Method and system to identify and recover from faults in non-safety targets and safety targets

1 . A method for reset in an integrated circuit, the method comprising:

receiving a fault signal indicative of a response to a request from an initiator not being received within a response time;

determining based on the fault signal a type of target from which the response is not received;

based on a determination at a first network interface unit (NIU) coupled to a network circuit that the response is not received from a non-safety target within the response time, stopping requests from an initiator from reaching the non-safety target, sending, by a dummy responder, to the initiator a response to the request that the non-safety target has not responded to, and resetting only the non-safety target; and

based on a determination at a second network interface unit coupled to the network circuit that the response is not received from a safety target within the response time, resetting the safety target and the non-safety target;

wherein the non-safety target is coupled to the first network interface unit, the safety target is coupled to the second network interface unit, the first network interface unit and the second network interface unit are coupled to the same network circuit, the non-safety target and the safety target communicate with the same network circuit over separate communication paths, and the dummy responder is arranged on a communication path between the first network interface unit and the non-safety target.

2 . The method of claim 1 , further comprising setting a respective bit in a configuration register to stop additional communication from the initiator from reaching the non-safety target and sending to the initiator the response to the request sent by the initiator to the non-safety target based on the response not being received from the non-safety target.

3 . The method of claim 2 , further comprising clearing the one or more bits set in the configuration register after the non-safety target is only reset.

4 . The method of claim 1 , wherein resetting the non-safety target further comprises not resetting an interconnect fabric of the integrated circuit when the non-safety target is reset.

5 . The method of claim 1 , wherein the fault signal is received from one of the NIUs arranged between a network circuit of an interconnect fabric and one of the non-safety target and safety target.

6 . The method of claim 1 , wherein the request is a read request and the response sent to the initiator is an error to the read request.

7 . The method of claim 1 , wherein a third NIU is arranged between the network circuit and the initiator, the method further comprising determining a fault in an interconnect fabric comprising the first NIU, the second NIU, the third NIU, and the network circuit based on a determination that the response to the request is received by one of the first NIU and second NIU and not the third NIU.

8 . A system on a chip (SoC) comprising:

an initiator;

a safety target and a non-safety target;

an interconnect fabric arranged to provide communication between the initiator and the safety target and the non-safety target;

a first network interface unit (NIU) coupled to a network circuit and arranged to receive a first request from an initiator; determine that a response to the first request is not received in a response time from the safety target; and send a fault signal to a fault collection and control circuit (FCCC) indicative of the response to the first request from the initiator not being received within the response time;

a second network interface unit (NIU) coupled to the network circuit and arranged to receive a second request from an initiator; determine that a response to the second request is not received in a response time from the non-safety target; and send a fault signal to the fault collection and control circuit (FCCC) indicative of the response to the second request from the initiator not being received within the response time;

fence logic arranged between the interconnect fabric and the non-safety target to stop requests from the initiator from reaching the non-safety target based on the response not being received from the non-safety target;

a dummy responder arranged to send to the initiator a response to the request that the non-safety target has not responded to, based on the response not being received from the non-safety target; and

the fault collection and control circuit (FCCC) arranged to receive the fault signal from the first NIU or the second NIU; determine based on the fault signal a type of target from which the response is not received; reset only the non-safety target based on the response not being received from the non-safety target; and reset the safety target and the non-safety target based on the response not being received from the safety target,

wherein the non-safety target is coupled to the second network interface unit, the safety target is coupled to the first network interface unit, the first network interface unit and the second network interface unit are coupled to the same network circuit of the interconnect fabric, the non-safety target and the safety target communicate with the same network circuit over separate communication paths, and the dummy responder is arranged on a communication path between the second network interface unit and the non-safety target.

9 . The SoC of claim 8 , wherein the FCCC is further arranged to set a respective bit in a configuration register to stop additional communication from the initiator from reaching the non-safety target and to cause the dummy responder to send to the initiator the response to the request.

10 . The SoC of claim 8 , wherein the FCCC is further arranged to clear the bits set in the configuration register after the non-safety target is reset.

11 . The SoC of claim 10 , wherein the FCCC is further arranged to not reset the interconnect fabric when the non-safety target is reset.

12 . The SoC of claim 11 , wherein the FCCC arranged to reset the safety target and the non-safety target further comprises the FCCC being further arranged to reset the interconnect fabric and the initiator.

13 . The SoC of claim 8 , wherein the request is a read request and the response is an error to the read request.

14 . The SoC of claim 8 , wherein the NIU is arranged between a network circuit of the interconnect fabric and one of the safety target and the non-safety target.

15 . The SoC of claim 8 , wherein the SoC further comprises a third NIU arranged between the network circuit and the initiator; and wherein the FCCC is further arranged to determine a fault in the interconnect fabric based on a determination that the response to the request is received by the first NIU or the second NIU and not the third NIU.

16 . The SoC of claim 8 , further comprising drain logic arranged between the interconnect fabric and the non-safety target to discard additional data sent by the initiator and associated with the request based on only the non-safety target being reset.

17 . A system on a chip (SoC) comprising:

a plurality of initiators;

a plurality of targets;

a plurality of interconnect fabrics arranged to facilitate communication between at least one initiator and at least one target; wherein a plurality of first non-safety and first safety targets is coupled to a first interconnect fabric; second non-safety targets are only coupled to a second interconnect fabric; and a third interconnect fabric is coupled to the first interconnect fabric and the second interconnect fabric by a respective communication path;

the third interconnect fabric further comprising a network interface unit (NIU) arranged to receive a communication from an initiator and provide a fault signal indicative of a fault detected to a fault collection and control circuit (FCCC) when a response to a request from the initiator is not received within a response time from one of the second non-safety targets;

fence logic arranged between the third interconnect fabric and the second interconnect fabric to stop the communication from the initiator from reaching the second interconnect fabric over the communication path between the second interconnect fabric and third interconnect fabric when the fault is detected; and

a dummy responder arranged between the third interconnect fabric and the second interconnect fabric to respond to the communication from the initiator when the one of the second non-safety targets does not respond and the fault is detected;

wherein the FCCC is arranged to determine based on the fault signal a type of target from which the response is not received and reset only one or more of the second non-safety targets based on the response not being received from the one of the second non-safety targets, wherein the FCCC is further arranged to reset the first safety targets, the first non-safety targets, and the second non-safety targets based on another fault signal indicating that a response is not received from one of the first safety targets;

wherein a first non-safety target of the plurality of first non-safety targets is coupled to a first network interface unit of the first interconnect fabric, a first safety target of the plurality of first safety targets is coupled to a second network interface unit of the first interconnect fabric, the first network interface unit of the first interconnect fabric and the second network interface unit of the first interconnect fabric is coupled to a same network circuit of the first interconnect fabric, and another dummy responder is arranged on a communication path between the first network interface unit of the first interconnect fabric and the first non-safety target coupled to the first network interface of the first interconnect fabric.

18 . The SoC of claim 17 , wherein fence logic and a dummy responder are arranged between the first non-safety targets and the first interconnect fabric.

19 . The SoC of claim 18 , wherein no fence logic and no dummy responder are arranged between the second non-safety targets and the second interconnect fabric.

20 . The SoC of claim 19 , wherein no fence logic and no dummy responder are arranged between the first interconnect fabric and the third interconnect fabric.