroelectronics SPC572L series Reference Manual

Nexus Aurora Router (NAR)
65.5.10.1 Error conditions
NAR errors can occur upon events such as illegal memory accesses, block-transfer
overloads, communication errors, or when the NAR is configured incorrectly. If the
configuration registers are programmed incorrectly, it is possible to put the NAR into an
illegal state. The NARs response varies depending on the error. For a minor error, the NAR
reverts to a default mode and continues operation, while major errors result in complete
shutdown.
In either case, a Nexus error message is sent to the Aurora out port, if its enabled, otherwise
to the trace memory bus port. The Nexus error message contains both ETYPE and ECODE
fields:
•
ETYPE is used to denote the basic type of error that has occurred
•
ECODE carries specific additional information about the error
All NAR error types are defined in this table.
ETYPE Description Classification
0000
Receive
(h'0) queue
overrun
Internal
0001
message
(h'1)
contention
Trace
memory bus
0011
port
(h'3)
read/write
access error
0101
Invalid NRR
(h'5)
1001
Config error
(h'9)
1100
HBDP error
(h'C)
1110
Block
transfer error
(h'E)
1111
Stall error
(h'F)
1. In the case where a request was successfully translated to a Trace memory bus transaction which itself returned an error,
that is conveyed in the STATUS field of the Nexus Target Response message subsequently returned to the debugger.
1898/2058

Table 1064. NAR error types

NAR action
Dropped
Minor
message
Dropped
Minor
message
Minor Access fails
Minor Access fails
See
Major
Table 1065
Minor N/A
Minor Dummy fill
Minor —
DocID027809 Rev 4
Comment
If a client sends a data word that the NAR cannot accept
(due to a full receive queue), the new data word is
dropped.
Internal errors/watchpoints occurred faster than NAR
could generate messages for them, causing lower-
priority message(s) to be dropped. The type of message
dropped is conveyed in the ECODE field, see
Table 1065.
Unable to send an Aurora-to-trace memory bus request.
Typical causes would be writes larger than the data
buffer or sending in a new request before the previous
(1)
one completes.
Tried to access an un-implemented NRR.
The specific configuration error is conveyed in bits 0–3 of
the ECODE field (Table
of Status Register.
A hard error is automatically reset the Aurora Link and
causes it to re-initialize. Although the Link is down, the
error message is queued and released as soon as the
Link comes back up.
If the Trace memory bus port becomes data-starved
during a block transfer, it fills the rest of the block with
IDLE messages.
RM0400
1065), as well as the NCE field
—