A life cycle management system and method based on UB interconnection protocol and IO interconnection chip

Abstract

The application discloses a life cycle management system based on a UB interconnection protocol, which is applied to a UB interconnection domain driven based on the UB interconnection protocol, and is arranged on an IO interconnection chip. The IO interconnection chip is arranged on an access path of a third-party access object to the UB interconnection domain. The life cycle management system comprises a receiving module, an execution module and a controlled management module. The receiving module is used for extracting a request type of a life cycle request and acquiring real-time state information of the third-party access object. The execution module is used for controlling an execution mode according to the request type and the real-time state information and executing a corresponding life cycle action. The controlled management module is used for performing controlled management on the third-party access object. The life cycle management system continuously performs controlled management on the third-party access object in a life cycle of the third-party access object. The application establishes a life cycle management chain of the third-party access object, improves running stability of the third-party access object and improves running efficiency of the IO interconnection chip.

Description

Technical Field

[0001] This invention relates to the field of data governance, and in particular to a lifecycle governance system, method, and I / O interconnect chip based on the UB interconnect protocol. Background Technology

[0002] With the explosive growth in AI computing power demand, chip interconnect technology is gradually evolving from traditional closed architectures to open interconnects. The UB (Lingqu) interconnect protocol breaks down the boundaries between bus and network, enabling pooled management and equal access to heterogeneous computing resources such as CPUs (Central Processing Units), NPUs (Neural Processing Units), memory, and storage, supporting large-scale expansion of tens of thousands of computing components. The responsibilities of the IO interconnect chip located at the third-party access boundary are no longer limited to traditional data transmission and basic isolation, but also need to further assume responsibilities such as object access governance, runtime governance, exception governance, recovery governance, and lifecycle governance within the UB interconnect domain.

[0003] However, existing lifecycle governance technologies for UB interconnect systems still have significant shortcomings. Current solutions handle various lifecycle actions of third-party access objects in a fragmented manner by different modules, failing to establish a constraint relationship between lifecycle actions and the real-time status of third-party access objects. This treats each action as an independent management operation, increasing the operational burden on the I / O interconnect chip. Furthermore, existing solutions lack a unified control mechanism for various lifecycle actions, making it difficult to mitigate the risks posed by high-risk lifecycle actions of third-party access objects. This hinders stable governance of third-party access objects throughout their entire lifecycle, making the I / O interconnect chip susceptible to disturbances. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention discloses a lifecycle governance system based on the UB interconnect protocol, applied to a UB interconnect domain driven by the UB interconnect protocol. The lifecycle governance system is deployed on an I / O interconnect chip, which is positioned on the access path from a third-party access object to the UB interconnect domain. The lifecycle governance system includes: a receiving module, used to receive lifecycle requests from the third-party access object, extract the request type of the lifecycle request, and obtain the real-time status information of the third-party access object; an execution module, connected to the receiving module, used to perform gating judgment based on the request type and the real-time status information to determine the corresponding control execution method for the lifecycle action, and execute the lifecycle action based on the control execution method; and a controlled governance module, connected to the execution module, used to perform controlled governance on the third-party access object based on the execution result of the control execution method. The lifecycle governance system continuously performs controlled governance on the third-party access object throughout its lifecycle.

[0005] Preferably, the real-time status information includes at least one of permission information, fault information, resource information, and recovery information; the control execution method includes any one of direct execution, proxy execution, delayed execution, and refusal execution; the execution module includes: a legality verification unit, used to verify the legality of the lifecycle request based on the permission information; a risk determination unit, connected to the legality verification unit, used to determine the risk level of the corresponding lifecycle action based on the request type when the lifecycle request is legal; the risk level includes a first risk level and a second risk level, the risk level of the first risk level being less than the risk level of the second risk level; and a status determination unit, connected to the legality verification unit, used to determine the status based on the fault information, the resource information, and the recovery information when the lifecycle request is legal. The governance status of a third-party access object is described; the governance status includes a normal control status and a restricted control status; a first execution unit, connected to the risk determination unit and the status determination unit, is used to directly execute the lifecycle action of the first risk level when the third-party access object is in the normal control status; a second execution unit, connected to the risk determination unit and the status determination unit, is used to proxy the execution of the lifecycle action of the second risk level when the third-party access object is in the normal control status, and to narrow the permission scope corresponding to the third-party access object; a third execution unit, connected to the status determination unit, is used to delay the execution of the lifecycle action when the third-party access object is in the restricted control status; a fourth execution unit, connected to the legality verification unit, is used to refuse to execute the lifecycle action when the lifecycle request is invalid.

[0006] Preferably, the second execution unit further includes a stage execution subunit, used to execute each of the sub-actions sequentially at a preset action execution frequency when the lifecycle action includes multiple sub-actions; the stage execution subunit executes the corresponding sub-action based on a preset whitelist; the stage execution subunit stops executing the corresponding sub-action when the third-party access object changes to the restricted control state, until the third-party access object returns to the normal control state and then re-executes the corresponding sub-action.

[0007] Preferably, the third execution unit includes: a storage subunit for temporarily storing the delayed lifecycle actions in a delay queue; an observation subunit for continuously observing the governance status of the third-party access object; an extraction subunit connected to the storage subunit and the observation subunit, for sequentially extracting the lifecycle actions from the delay queue when the governance status of the third-party access object is the normal control status; and an execution subunit connected to the extraction subunit for executing the extracted lifecycle actions according to the risk level.

[0008] Preferably, the controlled governance module includes: a result acquisition unit, used to determine whether the lifecycle action has been completed and to acquire the execution result after completion; a basic recovery unit, connected to the result acquisition unit, used to restore the basic permissions of the third-party access object after acquiring the execution result; the basic permissions include basic read / write, minimum management access, and necessary status reporting; a detection unit, connected to the result acquisition unit and the basic recovery unit, used to update the real-time status information of the corresponding third-party access object according to the execution result, and to continuously detect multiple real-time detection indicators in the real-time status information in multiple consecutive detection windows; a comparison unit, connected to the detection unit, used to compare each real-time detection indicator with the corresponding preset indicator threshold in each detection window; a permission recovery unit, connected to the comparison unit, used to restore the permission range of the third-party access object when each real-time detection indicator meets the corresponding preset indicator threshold in a consecutive preset number of detection windows; and a restricted control unit, connected to the comparison unit, used to update the governance status of the third-party access object to the restricted control status when the real-time detection indicator does not meet the corresponding preset indicator threshold.

[0009] Preferably, the permission scope is pre-divided into multiple sequentially ordered permission stages; the permission recovery unit restores one permission stage of the third-party access object in each detection window according to the order of the permission stages, and stops restoring the permission scope when the restricted management unit updates the governance status of the third-party access object to the restricted management status.

[0010] This invention discloses a lifecycle governance method based on the UB interconnection protocol, applied to the lifecycle governance system. The lifecycle governance method includes: step S1, receiving a lifecycle request from a third-party access object, extracting the request type of the lifecycle request, and obtaining the real-time status information of the third-party access object; step S2, performing gating judgment based on the request type and the real-time status information to determine the corresponding lifecycle action control execution mode, and executing the lifecycle action based on the control execution mode; step S3, performing controlled governance on the third-party access object based on the execution result of the control execution mode; and continuously executing steps S1 to S3 throughout the lifecycle of the third-party access object until the lifecycle ends.

[0011] Preferably, the real-time status information includes at least one of permission information, fault information, resource information, and recovery information; the control execution method includes any one of direct execution, proxy execution, delayed execution, and refusal execution; step S2 includes: step S21, verifying the legality of the lifecycle request based on the permission information; step S22a, determining the risk level of the corresponding lifecycle action based on the request type when the lifecycle action is legal; the risk level includes a first risk level and a second risk level, wherein the risk level of the first risk level is less than the risk level of the second risk level; step S22b, determining the control of the third-party access object based on the fault information, the resource information, and the recovery information when the lifecycle action is legal. The governance state includes a normal control state and a restricted control state; step S22c, when the lifecycle request is invalid, refuse to execute the lifecycle action; after executing steps S22a and S22b, proceed to step S23; step S23 includes steps S23a, S23b and S23c; step S23a, when the third-party access object is in a normal control state, directly execute the lifecycle action of the first risk level; step S23b, when the third-party access object is in a normal control state, proxy execute the lifecycle action of the second risk level and narrow the permission scope corresponding to the third-party access object; step S23c, when the third-party access object is in a restricted control state, delay the execution of the lifecycle action.

[0012] Preferably, step S3 includes: step S31, determining whether the lifecycle action has been completed and obtaining the execution result after completion; step S32, restoring the basic permissions of the third-party access object; the basic permissions include basic read / write, minimum management access, and necessary status reporting; step S33, updating the real-time status information of the corresponding third-party access object according to the execution result; continuously detecting multiple real-time detection indicators in the real-time status information in multiple consecutive detection windows; step S34, comparing each real-time detection indicator with the corresponding preset indicator threshold in each detection window; step S35a, restoring the permission range of the third-party access object when each real-time detection indicator meets the corresponding preset indicator threshold in a consecutive preset number of detection windows; step S35b, updating the governance status of the third-party access object to the restricted control status when the real-time detection indicator does not meet the corresponding preset indicator threshold.

[0013] This invention discloses an IO interconnect chip, which is set on the access path of a third-party access object to access the UB interconnect domain, and the IO interconnect chip is equipped with the lifecycle governance system.

[0014] The following beneficial effects can be obtained by using the present invention: 1. By acquiring the action type of lifecycle actions and the real-time status information of third-party access objects, gating judgment is performed based on the action type and real-time status information to determine the management and execution method of lifecycle actions. This eliminates the need to execute different lifecycle actions separately, establishes a lifecycle governance chain for third-party access objects, improves the operational stability of third-party access objects, and reduces the operational burden of IO interconnect chips.

[0015] 2. Based on the corresponding control and execution method, lifecycle actions are continuously executed throughout the lifecycle of the third-party access object, and the third-party access object is managed and governed to achieve stable governance throughout its entire lifecycle, making the IO interconnect chip less prone to disturbance. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the lifecycle governance system of the present invention; Figure 2 This is a schematic diagram of the execution module of the present invention; Figure 3 This is a schematic diagram of the structure of the third execution unit of the present invention; Figure 4 This is a schematic diagram of the controlled governance module of the present invention; Figure 5This is a flowchart illustrating the lifecycle governance method of the present invention; Figure 6 This is a schematic diagram of the process of executing lifecycle actions in this invention; Figure 7 This is a schematic diagram illustrating the controlled governance process for third-party access objects in this invention.

[0017] In the attached diagram: 1. Receiving module; 2. Execution module; 21. Legality verification unit; 22. Risk determination unit; 23. Status determination unit; 24. First execution unit; 25. Second execution unit; 26. Third execution unit; 261. Storage subunit; 262. Observation subunit; 263. Extraction subunit; 264. Execution subunit; 27. Fourth execution unit; 3. Controlled governance module; 31. Result acquisition unit; 32. Basic recovery unit; 33. Detection unit; 34. Comparison unit; 35. Permission recovery unit; 36. Restricted management unit. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0020] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0021] This invention discloses a lifecycle governance system based on the UB interconnect protocol, applied to a UB interconnect domain driven by the UB interconnect protocol. The lifecycle governance system is deployed on an I / O interconnect chip, which is positioned on the access path from third-party access objects to the UB interconnect domain; such as Figure 1As shown, the lifecycle governance system includes: a receiving module 1, used to receive lifecycle requests from third-party access objects, extract the request type of lifecycle actions, and obtain real-time status information of the third-party access objects; an execution module 2, connected to the receiving module 1, used to perform gating judgment based on the request type and real-time status information to determine the corresponding lifecycle action control execution method, and execute the lifecycle action based on the control execution method; and a controlled governance module 3, connected to the execution module 2, used to perform controlled governance on the third-party access objects based on the execution result of the control execution method. The lifecycle governance system continuously executes the lifecycle actions of the third-party access objects throughout their lifecycle and continuously performs controlled governance on them.

[0022] Specifically, the UB interconnection domain is used to receive transaction access, message interaction, maintenance operations and management access initiated by third-party access objects, and connects the shared resource domain and other service domains to send the above content to the shared resource domain or other service domains that are the target access objects.

[0023] Third-party access objects include, but are not limited to, AI accelerators, network accelerators, switching chips, storage controllers, board-level expansion modules, and other heterogeneous objects accessed through UB-related interfaces. These objects are used to access the IO interconnect chip and, after passing security verification, send request statements to the UB interconnect domain through the IO interconnect chip to access the shared resource domain, thereby performing resource acquisition, resource allocation, and other operations.

[0024] A lifecycle request is a transaction initiated by a third-party access object during its lifecycle. The request content specifically includes the lifecycle actions that need to be performed. At different stages of its lifecycle, the third-party access object needs to initiate corresponding lifecycle requests to the I / O interconnect chip to execute the corresponding lifecycle actions.

[0025] In this invention, lifecycle actions include: 1) discovery actions, such as performing link neighbor discovery, object existence detection, and object subject establishment for newly accessed objects; 2) enumeration actions, such as performing enumeration management, reading basic object information, reading resource representations, and establishing initial configuration for third-party accessed objects; 3) power-on actions, such as performing controlled power-on, de-silencing, or entering a manageable state for third-party accessed objects or their partial functional units; 4) power-off actions, such as performing controlled power-off, shutting down some functions, or exiting operation for third-party accessed objects; 5) reset actions, such as performing partial reset, soft reset, controlled reset, or state switching before and after reset for third-party accessed objects. 6) Upgrade actions, such as performing firmware upgrades, configuration updates, version switching, or image switching on third-party access objects; 7) Diagnostic actions, such as performing configuration checks, resource space checks, diagnostic read / write operations, status self-checks, or health checks on third-party access objects; 8) Suspension actions, such as performing transaction suspension, resource freezing, sharing qualification suspension, or pre-recovery preparation on third-party access objects; 9) Recovery actions, such as performing recovery observation, capability recovery, resource qualification recovery, and permission restoration on third-party access objects; 10) Exit actions, such as performing controlled offline, removal from the resource pool, deregistration from the management plane, or removal from the unified protocol domain release set on third-party access objects.

[0026] The real-time status information of third-party access objects includes: 1) Admission status, such as whether the third-party access object has passed trusted access and whether a governance entity has been established; 2) Object description status, such as whether the GUID (Globally Unique Identifier), Type, ClassCode, Version, configuration space, resource space, and governance template of the third-party access object have been established and bound; 3) Operational profile status, such as whether the third-party access object is currently in a normal, observation, restricted, isolated pending, recovery observation, or phased re-entry state; 4) Resource status, such as whether the third-party access object occupies a shared VC (Virtual Channel), shared Credit pool, shared queue, shared cache, shared management path, or shared recovery window, and whether the relevant shared resources currently allow the action to be executed; 5) Fault status, such as whether the third-party access object is a fault source, an affected object, a managed recovery object, or an isolated pending object; 6) Recovery status, such as whether the third-party access object is in the recovery observation period and whether it meets the conditions for entering the next recovery phase; 7) Management permission status, such as whether the entity triggering the action is UBFM (UBFabric). Whether the Manager (Lingqu Exchange Structure Manager) or User (User Management Entity) is accessed legally, and whether the identification information management flag matches the UPI (UB Path Identifier).

[0027] Furthermore, execution module 2 performs gating judgment on third-party access objects based on request type and real-time status information, and determines the corresponding lifecycle action control execution method. For example, when a third-party access object has not yet passed trusted access, it is allowed to perform restricted discovery and restricted enumeration, but not to perform full configuration writing or full power-on. As another example, when a third-party access object is in a recovery observation state, it is allowed to perform basic diagnostics and limited configuration checks, but not to directly restore full shared resource eligibility.

[0028] Furthermore, after implementing the controlled execution method, the scope of permissions for third-party access objects is correspondingly reduced. The controlled governance module 3 updates the real-time state information of a third-party access object after a lifecycle action is executed, and restores the reduced scope of permissions.

[0029] The lifecycle governance system disclosed in this invention continuously governs third-party access objects throughout their entire lifecycle, establishing a complete lifecycle governance chain to ensure the stable operation of third-party access objects.

[0030] In a preferred embodiment of the present invention, the real-time status information includes at least one of permission information, fault information, resource information, and recovery information; the control execution method includes any one of direct execution, proxy execution, delayed execution, and execution rejection; such as Figure 2 As shown, execution module 2 includes: a legality verification unit 21, used to verify the legality of lifecycle requests based on permission information; a risk determination unit 22, connected to the legality verification unit 21, used to determine the risk level of the corresponding lifecycle action based on the request type when the lifecycle request is legal; the risk level includes a first risk level and a second risk level, with the risk level of the first risk level being less than that of the second risk level; and a status determination unit 23, connected to the legality verification unit 21, used to determine the governance status of the third-party access object based on fault information, resource information, and recovery information when a lifecycle action request is made; the governance status includes a regular control status and a restricted control status; and the first execution unit 2. 4. Connecting the risk determination unit 22 and the status determination unit 23, the second execution unit 25 is used to directly execute the lifecycle action of the first risk level when the third-party access object is in a normal control state; the third execution unit 26 is connected to the status determination unit 23, and is used to proxy the execution of the lifecycle action of the second risk level when the third-party access object is in a normal control state, and to narrow the permission scope corresponding to the third-party access object; the fourth execution unit 27 is connected to the legality verification unit 21, and is used to refuse to execute the lifecycle action when the lifecycle request is invalid.

[0031] Specifically, permission-related information includes admission status, object description status, and management permission status. The legality verification unit 21 determines whether the third-party access object has completed trusted access based on its admission status. Subsequently, it determines the legality of the lifecycle request based on whether the object's GUID, Type, Class Code, Version, configuration space, resource space, and governance template have been established and bound, whether access to the relevant configuration space / resource space is legal, and whether the identification information management flag matches the UPI. Only when all judgment results are satisfactory is the lifecycle action deemed legal.

[0032] Furthermore, after determining that the lifecycle request is legitimate, the risk determination unit 22 determines the risk level of the corresponding lifecycle action. The first risk level corresponds to a low-risk action, and the second risk level corresponds to a high-risk action.

[0033] Specifically, lifecycle actions can be categorized into access actions, control actions, maintenance actions, and recovery actions. Within access actions, initial discovery and initial enumeration are considered low-risk actions, while re-enumeration and forced re-registration are considered high-risk actions. Within control actions, partial power-on, partial power-off, and temporary pause are considered low-risk actions, while global reset, global power-off, and forced isolation are considered high-risk actions. Within maintenance actions, basic diagnostics, configuration reading, and status self-checks are considered low-risk actions, while firmware upgrades, configuration rewriting, and version switching are considered high-risk actions. Within recovery actions, basic capability restoration and observation period status reporting are considered low-risk actions, while full recovery, controlled exit, and re-release are considered high-risk actions.

[0034] The status determination unit 23 determines the restricted control status based on information such as the third-party access object's occupation of the shared resource domain, whether it has become a source of failure or an affected object, and whether it has been isolated. This includes the isolation pending state and the observation and recovery state.

[0035] Furthermore, the first execution unit 24 is used to directly execute low-risk actions, which requires that the third-party access object be in a normal management and control state, the resources of the shared resource domain be accessible to the third-party access object normally, and the permission scope of the third-party access object be in line with the access to the shared resource domain.

[0036] The second execution unit 25 is used to perform proxy execution for high-risk actions. It needs to be executed after sequential proxy, rhythm proxy, whitelist proxy or intermediate state protection is implemented by the IO interconnect chip at the boundary side.

[0037] Specifically, the boundary side of the I / O interconnect chip serves as the access point for third-party access objects to connect to the I / O interconnect chip. Implementing proxy execution at the boundary side can effectively prevent high-risk actions from directly affecting the shared resource domain, causing shared resource domain failures, and subsequently spilling over to other third-party access objects.

[0038] The third execution unit 26 is used to delay the execution of lifecycle actions of third-party access objects that are in a restricted control state. For example, when the resource status indicates that the action is not allowed to be executed or the recovery status indicates that the recovery conditions for the next recovery stage are not met, the lifecycle action is placed in the delay queue until the real-time status information of the third-party access object updates its governance status to the normal control state before execution.

[0039] The fourth execution unit 27 is used to refuse to execute the corresponding lifecycle action when the lifecycle request is invalid.

[0040] In a preferred embodiment of the present invention, the second execution unit 25 further includes a stage execution subunit, which is used to execute each sub-action sequentially at a preset action execution frequency when the lifecycle action includes multiple sub-actions; the stage execution subunit executes the corresponding sub-action based on a preset whitelist; the stage execution subunit stops executing the corresponding sub-action when the third-party access object is switched to a restricted control state, and resumes executing the corresponding sub-action when the third-party access object is restored to a normal control state.

[0041] Specifically, lifecycle actions include individual actions (single actions) and step-by-step actions (step-by-step actions comprising multiple sub-actions). The stage execution sub-unit performs staged execution for step-by-step actions comprising multiple sub-actions. This is achieved by dividing multiple sub-actions into multiple sub-action stages according to their action sequence, and then proxying the execution of the corresponding sub-actions in each sub-action stage.

[0042] Furthermore, the proxy methods for phased execution sub-units include sequential proxy, whitelist proxy, rhythm proxy, resource protection proxy, and linkage proxy.

[0043] The default whitelist includes whitelists for lifecycle actions and whitelists for shared resource domains. During whitelist proxying, the stage execution subunit only executes lifecycle actions from the whitelist and accesses shared resource domains from the whitelist.

[0044] When the sub-action to be executed is not in the corresponding whitelist or the shared resource domain to be accessed is not in the corresponding whitelist, the execution of the corresponding sub-action is stopped and an alarm message is sent to the third-party access object. When the governance status of the third-party access object changes to restricted control status after the execution of a sub-action, the execution of the next sub-action is stopped, and the next sub-action is executed only after the third-party access object returns to normal control status. For example, if a lifecycle action has four sub-actions, and the pressure on the shared resource domain exceeds a preset pressure threshold after the second sub-action is executed, causing the third-party access object to be isolated and placed in restricted control status. The third-party access object is released from isolation only after the pressure on the shared resource domain decreases, and the phase execution sub-unit restarts the execution of the unexecuted sub-actions.

[0045] When performing rhythm proxy, the phase execution subunit controls the overall execution speed of lifecycle actions at the boundary side of the IO interconnect chip. By extending the execution interval between adjacent sub-actions and limiting the number of sub-actions executed per unit time, it avoids the hardware overload of third-party access objects or instantaneous congestion in shared resource domains caused by actions being executed too quickly.

[0046] When a retryable error occurs during the execution of a sub-action, the stage execution sub-unit schedules the retry timing in a unified manner according to the preset exponential backoff strategy to prevent frequent retries from amplifying system failures, while also avoiding invalid retries from consuming management path bandwidth.

[0047] When the phase execution subunit performs the corresponding lifecycle action within its lifecycle, a preset waiting period and verification step are inserted between each sub-action to prevent hardware timing abnormalities or inconsistent system states caused by excessively rapid state transitions.

[0048] When executing the resource protection agent, the phase execution subunit first shrinks the shared resource eligibility of the third-party access object, retains the resource eligibility necessary to execute the corresponding lifecycle actions, freezes the shared resource queue marked as a sensitive queue, and selectively postpones the shared recovery window according to the execution time of the resource protection agent.

[0049] The UB Interconnected Open System includes multiple different types of governance modules. The second execution unit 25, as a unit that acts as an agent to execute lifecycle actions, has corresponding linkages with other governance modules.

[0050] On the one hand, in conjunction with the trusted access mechanism, third-party access objects need to pass the trusted access judgment of the IO interconnect chip when accessing the IO interconnect chip. If the trusted access is not passed, the IO interconnect chip only allows the second execution unit to perform minimal probing or minimal enumeration, and does not allow it to enter the complete life cycle action chain, and updates the admission status.

[0051] On the other hand, in conjunction with the unified capability descriptor and runtime profile binding mechanism, the IO interconnect chip promotes the establishment of object subjects through discovery and enumeration actions; and promotes the update of object description versions, profile freezing or rebinding through upgrade, reset and recovery actions, and updates the object description status and profile update status.

[0052] On the other hand, in conjunction with the protocol semantic recognition, adjudication and proxy execution mechanism, when performing high-risk actions such as reset, upgrade and recovery, the transaction semantic permissions within the scope of the third-party access object are automatically contracted and switched to the boundary proxy mode, which is then executed by the second execution unit 25.

[0053] On the other hand, it is linked with trusted telemetry, health scoring and policy linkage mechanism. This mechanism is used to determine whether life cycle actions are legal, whether it is allowed to enter the next stage, and whether it is allowed to restore full capability. It is jointly constrained by health score, trend status and resource domain risk, and updates the recovery status.

[0054] On the other hand, in conjunction with the virtual fault domain isolation mechanism, before performing reset, pause or exit actions on the fault source object, it first switches to the restricted fault domain or isolation domain and updates the fault status.

[0055] On the other hand, in conjunction with the abnormal transaction management, proxy completion and recovery replay mechanism, before executing reset, power down, exit or version switch, unfinished transactions first enter the managed scanning and differentiated closure process, sort and execute the lifecycle actions that have not yet been executed, and allow exit after all lifecycle actions have been executed.

[0056] On the other hand, in conjunction with SLA and runtime degradation mechanisms, the service level and shared resource eligibility of objects can be temporarily reduced and the resource status updated during the execution of lifecycle actions.

[0057] On the other hand, in conjunction with the thermal / power consumption perception degradation and reentry mechanism, when the lifecycle action itself amplifies the thermal power consumption risk, priority is given to adopting phased actions with low thermal shock and low resource competition.

[0058] In a preferred embodiment of the present invention, such as Figure 3 As shown, the third execution unit 26 includes: a storage subunit 261, used to temporarily store delayed lifecycle actions in a delayed queue; an observation subunit 262, used to continuously observe the governance status of third-party access objects; an extraction subunit 263, connected to the storage subunit 261 and the observation subunit 262 respectively, used to extract lifecycle actions from the delayed queue sequentially when the governance status of the third-party access object is a normal control status; and an execution subunit 264, connected to the extraction subunit 263, used to execute the extracted lifecycle actions according to the risk level.

[0059] Specifically, lifecycle actions of third-party access objects will be continuously issued. If the third-party access object remains in a restricted control state, the lifecycle actions will be stored in a delay queue in the order they are issued until the third-party access object returns to a normal control state.

[0060] Furthermore, execution subunit 264 directly executes low-risk actions and delegates the execution of high-risk actions.

[0061] In a preferred embodiment of the present invention, such as Figure 4As shown, the controlled governance module 3 includes: a result acquisition unit 31, used to determine whether the lifecycle action has been completed and to acquire the execution result after completion; a basic recovery unit 32, connected to the result acquisition unit 31, used to restore the basic permissions of the third-party access object after acquiring the execution result; the basic permissions include basic read / write, minimum management access, and necessary status reporting; a detection unit 33, connected to the result acquisition unit 31 and the basic recovery unit 32, used to update the real-time status information of the corresponding third-party access object according to the execution result, and to continuously detect multiple real-time detection indicators in the real-time status information in multiple consecutive detection windows; a comparison unit 34, connected to the detection unit 33, used to compare each real-time detection indicator with the corresponding preset indicator threshold in each detection window; a permission recovery unit 35, connected to the comparison unit 34, used to restore the permission range of the third-party access object when each real-time detection indicator meets the corresponding preset indicator threshold in a consecutive preset number of detection windows; and a restricted control unit 36, connected to the comparison unit 34, used to update the governance status of the third-party access object to a restricted control status when the real-time detection indicator does not meet the corresponding preset indicator threshold.

[0062] Specifically, the result acquisition unit 31 is used to confirm that the lifecycle action was successfully executed and to obtain the execution result. If only a low-risk action is executed under normal control conditions and the corresponding permission scope is not narrowed, then there is no need to perform subsequent permission restoration and restricted control, and the system waits for the next lifecycle action to be executed.

[0063] After executing a high-risk action, the second execution unit 25 reduces the scope of permissions for the third-party access object, while retaining the permission scope to execute the corresponding lifecycle actions. The basic recovery unit 32 first restores the basic read / write, minimal management access, and necessary status reporting of the third-party access object.

[0064] The detection unit 33 determines the stabilization duration based on the recovery status in the real-time status information, determines the anomaly count based on the fault status, determines the health score of the configuration space / resource space object based on the running profile status, determines the management message access success rate based on the management permission status, and determines the shared resource occupancy ratio based on the resource status in multiple detection windows.

[0065] The comparison unit 34 determines in each detection window whether the stable duration is greater than the preset minimum stable duration, whether the abnormal count is less than the preset maximum allowed abnormal count, whether the configuration space / resource space reaches the preset object health score, whether the management message access success rate reaches the preset success rate, and whether the shared resource occupancy ratio is lower than the preset pressure threshold.

[0066] The permission recovery unit 35 gradually restores the restricted permission scope of the third-party access object when all the aforementioned real-time detection indicators meet the corresponding preset indicator thresholds. The complete permission scope is restored when each real-time detection indicator meets the corresponding detection indicator threshold within a consecutive preset number of detection windows.

[0067] During the recovery process, if a non-compliance occurs in a certain detection window, the restricted control unit 36 ​​stops the recovery of the permission scope, sets the governance status of the third-party access object to the restricted control status, and observes it in each detection window until no abnormalities occur in multiple detection windows, after which the governance status is set back to the normal control status.

[0068] In a preferred embodiment of the present invention, the scope of permissions is pre-divided into multiple sequentially ordered permission stages; the permission recovery unit 35 restores one permission stage of the third-party access object in each detection window according to the order of the permission stages, and stops restoring the scope of permissions when the restricted management unit 36 ​​updates the governance status of the third-party access object to the restricted management status.

[0069] Specifically, when the real-time detection indicators all meet the corresponding preset indicator thresholds, the permission recovery unit 35 re-enters the shrunken permission scope in real time in stages, and gradually restores the general transactions, restricted doorbell capabilities, necessary management access, shared resource qualifications and higher priority services of the third-party access object in multiple consecutive detection windows.

[0070] Specifically, general transactions include basic function calls and low-priority interrupts, as they have the lowest risk and can quickly verify whether the basic business path of an object is normal. Even if an anomaly occurs, it will not have a serious impact on the system. Higher-priority transactions include high-bandwidth data transmission services and services in dedicated shared resource domains, because they consume a lot of system resources and have a wide impact. They must be restored only after the object is completely stable to avoid system instability or recurrence of failures caused by service recovery.

[0071] This invention discloses a lifecycle governance method based on the UB interconnection protocol, applied to the aforementioned lifecycle governance system; such as Figure 5 As shown, the lifecycle governance method includes: Step S1, receiving a lifecycle request from a third-party access object, extracting the request type of the lifecycle request, and obtaining the real-time status information of the third-party access object; Step S2, performing gating judgment based on the request type and real-time status information to determine the control execution method for lifecycle actions, and executing the corresponding lifecycle actions based on the control execution method; Step S3, performing controlled governance on the third-party access object based on the execution result of the control execution method; Steps S1 to S3 are continuously executed within the lifecycle of the third-party access object until the lifecycle ends.

[0072] In a preferred embodiment of the present invention, the real-time status information includes at least one of permission information, fault information, resource information, and recovery information; the control execution method includes any one of direct execution, proxy execution, delayed execution, and execution rejection; such as Figure 6 As shown, step S2 includes: step S21, verifying the legality of the lifecycle request based on permission information; step S22a, determining the risk level of the lifecycle action based on the action type when the lifecycle request is legal; the risk level includes a first risk level and a second risk level, with the risk level of the first risk level being lower than that of the second risk level; step S22b, determining the governance status of the third-party access object based on fault information, resource information, and recovery information when the lifecycle request is legal; the governance status includes a normal control status and a restricted control status; step S22c, refusing to execute the lifecycle action when the lifecycle request is illegal; after executing steps S22a and S22b, proceeding to step S23; step S23 includes steps S23a, S23b, and S23c; step S23a, directly executing the lifecycle action of the first risk level when the third-party access object is in a normal control status; step S23b, proxying the execution of the lifecycle action of the second risk level when the third-party access object is in a normal control status, and narrowing the permission scope corresponding to the third-party access object; step S23c, delaying the execution of the lifecycle action when the third-party access object is in a restricted control status.

[0073] In a preferred embodiment of the present invention, such as Figure 7 As shown, step S3 includes: Step S31, determining whether the lifecycle action has been completed and obtaining the execution result after completion; Step S32, restoring the basic permissions of the third-party access object; the basic permissions include basic read / write, minimum management access, and necessary status reporting; Step S33, updating the real-time status information of the corresponding third-party access object according to the execution result; continuously detecting multiple real-time detection indicators in the real-time status information in multiple consecutive detection windows; Step S34, comparing each real-time detection indicator with the corresponding preset indicator threshold in each detection window; Step S35a, restoring the permission range of the third-party access object when each real-time detection indicator meets the corresponding preset indicator threshold in a consecutive preset number of detection windows; Step S35b, updating the governance status of the third-party access object to a restricted control status when the real-time detection indicator does not meet the corresponding preset indicator threshold.

[0074] This invention discloses an IO interconnect chip, which is set on the access path of a third-party access object to access the UB interconnect domain, and the IO interconnect chip is equipped with the aforementioned lifecycle governance system.

[0075] Those skilled in the art should understand that lifecycle actions can be triggered by UBFM or by a user or system software with legitimate permissions under constrained conditions; the validity verification of lifecycle requests can be performed by the hardware state machine or by the management microcontroller and hardware logic working together; high-risk actions can be performed by full proxy, or by phased proxy, delayed proxy, or whitelist proxy; recovery and reentry can be determined by referring only to the object's own state or by combining the shared resource domain state, health score, and recovery window state.

[0076] The above description is merely a preferred embodiment of the present invention and does not limit the implementation and protection scope of the present invention. Those skilled in the art should realize that any equivalent substitutions and obvious changes made based on the description and illustrations of the present invention should be included within the protection scope of the present invention.

Claims

1. A lifecycle governance system based on the UB interconnect protocol, applied to a UB interconnect domain driven by the UB interconnect protocol, characterized in that, The lifecycle governance system is deployed on the IO interconnect chip, which is set on the access path from the third-party access object to the UB interconnect domain; The lifecycle governance system includes: The receiving module is used to receive the lifecycle request of the third-party access object, extract the request type of the lifecycle request, and obtain the real-time status information of the third-party access object. An execution module, connected to the receiving module, is used to perform gating judgment based on the request type and the real-time status information to determine the corresponding lifecycle action control execution mode, and execute the lifecycle action based on the control execution mode; A controlled governance module, connected to the execution module, is used to perform controlled governance on the third-party access object based on the execution result of the control execution method; The lifecycle governance system continuously performs controlled governance on the third-party access object throughout its lifecycle; The real-time status information includes at least one of the following: permission information, fault information, resource information, and recovery information; The control execution method includes any one of direct execution, proxy execution, delayed execution, and refusal to execute; The execution module includes: A legality verification unit is used to verify the legality of the lifecycle request based on the permission class information; A risk determination unit, connected to the legality verification unit, is used to determine the risk level of the corresponding lifecycle action based on the request type when the lifecycle request is legal. The risk levels include a first risk level and a second risk level, wherein the risk level of the first risk level is less than the risk level of the second risk level; A status determination unit, connected to the legality verification unit, is used to determine the governance status of the third-party access object based on the fault class information, the resource class information, and the recovery class information when the lifecycle request is legal. The governance status includes regular control status and restricted control status; The first execution unit, connected to the risk determination unit and the status determination unit, is used to directly execute the lifecycle action of the first risk level when the third-party access object is in a normal management and control state; The second execution unit, connected to the risk determination unit and the status determination unit, is used to proxy the execution of the lifecycle action of the second risk level when the third-party access object is in a normal management and control state, and to narrow the permission scope corresponding to the third-party access object; The third execution unit, connected to the state determination unit, is used to delay the execution of the lifecycle action when the third-party access object is in a restricted control state. The fourth execution unit, connected to the legality verification unit, is used to refuse to execute the lifecycle action when the lifecycle request is invalid.

2. The lifecycle governance system according to claim 1, characterized in that, The second execution unit further includes a stage execution subunit, used to execute each of the sub-actions sequentially at a preset action execution frequency when the lifecycle action includes multiple sub-actions; The stage execution subunit executes the corresponding sub-action based on a preset whitelist list; The phase execution subunit stops executing the corresponding sub-action when the third-party access object changes to the restricted control state, and resumes executing the corresponding sub-action when the third-party access object returns to the normal control state.

3. The lifecycle governance system according to claim 1, characterized in that, The third execution unit includes: A storage subunit is used to temporarily store the delayed lifecycle actions in a delay queue; An observation subunit is used to continuously observe the governance status of the third-party access object. An extraction subunit is connected to the storage subunit and the observation subunit, respectively, and is used to extract the lifecycle actions in the delay queue sequentially when the governance status of the third-party access object is the normal management and control status; An execution subunit, connected to the extraction subunit, is used to execute the extracted lifecycle actions according to the risk level.

4. The lifecycle governance system according to claim 1, characterized in that, The controlled governance module includes: The result acquisition unit is used to determine whether the lifecycle action has been completed and to acquire the execution result after completion. A basic recovery unit, connected to the result acquisition unit, is used to restore the basic permissions of the third-party access object after acquiring the execution result; The basic permissions include basic read and write, minimal management access, and necessary status reporting; The detection unit, connected to the result acquisition unit and the basic recovery unit, is used to update the real-time status information of the corresponding third-party access object according to the execution result, and to continuously detect multiple real-time detection indicators in the real-time status information in multiple consecutive detection windows. The comparison unit, connected to the detection unit, is used to compare each of the real-time detection indicators with the corresponding preset indicator threshold in each detection window; A permission restoration unit, connected to the comparison unit, is used to restore the permission range of the third-party access object when each of the real-time detection indicators meets the corresponding preset indicator threshold in a consecutive preset number of detection windows. The restricted control unit, connected to the comparison unit, is used to update the governance status of the third-party access object to the restricted control status when the real-time detection indicator does not meet the corresponding preset indicator threshold.

5. The lifecycle governance system according to claim 4, characterized in that, The scope of permissions is pre-divided into multiple permission stages that are ordered sequentially; The permission restoration unit restores one permission stage of the third-party access object sequentially in each detection window according to the order of the permission stages, and stops restoring the permission scope when the restricted management unit updates the governance status of the third-party access object to the restricted management status.

6. A lifecycle governance method based on the UB interconnection protocol, characterized in that, Applied to the lifecycle governance system as described in any one of claims 1-5; The lifecycle governance method includes: Step S1: Receive the lifecycle request of the third-party access object, extract the request type of the lifecycle request, and obtain the real-time status information of the third-party access object; Step S2: Based on the request type and the real-time status information, perform gating judgment to determine the corresponding lifecycle action control execution method, and execute the lifecycle action based on the control execution method; Step S3: Perform controlled management on the third-party access object based on the execution result of the control execution method; Steps S1 to S3 are continuously executed throughout the lifecycle of the third-party access object until the lifecycle ends.

7. The lifecycle governance method according to claim 6, characterized in that, The real-time status information includes at least one of the following: permission information, fault information, resource information, and recovery information; The control execution method includes any one of direct execution, proxy execution, delayed execution, and refusal to execute; Step S2 includes: Step S21: Verify the legality of the lifecycle request based on the permission type information; Step S22a: When the lifecycle request is valid, determine the risk level of the corresponding lifecycle action based on the request type; The risk levels include a first risk level and a second risk level, wherein the risk level of the first risk level is less than the risk level of the second risk level; Step S22b: When the lifecycle request is valid, determine the governance status of the third-party access object based on the fault class information, the resource class information, and the recovery class information; The governance status includes regular control status and restricted control status; Step S22c: If the lifecycle request is invalid, refuse to execute the lifecycle action; After performing steps S22a and S22b, proceed to step S23; Step S23 includes steps S23a, S23b and S23c; Step S23a: When the third-party access object is in a normal management and control state, directly execute the lifecycle action of the first risk level; Step S23b: When the third-party access object is in a normal management state, the proxy performs the lifecycle action of the second risk level and narrows the permission scope corresponding to the third-party access object; Step S23c: Delay the execution of the lifecycle action when the third-party access object is in a restricted control state.

8. The lifecycle governance method according to claim 7, characterized in that, Step S3 includes: Step S31: Determine whether the lifecycle action has been completed, and obtain the execution result after completion; Step S32: Restore the basic permissions of the third-party access object; The basic permissions include basic read and write, minimal management access, and necessary status reporting; Step S33: Update the real-time status information of the corresponding third-party access object according to the execution result; Multiple real-time detection indicators in the real-time status information are continuously detected in multiple consecutive detection windows; Step S34: In each detection window, compare each real-time detection indicator with the corresponding preset indicator threshold. Step S35a: When each of the real-time detection indicators meets the corresponding preset indicator threshold in a consecutive preset number of detection windows, the permission range of the third-party access object is restored; Step S35b: When the real-time detection indicator does not meet the corresponding preset indicator threshold, update the governance status of the third-party access object to the restricted control status.

9. An I / O interconnect chip, characterized in that, The IO interconnect chip is located on the access path of a third-party access object to access the UB interconnect domain, and the IO interconnect chip is equipped with the lifecycle governance system as described in any one of claims 1-5.

Images