69rlq62d-f714peg4 * Memec (Headquar
ter
s) - Unique
Tec
h,
Insight,
Impact
MAR
VELL CONFIDENTIAL,
UNDER ND
A# 12101050
69rlq62d-f714peg4 * Memec (Headquar
ter
s) - Unique
Tec
h,
Insight,
Impact
MAR
VELL CONFIDENTIAL,
UNDER ND
A# 12101050
69r
lq62d-f714peg4 * Memec (Headquar
ters) - Unique
T
ech, Insight, Impact * UNDER ND
A# 12101050
MAR
VELL CONFIDENTIAL - UNA
UTHORIZED DISTRIB
UTION OR USE STRICTL
Y PR
OHIBITED
Slave Power Management Unit
Copyright © 12/13/06 Marvell
CONFIDENTIAL
Doc. No. MV-TBD-00 Rev. A
December 13, 2006
Document Classification: Proprietary Information
Page 253
Not approved by Document Control. For review only.
•
The internal application subsystem resets asserted when entering D3 state are de-asserted.
•
The state configuration is cleared in the PWRMODE register, refer to
Section 4.5.16, “Core PWRMODE
Register (CP14 Register 7)” on page 4-62
.
•
The core begins a normal boot sequence
•
The DDR SDRAM must be brought out of self-refresh mode, which requires the DDR SDRAM controller to
be transitioned to its idle state. See the Memory Controller section for details on configuring the DDR
SDRAM interface.
Normal boot-up sequencing begins with all units in the application subsystem beginning with their predefined
reset conditions. Core software must examine the
“Application Subsystem Reset Status Register (ARSR)” on
to determine the reset source was an S3 low-power state exit reset. Core must examine the
Section 9.3.1, “Application Subsystem Power Status/Configuration Register (ASCR)” on page 9-257
determine that the application subsystem was in D3 state previously. If the Power Management Unit Scratch-Pad
register (PSPR) in the MPMU was used for saving any general processor state during D3 state, the state can be
recovered.
9.2.2.1.5
D4 State
D4 power state offers even lower power consumption by powering down all units in the application subsystem
including the BPMU, clocks, OS timers, mini-LCD, and mini-SRAM controller. The penalty for this low-power
state is that all state is lost and there is no activity inside the application subsystem. Because internal activity has
stopped, recovery from D4 state must be through an external wake-up event or a RTC event detected by the
MPMU. Because all state has been lost in the application subsystem, the state of the core and all peripheral units
are reset and recovery begins from the core reset vector.
D4 state can be entered in two ways: through software control by setting the mode bits in the
Section 4.5.16,
“Core PWRMODE Register (CP14 Register 7)” on page 4-62
to S3 or by hardware control due to the assertion
of nBATT_FAULT with BIE bit clear. If entry to D4 is through software control, the application subsystem D4
state is included as a subset of the MPMU S3 state. The application subsystem can then initiate D4 entry by
setting the mode bits in the
Section 4.5.16, “Core PWRMODE Register (CP14 Register 7)” on page 4-62
to S3.
When the BPMU enters D4 state, the MPMU also enters S3 state.All supplies except VCC_BBATT may be
powered down for maximum power savings. Exit from D4 is initiated by a MPMU wake-up event. shows the
state of the power domains within the application subsystem when in D4 state.
Preparation for D4 State
The following steps must be taken by system software before entering D4 state:
•
The appropriate registers in the MPMU must be initialized to determine the wake-up enable sources from S3
state:
— Power Manager Wake-Up Enable register (PWER) (refer to the MPMU Internal Architecture Spec) for
exit from S3 state
•
The memory controller must be properly configured to ensure DDR SDRAM contents are maintained during
D4 state. See the Memory Controller section for more details.
•
All peripherals unit must be stopped/disabled, including the LCD controller and mini-LCD unit. LCD
operation during D4 state is only possible with an external LCD panel with a built-in frame buffer. The
clocks to all peripherals are stopped when D4 state is entered and they will not function normally.
•