System for configurable error handling

By detecting I/O hardware endpoint errors in the operating system and triggering interrupts, determining the use of operating system or firmware to handle errors, the problem of difficulty in updating error propagation schemes in the prior art is solved, and fast and flexible error handling is achieved, reducing the number of System downtime.

CN112970002BActive Publication Date: 2024-11-01MICROSOFT TECHNOLOGY LICENSING LLC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN201980073582.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2018-11-08
Filing Date
2019-11-01
Publication Date
2024-11-01
Estimated Expiration
2039-11-01

AI Technical Summary

Technical Problem

The existing technology lacks a flexible error handling mechanism and cannot update the error propagation scheme as needed without reinitializing the hardware, resulting in too long system downtime.

Method used

By detecting errors on I/O hardware endpoints in the operating system and triggering corresponding interrupts, the advanced configuration and power interface system driver runtime components determine whether to use the operating system or firmware to handle errors without changing the platform BIOS Update the error propagation plan below.

Benefits of technology

It realizes rapid update of error propagation schemes without reinitializing hardware, reduces system downtime, and improves system flexibility and configurability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112970002B_ABST
    Figure CN112970002B_ABST
Patent Text Reader

Abstract

The error handling system provides detection of an error on an I / O hardware endpoint, triggering of an operating system interrupt in response to a detected error, receipt of an interrupt at an operating system component, determination of whether to handle the error using an operating system handler or a firmware error handler associated with the I / O hardware endpoint in response to the received interrupt, and triggering of a firmware interrupt associated with the firmware runtime error handler if it is determined that the error is to be handled using a firmware runtime error handler associated with the I / O hardware endpoint.
Need to check novelty before this filing date? Find Prior Art

Description

Background Art

[0001] Traditional computer hardware devices are able to detect errors as they operate and report them to the computing platform on which they are installed. The errors may then be propagated through the platform software stack for appropriate handling. The errors may be propagated to the firmware or directly to the operating system (O / S). If the error is first propagated to the firmware, the BIOS system management mode component receives control for handling the error, and control is then passed to the O / S.

[0002] During the pre-boot process, the platform is initialized to specify an error propagation scheme (e.g., firmware first or O / S first) for each installed hardware device. Such hardware devices may include memory modules and peripheral component interconnect express (PCIe) devices for accelerated networking, compute offload, storage, or other services. The error propagation scheme specified for a device cannot be changed during runtime.

[0003] A flexible error handling mechanism is desired by which the O / S can determine whether to handle the error in an appropriate manner or provide control back to the BIOS for typical firmware-first error handling. On-demand updates to error propagation schemes for specific hardware endpoints are also desired, especially updates that do not require hardware reinitialization. Technical Field

[0004] Example embodiments of the present disclosure relate generally to the field of computer technology, and more particularly to a system for configurable error handling. Summary of the invention

[0005] In a first aspect of the present disclosure, a computing system is proposed, the computer system comprising: an I / O hardware endpoint; an I / O hardware endpoint controller, which communicates with the I / O hardware endpoint and is used to: detect an error on the I / O hardware endpoint; and trigger an interrupt associated with the I / O hardware endpoint in response to the detected error; and a processor, which is used to execute a runtime component executed by an operating system of the computing system to: receive the interrupt; in response to the received interrupt, determine whether the operating system includes a device-specific method associated with the interrupt; and if it is determined that the operating system includes a device-specific method associated with the interrupt, determine to use an operating system processor associated with the I / O hardware endpoint to handle the error; and if it is determined that the operating system does not include a device-specific method associated with the interrupt: determine to use a firmware runtime error handler associated with the I / O hardware endpoint to handle the error; and trigger a firmware interrupt associated with the firmware runtime error handler.

[0006] In a second aspect of the present disclosure, a computer-implemented method is proposed, comprising: detecting an error on an I / O hardware endpoint; triggering an operating system interrupt associated with the I / O hardware endpoint in response to the detected error; receiving the interrupt at a runtime component executed by the operating system; in response to the received interrupt, determining at the runtime component whether the operating system includes a device-specific method associated with the interrupt; if it is determined that the operating system includes a device-specific method associated with the interrupt, determining to use an operating system processor associated with the I / O hardware endpoint to handle the error; and if it is determined that the operating system does not include a device-specific method associated with the interrupt: determining to use a firmware runtime error handler associated with the I / O hardware endpoint to handle the error; and triggering a firmware interrupt associated with the firmware runtime error handler.

[0007] In a third aspect of the present disclosure, a computing platform providing a runtime architecture is proposed, comprising: an I / O hardware endpoint; an I / O hardware endpoint controller, which communicates with the I / O hardware endpoint; an advanced configuration and power interface system driver runtime component; one or more device-specific methods associated with the advanced configuration and power interface system driver runtime component; a system management mode runtime component; one or more system management mode error handlers associated with the system management mode runtime component; and a processor for detecting an error on the I / O hardware endpoint and triggering an interrupt associated with the I / O hardware endpoint in response to the detected error, wherein the advanced configuration and power interface system driver runtime component The runtime component is used to: receive the interrupt, and in response to the received interrupt, determine whether the one or more device-specific methods include a device-specific method associated with the received interrupt; if it is determined that the one or more device-specific methods include a device-specific method associated with the received interrupt, determine to use a system management mode processor associated with the I / O hardware endpoint to handle the error; and if it is determined that the one or more device-specific methods do not include a device-specific method associated with the received interrupt: determine to use a firmware runtime error handler associated with the I / O hardware endpoint to handle the error; and trigger a firmware interrupt associated with the firmware runtime error handler. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 Error propagation in a platform architecture according to some embodiments is shown.

[0009] Figure 2 is a block diagram of a hardware architecture of a server platform according to some embodiments.

[0010] Figure 3is a flow chart of a process to propagate memory errors according to some embodiments.

[0011] Figure 4 is a flow chart of a process for propagating PCIe errors according to some embodiments.

[0012] Figure 5 is a flow chart of a process for propagating PCIe errors according to some embodiments.

[0013] Figure 6 is a block diagram illustrating an alternative I / O endpoint according to some embodiments.

[0014] Figure 7 is a flow diagram of a process to change the error propagation scheme for a replaced I / O endpoint in a server platform without requiring a hard reboot, according to some embodiments. DETAILED DESCRIPTION

[0015] The following description is provided to enable anyone skilled in the art to make and use the described embodiments. However, various modifications will still be readily apparent to those skilled in the art.

[0016] Some embodiments provide a technical solution to the technical problem of fixing error propagation schemes by determining at the O / S whether to handle errors in the O / S or in firmware based on updateable filtering logic. Embodiments also provide a technical solution to the problem of excessive downtime during platform customization / updates by allowing updates to filtering logic without hardware reinitialization.

[0017] In general, some embodiments operate to detect hardware endpoint errors and trigger a software interrupt received by an Advanced Configuration and Power Interface (ACPI) runtime component. The ACPI runtime component determines whether the error is specified to be handled by a firmware-based error handler, and if so, triggers a software interrupt to the firmware component for handling the error. If the error is not specified to be handled by a firmware-based error handler, the error is handled by an appropriate O / S error handler.

[0018] The determination of whether an error is designated as being handled by a firmware-based error handler can be based on whether the corresponding ACPI device-specific method is present or absent in the O / S layer. Advantageously, and in accordance with some embodiments, the error propagation scheme for a hardware endpoint can be changed without changing the platform BIOS, but only by updating one or more ACPI device-specific methods and the O / S handler. In some instances, the ACPI device-specific scheme and the O / S handler can be updated (and the error propagation scheme thereby changed) without reinitializing the platform, thereby minimizing system downtime.

[0019] Figure 1 1 is a block diagram of a platform architecture 100 according to some embodiments. The platform architecture 100 may represent a server runtime architecture. The architecture 100 may support user mode applications and / or virtual machines as known in the art.

[0020] The hardware layer 110 of the architecture 100 includes a central processing unit (CPU) 112, which is configured to detect errors from various hardware endpoints 114-117. According to some embodiments, hardware endpoint 114-117 errors are detected by a controller or port of the CPU 112. An embodiment may include more than one CPU, and each of such CPUs may receive error notifications from any number of hardware endpoints.

[0021] The hardware layer 110 also includes a service processor 118. The service processor 118 may include a microcontroller that is distinct from the CPU 112 and manages platform-specific functions such as monitoring environmental conditions and handling specific error conditions. The service processor 118 may maintain error records in a system event log for post-crash analysis and may include, for example, a component of a baseboard management controller.

[0022] The BIOS layer 120 includes runtime components that are initialized during BIOS boot. Specifically, the system management mode (SMM) runtime 122 is initialized and associated with the device specific error manager 124. The ACPI runtime 132 is also initialized during BIOS boot.

[0023] BIOS boot may also include configuration of error signaling such as ACPI system control interrupts (SCI) as opposed to SMIs (system management interrupts) propagated to SMM runtime 122. This configuration may be specified via general purpose IO (GPIO) settings of a platform controller hub (not shown) as known in the art.

[0024] O / S layer 130 includes runtime components that are instantiated during boot of the O / S following BIOS boot. Error handling filter 133 may include device specific methods (_DSM) associated with one or more hardware endpoints. Use of error handling filter 133 will be described in detail below.

[0025] Error handler 134, kernel 136, and O / S event log 137 of the O / S layer may operate as known in the art. For example, error handler 134 corresponding to hardware having an error condition creates an error packet and forwards the packet to kernel 136. Kernel 136 creates an error record based on the error packet and may log the error record among error records 138 of O / S event log 137. In the case of a non-fatal uncorrected hardware error, kernel 136 may attempt to correct the error condition.

[0026] Figure 2 is a block diagram of a platform 200 according to some embodiments. The platform 200 may implement the architecture 100, but embodiments are not limited thereto.

[0027] The platform 200 includes two CPUs 202 and 204, which may include any type of CPU known or that will become known. Embodiments are not limited to platforms including any particular number of CPUs. Each CPU 202 and 204 is linked to various I / O interfaces, into which corresponding hardware I / O endpoints may be installed.

[0028] For example, CPU 202 is coupled to dual in-line memory modules (DIMMs) 212 and 214. CPU 202 includes an integrated memory controller 203 to interface with DIMMs 212. Thus, memory controller 203 operates to detect errors in DIMMs 212 and 214 and take appropriate corrective action if possible. Memory controller 203 may also log error information into error context registers such as a structure of memory specific registers (MSRs) of CPU 202. CPU 203 is directly coupled to DIMMs 213 and 215 and includes a memory controller 205 that may operate as described above.

[0029] CPU 202 is also directly coupled to PCIe slots 222, 224, and 226. PCIe slots 222, 224, and 226 can be installed with expansion cards that provide networking, graphics processing, storage, or other functions. Each expansion card can communicate directly with the PCIe root port of CPU 202 through its associated PCIe slot 222, 224, or 226 and the bus connecting the slot to CPU 202. The PCIe root port of CPU 202 can detect errors on any PCIe endpoints to which it is connected. Again, CPU 204 is also coupled to PCIe slots 225 and 227, and can include a PCIe root port to detect errors from the PCIe endpoints installed therein.

[0030] CPU 202 is coupled to a platform controller hub (PCH) 230 as is known in the art. PCH 230 controls certain data paths and supports functions used in conjunction with CPUs 202 and 204. PCH 230 is coupled to flash memory 239, which stores BIOS, and to baseboard management controller (BMC) 240. BMC 240 is a dedicated service processor that monitors the physical state of platform 200 and can support out-of-band communications with system administrators through management interface 246.

[0031] Figure 3 A flow chart of a process 300 according to some embodiments is included. In some embodiments, a processing unit (e.g., one or more processors, processing cores, processor threads) of a computing device (e.g., a computer server) executes a program code so that the device performs the process 300. Process 300 and all other processes mentioned herein can be embodied in a processor executable program code read from one or more non-transient computer-readable media such as a hard disk, flash memory, and then stored in a compressed, uncompiled and / or encrypted format. In some embodiments, a hard-wired circuit can be used in place of the program code or in conjunction with the program code for implementation of the process according to some embodiments. The embodiments are therefore not limited to any particular combination of hardware and software.

[0032] Process 300 begins with the detection of a memory error at S310. As described above, a platform memory controller (which may or may not be integrated into the platform CPU) may detect a memory error at S310 using a consistent error signaling and detection protocol. The memory controller may take corrective action and may also log corresponding error information into an error context register built into the platform CPU.

[0033] The error information recorded in the log causes the CPU to trigger an interrupt at S320. The interrupt may include: Figure 1 The SCI received by the ACPI runtime component is shown in . Next, at S330, the ACPI runtime component determines whether the error should be handled by a firmware-based error handler in the BIOS runtime or by an O / S error handler.

[0034] The determination at S330 may include determining whether the O / S includes a device-specific method associated with the error source. If no such method exists, the flow proceeds to S340 to process the interrupt using an O / S error handler (e.g., processor 134). Advantageously, such error handling may proceed to the processing thread that received the error, leaving other processing threads to perform O / S functions.

[0035] If a device specific method exists, the interrupt is handled by an error handler registered with the BIOS runtime (e.g., with the system management mode runtime) at S350. In this regard, the ACPI runtime may be as follows: Figure 1 As shown in , a system management interrupt (SMI) is initiated to trigger the SMM error handler. The flow then continues from S350 to S340 to transfer control to the O / S error handler.

[0036] Process 300 need not be limited to memory errors, but can be extended to handle errors detected for any hardware endpoint.The specific implementation of error detection and initiation of the SCI may differ depending on the hardware endpoint. Figure 4 Process 400 depicts handling of detected PCIe errors according to some embodiments.

[0037] PCIe errors are detected at S410. Each PCIe endpoint installed in the platform is associated with a PCIe root port. The PCIe root port can detect errors from its associated PCIe device. The PCIe root endpoint can be integrated into the CPU, PCH, or any suitable hardware component.

[0038] In response to the detection of the error, the PCIe root endpoint triggers a PCIe Advanced Error Reporting (AER) message signaled interrupt (MSI) at S420. The MSI is received by the O / S root endpoint error handler, which triggers the SCI in response thereto. As previously described, the SCI is received by the ACPI runtime component, which determines at S430 whether the error should be handled by a firmware-based error handler in the BIOS runtime.

[0039] The determination at S430 may be similar to that described above with respect to S330. Specifically, if a device-specific method associated with the error source exists, the ACPI runtime may initiate an SMI at S440 to trigger a corresponding SMM error handler to handle the error. The flow then proceeds to S450 to return control to the O / S root port error handler. If a corresponding device-specific method is not identified at S430, the error is returned to the O / S root port error handler at S450 without first returning control to the BIOS.

[0040] Figure 5 Process 500 shows an alternative way to handle PCIe errors according to some embodiments. S510, S520, S530 and S540 are similar to those described above in the case where it is determined that the detected PCIe error is to be handled exclusively by the O / S root port processor. S410, S420, S430 and S440 are performed.

[0041] However, if it is determined at S530 that the detected PCIe error will be handled by the SMM processor, the process first advances to S550 to use the SMM processor to handle the error. After the SMM processor completes its exposure, it is determined at S560 whether the error is a correctable error. If so, the SCI is triggered at S570 to return a notification to the ACPI runtime. If the error is determined to be uncorrectable at S560, a non-maskable interrupt (NMI) is triggered at S580. The NMI is reported to the O / S regardless of the current interrupt priority level of the processor. The NMI generally indicates a fatal hardware error condition and is handled appropriately by the O / S.

[0042] Figure 6 Shown is the replacement of a PCIe endpoint installed in slot 222 of platform 200 with an unidentified (eg, upgraded) PCIe endpoint. Such replacement of an endpoint may typically require a firmware update or at least a system reboot to ensure proper error handling.

[0043] Figure 7 A process 700 for responding to a replaced hardware endpoint according to some embodiments is shown. At S710, an endpoint is replaced during O / S execution. For example, during replacement of an endpoint within a platform (e.g., a server) of S710, the platform is fully booted into its O / S, and the O / S runs to service requests received from applications.

[0044] Next, at S720, the O / S is updated to disassociate the endpoint using the BIOS error handler. It is therefore assumed that the endpoint was previously associated with the SMM error handler, and its errors were first handled by the SMM error handler as described with respect to processes 400 and 500. As also previously described, some embodiments may use ACPI device-specific methods to associate the error source with the SMM error handler. Therefore, S720 may include disabling or uninstalling the ACPI device-specific methods that associated the previous endpoint with the SMM error handler. The O / S is then updated at S730 to include the O / S error handler associated with the replacement endpoint.

[0045] At S740, the platform is rebooted to the now updated O / S without executing the BIOS. In some embodiments, S740 includes a kernel soft reboot (KSR) that reboots the O / S without reinitializing the platform hardware. For example, control is passed to a driver instead of firmware after shutdown and proceeds directly to kernel initialization while bypassing platform initialization, boot manager, and O / S loader. KSR can advantageously allow execution of the updated O / S while experiencing minimal platform downtime.

[0046] As a result of the foregoing, and in accordance with processes 400 and 500, detected errors from the new endpoint will be handled by the newly included O / S processor rather than by the SMM processor associated with the previous endpoint. Such an arrangement allows new endpoint hardware error handling to be introduced into the platform without requiring firmware updates. Furthermore, using KSR or other similar techniques, platform downtime is significantly reduced.

[0047] Each functional component described herein may be implemented in computer hardware (integrated and / or discrete circuit components), in program code, and / or in one or more computing systems that execute such program code as known in the prior art. Such a computing system may include one or more processing units that execute executable program code stored in a memory system.

[0048] The aforementioned accompanying drawings represent the logical architecture for describing the process according to some embodiments, and in practice you can include more or different components that are arranged in other ways. Other topologies can be connected to other embodiments and be used. In addition, each component or device described herein can be realized by any number of devices via any number of other public and / or private network communications. Two or more computing devices in such computing devices can be positioned away from each other, and can communicate with each other via any known means of (multiple) networks and / or dedicated links. Each component or device can include any number of hardware and / or software elements that are suitable for providing the functions described herein and any other functions.

[0049] The embodiments described herein are for illustrative purposes only. Those skilled in the art will recognize that other embodiments can be practiced using modifications or changes to the embodiments described above.

Claims

1. A computing system, include: I / O hardware endpoints; An I / O hardware endpoint controller, which communicates with the I / O hardware endpoint and is used to: detecting an error on the I / O hardware endpoint; as well as triggering an interrupt associated with the I / O hardware endpoint in response to the detected error; as well as a processor to execute a runtime component executed by an operating system of the computing system to: receiving the interrupt; In response to receiving the interrupt, determining whether the operating system includes a device specific method associated with the interrupt; and If it is determined that the operating system includes a device specific method associated with the interrupt, determining to use an operating system handler associated with the I / O hardware endpoint to handle the error; as well as If it is determined that the operating system does not include a device specific method associated with the interrupt: determining to handle the error using a firmware runtime error handler associated with the I / O hardware endpoint; as well as A firmware interrupt associated with the firmware runtime error handler is triggered.

2. The computing system of claim 1, wherein the interrupt is an Advanced Configuration and Power Interface (ACPI) system control interrupt and the runtime component is an Advanced Configuration and Power Interface (ACPI) software driver.

3. The computing system of claim 2, wherein the firmware interrupt is a system management interrupt and the firmware runtime error handler is a system management mode error handler.

4. The computing system of claim 2, wherein the determination of handling the error using a firmware error handler associated with the I / O hardware endpoint include: Identification of advanced configuration and power interface device-specific methods registered with the operating system and associated with the I / O hardware endpoint and the firmware error handler. 5 . The computing system of claim 1 , the processor further configured to execute the firmware error handler and, after execution of the firmware error handler, execute an operating system handler associated with the I / O hardware endpoint.

6. The computing system of claim 1 , wherein the I / O hardware endpoint is a memory device, wherein the I / O hardware endpoint controller is a memory controller within the processor, and wherein the interrupt is triggered in response to the memory controller writing error information into one or more machine-specific registers of the processor.

7. The computing system of claim 1, wherein the I / O hardware endpoint is a Peripheral Component Interconnect Express device, wherein the I / O hardware endpoint controller is a root port controller, and wherein the interrupt is triggered to a root port error handler.

8. A computer-implemented method, include: Detect errors on I / O hardware endpoints; triggering an operating system interrupt associated with the I / O hardware endpoint in response to the detected error; receiving the interrupt at a runtime component executed by the operating system; In response to receiving the interrupt, determining at the runtime component whether the operating system includes a device specific method associated with the interrupt; If it is determined that the operating system includes a device specific method associated with the interrupt, determining to use an operating system handler associated with the I / O hardware endpoint to handle the error; as well as If it is determined that the operating system does not include a device specific method associated with the interrupt: determining to handle the error using a firmware runtime error handler associated with the I / O hardware endpoint; as well as A firmware interrupt associated with the firmware runtime error handler is triggered.

9. The computer implemented method of claim 8, wherein the interrupt is an Advanced Configuration and Power Interface system control interrupt and the runtime component is an Advanced Configuration and Power Interface software driver.

10. The computer implemented method of claim 9, wherein the firmware interrupt is a system management interrupt and the firmware runtime error handler is a system management mode error handler.

11. The computer-implemented method of claim 9, wherein determining to handle the error using a firmware error handler associated with the I / O hardware endpoint include: An advanced configuration and power interface device specific method registered with the operating system and associated with the I / O hardware endpoint and the firmware error handler is identified.

12. The computer-implemented method of claim 8, further comprising: include: executing the firmware error handler; as well as After execution of the firmware error handler, an operating system handler associated with the I / O hardware endpoint is executed.

13. The computer-implemented method of claim 8, wherein the I / O hardware endpoint is a memory device, wherein the I / O hardware endpoint controller is a memory controller, and wherein the interrupt is triggered by a processor writing to one or more machine-specific registers of the processor in response to error information.

14. The computer-implemented method of claim 8, wherein the I / O hardware endpoint is a Peripheral Component Interconnect Express device, wherein the I / O hardware endpoint controller is a root port controller, and wherein the interrupt is triggered to a root port error handler.

15. A computing platform providing a runtime architecture, include: I / O hardware endpoints; An I / O hardware endpoint controller that communicates with the I / O hardware endpoint; Advanced Configuration and Power Interface system driver runtime components; one or more device-specific methods associated with the advanced configuration and power interface system driver runtime component; System management mode runtime components; one or more system management mode error handlers associated with the system management mode runtime component; as well as a processor to detect an error on the I / O hardware endpoint and, in response to the detected error, trigger an interrupt associated with the I / O hardware endpoint, wherein the advanced configuration and power interface system driver runtime component is to: receiving the interrupt, and in response to the received interrupt, determining whether the one or more device specific methods include a device specific method associated with the received interrupt; If it is determined that the one or more device specific methods include a device specific method associated with the received interrupt, determining to handle the error using a system management mode processor associated with the I / O hardware endpoint; as well as If it is determined that the one or more device-specific methods do not include a device-specific method associated with the received interrupt: determining to handle the error using a firmware runtime error handler associated with the I / O hardware endpoint; as well as A firmware interrupt associated with the firmware runtime error handler is triggered.

16. The computing platform of claim 15, wherein the ACI system driver runtime component triggers a system management interrupt if a system management mode error handler of the one or more system management mode error handlers determines to handle the error using a firmware runtime error handler.

17. The computing platform of claim 15, wherein the determination of the error is processed using a firmware error handler associated with the I / O hardware endpoint. include: A determination that one of the advanced configuration and power interface device specific methods is associated with the I / O hardware endpoint and one of the one or more system management mode error handlers.

18. The computing platform of claim 17, wherein the processor further executes the one of the one or more system management mode error handlers and, after execution of the one of the one or more system management mode error handlers, executes an operating system processor associated with the I / O hardware endpoint.

19. The computing platform of claim 15, wherein the I / O hardware endpoint is a memory device, wherein the I / O hardware endpoint controller is a memory controller within the processor, and wherein the interrupt is triggered in response to the memory controller writing error information to one or more machine-specific registers of the processor.

20. The computing platform of claim 15, wherein the I / O hardware endpoint is a Peripheral Component Interconnect Express device, wherein the I / O hardware endpoint controller is a root port controller, and wherein the interrupt is triggered to a root port error handler.