3.2.5.2. Extended Reset Control
The figure shows multiple connections from the Dedicated Reset Controller to the respective modules. The associated reset signals are categorized based on color, and the table below explains their respective functions.
Modules |
Reset Signals | Direction (Respect to Dedicated Reset Controller) |
Description |
---|---|---|---|
Power-On Reset |
|||
Host CPU | reset | Output | Generated by Dedicated Reset Controller upon a Power-On Reset. |
fRSmartComp | ARSTn | Output | |
Agent CPU | reset_r | Output | |
Warm Reset |
|||
Host CPU |
|
|
Upon a Warm Reset from either fRSmartComp or other units, Dedicated Reset Controller:
|
fRSmartComp | RREQL | Input | Dedicated Reset Controller receives the Host CPU Warm Reset request from fRSmartComp. |
RREQR | Input | Dedicated Reset Controller receives the Agent CPU Warm Reset request from fRSmartComp. |
|
RACKL | Output | Dedicated Reset Controller sends an acknowledgement to the fRSmartComp. |
|
RACKR | Output | Dedicated Reset Controller sends an acknowledgement to the fRSmartComp. |
|
Agent CPU |
|
|
Upon a Warm Reset from either fRSmartComp or other units, the Dedicated Reset Controller:
|
When the fRSmartComp wants to reset the CPUs (Host, Agent or both), it activates the request signals (RREQL and/or RREQR). Then, the fRSmartComp waits for an acknowledgment (RACKL or RACKR or both) from the Dedicated Reset Controller. When reset acknowledgement is received, it automatically removes the reset request. The fRSmartComp always asserts and de-asserts RREQL together with RREQR (same clock cycle).
In addition to that, the Agent CPU is brought out of asynchronous reset at least 2 clock cycles more after the Host CPU (to manage the Time Diversity).
Altera recommends the following best practices in using the Dedicated Reset Controller:
- In the Dedicated Reset Controller, assert the CPU reset request until the CPU asserts the acknowledgment signal.
- The Dedicated Reset Controller must always signals the fRSmartComp when a specific CPU is being reset by activating the corresponding RACK*, even if the Warm Reset request originates from other units instead of the fRSmartComp. This ensures that the fRSmartComp is aware of any reset actions, allowing it to interpret the signals from the processor correctly.