CN118426794A - Acceleration card upgrading method, device and system, electronic equipment and storage medium - Google Patents

Abstract

The invention discloses an accelerating card upgrading method, wherein an accelerating card comprises a normal version image and a recovered version image, the normal version image and the recovered version image are isolated from each other, and the recovered version image comprises a firmware partition for providing re-upgrading capability; acquiring an upgrade state of a normal version image, wherein the upgrade state comprises upgrade failure or upgrade success; if the upgrade status is upgrade failure, the method enters a recovered version image and carries out next online automatic upgrade through the firmware partition. By setting the normal version image and the recovery version image which are isolated from each other in the accelerator card, the operation of the accelerator card can be supported by the recovery version image when the normal version image fails to upgrade, and the accelerator card entering the recovery version image can be automatically upgraded again on line due to the firmware partition providing the re-upgrading capability in the recovery version image, so that the accelerator card can be maintained without manual factory return, and the maintenance cost of the accelerator card is reduced.

Description

Accelerator card upgrade method, device, system, electronic device and storage medium

Technical Field

The present invention relates to the technical field of accelerator card upgrade, and in particular to an accelerator card upgrade method, device, system, electronic equipment and storage medium.

Background technique

The system upgrade solution for AI smart acceleration cards generally uses the recovery (data recovery) mode to perform OTA online automatic upgrades. The recovery mode provides a minimum system (recovery system) to prevent the normal system from failing to start after the upgrade. When the normal system fails to start after the upgrade, the minimum system is entered for startup and operation. However, the minimum system does not have the ability to automatically upgrade again and can only be manually returned to the factory for repair, which increases the maintenance cost of the smart acceleration card. Therefore, in the existing system upgrade mode, after the smart acceleration card upgrade fails, there is a problem that the smart acceleration card cannot be automatically upgraded again.

Summary of the invention

The embodiment of the present invention provides a video processing method, which aims to solve the problem that in the existing system upgrade mode, after the smart acceleration card fails to be upgraded, the smart acceleration card cannot be automatically upgraded again. By setting a normal version image and a recovery version image that are isolated from each other in the acceleration card, the operation of the acceleration card can be supported by the recovery version image when the normal version image fails to be upgraded. Since the recovery version image is provided with a firmware partition that provides re-upgrade capability, the acceleration card that enters the recovery version image can be automatically upgraded online again, and there is no need to return the card to the factory for repair manually, thereby reducing the maintenance cost of the acceleration card.

In a first aspect, the main purpose of the present invention is to provide an accelerator card upgrade method, wherein the accelerator card includes a normal version image and a recovery version image, the normal version image and the recovery version image are isolated from each other, and the recovery version image includes a firmware partition that provides re-upgrade capability, and the method includes:

Obtaining the upgrade status of the normal version image, wherein the upgrade status includes upgrade failure or upgrade success;

If the upgrade status is upgrade failure, enter the recovery version image and perform the next online automatic upgrade through the firmware partition.

Optionally, before obtaining the upgrade status of online automatic upgrade of the normal version image, the method further includes:

Receive the normal version upgrade package sent by the master end;

The normal version image is automatically upgraded online through the normal version upgrade package.

Optionally, before the normal version image is upgraded or after the normal version image is successfully upgraded, the method further includes:

Run the normal version image or the normal version image after successful upgrade, and upgrade the recovery version image.

Optionally, obtaining the upgrade status of the normal version image includes:

Obtaining the startup status of the upgraded normal version image through the program state machine;

According to the startup status, the upgrade status of the normal version image is determined.

Optionally, determining the upgrade status of the normal version image according to the startup status includes:

When the normal version image after the upgrade is started normally, if the number of normal startup failures is greater than a preset value, determining that the upgrade status of the normal version image is upgrade failure;

When the normal version image after the upgrade is started for the upgrade, if it is not the first startup after the upgrade, it is determined that the upgrade status of the normal version image is upgrade failure.

Optionally, the communication link between the master end and the acceleration card includes a first communication link and a second communication link, the first communication link and the second communication link include different communication subjects, and before performing the next online automatic upgrade by restoring the version image, the method further includes:

If the first communication link is normal, controlling the acceleration card to enter the recovery version image through the first communication link;

If the first communication link is abnormal, the acceleration card is controlled to enter the recovery version image through the second communication link.

In a second aspect, an embodiment of the present invention provides an accelerator card upgrade device, wherein the accelerator card includes a normal version image and a recovery version image, the normal version image and the recovery version image are isolated from each other, and the recovery version image includes a firmware partition that provides a re-upgrade capability, and the accelerator card upgrade device includes:

An acquisition module, used to acquire the upgrade status of the normal version image, wherein the upgrade status includes upgrade failure or upgrade success;

The first upgrade module is used to enter the recovery version image and perform the next online automatic upgrade through the firmware partition if the upgrade status is upgrade failure.

In a third aspect, the system includes: an acceleration card and a host end, the acceleration card includes an acceleration card upgrade device, a normal version image and a recovery version image, and the acceleration card upgrade device is used to execute any acceleration card upgrade method described in the embodiments of the present invention.

In a fourth aspect, an embodiment of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the acceleration card upgrade method as described above when executing the computer program.

In a fifth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the acceleration card upgrade method as described above are implemented.

In an embodiment of the present invention, the accelerator card includes a normal version image and a recovery version image, the normal version image and the recovery version image are isolated from each other, and the recovery version image includes a firmware partition that provides re-upgrade capability; the upgrade status of the normal version image is obtained, and the upgrade status includes upgrade failure or upgrade success; if the upgrade status is upgrade failure, the recovery version image is entered and the next online automatic upgrade is performed through the firmware partition. By setting the normal version image and the recovery version image that are isolated from each other in the accelerator card, the operation of the accelerator card can be supported by the recovery version image when the normal version image fails to be upgraded. Since the recovery version image is provided with a firmware partition that provides re-upgrade capability, the accelerator card that enters the recovery version image can be automatically upgraded online again, and does not need to be manually returned to the factory for repair, thereby reducing the maintenance cost of the accelerator card.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

FIG1 is an architecture diagram of an accelerator card upgrade system provided by an embodiment of the present invention;

FIG2 is a flow chart of an accelerator card upgrade method provided by an embodiment of the present invention;

FIG3 is a schematic diagram of the structure of an accelerator card upgrade device provided by an embodiment of the present invention;

FIG. 4 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention.

Detailed ways

The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

Please refer to Figure 1, which is an architecture diagram of an acceleration card upgrade system provided by an embodiment of the present invention. As shown in Figure 1, the acceleration card upgrade system includes: an acceleration card and a host end, the acceleration card and the host end are communicated with each other, and the acceleration card includes an acceleration card upgrade device, a normal version image, and a recovery version image.

The host end can be a server or a host connected to the accelerator card. The host end can send upgrade operations to the accelerator card through a packaged interface component, and perform message transmission and file transmission with the accelerator card through the communication interface provided by the communication library plug-in. The interface component can be a DCMI (Device Control Management Interface) interface component.

The above-mentioned acceleration card can be an AI smart acceleration card or an AI chip. The above-mentioned acceleration card is mainly used for accelerated computing of visual processing. The above-mentioned acceleration card can call the local upgrade library through the acceleration card upgrade device to perform upgrades and log management.

Furthermore, the acceleration card may be a pcie acceleration card, the host may be a pcie host, and the pcie acceleration card and the pcie host may be communicatively connected via a pcie high-speed peripheral bus. In a possible embodiment, the processor in the pcie acceleration card further includes a core for communication, the pcie host further includes a microcontroller, and the core for communication and the microcontroller may be communicatively connected via a bus, specifically, the core for communication and the microcontroller may be communicatively connected via a Smbus bus.

The above image is a system image, which is used to provide system support for the accelerator card; the above normal version image can be understood as a commonly used version image or a default version image, which corresponds to the commonly used system of the accelerator card; the above recovery version image can be understood as an uncommon version image or a spare version image, which is used to replace the normal version image to provide spare system support for the accelerator card when the system startup or system upgrade of the accelerator card fails. The recovery version image is a streamlined version image, which occupies a smaller storage space in the accelerator card.

The master end publishes normal version image upgrade packages and recovery version image upgrade packages. When the accelerator card needs to be upgraded, it sends a version image upgrade request to the master end. The master end sends the upgrade operation to the accelerator card through the encapsulated interface component and sends the version image upgrade package to the accelerator card through the communication interface provided by the communication library plug-in.

When the normal version image is automatically upgraded online using OTA (Over the Air Technology), the accelerator card upgrade is divided into two stages. The first stage is mainly to prepare for the upgrade. The first stage is under the operation of the normal version image, that is, the first stage is carried out under the operation of the normal version system. The accelerator card receives the normal version image upgrade package from the master end, checks and verifies the upgrade package, and after the preparation is completed, notifies the master end, which sends a reset operation to the accelerator card through the packaged interface component, and the accelerator card enters the recovery mode through the reset operation. The second stage is to upgrade the normal version image in the recovery mode. The second stage is under the operation of the recovery version image, and the upgrade package is written to the storage of the accelerator card to complete the upgrade of the normal version image, and then notifies the master end. The master end sends a reset operation to the accelerator card again through the packaged interface component, and the accelerator card starts the normal version image through the reset operation and enters the upgraded normal version system.

When the normal version system is running normally, if the master end releases a recovery version image upgrade package, it can notify the accelerator card, and the accelerator card sends a recovery version image upgrade request to the master end. The master end sends the recovery version image upgrade package to the accelerator card through the communication interface provided by the communication library plug-in, and writes the recovery version image to the storage of the accelerator card to complete the upgrade of the recovery version image, and then notifies the master end.

After the normal version system upgrade is completed, the upgrade status of the normal version image can be detected. If the upgrade status is successful, the result of the successful upgrade is sent to the master end. If the upgrade status is failed, the result of the failed upgrade is sent to the master end. The master end sends a reset operation to the accelerator card through the packaged interface component. The reset operation makes the accelerator card enter the recovery mode. At this time, the recovery mode will enable the firmware partition in the recovery version image, and the next online automatic upgrade will be performed through the firmware partition in the recovery version image.

The firmware partition may be an upgrade_firmware partition, which may be used to provide a function of continuing to upgrade the version image, so that the online automatic upgrade of the version image may be continued through the upgrade_firmware partition in the recovery version.

Specifically, based on the upgrade in recovery mode, a new upgrade_firmware partition is added to provide the function of continuing OTA upgrade in recovery mode. When the normal upgrade enters recovery mode, the upgrade_firmware firmware in the recovery version image will not be triggered. Only when the OTA upgrade fails and the accelerator card enters recovery mode, the upgrade_firmware firmware in the recovery version image will be triggered, thus ensuring that the next OTA upgrade can be performed.

Please refer to FIG. 2, which is a flow chart of an accelerator card upgrade method provided by an embodiment of the present invention. As shown in FIG. 2, the accelerator card upgrade method is applied to the accelerator card, and is specifically applied to an accelerator card upgrade device in the accelerator card. The accelerator card includes a normal version image and a recovery version image. The normal version image and the recovery version image are isolated from each other. The recovery version image includes a firmware partition that provides re-upgrade capability. The accelerator card upgrade method specifically includes the following steps:

101. Get the upgrade status of the normal version image.

In the embodiment of the present invention, the above image is a system image, which is used to provide system support for the accelerator card; the above normal version image can be understood as a commonly used version image or a default version image, and the above normal version image corresponds to a commonly used system of the accelerator card, and one version image corresponds to one version system. Generally speaking, upgrading the system version refers to upgrading the normal version image, and the upgrade method can be OTA online automatic upgrade.

The upgrade status includes upgrade failure or upgrade success. After upgrading the normal version image, the upgrade status of the normal version image can be determined by starting the normal version image. Generally speaking, startup failure indicates upgrade failure, and startup success indicates upgrade success. After the normal version image is successfully started, the corresponding normal version system can be entered to provide corresponding system support. After the normal version image fails to start, the normal version system cannot be entered and the corresponding system support cannot be provided. Alternatively, the upgrade status of the normal version image can be determined by querying the status result, or by the upgrade log.

102. If the upgrade status is upgrade failure, enter the recovery version image and perform the next online automatic upgrade through the firmware partition.

In the embodiment of the present invention, after obtaining the upgrade status of the normal version image, if the upgrade status is upgrade success, the normal version system can be entered through the normal version image. If the upgrade status is upgrade failure, the normal version system cannot be entered. If there is no recovery version image to enter the corresponding standby version system, the accelerator card cannot be supported by the system, the corresponding system function cannot be realized, and the accelerator card cannot be used, which is the so-called accelerator card bricking.

The above-mentioned accelerator card includes a recovery version image. When the upgrade status of the normal version image is upgrade failure, the recovery version image can be started in recovery mode to enter the corresponding backup version system, so that the accelerator card can be supported by the backup system and realize the corresponding system functions. At the same time, after the normal version image upgrade fails, the firmware partition in the recovery version image will be triggered when entering the recovery mode to provide hardware resources for the recovery version image to continue upgrading, so as to perform the next online automatic upgrade through the firmware partition in the recovery version image.

The above-mentioned recovery version image can be understood as an uncommon version image or a spare version image. The above-mentioned recovery version image is used to replace the normal version image to provide backup system support for the accelerator card when the system startup or system upgrade fails. The recovery version image is a streamlined version image, which occupies a smaller storage space in the accelerator card.

The firmware partition may be an upgrade_firmware partition, which may be used to provide a function of continuing to upgrade the version image, so that the online automatic upgrade of the version image may be continued through the upgrade_firmware partition in the recovery version.

Specifically, based on the upgrade in recovery mode, a new upgrade_firmware partition is added to provide the function of continuing OTA upgrade in recovery mode. When the normal upgrade enters recovery mode, the upgrade_firmware firmware in the recovery version image will not be triggered. Only when the OTA upgrade fails and the accelerator card enters recovery mode, the upgrade_firmware firmware in the recovery version image will be triggered, thus ensuring that the next OTA upgrade can be performed.

In an embodiment of the present invention, the accelerator card includes a normal version image and a recovery version image, the normal version image and the recovery version image are isolated from each other, and the recovery version image includes a firmware partition that provides re-upgrade capability; the upgrade status of the normal version image is obtained, and the upgrade status includes upgrade failure or upgrade success; if the upgrade status is upgrade failure, the recovery version image is entered and the next online automatic upgrade is performed through the firmware partition. By setting the normal version image and the recovery version image that are isolated from each other in the accelerator card, the operation of the accelerator card can be supported by the recovery version image when the normal version image fails to be upgraded. Since the recovery version image is provided with a firmware partition that provides re-upgrade capability, the accelerator card that enters the recovery version image can be automatically upgraded online again, and does not need to be manually returned to the factory for repair, thereby reducing the maintenance cost of the accelerator card.

Optionally, before the step of obtaining the upgrade status of the online automatic upgrade of the normal version image, a normal version upgrade package sent by the master end may be received; and the normal version image may be automatically upgraded online through the normal version upgrade package.

In an embodiment of the present invention, a normal version image upgrade package and a recovery version image upgrade package are published on the master end. When the accelerator card needs to be upgraded, a version image upgrade request is sent to the master end. The master end sends the upgrade operation to the accelerator card through the encapsulated interface component and sends the version image upgrade package to the accelerator card through the communication interface provided by the communication library plug-in.

When the normal version image is automatically upgraded online via OTA, the accelerator card upgrade is divided into two stages. The first stage is mainly for preparations related to the upgrade. The first stage is under the operation of the normal version image, that is, the first stage is carried out under the operation of the normal version system. The accelerator card receives the normal version image upgrade package from the master, checks and verifies the upgrade package, and after the preparation is completed, notifies the master. The master sends a reset operation to the accelerator card through the packaged interface component, and the accelerator card enters the recovery mode through the reset operation. The second stage is to upgrade the normal version image in the recovery mode. The second stage is under the operation of the recovery version image, and the upgrade package is written to the storage of the accelerator card to complete the upgrade of the normal version image. Then the master is notified. The master sends a reset operation to the accelerator card again through the packaged interface component. The accelerator card starts the normal version image through the reset operation and enters the upgraded normal version system.

Optionally, before the normal version image is upgraded or after the normal version image is successfully upgraded, the normal version image or the normal version image after the successful upgrade may be run to upgrade the recovery version image.

In an embodiment of the present invention, after starting the normal version image and entering the normal version system, the accelerator card can query whether the master end has released a recovery version image upgrade package. If the master end has a recovery version image upgrade package, a recovery version image upgrade request can be sent to the master end. The master end sends the recovery version image upgrade package to the accelerator card through the communication interface provided by the communication library plug-in, and writes the recovery version image to the storage of the accelerator card to complete the upgrade of the recovery version image, and then notifies the master end.

The recovery version image can be upgraded by upgrading the recovery version image under the normal operation of the normal version system. It should be noted that the above recovery version image is a simplified version image, and the above recovery version image upgrade package can be tested on the test device first. When the recovery version image upgrade package runs without error, it can be released to the master end.

Optionally, in the step of obtaining the upgrade status of the normal version image, the startup status of the upgraded normal version image may be obtained; and the upgrade status of the normal version image may be determined according to the startup status.

In an embodiment of the present invention, the above-mentioned startup status can be understood as whether the startup of the normal version image is successful. If the startup of the normal version image is successful, the normal version system will be entered. If the startup of the normal version image fails, the normal version system cannot be entered. If there is no backup system, the acceleration card will become bricked and cannot be used.

Specifically, if the startup status of the upgraded normal version image is startup success, the upgrade status of the normal version image can be determined to be upgrade success; if the startup status of the normal version image is startup failure, the upgrade status of the normal version image can be determined to be upgrade failure.

Optionally, in the step of determining the upgrade status of the normal version image according to the startup status, when the upgraded normal version image is started normally, if the number of normal startup failures is greater than a preset value, the upgrade status of the normal version image is determined to be upgrade failure; when the upgraded normal version image is started for upgrade, if it is not the first startup after the upgrade, the upgrade status of the normal version image is determined to be upgrade failure.

In the embodiment of the present invention, the above-mentioned normal startup can be understood as starting the normal version image in a non-upgraded state, and can be understood as starting after a period of time after the upgrade. It can be judged whether it is a normal startup. If it is a normal startup, it is further judged whether the normal startup is successful. If the normal startup is successful, it can be determined that the startup state is a successful startup. If the normal startup fails, it can be judged whether it is an upgrade failure based on the number of normal startup failures. If the number of normal startup failures is greater than the preset value, it can be determined that the upgrade has failed. If the number of normal startup failures is less than or equal to the preset value, the startup will continue to be tried until it is determined that the upgrade has failed or the startup is successful.

The above-mentioned upgrade startup can be understood as startup within a period of time after the upgrade. When it is determined that it is not a normal startup, it can be further determined whether it is an upgrade startup. If it is an upgrade startup, it can be further determined whether it is the first startup after the upgrade. If it is the first startup after the upgrade, the first startup after the upgrade is successful, then it can be determined that the upgrade is successful. If it is the first startup after the upgrade fails, then it can be determined that the upgrade failed.

If it is neither a normal startup nor an upgrade startup, it can be a forced startup. During the forced startup, you will be prompted whether to enter the recovery mode. If you choose yes, you can force enter the recovery mode and start the corresponding recovery version image.

When the upgrade status is upgrade failure, the master end can be notified, and the master end sends a reset operation to the accelerator card through the packaged interface component, and the accelerator card enters the recovery mode through the reset operation.

Optionally, the communication link between the main end and the acceleration card includes a first communication link and a second communication link, and the first communication link and the second communication link include different communication subjects. Before the next online automatic upgrade step through the recovery version image, if the first communication link is normal, the acceleration card is controlled to enter the recovery version image through the first communication link; if the first communication link is abnormal, the acceleration card is controlled to enter the recovery version image through the second communication link.

In an embodiment of the present invention, the acceleration card may be a pcie acceleration card, the host may be a pcie host, the first communication link may be a high-speed peripheral bus pcie link, and the pcie acceleration card and the pcie host may be communicatively connected via the pcie link. The processor in the pcie acceleration card also includes a core for communication, and the pcie host also includes a microcontroller. The second communication link may be used for the communication core to communicate with the BMC (Baseboard Management Controller) microcontroller. Specifically, the second communication link may be an Smbus link based on the Smbus bus, and the core for communication may be communicatively connected with the microcontroller via the Smbus bus. It is understandable that the communication subject of the pcie link is the pcie link interface component between the pcie acceleration card and the pcie host, and the communication subject of the Smbus link is a small core in the pcie acceleration card processor and the BMC microcontroller in the pcie host.

In the process of switching between the normal version image and the recovery version image, the program state machine can control the startup of the normal version image and the recovery version image. Specifically, when the normal version image is upgraded, the initial state of the program state machine is changed to the upgraded state, and when the upgraded normal version image is successfully started, the upgraded state of the program state machine is restored to the initial state. In this way, it can be ensured that in the normal startup scenario, when a startup abnormality occurs, it can switch to the recovery mode and start the corresponding recovery version image.

Specifically, the pcie accelerator card also includes a watchdog timeout interrupt. When the normal version image fails to start, if the pcie link is normal, a message can be sent to the host master through the watchdog timeout interrupt. After the host master receives the message, it operates pcie_hot_reset to reset the accelerator card and put the accelerator card into recovery mode.

When the normal version image fails to start, if the pcie link is abnormal and pcie communication cannot be performed, it can be connected to the Host master with the BMC microcontroller through the Smbus link. The accelerator card uses the Smbus link to communicate with the small core (core used for communication) in the processor to reset the accelerator card.

Optionally, please refer to FIG. 3, which is a schematic diagram of the structure of an accelerator card upgrade device provided by an embodiment of the present invention. As shown in FIG. 3, the accelerator card includes a normal version image and a recovery version image, the normal version image and the recovery version image are isolated from each other, the recovery version image includes a firmware partition that provides re-upgrade capability, and the device includes:

The acquisition module 201 is used to acquire the upgrade status of the normal version image, where the upgrade status includes upgrade failure or upgrade success;

The first upgrade module 202 is used to enter the recovery version image and perform the next online automatic upgrade through the firmware partition if the upgrade status is upgrade failure.

Optionally, the device further comprises:

The receiving module is used to receive the normal version upgrade package sent by the master end;

The second upgrade module is used to automatically upgrade the normal version image online through the normal version upgrade package.

Optionally, the device further comprises:

The third upgrade module is used to run the normal version image or the normal version image after successful upgrade to upgrade the recovery version image.

Optionally, the acquisition module 201 includes:

A status acquisition submodule, used for acquiring the startup status of the upgraded normal version image through a program state machine;

The state determination submodule is used to determine the upgrade state of the normal version image according to the startup state.

Optionally, the state determination submodule includes:

A first determining unit is used to determine that the upgrade status of the normal version image is an upgrade failure when the normal version image after the upgrade is normally started and the number of normal startup failures is greater than a preset value;

The second determining unit is configured to determine that the upgrade status of the normal version image is upgrade failure when the normal version image after the upgrade is started for upgrade, if it is not the first startup after the upgrade.

Optionally, the communication link between the master end and the acceleration card includes a first communication link and a second communication link, the first communication link and the second communication link include different communication entities, and the device further includes:

A first processing module, configured to control the acceleration card to enter the recovery version image through the first communication link if the first communication link is normal;

The second processing module is used to control the acceleration card to enter the recovery version image through the second communication link if the first communication link is abnormal.

It should be noted that the river monitoring device provided in the embodiment of the present invention can be applied to smart cameras, smart phones, computers, servers and other devices that can perform the acceleration card upgrade method.

The accelerator card upgrade device provided in the embodiment of the present invention can implement each process implemented by the accelerator card upgrade method in the above method embodiment and can achieve the same beneficial effect. To avoid repetition, it will not be described here.

Referring to FIG. 4 , FIG. 4 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. As shown in FIG. 4 , the electronic device includes: a memory 302, a processor 301, and a computer program of an accelerator card upgrade method stored in the memory 302 and executable on the processor 301, wherein:

The accelerator card includes a normal version image and a recovery version image, the normal version image and the recovery version image are isolated from each other, the recovery version image includes a firmware partition providing re-upgrade capability, the processor 301 is used to call the computer program stored in the memory 302, and execute the following steps:

Obtaining the upgrade status of the normal version image, wherein the upgrade status includes upgrade failure or upgrade success;

If the upgrade status is upgrade failure, enter the recovery version image and perform the next online automatic upgrade through the firmware partition.

Optionally, before obtaining the upgrade status of online automatic upgrade of the normal version image, the method executed by the processor 301 further includes:

Receive the normal version upgrade package sent by the master end;

The normal version image is automatically upgraded online through the normal version upgrade package.

Optionally, before the normal version image is upgraded or after the normal version image is successfully upgraded, the method executed by the processor 301 further includes:

Run the normal version image or the normal version image after successful upgrade, and upgrade the recovery version image.

Optionally, the obtaining of the upgrade status of the normal version image performed by the processor 301 includes:

Obtaining the startup status of the upgraded normal version image through the program state machine;

According to the startup status, the upgrade status of the normal version image is determined.

Optionally, the determining, according to the startup state, the upgrade state of the normal version image performed by the processor 301 includes:

When the normal version image after the upgrade is started normally, if the number of normal startup failures is greater than a preset value, determining that the upgrade status of the normal version image is upgrade failure;

When the normal version image after the upgrade is started for the upgrade, if it is not the first startup after the upgrade, it is determined that the upgrade status of the normal version image is upgrade failure.

Optionally, the communication link between the master end and the acceleration card includes a first communication link and a second communication link, the first communication link and the second communication link include different communication entities, and before the next online automatic upgrade is performed by restoring the version image, the method executed by the processor 301 further includes:

If the first communication link is normal, controlling the acceleration card to enter the recovery version image through the first communication link;

If the first communication link is abnormal, the acceleration card is controlled to enter the recovery version image through the second communication link.

The electronic device provided in the embodiment of the present invention can implement the various processes implemented by the acceleration card upgrade method in the above method embodiment, and can achieve the same beneficial effects. To avoid repetition, it will not be repeated here. The computer-readable storage medium provided in the embodiment of the present invention stores a computer program on the computer-readable storage medium. When the computer program is executed by the processor, it implements the various processes of the acceleration card upgrade method or the application-side acceleration card upgrade method provided in the embodiment of the present invention, and can achieve the same technical effects. To avoid repetition, it will not be repeated here.

Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the concept of the present invention, or directly/indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims (10)

1. An accelerator card upgrade method, wherein the accelerator card includes a normal version image and a recovered version image, the normal version image and the recovered version image being isolated from each other, the recovered version image including a firmware partition providing re-upgrade capability, the method comprising:

Acquiring an upgrade state of a normal version image, wherein the upgrade state comprises upgrade failure or upgrade success;

if the upgrade status is upgrade failure, the method enters a recovered version image and carries out next online automatic upgrade through the firmware partition.

2. The accelerator card upgrade method according to claim 1, wherein before the acquiring the upgrade status for online automatic upgrade of the normal version image, the method further comprises:

Receiving a normal version upgrade package issued by a main terminal;

And carrying out online automatic upgrading on the normal version image through the normal version upgrading package.

3. The accelerator card upgrade method according to claim 2, wherein before the normal version image upgrade or after the normal version image upgrade is successful, the method further comprises:

and operating the normal version image or the normal version image after the successful upgrading is carried out, and upgrading the recovered version image.

4. The method for upgrading an accelerator card according to claim 2, wherein the obtaining the upgrade status of the normal version image includes:

acquiring the starting state of the updated normal version image through a program state machine;

And determining the upgrading state of the normal version image according to the starting state.

5. The method for upgrading an accelerator card according to claim 3, wherein determining an upgrade status of the normal version image according to the start-up status comprises:

when the updated normal version image is normally started, if the normal start failure times are larger than a preset value, determining that the upgrading state of the normal version image is upgrading failure;

when the updated normal version image is started after upgrading, if the updated normal version image is not started for the first time after upgrading, the upgrading state of the normal version image is determined to be failed in upgrading.

6. The method of claim 5, wherein the communication link between the host and the accelerator card comprises a first communication link and a second communication link, the first communication link and the second communication link comprising different communication agents, the method further comprising, prior to the next online automatic upgrade via a recovery version image:

if the first communication link is normal, controlling the accelerator card to enter the recovery version mirror image through the first communication link;

and if the first communication link is abnormal, controlling the accelerator card to enter the recovery version mirror image through the second communication link.

7. An accelerator card upgrade apparatus, wherein the accelerator card comprises a normal version image and a recovered version image, the normal version image and the recovered version image being isolated from each other, the recovered version image comprising a firmware partition providing re-upgrade capability, the apparatus comprising:

the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring the upgrading state of the normal version image, and the upgrading state comprises upgrading failure or upgrading success;

And the first upgrading module is used for entering a recovered version image and carrying out the next online automatic upgrading through the firmware partition if the upgrading state is that the upgrading fails.

8. An accelerator card upgrade system, the system comprising: the accelerator card comprises an accelerator card upgrading device, a normal version image and a recovery version image, wherein the accelerator card upgrading device is used for executing the accelerator card upgrading method according to any of claims 1 to 6.

9. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the accelerator card upgrade method according to any one of claims 1 to 6 when the computer program is executed by the processor.

10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the steps of the accelerator card upgrade method of any one of claims 1 to 6.