Virtual machine full life cycle instance static measurement method based on TPCM

Abstract

The invention relates to the technical field of virtualization and trusted computing, and discloses a TPCM-based virtual machine full life cycle instance static measurement method. A novel virtual machine full life cycle model is defined, and a request instance measurement module is added in each life stage processing function of a virtual machine based on the model and is used for realizing a measurement request initiation function; a strategy file located in a local protection component is constructed and used for specifying how to organize instance core data used for measurement, and a hardware trusted cryptography module located in the protection component is called to perform hash operation on the instance core data. According to the method, the following three problems existing in an existing virtual machine instance static measurement scheme are solved: (1) virtual machine instance static measurement operation cannot cover the full life cycle of a virtual machine; (2) the measurement overhead caused by virtual machine instance expansion is too large; and (3) the trusted measurement service provided by the TPM has the risk of being bypassed, so that the security of measurement operation is difficult to guarantee.

Description

Technical Field

[0001] The present invention relates to the technical field of virtualization and trusted computing, and in particular to a static measurement method for a virtual machine full life cycle instance based on a TPCM (Trusted Platform Control Module). Background Art

[0002] Trusted computing technology uses an unalterable physical chip as the hardware root of trust, and builds a trust chain from computer power-on to computer operation through step-by-step measurement and trust chain transmission. Taking Trusted Computing 2.0 as an example, it uses the Trusted Platform Module (TPM) as the physical root of trust, and builds a trust chain from the bottom up to ultimately ensure the trustworthiness of the entire computing system. However, in its design concept, TPM itself is a mounted component of a computing system. Its passive call method has the risk of being maliciously bypassed, and lacks the ability to actively control the computing platform. Therefore, my country has proposed the concept of Trusted Computing 3.0 - dividing the computing system into computing components and protection components, which are completely separated. TPCM, as the protection component, starts before the computing component, and performs security protection and control on the computing component by measuring and starting at the same time. The two communicate through the shared memory area located in the computing component - this solves the security risks of the above-mentioned TPM, and is more in line with my country's independent and controllable trust creation concept.

[0003] QCOW2 (Qemu Copy On Write Version 2) virtual disk format is a common virtual machine instance format used by QEMU (Quick Emulator) / KVM (Kernel-based Virtual Machine). Its internal data is composed of fixed-size data clusters, which can be divided into: QCOW2 file header cluster, partition table cluster, and data cluster.

[0004] The cloud management center is the control center of the entire cloud platform. One of its important tasks is to issue policy files - policy files are issued to the protection components in the computing system, and the protection components perform refined management actions on the computing components according to the specific provisions of the policy files.

[0005] The virtual machine lifecycle model defines the various stages that a virtual machine may currently be in. The existing definition of the lifecycle model generally includes the startup, running, and shutdown states of the virtual machine, and does not include the creation, suspension, and suspension recovery states of the virtual machine. In particular, the above virtual machine states can be divided into two categories: dynamic and static. Dynamic refers to the state in which the virtual machine occupies the host machine's CPU, memory and other resources to perform operations such as disk reading and writing, mathematical calculations, etc.; in contrast, static refers to the state in which the virtual machine gives up the host machine's CPU, memory and other resources and stops all operations. Therefore, the state change of the virtual machine can be divided into dynamic to static and static to dynamic.

[0006] The static measurement of virtual machine instances based on the switching of virtual machine life states is defined as: when the virtual machine life state switches from static to dynamic, the virtual machine instance file is measured using a hash function, and the obtained measurement value is compared and verified with the baseline value stored in the security area to confirm whether the file has been damaged; in particular, when the virtual machine life state changes from dynamic to static, the newly generated measurement value does not perform the baseline value comparison operation, but updates the baseline value, that is, replaces the original baseline value with the new measurement value. The initial baseline value of the virtual machine instance is generally provided by the cloud service provider.

[0007] The existing public static measurement protection solutions for virtual machine instance files mainly have the following problems:

[0008] (1) The current virtual machine instance protection scheme only performs static measurement operations on the virtual machine when it is in the startup and shutdown states, ignoring the life states of the virtual machine such as creation, operation, suspension, recovery, and destruction, and fails to achieve virtual machine instance measurement and protection based on the entire life cycle.

[0009] (2) The measurement operations adopted by the existing static measurement schemes for virtual machine instance files are mostly implemented based on the cryptographic module of TPM2.0 or by calling the hash function provided by the host operating system. There is a problem that the hash operation process is easily hijacked.

[0010] (3) The existing static measurement scheme for virtual machine instances adopts an overall measurement method for virtual machine instance files. This method does not take into account the actual situation that the instance file size quickly expands due to the long-term operation of the virtual machine, and the resulting problems include low measurement efficiency and excessive system resources consumed by the measurement operation. Summary of the invention

[0011] The purpose of the present invention is to provide a static measurement method for virtual machine instances throughout their life cycle based on TPCM, which solves the problems that static measurement of virtual machine instance files fails to achieve full life cycle coverage of virtual machines, measurement operations are susceptible to hijacking, and hash operations on the entire file cause low efficiency and system performance loss. The technical solution is as follows:

[0012] A static measurement method for a virtual machine full life cycle instance based on TPCM includes the following steps:

[0013] Step 1: In the processing functions of each stage of the virtual machine life cycle, add a function module for initiating instance measurement; the function module is used to implement the function of requesting measurement of virtual machine instances throughout the entire life cycle, including creation, startup, operation, suspension, suspension recovery and shutdown of the virtual machine;

[0014] Step 2: Generate and maintain a policy file in the local protection component to specify data reading and organization rules for virtual machine instance files;

[0015] Step 3: According to the provisions of the policy file in the local protection component, organize the core data of the instance to be measured, place it in the shared memory, and call the SM3 algorithm of the hardware trusted cryptographic module in the protection component to measure the data in the shared memory area and generate a measurement value;

[0016] Step 4: The generated metric values ​​are compared with the baseline or updated according to the dynamic and static conversion conditions of the life cycle of the virtual machine, and the current virtual machine is controlled according to the comparison results.

[0017] Furthermore, the step 1 is specifically as follows:

[0018] Step 1.1: Before executing the function logic of virtual machine creation, add a function module for requesting virtual machine instance metrics to initiate a metric request for the instance before the virtual machine is created;

[0019] Step 1.2: Before executing the function logic of virtual machine startup, add a function module for requesting virtual machine instance metrics, which is used to initiate a metric request for the instance before the virtual machine is started;

[0020] Step 1.3: Before executing the function logic of suspending and resuming the virtual machine, add a function module for requesting the virtual machine instance measurement, which is used to initiate a measurement request for the instance before the virtual machine is suspended and resumed;

[0021] Step 1.4: After executing the virtual machine suspension function logic, add a function module for requesting virtual machine instance metrics, which is used to initiate a metric request for the instance after the virtual machine is suspended;

[0022] Step 1.5: After executing the virtual machine shutdown function logic, add a function module for requesting virtual machine instance metrics, which is used to initiate a metric request for the instance after the virtual machine completes the shutdown operation.

[0023] Furthermore, in step 2, the policy file includes:

[0024] Cluster_num, a parameter issued by the cloud management center to specify how many groups the data cluster part in the virtual machine instance file can be divided into;

[0025] A random number N generated by a trusted cryptographic module TCM in the local protection component, which is used to specify which cluster is selected from each group to form the measurement data;

[0026] An algorithm for random cluster selection based on the parameters Cluster_num and the random number N.

[0027] Furthermore, the step 2 is specifically as follows:

[0028] Step 2.1: The cloud management center sends a policy file to the protection component of the computing system. The policy file carries the parameter Cluster_num that controls the parsing of the virtual machine instance in the QCOW2 format.

[0029] Step 2.2: The hardware trusted cryptographic module TCM located in the protection component generates a random number N, and extends the random number N into the policy file in step 2.1;

[0030] Step 2.3: Determine an algorithm for reading random N clusters and using them as instance core data for measurement, where the input parameters are the parameter Cluster_num and the random number N generated in steps 2.1 and 2.2, respectively, and the algorithm output is the starting address of the cluster to be selected in the instance file; this algorithm will also be extended to the strategy file.

[0031] Furthermore, the specific steps of the algorithm in step 2.3 are divided into:

[0032] Step 2.3.1: accept the parameters Cluste_num and the random number N generated in steps 2.1 and 2.2 and the address of the instance file in memory as three input parameters;

[0033] Step 2.3.2: Divide the data cluster part of the instance file into Cluster_num groups;

[0034] Step 2.3.3: Return the address of the Nth cluster in the Cluster_num group in memory.

[0035] Furthermore, the step 3 is specifically as follows:

[0036] Step 3.1: Add an instance parsing module in the computing component to parse the file header cluster, partition table cluster, and data cluster of the virtual machine instance in QCOW2 format; and divide the data cluster according to the parameter Cluster_num specified in the policy file;

[0037] Step 3.2: Take out the Nth data cluster from the divided Cluster_num groups, and combine the data cluster with the file header cluster and partition table cluster of the virtual machine instance in QCOW2 format to form instance core data, and store them in the memory area;

[0038] Step 3.3: Organize the instance measurement command and notify the hardware cryptographic module TCM in the local protection component to perform SM3 hash operation on the instance core data located in the shared memory to generate the instance measurement value.

[0039] Furthermore, the step 4 is specifically as follows:

[0040] Step 4.1: For instance measurement requests involving the transition of a virtual machine from static to dynamic, the instance measurement value generated in step 3.3 is compared with the instance reference value stored in the protection component to determine whether the current instance has been tampered with. If not, the virtual machine continues to run, otherwise the virtual machine is controlled;

[0041] Step 4.2: For instance measurement requests involving the virtual machine changing from dynamic to static, a baseline value update operation is performed on the instance measurement value generated in step 3.3, and the original instance measurement value in the baseline library is replaced with the new instance measurement value generated this time, and the virtual machine continues to run.

[0042] The beneficial effects of the present invention are:

[0043] 1. The present invention implements an instance measurement function covering the entire life cycle of a virtual machine by inserting code in each life cycle processing function of the virtual machine;

[0044] 2. The present invention generates and maintains a policy file in a local protection component to define the organization process of the core data of a virtual machine instance, thereby significantly reducing the volume of data to be measured and effectively improving the measurement efficiency;

[0045] 3. The present invention performs measurement operations on the core data of the virtual machine instance by calling the SM3 hash algorithm provided by the hardware cryptographic module located in the protection component, thereby ensuring the security and reliability of the measurement operation itself. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of virtual machine instance measurement in the present invention.

[0047] Figure 2 It is a schematic diagram of the entire life cycle state of the virtual machine in the present invention.

[0048] Figure 3 This is a schematic diagram of constructing a strategy file in the present invention.

[0049] Figure 4 This is a schematic diagram of the core data organization of the virtual machine instance in the present invention. DETAILED DESCRIPTION

[0050] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementations.

[0051] The static measurement method of virtual machine full life cycle instance based on TPCM proposed in the present invention is a static measurement scheme of virtual machine instance based on the new virtual machine full life cycle model, integrating trusted computing 3.0 technology and combining QCOW2 virtual disk file format. Firstly, the function of initiating virtual machine instance measurement request is realized by inserting code in each stage processing function of the new virtual machine life cycle model defined in this article; secondly, a policy file located in the local protection component is generated and maintained to specify the data reading and organization rules of the virtual machine instance file; then, according to the provisions of the policy file in the local protection component, the instance core data to be measured is organized and placed in the shared memory; then the SM3 algorithm of the hardware trusted cryptographic module (TCM) in the protection component is called to measure the data located in the shared memory area and generate a measurement value; finally, the newly generated instance measurement value is compared with the reference value stored in the protection component, and the current virtual machine is controlled according to the comparison result.

[0052] like Figure 1 As shown, the example measurement process diagram of the present invention, its basic process is as follows:

[0053] (1) The virtual machine lifecycle state processing function initiates an instance measurement request;

[0054] (2) The instance parsing module parses the virtual machine instance to obtain the header partition cluster, partition table cluster, and data cluster;

[0055] (3) Organize the instance core data according to the policy file stored in the local protection component and place it in the shared memory area;

[0056] (4) Calling the hardware cryptographic module TCM located in the local protection component to perform measurement operations on the instance core data;

[0057] (5) Determine whether the instance measurement is to be compared with the benchmark value. If so, proceed to step (6) and step (7); otherwise, proceed to step (8);

[0058] (6) Perform baseline verification on the obtained measurement values ​​to determine whether the instance file has been tampered with;

[0059] (7) Interfere with the currently running virtual machine based on the verification results. If it is proven that the instance file has not been tampered with, the virtual machine continues to run; otherwise, the virtual machine is controlled accordingly.

[0060] (8) The baseline value is updated and the virtual machine continues to run.

[0061] The specific implementation steps are as follows:

[0062] Step 1: Add a function module for requesting instance metrics in the function logic of each life stage of the virtual machine.

[0063] After analyzing the source code of the virtualization software QEMU, it was found that it is only necessary to initiate a measurement request for the virtual machine instance before the virtual machine executes the function logic of creation, startup, suspension and recovery, and after the virtual machine shutdown and suspend function logic, so as to achieve the instance protection effect for the entire life cycle of the virtual machine, including virtual machine creation, startup, suspension, suspension recovery, and shutdown.

[0064] Step 1.1: Before executing the function logic for virtual machine creation, add a function module for requesting virtual machine instance metrics to initiate a metric request for the instance before the virtual machine is created.

[0065] Step 1.2: Before executing the function logic of virtual machine startup, add a function module for requesting virtual machine instance metrics to initiate a metric request for the instance before the virtual machine is started.

[0066] Step 1.3: Before executing the function logic of suspending and resuming the virtual machine, add a function module for requesting virtual machine instance metrics to initiate a metric request for the instance before the virtual machine is suspended and resumed.

[0067] Step 1.4: After executing the virtual machine suspension function logic, add a function module for requesting virtual machine instance metrics, which is used to initiate a metric request for the instance after the virtual machine is suspended.

[0068] Step 1.5: After executing the virtual machine shutdown function logic, add a function module for requesting virtual machine instance metrics, which is used to initiate a metric request for the instance after the virtual machine completes the shutdown operation.

[0069] like Figure 2 As shown in the figure, the virtual machine full life cycle state diagram defined in the present invention has the following basic principles:

[0070] The entire life cycle of a virtual machine includes six states: creation, startup, operation, suspension, suspension recovery, and shutdown. In addition to the creation and shutdown states, the other states have a linear relationship - the transition between states must be carried out in order.

[0071] Step 2: Build a policy file that is stored in the protection component and is used to control the reading of data from virtual machine instances.

[0072] The policy file mainly consists of three parts: the first is the parameter Cluster_num issued by the cloud management center, which is used to specify how many groups the virtual machine instance file QCOW2 virtual disk file can be divided into; the second is the random number N randomly generated by the trusted cryptographic module TCM located in the local protection component, which is used to specify which cluster in each group is used to measure the core data; the third is the algorithm for randomly selecting clusters based on the parameter Cluster_num and the random number N.

[0073] Step 2.1: The cloud management center sends a policy file to the protection component of the computing system. The policy file carries the parameter Cluster_num for controlling the resolution of virtual machine instances (QCOW2 format) - it specifies the number of groups that the virtual machine instance should be resolved into.

[0074] Step 2.2: The hardware trusted cryptographic module TCM located in the protection component generates a random number N, which is used to control the offset of the cluster that needs to be read for instance measurement, and extends the random number into the policy file in step 2.1.

[0075] Step 2.3: Determine the algorithm for reading random N clusters as instance core data for measurement, where the input parameters are Cluster_num and N generated in steps 2.1 and 2.2 respectively, and the algorithm output is the starting address of the cluster to be selected in the instance file. This algorithm will also be extended to the strategy file in step 2.1.

[0076] The specific steps of the algorithm are as follows:

[0077] Step 2.3.1: accept the parameters Cluste_num and the random number N generated in steps 2.1 and 2.2 and the address of the instance file in memory as three input parameters;

[0078] Step 2.3.2: Divide the data cluster part of the instance file into Cluster_num groups;

[0079] Step 2.3.3: Return the address of the Nth cluster in the Cluster_num group in memory.

[0080] Step 3: Organize the core data of the VM instance for measurement according to the policy file generated in Step 2.

[0081] The virtual machine instance file in QCOW2 format is divided into data cluster groups according to the provisions of the policy file in step 2, and then a random number of clusters are selected from each group to form the instance core data, and the clusters are placed in the shared memory area that communicates with the local protection component; the hardware trusted cryptographic module in the protection component reads the instance core data located in the shared memory through direct memory access technology, and calls the SM3 hash algorithm to perform hash operations on it to generate measurement values.

[0082] like Figure 3 As shown, the schematic diagram of constructing the strategy file in the present invention, the basic principle is as follows:

[0083] The policy file stored in the local protection component is used to indicate how to organize the instance core data in order to solve the problem of excessive overhead caused by measuring the entire instance file. The policy file consists of three parts: (1) the parameter Cluster_num issued by the cloud management center to set the number of groups into which the data clusters in the original instance are divided; (2) the random number N generated by the hardware TCM, which is used to randomly select the Nth data cluster from the divided Cluster_num groups; (3) the algorithm for random cluster selection based on parameters (1) and (2), and the algorithm output is the starting address of the selected cluster in the memory address space. The specific steps are as follows:

[0084] Step 3.1: Add an instance parsing module in the computing component. Its main functions are: (1) parsing the file header cluster and partition table cluster of the QCOW2 instance; (2) partitioning the remaining data clusters according to the Cluster_num parameter specified in the strategy file in step 2.1.

[0085] Step 3.2: From the divided Cluster_num groups, according to the provisions of the policy file in step 2.2, take out the Nth data cluster, combine it with the file header cluster and partition table cluster of the QCOW2 instance to form the "instance core data", and store it in the memory area.

[0086] Step 3.3: Organize the instance measurement command and notify the hardware cryptographic module TCM in the local protection component to perform SM3 hash operation on the "instance core data" located in the shared memory to generate a measurement value.

[0087] Step 4: Perform a benchmark comparison operation or a metric update operation on the generated metric values ​​according to the dynamic and static conditions of the virtual machine's life cycle. The benchmark comparison is used to verify whether the current virtual machine instance has been tampered with; the metric update operation is to replace the benchmark value in the benchmark library with the newly generated metric value.

[0088] Step 4.1: The virtual machine instance measurement request initiated by the above steps 1.1, 1.2, and 1.3 involves the above-mentioned situation where the virtual machine changes from static to dynamic. Therefore, the measurement value of the instance generated in the above step 3.3 will be compared with the instance baseline value stored in the protection component to determine whether the current virtual machine instance has been tampered with. If it has not been tampered with, the virtual machine continues to run, otherwise the virtual machine is controlled.

[0089] Step 4.2: The instance measurement request initiated by the above steps 1.4 and 1.5 involves the situation mentioned above where the virtual machine is converted from dynamic to static. Therefore, the instance measurement value generated in the above step 3.3 will perform the baseline value update operation - the new instance measurement value generated this time will replace the original instance measurement value in the baseline library, and the virtual machine continues to execute.

[0090] like Figure 4 This is a schematic diagram of the core data organization of the example in the present invention, and its basic principles are as follows:

[0091] (1) First, identify the header partition cluster, partition table cluster, and data cluster of the instance file in the QCOW2 virtual disk format; (2) divide and select the data cluster according to the policy file; (3) organize the header partition cluster, partition table cluster, and selected cluster into "instance core data".

Claims

1. A static measurement method for virtual machine full life cycle instances based on TPCM, characterized in that: The following steps are involved: Step 1: In the processing functions of each stage of the virtual machine life cycle, add a function module for initiating instance measurement; the function module is used to implement the function of requesting measurement of virtual machine instances throughout the entire life cycle, including creation, startup, operation, suspension, suspension recovery and shutdown of the virtual machine; Step 2: Generate and maintain a policy file in the local protection component to specify data reading and organization rules for virtual machine instance files; Step 3: According to the provisions of the policy file in the local protection component, organize the core data of the instance to be measured, place it in the shared memory, and call the SM3 algorithm of the hardware trusted cryptographic module in the protection component to measure the data in the shared memory area and generate a measurement value; Step 4: The generated metric values ​​are compared with the baseline values ​​or updated according to the dynamic and static conversion conditions of the life cycle of the virtual machine, and the current virtual machine is controlled according to the comparison results.

2. The static measurement method for virtual machine full life cycle instance based on TPCM according to claim 1 is characterized in that: The step 1 is specifically as follows: Step 1.1: Before executing the function logic of virtual machine creation, add a function module for requesting virtual machine instance metrics to initiate a metric request for the instance before the virtual machine is created; Step 1.2: Before executing the function logic of virtual machine startup, add a function module for requesting virtual machine instance metrics, which is used to initiate a metric request for the instance before the virtual machine is started; Step 1.3: Before executing the function logic of suspending and resuming the virtual machine, add a function module for requesting the virtual machine instance measurement, which is used to initiate a measurement request for the instance before the virtual machine is suspended and resumed; Step 1.4: After executing the virtual machine suspension function logic, add a function module for requesting virtual machine instance metrics, which is used to initiate a metric request for the instance after the virtual machine is suspended; Step 1.5: After executing the virtual machine shutdown function logic, add a function module for requesting virtual machine instance metrics, which is used to initiate a metric request for the instance after the virtual machine completes the shutdown operation.

3. The static measurement method for virtual machine full life cycle instance based on TPCM according to claim 1 is characterized in that: In step 2, the policy file includes: Cluster_num, a parameter issued by the cloud management center to specify how many groups the data cluster part in the virtual machine instance file can be divided into; A random number N generated by a trusted cryptographic module TCM in the local protection component, which is used to specify which cluster is selected from each group to form the measurement data; An algorithm for random cluster selection based on the parameters Cluster_num and the random number N.

4. The static measurement method for virtual machine full life cycle instance based on TPCM according to claim 2 is characterized in that: The step 2 is specifically as follows: Step 2.1: The cloud management center sends a policy file to the protection component of the computing system. The policy file carries the parameter Cluster_num that controls the parsing of the virtual machine instance in the QCOW2 format. Step 2.2: The hardware trusted cryptographic module TCM located in the protection component generates a random number N, and extends the random number N into the policy file in step 2.1; Step 2.3: Determine an algorithm for reading random N clusters and using them as instance core data for measurement, where the input parameters are the parameter Cluster_num and the random number N generated in steps 2.1 and 2.2, respectively, and the algorithm output is the starting address of the cluster to be selected in the instance file; this algorithm will also be extended to the strategy file.

5. The TPCM-based static measurement method for virtual machine full life cycle instances according to claim 4 is characterized in that: The specific steps of the algorithm in step 2.3 are as follows: Step 2.3.1: accept the parameters Cluste_num and the random number N generated in steps 2.1 and 2.2 and the address of the instance file in memory as three input parameters; Step 2.3.2: Divide the data cluster part of the instance file into Cluster_num groups; Step 2.3.3: Return the address of the Nth cluster in the Cluster_num group in memory.

6. The TPCM-based static measurement method for virtual machine full life cycle instances according to claim 4 is characterized in that: The step 3 is specifically as follows: Step 3.1: Add an instance parsing module in the computing component to parse the file header cluster, partition table cluster, and data cluster of the virtual machine instance in QCOW2 format; and divide the data cluster according to the parameter Cluster_num specified in the policy file; Step 3.2: Take out the Nth data cluster from the divided Cluster_num groups, and combine the data cluster with the file header cluster and partition table cluster of the virtual machine instance in QCOW2 format to form instance core data, and store them in the memory area; Step 3.3: Organize the instance measurement command and notify the hardware cryptographic module TCM in the local protection component to perform SM3 hash operation on the instance core data located in the shared memory to generate the instance measurement value.

7. The TPCM-based static measurement method for virtual machine full life cycle instances according to claim 6 is characterized in that: The step 4 is specifically as follows: Step 4.1: For instance measurement requests involving the transition of a virtual machine from static to dynamic, the instance measurement value generated in step 3.3 is compared with the instance reference value stored in the protection component to determine whether the current instance has been tampered with. If not, the virtual machine continues to run, otherwise the virtual machine is controlled; Step 4.2: For instance measurement requests involving the virtual machine changing from dynamic to static, a baseline value update operation is performed on the instance measurement value generated in step 3.3, and the original instance measurement value in the baseline library is replaced with the new instance measurement value generated this time, and the virtual machine continues to run.

Images