Black box circuit for data processing device
By integrating black-box circuitry into the data processing device to capture and sign event logs of telemetry data, the vulnerability and difficulty in diagnosing of data processing devices are solved, enabling rapid and reliable attack diagnosis and remediation.
Patent Information
- Application Number
- CN202580027908.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-04-25
- Filing Date
- 2025-03-05
- Publication Date
- 2026-11-17
AI Technical Summary
Data processing devices are vulnerable to software attacks, which can lead to the loss of security-critical functions and make it difficult to detect and diagnose the cause of the attack, resulting in prolonged system downtime.
A black-box circuit is integrated into the data processing device to capture event logs of telemetry data, record behaviors before and after abnormal events, and ensure the credibility and integrity of the event logs through signatures and timestamps, supporting forensic analysis.
It improves the ability to diagnose software attacks, reduces system downtime, enhances the reliability and robustness of event logs, and helps to quickly identify and patch vulnerabilities.
Smart Images

Figure CN123125594A_ABST
Abstract
Description
[0001] This technology relates to the field of data processing.
[0002] Data processing apparatus may include processing circuitry for performing data processing and memory circuitry for storing data and instructions for processing by the processing circuitry. Such data processing apparatus may sometimes be vulnerable to software attacks, in which an attacker may attempt to exploit vulnerabilities in the software processed by the processing circuitry to cause the software to malfunction or to gain access to information stored in memory that the attacker should not otherwise have access to.
[0003] At least some examples of this technology provide a data processing device implemented on one or more chiplets, the data processing device comprising: a processing circuit for performing data processing; a memory storage circuit for storing data and instructions for processing by the processing circuit; and a black box circuit that, in response to detecting the occurrence of an abnormal event indicating a risk of loss of safety-critical functions of the data processing device, captures an event log of telemetry data, the event log indicating the behavior of the data processing device associated with a point in time before or after the occurrence of the abnormal event; wherein the black box circuit is integrated on at least one of the one or more chiplets of the data processing device.
[0004] At least some examples of this technology provide a system comprising: the data processing apparatus described above, the data processing apparatus being implemented in at least one packaged chip; at least one system component; and a board, wherein at least one packaged chip and at least one system component are assembled on the board.
[0005] At least some examples of this technology provide a chip-integrated product that includes the system described above, wherein the system is assembled on an additional board with at least one other product component.
[0006] At least some examples of this technology provide computer-readable code for manufacturing a data processing apparatus as described above. This computer-readable code can be stored on a computer-readable storage medium. The storage medium can be a non-transitory storage medium.
[0007] At least some examples of this technology provide a method comprising: performing data processing on a data processing device implemented on one or more chiplets, the data processing device including: processing circuitry for performing data processing, and memory storage circuitry for storing data and instructions for processing by the processing circuitry; and capturing an event log of telemetry data in response to detecting the occurrence of an abnormal event indicating a risk of loss of safety-critical functions of the data processing device, the event log indicating the behavior of the data processing device associated with a time point before or after the occurrence of the abnormal event; wherein the event log is captured using black-box circuitry on at least one of the one or more chiplets integrated into the data processing device.
[0008] At least some examples of this technology provide a chiplet for a data processing device, the chiplet comprising: a black box circuit that, in response to detecting the occurrence of an abnormal event indicating a risk of loss of safety-critical functions of the data processing device, captures an event log of telemetry data, the event log indicating the behavior of the data processing device associated with a point in time before or after the abnormal event; and an internal signal path for communication between the black box circuit and other parts of the data processing device, the internal signal path being separate from input / output interface circuitry used by the data processing device to communicate with external devices outside the data processing device according to an input / output interface protocol.
[0009] Further aspects, features, and advantages of this technology will become apparent from the following description of an example read in conjunction with the accompanying drawings, in which:
[0010] Figure 1 An example of a data processing device including black-box circuitry is shown;
[0011] Figure 2 An example of implementing a data processing device on multiple chiplets is shown;
[0012] Figure 3 An example illustrating the functionality of a black-box circuit in pre-event mode is provided.
[0013] Figure 4 An example is shown illustrating the functionality of a black-box circuit in post-event mode;
[0014] Figure 5 The steps for capturing an event log in response to the detection of an abnormal event are illustrated; and
[0015] Figure 6 A specific example of the steps for capturing event logs using black-box circuitry is illustrated; and
[0016] Figure 7 Examples of systems and products containing chips are shown.
[0017] A data processing device is implemented on one or more small chips and includes: processing circuitry for performing data processing; and memory storage circuitry for storing data and instructions for processing by the processing circuitry. A malicious party may attempt to launch a software attack on the device, causing it to malfunction. Various attacks are possible, such as back-directed programming attacks and denial-of-service attacks. In safety-critical processing systems designed to perform safety-critical functions, the consequences of such attacks can be severe. For example, for processors designed for automotive, medical, or public infrastructure applications, a successful attack could compromise the safety-critical functions of the processing device. However, in practice, if an attack occurs and succeeds, it can be difficult for system operators to determine how and why the attack was carried out after the incident. Therefore, a successful software attack on a safety-critical system can lead to prolonged system downtime, during which operators attempt to detect and eliminate the attack.
[0018] In the examples discussed below, the data processing device includes a black-box circuit that, in response to the detection of an anomalous event indicating a risk of loss of safety-critical functions of the data processing device, captures an event log of telemetry data indicating the behavior of the data processing device relative to points in time before or after the anomalous event. This black-box circuit is integrated onto at least one of the one or more chiplets of the data processing device. Thus, the black-box circuit is designed as a component of a processor chip product (e.g., as a single chiplet used as a system-on-a-chip, or as a collection of multiple chiplets packaged in a semiconductor package). The black-box circuit captures an event log that allows system researchers to understand what happened on the chip / chiplet around the time of the anomalous event. This can provide useful information for identifying the cause of software attacks that compromise safety-critical functions. For example, the event log can help pinpoint the specific application or part of an application that was compromised to carry out the attack. Therefore, providing integrated black-box circuitry to the data processing device can better diagnose malware attacks, thereby helping to reduce downtime in safety-critical systems.
[0019] In some examples, black-box circuits can sign event logs to enable proof of their source. This allows investigating operators who might be attacking to demonstrate that the event logs used for forensic analysis were generated by a trusted circuit, rather than a forged log that an attacker might have attempted to replace the genuine event log. Therefore, by using signed proofs, the credibility of event logs is increased, thereby supporting forensic analysis after anomalies occur.
[0020] For example, a signature can be based on applying a cryptographic signature to an event log. Alternatively, a signature can be based on a public key infrastructure. For instance, a black-box circuit can sign an event log using a private key unique to the black-box circuit, and the signature can be verified using a corresponding public key associated with a digital certificate used to prove the identity of the black-box circuit.
[0021] In some examples, black-box circuitry can perform immutability protection to safeguard event logs from tampering. This increases the likelihood that an attacker will fail to replace real data with forged event log data in an attempt to evade detection.
[0022] For example, a black-box circuit can generate at least one event log digest associated with an event log, each digest derived from at least a portion of the event log, and providing a verification value for detecting tampering with the event log. The function used to derive the digest from that portion of the event log can be, for example, a cryptographic function. For instance, a digest generation function can be used to generate the digest, with a probability that an attacker could guess the correct digest value for a given input dataset that is negligible. Therefore, it can be assumed that if the digest value generated from a given event log at analysis time still matches the corresponding digest value generated by the black-box circuit when the event log was generated, the event log can be believed to have remained tampered with since its generation. This makes it more difficult for an attacker to evade detection.
[0023] In some examples, the black-box circuitry can send at least one event log digest to an external device outside the data processing device, or to at least one adjacent chip within the data processing device other than the chip containing the black-box circuitry. This can help improve robustness against attacks because if an attacker has compromised the circuitry on one chip, the likelihood of them also compromising adjacent chips or external devices is low. Therefore, by sending the event log digest to at least adjacent chips or external devices, the likelihood of an attacker simultaneously intercepting or replacing the event log digest is low, thus providing stronger tamper protection.
[0024] In some examples, black-box circuitry can associate timestamps with event logs to achieve synchronization with points in time associated with different telemetry data streams captured in the event logs. Different telemetry data sources may reside in portions of a chiplet / multi-chiplet system that operate without any synchronized clock source; therefore, timestamping helps ensure relevance. For example, black-box circuitry can use its own timing mechanism to determine a timestamp value and transmit that timestamp value to other telemetry data sources, allowing for synchronization of the current time across multiple telemetry data sources.
[0025] In some examples, black-box circuits are configured to include pre-event telemetry data in the event log that is associated with a point in time prior to the occurrence of the anomalous event. Including pre-event telemetry data can help allow for better diagnosis of the cause of an attack.
[0026] To capture pre-event telemetry data, the black-box circuitry can be configured to write the captured telemetry data into a first-in-first-out (FIFO) buffer, even when no anomalous events are detected. Therefore, the black-box circuitry continuously logs incoming telemetry data into its FIFO buffer; if there is insufficient capacity to write new data without overwriting any previous data, the new data will overwrite the oldest data. In response to the detection of an anomalous event, the black-box circuitry can capture the contents of the FIFO buffer corresponding to the nearest window before the anomalous event as pre-event telemetry data to be included in the event log, preventing the pre-event telemetry data from being overwritten by subsequently captured telemetry data. Thus, detecting an anomalous event causes the current contents of the FIFO buffer to be "frozen," providing a snapshot of the processing activity being performed in the immediate period preceding the anomalous event.
[0027] As a supplement or alternative to including pre-event telemetry data in the event log, black-box circuitry can record post-event telemetry data in the event log. Therefore, in response to the detection of an anomalous event, black-box circuitry is configured to capture post-event telemetry data associated with a point in time after the anomalous event and include this post-event telemetry data in the event log. This provides additional useful diagnostic information.
[0028] While pre-event telemetry data may often be more valuable than post-event telemetry data for diagnostic purposes, post-event telemetry data can still be useful in diagnosing the cause of an attack. In practice, limitations in circuit area and power budgets may restrict the amount of storage capacity and telemetry data capture overhead that system designers might be willing to incur for speculatively capturing pre-event telemetry data without detecting any specific anomalous events. Therefore, in some examples, providing both pre-event and post-event telemetry data in the event log offers an improved balance between diagnostic capabilities and the circuit area / power costs associated with black-box circuitry.
[0029] In some examples, in response to the detection of an anomalous event, the black-box circuitry is configured to trigger the capture of at least one type of post-event telemetry data that was not previously captured before the annomalous event was detected, and to include said at least one type of post-event telemetry data in the event log. This recognizes that while the increased storage capacity and power consumption costs of capturing larger, more detailed telemetry data once an anomalous event is detected may be reasonable, the cost is not justified if speculative telemetry data is captured simply in case of an event requiring pre-event data collection. Therefore, it can be particularly useful if the black-box circuitry activates one or more additional post-event telemetry data sources in response to the occurrence of an anomalous event, thereby increasing the amount of telemetry data captured after the anomalous event compared to the amount captured before the anomalous event. This provides a better balance between security and power and circuit area overhead.
[0030] In some examples, the black-box circuitry can capture event logs in a separate black-box memory storage circuit, distinct from the memory storage circuitry used to store data and instructions processed by the processing circuitry. By providing the black-box circuitry with its own dedicated memory storage for capturing event log data, the security and reliability of the event logs are improved, as the event log data is likely less susceptible to tampering by attackers compared to storing it in generally accessible system memory. Furthermore, this helps reduce the performance impact of capturing event logs during the pre-event execution of the black-box circuitry, because by capturing the pre-event data in a dedicated black-box memory storage circuitry, rather than in the general-purpose memory system accessible to the software executing on the processing circuitry, the storage of pre-event data does not consume main memory system bandwidth that would otherwise be available for regular memory access triggered by the processing circuitry.
[0031] In some examples, the black-box circuitry can operate in a black-box power domain that is separate from at least one power domain associated with the processing circuitry and memory storage circuitry. For example, the black-box power domain may have a dedicated, normally-on power supply, such as a backup battery separate from the main power supply used for other parts of the processing device. This provides increased reliability to the black-box power domain, allowing the black-box circuitry to continue operating even if power supply voltage changes or other power outage events affect the operation of the main processing system. For example, it allows event logging operations, such as the calculation of event log signatures and / or summaries, to continue even if the main system is not operating.
[0032] Black-box circuitry can be tightly integrated into data processing systems at the semiconductor level (e.g., the silicon level). This differs from the classic "black boxes" associated with aircraft or other infrastructure, which may record information after a crash or other loss of function. In these systems, the black box is not integrated into the semiconductor chip of the processor system but is a separate device connected to the processor system by the end user via external input / output ports.
[0033] In some examples, the black-box circuitry is housed within the same semiconductor package as the processing and memory storage circuitry. The semiconductor package provides a housing (e.g., made of metal, plastic, glass, or ceramic) that protects the individual semiconductor components implemented on one or more small chips within the package. Integrating the black-box circuitry with the processing and memory storage circuitry, which are constrained by the diagnostic capabilities of the black-box circuitry, into the same semiconductor package can be extremely helpful in diagnosing the possible causes of software attacks, something difficult to achieve using external circuitry because the information available to external devices via input / output mechanisms may be limited.
[0034] More specifically, in some examples, the black-box circuitry may be disposed on the same chip as at least a portion of at least one of the processing circuitry and the memory storage circuitry. In this case, a single integrated circuit comprising the processing circuitry, the memory storage circuitry, and the black-box circuitry can be implemented on a semiconductor wafer.
[0035] In other examples, the black-box circuitry may be implemented as its own dedicated chiplet, separate from one or more chipslets providing processing and memory storage circuitry, the operation of which is monitored by the black-box circuitry. The chiplets providing the black-box circuitry and other parts of the data processing device may be integrated into a single semiconductor package (e.g., using an interpolator).
[0036] Data processing devices may include input / output (I / O) interface circuitry for communicating with external devices outside the data processing device according to I / O interface protocols. Examples of I / O interface protocols include, for example, PCIe and CXL. I / O mechanisms are typically used for communication with the external world, such as with user input / output devices, external data storage units, networking devices, etc. Communication via I / O interfaces is often relatively slower and has limited bandwidth compared to internal signal paths (e.g., buses) within the data processing device. Furthermore, I / O mechanisms may limit what information external devices can access from within the information processing device.
[0037] In contrast, communication between the black-box circuitry and other parts of the data processing unit can be achieved via internal signal paths separate from the input / output interface circuitry. Therefore, the black-box circuitry is integrated into the information processing unit, rather than being accessed via external devices through I / O mechanisms. This tight integration allows for broader access to a variety of telemetry data sources (some of which might be impractical to access via I / O mechanisms) and allows for faster response to anomaly detection.
[0038] An anomaly can be any unusual event and can be a potential indicator of the loss of safety-critical functions of a data processing device. Some examples of such anomalies may include:
[0039] • System crash events. If a system has crashed, safety-critical functions are likely compromised, so diagnostic information about the cause of the crash can help identify the appropriate response. For example, system crash events can be detected based on performance metrics or other monitoring information associated with the operation of processing circuitry (detected by the black-box circuitry itself or by separate monitoring circuitry). For instance, if a deadlock occurs in the system, instruction execution rate may decrease, so performance metrics about instruction execution rate can be used to detect system crash events. System crashes can also be detected based on assertions of reset events or other indicators of abnormal system operation.
[0040] • A potential attack by a malicious party against the data processing device has been detected. Some systems may have mechanisms to enable attack detection. For example, a network security processor included in the data processing device may receive telemetry data from multiple sources (which may be the same telemetry data used by the black-box circuitry, or a different subset of telemetry data used by the black-box circuitry), and may process the received telemetry data using a predictive model (e.g., a machine learning model) trained to detect signatures of known attacks, thereby generating an attack detection signal indicating whether any attack has been detected. The black-box circuitry may detect the network security processor's assertion of the attack detection signal as an anomalous event, leading to the capture of an event log.
[0041] • Abnormal interruption or failure modes. If processing circuitry repeatedly encounters interruptions or failures, this could be an indication of a denial-of-service attack or programming error that could disrupt the normal operation of a safety-critical system. Therefore, metrics that track the frequency or rate of failures (whether general failures or one or more specific types of failures) can be used to detect anomalous events that trigger black-box circuitry to capture event logs.
[0042] • Anomalous Modular Redundancy Error Patterns. Some safety-critical systems may employ modular redundancy, where the same processing is executed redundantly more than once to allow for error detection when divergence is detected between two or more redundant instances of the same processing. Redundancy can be provided at the hardware level (where multiple redundant processor cores operate on the same instructions in a lockstep manner and divergence is detected by comparing their outputs) or at the software level (where the operating system manages the execution of multiple threads of the same processing). Especially in systems where modular redundancy is provided at the software level, software attacks may find it difficult to successfully attack each redundant version of the same processing in the same way, so divergence may occur between redundant threads if an attack attempt is made. Therefore, information about anomalous modular redundancy error patterns can also be a sign of potential attacks and may prompt black-box circuitry to capture event logs.
[0043] • An abnormal processing performance associated with the data processing device was detected.
[0044] • Changes to the system configuration and / or integrity that may compromise safety (e.g., disabling security devices or shutting down protection systems).
[0045] • Remove physical protective measures, such as covers or shields that cover data processing devices.
[0046] In a given specific implementation of a black-box circuit, not all of these examples need to be supported. Some examples may only support one or a subset of the described anomalous event types that might trigger event log capture. Other examples may support the detection of any event among these types. It should be understood that other event types not described may also be supported.
[0047] Event logs can include various types of telemetry data and often include any information that can be used to diagnose the possible causes of anomalous events.
[0048] Event logs can include data from portions of the data processing device that are not accessible via external pins of the one or more chipsets within the data processing device. This can help allow for more detailed questioning of the data processing device's operation, which would be impractical without onboard black-box circuitry.
[0049] For example, event logs may include at least one of the following types of telemetry data:
[0050] • Context information indicating the active processing context when the abnormal event occurred;
[0051] • Software version information indicating the version of the software being processed when the abnormal event occurred;
[0052] • Fingerprint data indicating the active fingerprint on the software processing device;
[0053] • Data interruption;
[0054] •Performance monitoring data;
[0055] • Accessing data from memory;
[0056] • The memory contains built-in self-test data;
[0057] • Prefetcher data;
[0058] • Branch predictor data;
[0059] • Data transmitted from one or more adjacent chips, excluding one or more chips that include the black box circuit, to adjacent chips of the black box circuit.
[0060] • Power management data associated with the power management of the data processing device; and
[0061] • Modular redundancy check information.
[0062] Similarly, it should be understood that not all of these types can be collected in a given system. In practice, as discussed above, pre-event telemetry data can include a smaller subset of data types than post-event telemetry data.
[0063] Information identifying the specific software version being executed at the time of the incident can be particularly useful for forensic analysis in identifying potentially compromised software that may have been attacked. For other diagnostic mechanisms typically supported in data processing systems (e.g., trace capture tools that provide tracing of the behavior of a specific program to assist in its software development or performance optimization), the captured trace data is unlikely to specify the executing software. This is because such trace capture tools are typically used to diagnose errors or inefficiencies in the operation of specific software; therefore, the identity of the executing software is already known to the diagnostic analyzer and does not need to be captured by the trace circuitry. Another difference with this trace circuitry is that diagnostic trace capture can be continuous, rather than capturing event logs in response to the detection of a safety-critical event that risks the loss of safety-critical functionality.
[0064] In some examples, if other diagnostic tools that also capture diagnostic information are provided, the black-box circuitry may share some hardware (e.g., memory storage devices) with those other diagnostic tools. For example, in the absence of an anomalous event, the black-box circuitry may capture both safety-critical and non-safety-critical telemetry data. In response to an anomalous event, the black-box circuitry may prioritize retaining safety-critical telemetry data over non-safety-critical telemetry data. This allows for the saving of some hardware in the system while still capturing telemetry data for non-safety-critical reasons. For example, safety-critical telemetry data may include contextual information, software version information, interrupt data, operating system-provided fingerprint data, etc., which may be useful for forensic analysis of potential software attacks, while non-safety-critical data may include other information more relevant to the potential loss of diagnostic software, such as information about branch misprediction rates, memory address faults, etc.
[0065] Safety-critical functions that may be at risk of disruption due to abnormal events can include a wide variety of safety-critical functions. As an example, in a non-exhaustive list, safety-critical functions may include controls over medical equipment, if the equipment stops operating, the health of patients will be at risk; controls over vehicles (such as cars, airplanes, or spacecraft); controls over power plants; controls over processes within factories or other industrial facilities; controls over public infrastructure (such as traffic lights, air traffic control, or train signaling systems); and so on.
[0066] Some examples provide a chiplet including the black-box circuitry described above. In some examples, the chiplet may also include processing circuitry and memory storage circuitry. In other examples, the chiplet including the black-box circuitry may be separate from the chiplet including the processing circuitry and / or memory storage circuitry.
[0067] Figure 1 An example of a data processing device 2 implemented on one or more chiplets (integrated circuits) provided within a semiconductor package is illustrated. Device 2 has processing circuitry 4 for performing data processing. For example, processing circuitry 4 may include one or more processing elements (PEs). PEs may include, for example, a central processing unit (CPU), a graphics processing unit (GPU), or another more specialized type of processing unit, such as a neural processing unit (NPU) that provides acceleration for machine learning applications. At least a portion of processing circuitry 4 (e.g., one or more CPUs) is capable of executing program instructions defined according to an instruction set architecture to perform data processing operations represented by those instructions. The instructions executed by processing circuitry 4 are fetched from memory storage circuitry 6. Memory storage circuitry 6 also stores data values processed by processing circuitry 4. Resulting data generated by processing circuitry 4 may be written back to memory storage circuitry 6. Although Figure 1Although not shown, it should be understood that the processing circuit 4 may have one or more caches for caching a subset of data from the memory storage circuit 6 so that the processing circuit 4 can access it more quickly.
[0068] Processing circuit 4 and memory storage circuit 6 communicate via an internal bus and / or signal path 10, for example via system interconnects that communicate according to a memory system bus protocol. Device 2 also has input / output (I / O) interface circuitry 8 for exchanging communication between the internal components of device 2 and the external world. For example, I / O interface circuitry 8 can handle communication with peripheral devices such as user input / output devices, network or radio communication equipment, and external mass data storage devices.
[0069] Device 2 includes a black box circuit 20 coupled to the internal bus and signal path 10 of device 2. The black box circuit 20 can receive telemetry data from various sources within device 2, thereby providing information about the operation of processing circuit 4 and / or memory storage circuit 6 and / or other components of device 2. For example, the telemetry sources can provide information about which software processes are currently running, which memory regions are being accessed, and the types of operations performed by these processes.
[0070] The black box circuit 20 provides additional hardware designed to be integrated into the same semiconductor product as the processing circuit 4 monitored by the black box circuit 20, so as to allow the entire device 2 to perform secure, self-immutable logging of diagnostic data in the event of a malware attack.
[0071] In response to the detection of an anomaly indicating a risk of loss of security-critical functions of data processing device 2, black-box circuitry 20 captures an event log of telemetry data. This event log indicates the behavior of the data processing device associated with points in time before or after the anomaly. The event log can be used for forensic analysis after the event to help diagnose the cause of the event and identify possible mitigation measures. For example, the event log can identify a specific software version that may have been compromised by an attacker, allowing developers to patch any vulnerabilities and reduce the risk of future such attacks.
[0072] like Figure 1 As shown, the black box circuit 20 can operate in its own dedicated black box power domain 22, which is separate from one or more power domains used to power other components 4, 6, 10, 8 of the device 2. For example, the black box power domain 22 may have a dedicated backup battery not used for any other part of the device 2, which can continue to power the black box circuit 20 even if other parts of the device 2 become inoperable due to a power outage. This helps ensure that event log data captured by the black box circuit 20 is still retained even in the event of a power failure.
[0073] like Figure 2 As shown, in some specific embodiments, device 2 may include a plurality of chiplets 50, each chiplet 50 including a separate integrated circuit on which a subset of the components of device 2 are implemented. The chiplets may communicate via one or more inter-chiplet bridging mechanisms (e.g., via a communication path provided by an interpolator). The group of chiplets 50 may be integrally packaged into a single semiconductor package, wherein inter-chiplet communication is hidden from exposure to the outside world, and any external communication between the chiplets 50 and devices outside the semiconductor package is limited to a limited set of integrated circuit pins (e.g., debug ports for debugging communication) and I / O communication performed via I / O interface circuitry 8 according to a given I / O protocol (such as PCIe or CXL).
[0074] It should be understood that the specific partitioning of circuit components onto a particular chiplet in a multi-chiplet system can vary depending on the design requirements of the specific system. Therefore, Figure 2 The specific examples shown are for illustrative purposes only. In some cases, a given chip 50 may include a portion of processing circuitry 4 (e.g., one or more processing elements) and a memory storage device 6 (e.g., see [link to relevant documentation]). Figure 2 Chips 0, 2, and 3 in the chip. Some chips including processing circuitry 4 may also include I / O ports 8 (e.g., see chip 2). Chip 50 may also function as an I / O hub chip without any processing elements 4. Other components that may be included on one or more chips may include:
[0075] • A network security processor 32 is responsible for detecting software attacks or other network security risks. For example, the network security processor 32 may run a machine learning model that uses a trained machine learning model (trained to recognize telemetry patterns associated with certain known software attacks) to process telemetry inputs obtained from various telemetry sources 30 to generate an attack probability prediction of whether the system is likely to be attacked. For example, the network security processor 32 may detect attack risks based on a method for detecting anomalous behavior of electronic devices described in European Patent Application 23306870.9 filed on 27 October 2023 by Arm Limited of Cambridge, UK, the contents of which are incorporated herein by reference in their entirety.
[0076] • The memory has a built-in self-test (BIST) circuit 34, which performs diagnostic operations on the associated memory storage circuit 6. The MBIST circuit 34 can run a self-test operation on the memory to detect the occurrence of random hardware faults (such as a memory cell getting stuck at a fixed value regardless of what data is written to it), which may prevent the memory from functioning properly; and / or
[0077] • Power management circuit 36, which manages the power conversion of its own chiplet and / or other adjacent chipslets.
[0078] like Figure 2 As shown, at least one of these chiplets (chiplets 0 and 3 in this example) includes a digital black box (DBB) circuit 20 for capturing an event log as discussed above in response to the detection of an anomalous event. The event log includes telemetry data captured by at least some of the telemetry sources 30, associated with time points near (before and / or after) the time the event was detected. A wide variety of event types can be detected as anomalous events, triggering the capture of the event log. For example, such events may include system crash events; detection of suspicious interrupt patterns; detection of divergences between the outputs of redundant software components executing on the processing circuitry; and / or the network security processor 32 detecting a software attack pattern. See below for reference. Figure 3 and Figure 4 Describe an example of telemetry data that can be captured in the event log.
[0079] Although Figure 2 An example is shown in which the black box circuit 20 is located on the same chip 50 as the processing circuit 4 and memory storage circuit 6 of the monitored device 2, but the black box circuit 20 may also be provided as a separate chip from the chip including the processing circuit 4 and memory storage circuit 6.
[0080] Black-box circuitry 20 provides a secure (e.g., immutable) way to store silicon die-level context and telemetry data in case of attacks or anomalous behavior in safety-critical systems. This is typically done at the system or system-to-system level (e.g., a black box in an aircraft), where the black box communicates with the system it is monitoring via I / O mechanisms. In contrast, black-box circuitry 20 is integrated at the silicon level into the processing system it monitors, and is a component of the semiconductor product it is monitoring (e.g., a system-on-a-chip or a chiplet processor system).
[0081] Telemetry data, such as die “context” information defining an application running at a given time and operational telemetry about the application’s activities, can be streamed to an embedded memory within black-box circuitry 20 that consistently stores on-chip telemetry data for the past X seconds (X being a design parameter to be selected by the system designer). Once an event is detected by an internal / external chip security or security monitoring system or an operating system executing on processing circuitry 4, the chipset enters digital black-box (DBB) mode, freezing the capture of telemetry data prior to the event to retain the storage for the past X seconds before the event, and begins recording new data. Black-box circuitry 20 creates an event block that can be used for forensic analysis, and this event block is trustworthy due to the adoption of this process and the use of CCA or TEE-signed proof. This allows for the access to data from blocks within the integrated circuit (chiplet 50) that are not directly connected to external pins accessible via I / O interface 8 when performing such forensic analysis of the cause of the loss of security-critical functions.
[0082] Figure 3 An example of components of the black-box circuit 20 operating in pre-event mode prior to any abnormal event occurring is illustrated. For example... Figure 3 As shown, the black box circuit 20 includes an ingress bus 62 for receiving telemetry data 60 from various telemetry sources, a compression / encryption circuit 64 for compressing and / or encrypting the telemetry data 60, a switch 66 for selecting telemetry data and directing it to other components of the black box circuit 20, a black box controller (DBB controller) 68 for controlling the operation of the black box circuit 20, an event detector 70 for detecting the occurrence of abnormal events, and dedicated black box memory storage devices 72 and 74 separate from the main memory storage circuit 6 used by the processing circuit 4. The black box memory storage device includes a normally open portion 72 for recording telemetry data in pre-event mode, and an event trigger portion 74 of the memory storage device, which is in a power-saving state during pre-event mode but is powered on in response to the occurrence of an abnormal event to provide additional storage space for storing a larger amount of telemetry data than that that can be held by the normally open memory storage device 72.
[0083] Telemetry data can be obtained from various sources 60. For example, Figure 3 Examples of any of the following types of telemetry data 60 are shown:
[0084] - Operational telemetry from system components such as power manager 36;
[0085] - Dedicated telemetry for black-box operation (e.g., telemetry from delay monitors, process detectors, voltage drop detectors, and / or hotspot analyzers).
[0086] - Security tracking information, which provides tracking of the activities of one or more applications running on the processing device 2;
[0087] - Interrupt data, which provides information about interrupt patterns (e.g., interrupt type and interrupt frequency) that occur during processing performed by the processing device 2;
[0088] - Operating system (OS) fingerprint information, which provides a snapshot of OS activity;
[0089] - Hardware / software fingerprints provide a snapshot of other hardware or software activity. For example, fingerprint data may include the instruction address of an executed instruction or a summary of the data address of data accessed by processing circuitry 4 in memory 6;
[0090] - Identifiers of the applications running on processing circuit 4, including version-specific information that identifies the specific software version being executed;
[0091] - BIST data provided by the memory built into the self-test circuit 34;
[0092] -Information transmitted from telemetry sources on adjacent chip 50, excluding the chip 50 containing black box circuitry 20;
[0093] It should be understood that this list is not exhaustive, and it is not necessary to consider all of these types of telemetry data in a given specific implementation. In addition, in pre-event mode, only a subset of these telemetry sources may be active (some sources may only provide telemetry in post-event mode).
[0094] One or more streams of such telemetry data are received at ingress bus 62 and may be losslessly compressed and / or encrypted by compression / encryption circuitry 64. Switch 66 selects telemetry data from one or more streams for routing to black-box controller 68 and / or other components.
[0095] The black-box controller 68 supports an authentication service 76 for digitally signing the received telemetry data to provide a signature that enables third parties to authenticate the identity of the telemetry data source. Therefore, the authentication service 76 generates a digital signature corresponding to pre-event log data generated based on the received telemetry data. Additionally, the black-box controller 68 supports a timestamp service 78, which periodically inserts timestamps (defined according to the local clock scheme used by the black-box circuit 20) into the pre-event telemetry data stream to help synchronize information from different telemetry sources.
[0096] Telemetry data, with signatures and timestamps, is written to normally open memory 72 to provide a buffer for pre-event telemetry data. Normally open memory 72 is managed as a first-in, first-out (FIFO) buffer, so that once its capacity is fully utilized, the oldest telemetry data item is overwritten by a newer one. Therefore, black-box controller 68 controls normally open memory 72 to capture the recording of telemetry data from the most recent time window (e.g., the last X seconds – the capacity of normally open memory can be selected based on the length of time before the event-pre-event log data to be captured).
[0097] Event detector 70 detects whether any abnormal events have occurred. Figure 3 In the example shown, event detector 70 may read telemetry data from normally open memory 72 and detect the occurrence of anomalies based on the telemetry data (e.g., interrupt data provided as a telemetry source, which may indicate a suspicious interrupt pattern). In other examples, event detector 70 may receive event signals from other parts of the system, rather than detecting anomalies internally based on the analysis of the received telemetry data. For example, event detector 70 may receive signals from network security processor 32 indicating whether any risk of a software attack has been detected, or from a system crash monitor that monitors system activity to detect scenarios where the system has crashed.
[0098] When an abnormal event occurs, the black box controller 68 switches the operation of the black box circuit to the post-event mode. Figure 4 The function of black box circuit 20 in post-event mode is illustrated.
[0099] In post-event mode, the black box controller 68 communicates with the telemetry source 30 to activate one or more additional telemetry sources 30 that did not previously transmit telemetry to the black box circuit 20 during pre-event mode.
[0100] Additionally, the black-box controller 68 freezes the contents of the normally open memory 72 to capture telemetry data 60 recorded in the FIFO buffer during the period preceding the occurrence of an abnormal event as a pre-event log. The normally open memory 72 becomes read-only.
[0101] In addition, the black-box controller 68 activates the event-triggered memory 74 by powering it on, so that the event-triggered memory 74 can begin recording the telemetry data 60 received in post-event mode. In post-event mode, the black-box controller 68's authentication service 76 and timestamp service 78 continue to sign and timestamp the incoming telemetry data, and write the signed / timestamped data into the event-triggered memory 74.
[0102] The black-box controller 68 also supports a hash service 80, which generates a digest of a given portion of the event log data based on a secret key using a given cryptographic function (e.g., a variant of SHA, QARMA, etc.). This cryptographic function is designed to make it statistically impossible for an attacker who does not know the secret key to guess the correct digest value of a given block of the event log data. This digest can be generated based on pre-event log data stored in normally open memory 72, or on post-event telemetry data stored in event-triggered memory 74 after the event is detected.
[0103] The black-box controller 68 also supports a broadcast service 82, through which summaries and / or captured event log data (data recorded before or after an event) can be sent to adjacent chips 50 in the processing system and / or to external devices via I / O circuits 8. For example, to prevent attackers from intercepting event log data and replacing it with other data during reading, a summary can be broadcast to other chips 50 and / or external devices immediately after the event log is generated, providing a tamper-proof protection mechanism.
[0104] The black box circuit 20 may continue to capture post-event telemetry data until the system is reset or until the black box circuit 20 is otherwise instructed (e.g., by trusted system software or by hardware) to stop capturing event log data.
[0105] Captured event log data can be read from event-triggered memory 74 through various mechanisms. In some cases, the only mechanism for retrieving event memory 74 may be via broadcast service 82, which can write the captured event log data to a memory storage device on another chip 50, or via I / O mechanism 8, which can write it to an external device. Other examples may support mapping black-box memories 72, 74 into the physical address space of device 2, allowing the event log to be read by software, and / or providing a dedicated hardware path (e.g., certain black-box read pins) through which an external device can read the captured event log data from black-box memories 72, 74.
[0106] A specific example of the process used to capture event logs is as follows, to obtain telemetry data with signatures and timestamps stored immutably by black-box circuit 20.
[0107] 1. Triggering DBB Startup: The DBB controller detects an anomaly (e.g., SoC crash, suspicious interrupt / failure mode) or receives a signal from a network security processor, firewall, etc. The DBB controller triggers the generation of a timestamp event request.
[0108] a.DBB's embedded memory stores streaming data from all telemetry data from the SoC over the past X seconds during runtime. This embedded memory is frozen (to prevent it from being overwritten by other telemetry data) and timestamped to generate a "pre-event log".
[0109] b. The DBB controller sends a signal to the DBB receiver (e.g., on-chip telemetry sensor, BIST, memory controller, adjacent chiplet, operating system, etc.) containing a timestamp of the time when the anomaly occurred. This timestamp can be inserted into the stream generated by each telemetry data source to ensure time alignment between telemetry data (because system clocks may not be perfectly aligned between different parts of the system).
[0110] d. Therefore, for a given set of components or subcomponents of the system, obtain a telemetry stream with timestamps and event ID stamps.
[0111] e.DBB circuit 20 can request certain sub-components that were not previously in continuous monitoring mode to now begin monitoring and generate telemetry streams.
[0112] 2. Collect event data: Switch the entire system to black box logger mode and enable data proof and security mode.
[0113] The DBB controller signs all SOC transactions / inputs / outputs, as well as other collectable data / metadata. The DBB controller generates hash digests of event data to prevent future malicious data tampering. The device subcomponent telemetry source continuously stores and streams historical data from the past X seconds and new data to the black-box controller.
[0114] 3. Verify and create new events:
[0115] a. When a DBB event occurs, the DBB controller begins broadcasting the hashes of the subcomponent telemetry events and the global event hash to DBB agents (e.g., on different chiplets or at external devices). This will support future verification and protect against potential attacks on the system, such as forged event data.
[0116] 4. Trusted Event Proof: The DBB controller records the submitter's digitally signed input and the digitally signed transaction responses from each subcomponent in a protected memory.
[0117] 5. Review the incident for forensic investigation: Later, personnel or artificial intelligence (AI) tools can review the incident log to identify the cause of the incident.
[0118] Figure 5The steps for capturing an event log using black-box circuitry 20 are illustrated. At step 100, data processing is performed on data processing device 2. At step 102, black-box circuitry 20 detects whether an abnormal event has occurred. When an abnormal event occurs, at step 104, black-box circuitry 20 (integrated into at least one chiplet 50 of device 2) captures an event log of telemetry data 30, which indicates the behavior of the device associated with a point in time before or after the abnormal event.
[0119] Figure 6 The steps for capturing the event log are illustrated in more detail. At step 150, the black-box circuit 20 captures pre-event telemetry data into a FIFO buffer structure stored in its dedicated black-box memory storage device (e.g., in normally open memory 72). The capture of pre-event telemetry data continues (overwriting older data with newer data if necessary) until an anomalous event is detected at step 152.
[0120] In response to the detection of an abnormal safety-critical event, at step 154, black-box circuitry 20 freezes the contents of the pre-event log FIFO structure (to prevent further updates), timestamps the captured pre-event log data, and generates a summary of the pre-event telemetry data. At step 156, black-box circuitry 20 activates one or more additional post-event telemetry data streams that were inactive before the event detected at step 152 (one or more telemetry data streams used before the event may also continue to generate telemetry data after the event). At step 158, black-box circuitry 20 timestamps the post-event telemetry data and captures the post-event telemetry data into the event triggering section 74 of the black-box memory storage device. At step 160, a summary of the post-event telemetry data is generated.
[0121] If additional post-event telemetry data needs to be captured (Yes at step 162), the method continues to loop through steps 158 to 162 until it is determined at step 162 that capturing post-event telemetry data can be stopped, at which point the capture ends at step 164. For example, telemetry data capture can be stopped when the processing system is reset or when a command to stop capturing telemetry data is sent to black box circuit 20.
[0122] Simultaneously, at step 164, a given portion of the event log data captured at step 158 and / or the digest generated at step 160 is periodically cryptographically signed (to provide a provable root of trust for the event log data and the digest). This provides both proof of the event log data and tamper protection. The timing of signing / hashing the event log data may vary in different implementations. For example, this may be done at fixed time intervals or as needed, such as based on capacity constraints in the event-triggered memory 74. At step 166, at least the digest (and optionally the event log data itself) may be sent to an adjacent chiplet and / or to an external device for external storage. Sending at least the digest helps provide a value stored by a “witness” that can be compared with the hash of the event log data read from the black-box memories 72, 74, thereby providing tamper detection.
[0123] Although Figure 6 The flowchart shown illustrates steps performed in a specific order; however, it should be understood that the same function may be performed in a different order than that shown, or some of the steps shown in the order may be performed in parallel, at least partially.
[0124] The concepts described herein can be specifically embodied in a system including at least one packaged chip. The previously described data processing device 2 is implemented in at least one packaged chip (either in a specific chip of the system or distributed across more than one packaged chip). At least one packaged chip is assembled on a board together with at least one system component. Chip-containing products may include systems assembled on a separate board together with at least one other product component. The system or chip-containing product may be assembled into a housing or onto a structural support (such as a frame or blade).
[0125] like Figure 7 As shown, one or more packaged chips 400 are manufactured by a semiconductor chip manufacturer, wherein the device 2 described above is implemented on a single chip or distributed across two or more chips. In some examples, the chip product 400 manufactured by the semiconductor chip manufacturer may be provided as a semiconductor package, which includes a protective housing (e.g., made of metal, plastic, glass, or ceramic) housing the semiconductor device implementing the above-described device, and connectors such as pads, solder balls, or pins for connecting the semiconductor device to the external environment. When more than one chip 400 is provided, these chips may be provided as separate integrated circuits (as separate packages), or may be packaged by a semiconductor provider into a multi-chip semiconductor package (e.g., using interpolators, or by using three-dimensional integration to provide a multi-layer chip product comprising two or more vertically stacked integrated circuit layers).
[0126] In some examples, a collection of chiplets (i.e., small modular chips with specific functionalities) may be referred to as a chip in itself. Chipslets may be individually packaged in semiconductor packages and / or packaged together with other chiplets in multi-chiplet semiconductor packages (e.g., using interpolators, or by using three-dimensional integration to provide multi-layer chiplet products comprising two or more vertically stacked integrated circuit layers).
[0127] One or more packaged chips 400 are assembled on a board 402 together with at least one system component 404 to provide system 406. For example, the board may include a printed circuit board. The board substrate may be made of any of a variety of materials, such as plastic, glass, ceramic, or flexible substrate materials such as paper, plastic, or textile materials. At least one system component 404 includes one or more external components that are not part of the one or more packaged chips 400. For example, at least one system component 404 may include any one or more of the following: another packaged chip (e.g., supplied by a different manufacturer or manufactured at a different process node), an interface module, a resistor, a capacitor, an inductor, a transformer, a diode, a transistor, and / or a sensor.
[0128] A chip-containing product 416 is manufactured, comprising a system 406 (including a board 402, one or more chips 400, and at least one system component 404) and one or more product components 412. Product components 412 include one or more additional components that are not part of system 406. As an example, in a non-exhaustive list, one or more product components 412 may include user input / output devices such as keypads, touchscreens, microphones, speakers, displays, haptic devices, etc.; wireless communication transmitters / receivers; sensors; actuators for actuating mechanical motion; thermal control devices; additional packaged chips; interface modules; resistors; capacitors; inductors; transformers; diodes; and / or transistors. System 406 and one or more product components 412 may be assembled onto an additional board 414.
[0129] Plate 402 or another plate 414 may be disposed on or within the equipment housing or other structural support (e.g., frame or blade) to provide a product that can be disposed of by a user and / or intended for operational use by personnel or company.
[0130] System 406 or chip-containing product 416 can be at least one of the following: end-user product, machine, medical device, computing or telecommunications infrastructure product, or automated control system. For example, as a non-exhaustive list, a chip-containing product can be any of the following: telecommunications equipment, mobile phone, tablet computer, laptop computer, computer, server (e.g., rack server or blade server), infrastructure equipment, networking equipment, vehicle or other automotive product, industrial machine, consumer device, smart card, credit card, smart glasses, avionics equipment, robotic equipment, camera, television, smart TV, DVD player, set-top box, wearable device, home appliance, smart meter, medical device, heating / lighting control equipment, sensor, and / or control system for controlling public infrastructure equipment (such as smart highways or traffic lights).
[0131] The concepts described herein may be embodied in computer-readable code used to manufacture devices embodying the described concepts. For example, the computer-readable code may be used in one or more stages of the semiconductor design and manufacturing process, including the electronic design automation (EDA) stage, to manufacture integrated circuits including devices embodying these concepts. The aforementioned computer-readable code may additionally or alternatively enable the definition, modeling, simulation, verification, and / or testing of devices embodying the concepts described herein.
[0132] For example, computer-readable code for manufacturing a device embodying the concepts described herein may be embodied in code that defines the hardware description language (HDL) representation of these concepts. For instance, the code may define a register-transfer level (RTL) abstraction of one or more logic circuits for defining a device embodying these concepts. The code may define an HDL representation of one or more logic circuits embodying the device using Verilog, SystemVerilog, Chisel, or VHDL (Very High Speed Integrated Circuit Hardware Description Language) and intermediate representations such as FIRRTL. Computer-readable code may provide definitions of the concepts or other behavioral representations of the concepts embodying the concepts using system-level modeling languages such as SystemC and SystemVerilog, which can be interpreted by a computer to enable simulation, functional and / or formal verification and testing of the concepts.
[0133] Additionally or alternatively, computer-readable code may define a low-level description of an integrated circuit component embodying the concepts described herein, such as one or more netlists or integrated circuit layout definitions, including representations such as GDSII. One or more netlists or other computer-readable representations of the integrated circuit component may be generated by applying one or more logic synthesis processes to the RTL representation to generate a definition for manufacturing a device embodying the invention. Alternatively or additionally, one or more logic synthesis processes may generate a bitstream from the computer-readable code to be loaded into a field-programmable gate array (FPGA) to configure the FPGA to embody the described concepts. The FPGA may be deployed for the purpose of verifying and testing the concepts prior to manufacturing integrated circuits, or the FPGA may be deployed directly in a product.
[0134] Computer-readable code may include a mixture of code representations for manufacturing apparatus, such as one or more of RTL representations, netlist representations, or other computer-readable definitions used in the semiconductor design and manufacturing process for manufacturing apparatus embodying the present invention. Alternatively or additionally, the concept may be defined in a combination of computer-readable definitions used in the semiconductor design and manufacturing process for manufacturing apparatus and computer-readable code defining instructions that will be executed by the defined apparatus once manufactured.
[0135] This computer-readable code can be set in any known transient computer-readable medium (such as wired or wireless transmission of code over a network) or non-transient computer-readable medium such as semiconductors, disks, or optical discs. Integrated circuits manufactured using computer-readable code may include components such as one or more of the following: a central processing unit, a graphics processing unit, a neural processing unit, a digital signal processor, or other components that embody the concept independently or collectively.
[0136] In this application, the phrase "configured as..." is used to mean that the elements of the device have a configuration capable of performing the defined operation. In this context, "configuration" means the arrangement or manner of interconnection of hardware or software. For example, the device may have dedicated hardware that provides the defined operation, or a processor or other processing device may be programmed to perform the function. "Configured as" does not mean that the elements of the device need to be changed in any way to provide the defined operation.
[0137] In this application, a list of features beginning with the phrase “at least one of” means that any one or more of those features may be provided independently or in combination. For example, “at least one of [A], [B], and [C]” covers any of the following options: A only (without B or C), B only (without A or C), C only (without A or B), a combination of A and B (without C), a combination of A and C (without B), a combination of B and C (without A), or a combination of A, B, and C.
[0138] While exemplary embodiments of the invention have been described in detail herein with reference to the accompanying drawings, it should be understood that the invention is not limited to those precise embodiments, and various changes and modifications can be made therein by those skilled in the art without departing from the scope of the invention as defined in the appended claims.
Claims
1. A data processing apparatus implemented on one or more chiplets, the data processing apparatus comprising: Processing circuitry, the processing circuitry being used to perform data processing; A memory storage circuit, wherein the memory storage circuit is used to store data and instructions for processing by the processing circuit; and A black-box circuit, in response to detecting the occurrence of an abnormal event indicating a risk of loss of a safety-critical function of the data processing device, captures an event log of telemetry data, the event log indicating the behavior of the data processing device associated with a time point before or after the occurrence of the abnormal event. The black box circuitry is integrated onto at least one of the one or more chips in the data processing device.
2. The data processing apparatus of claim 1, wherein the black-box circuitry is configured to sign the event log to enable the verification of the source of the event log.
3. The data processing apparatus according to any one of claims 1 and 2, wherein the black box circuit is configured to perform an immutability protection action to protect the event log from tampering.
4. The data processing apparatus according to any preceding claim, wherein the black-box circuitry is configured to generate at least one event log summary associated with the event log, each event log summary being derived from at least a portion of the event log and providing a verification value for detecting tampering with the event log.
5. The data processing apparatus of claim 4, wherein the black box circuit is configured to send the at least one event log summary to an external device outside the data processing apparatus, or to at least one adjacent chip in the data processing apparatus other than the at least one chip including the black box circuit.
6. The data processing apparatus according to any of the preceding claims, wherein the black box circuitry is configured to associate a timestamp with the event log to achieve synchronization with time points associated with different telemetry data streams captured in the event log.
7. The data processing apparatus according to any of the preceding claims, wherein the black box circuitry is configured to include pre-event telemetry data in the event log associated with a time point prior to the occurrence of the anomalous event.
8. The data processing apparatus according to any preceding claim, wherein, in the absence of the aforementioned abnormal event, the black-box circuitry is configured to write the captured telemetry data into a first-in-first-out (FIFO) buffer for telemetry data; and In response to the detection of the anomaly, the black-box circuitry is configured to capture the contents of the first-in-first-out buffer corresponding to the nearest window before the anomaly occurred as pre-event telemetry data to be included in the event log, and to prevent the pre-event telemetry data from being overwritten by subsequently captured telemetry data.
9. The data processing apparatus according to any preceding claim, wherein in response to detecting the occurrence of the anomalous event, the black box circuitry is configured to capture post-event telemetry data associated with a time point after the occurrence of the anomalous event, and to include the post-event telemetry data in the event log.
10. The data processing apparatus according to any preceding claim, wherein in response to detecting the occurrence of the anomalous event, the black box circuitry is configured to trigger the capture of at least one type of post-event telemetry data that was not previously captured before the occurrence of the anomalous event, and to include the at least one type of post-event telemetry data in the event log.
11. The data processing apparatus according to any of the preceding claims, wherein the black box circuitry is configured to capture the event log in a black box memory storage circuitry separate from the memory storage circuitry.
12. The data processing apparatus according to any of the preceding claims, wherein the black box circuitry is configured to operate in a black box power domain, the black box power domain being separate from at least one power domain associated with the processing circuitry and the memory storage circuitry.
13. The data processing apparatus according to any of the preceding claims, wherein the black box circuit is disposed within the same semiconductor package as the processing circuit and the memory storage circuit.
14. The data processing apparatus according to any of the preceding claims, wherein the black box circuit and at least a portion thereof are disposed on the same chip as at least one of the processing circuit and the memory storage circuit.
15. The data processing apparatus according to any preceding claim, the data processing apparatus comprising an input / output interface circuit for communicating with an external device outside the data processing apparatus according to an input / output interface protocol; The communication between the black box circuit and other parts of the data processing device is performed via an internal signal path separate from the input / output interface circuit.
16. The data processing apparatus according to any preceding claim, wherein the abnormal event includes at least one of the following: System crash event; A possible attack launched by a malicious party against the data processing device has been detected; Abnormal interruption or failure mode; Abnormal modular redundancy error mode; An abnormal processing performance associated with the data processing device was detected; The system's configuration and / or integrity may be subject to changes that could compromise security. Remove physical protective measures.
17. The data processing apparatus according to any preceding claim, wherein the event log includes data from portions of the data processing apparatus that are not accessible via external pins of the one or more chips of the data processing apparatus.
18. The data processing apparatus according to any preceding claim, wherein the event log comprises at least one of the following: Context information indicating the processing context that was active when the abnormal event occurred; Software version information indicating the version of the software being processed when the abnormal event occurred; Fingerprint data indicating the active fingerprint on the data processing device; Interrupted data; Performance monitoring data; Accessing data from memory; The memory contains built-in self-test data; Prefetcher data; Branch predictor data; Data transmitted from one or more adjacent chips other than the one or more chips that include the black box circuit to adjacent chip data of the black box circuit; Power management data associated with the power management of the data processing device; and Modular redundancy verification information.
19. The data processing apparatus according to any preceding claim, wherein, in the absence of the aforementioned anomalous event, the black-box circuitry is configured to capture safety-critical telemetry data and non-safety-critical telemetry data; and In response to the occurrence of the abnormal event, the black box circuit is configured to prioritize retaining the safety-critical telemetry data rather than retaining the non-safety-critical telemetry data.
20. A system comprising: A data processing apparatus according to any of the preceding claims, implemented in at least one packaged chip; At least one system component; and plate, The at least one packaged chip and the at least one system component are assembled on the board.
21. A chip-containing product, the chip-containing product comprising the system of claim 20, wherein the system is assembled on an additional board together with at least one other product component.
22. A computer-readable code for manufacturing a data processing apparatus according to any of the preceding claims.
23. A method, the method comprising: Data processing is performed on a data processing device implemented on one or more small chips, the data processing device including processing circuitry and memory storage circuitry, the processing circuitry being used to perform the data processing, and the memory storage circuitry being used to store data and instructions for processing by the processing circuitry. as well as In response to the detection of an abnormal event indicating a risk of loss of safety-critical functions of the data processing device, an event log of telemetry data is captured, the event log indicating the behavior of the data processing device associated with time points before or after the occurrence of the abnormal event; The event log is captured using a black-box circuit on at least one of the one or more chips integrated into the data processing device.
24. A chiplet for a data processing apparatus, the chiplet comprising: A black-box circuit, in response to detecting the occurrence of an abnormal event indicating a risk of loss of safety-critical functions of a data processing device, captures an event log of telemetry data, the event log indicating the behavior of the data processing device associated with time points before or after the occurrence of the abnormal event. and An internal signal path is used for communication between the black box circuit and other parts of the data processing device. This internal signal path is separate from the input / output interface circuitry used by the data processing device to communicate with external devices outside the data processing device according to an input / output interface protocol.
