Intel
®
81341 and 81342 I/O Processors
December 2007
Developer’s Manual
Order Number: 315037-002US
619
DDR SDRAM Memory Controller—Intel
®
81341 and 81342
7.4.3
Memory Controller Response to P_RST#
The memory controller assumes a power failure condition whenever
P_RST#
is
asserted.
P_RST#
assertion following an initial power up sequence results in the
following sequence of events:
1. Clock and Reset Unit (CRU) requests that DMCU run the power fail sequence depicted
in
. The DMCU takes the following steps to execute the power fail sequence:
a. Gracefully terminate the current transaction.
b. Deactivate all DDR SDRAM leaves with the
precharge-all
command.
c. After Trp, the DMCU issues a
self-refresh
command to the DDR SDRAM devices one
bank at a time and continue to deassert CKE[1:0]
.
2. The DMCU notifies the CRU that the power fail sequence has completed on the
memory bus.
3. The CRU then asserts the internal bus reset to reinitialize all internal bus agents
including the DMCU.
Note:
Internal bus reset is asserted in response to the assertion of
P_RST#
. When
P_RST#
indicates a true power failure, then battery-backup power is supplied to the DDR SDRAM
array. A true power failure must be determined by external circuitry based on
P_RST#
and also by other external logic monitoring the system power.
Refer to
for a high-level state machine representation illustrating the
memory controller’s behavior during a power failure condition.
Note:
Following the request from the CRU to run the power fail sequence, any data that is in
the DMCUs 1024 byte posted write buffer is discarded.
Figure 97. Power Failure State Machine
RESET
Power Fail State
Machine IDLE
PRECHARGE ALL
To both banks
Power failure detected and
ALL DMCU operations
gracefully returned to idle.
SELF REFRESH
To Bank0
tRP satisfied
DONE
(continue to deassert
CK, signal DONE to CRU)
m_clk
m_clk
SELF REFRESH
To Bank1
B6264-01