Bus interconnection matrix
RM0082
104/844
Doc ID 018672 Rev 1
9.1 ICM
The AMBA system has ten programmable Multi-layer Interconnection Matrix (ICM). The ICM
allows multiple master layers to access a slave (See
Figure 6.
ICM block diagram
A layer is referred to as one or more masters that complete together with one master
winning ownership of the slave.
When there is more that one layer looking for access to the slave at the sametime, this is
referred to as a clash of requests. Whenever a clash is detected, only one layer can gain
access to the slave. The layer that do not gain access to the slave need to have their
address and control signals stored into their input stage. When address and control signals
are stored into an input stage, then the stored transfer controls the request and lock
generation circuitry. When a lower priority layer is in the middle of a burst transfer and a
higher priority layer issues a transfer, the higher priority layer is stored and then held off until
the lower priority layer completes the transfer.
shows the Master Layer on each ICM input stages while
lists the ICM
slaves.
Table 51.
ICM master layers (Initiator)
ICM
L0
L1
L2
L3
1
Processor
RAS_H
DMA#1
2
Processor
RAS_H
DMA#1
3
Processor
RAS_H
4
Processor
RAS_H
5
RAS_H
DMA#1
6
Ethernet MAC
USB(Hosts - Device
7
Ras_E
DMA#2
Ethernet MAC
C3
8
C3
USB(Hosts - Device)
ICM
AHB
MASTER
Arbiter
priority
AHB
(Layer 0)
AHB MASTER
(Layer N)
SLAVE