KR20240044287A - Electronic device and method for protecting data using the same - Google Patents

Abstract

An electronic device according to various embodiments of the present disclosure includes a memory and a processor operatively connected to the memory, wherein when the memory is executed, the processor performs an operation according to application execution based on a model data file. When started, the model data file is loaded from the memory, a hash value for the model data file is calculated in the hypervisor space, stored together with model-related information about the model data file, and based on a timer created in the kernel space. Calculate the hash value of the loaded model data file in the hypervisor space, compare the calculated hash value with the stored hash value to confirm identity, and if the hash values are not the same according to the comparison, the model You can store instructions that determine whether a data file has been modified. Various other embodiments other than the various embodiments of the present disclosure are possible.

Description

Electronic devices and data protection methods using the same {ELECTRONIC DEVICE AND METHOD FOR PROTECTING DATA USING THE SAME}

Various embodiments of the present invention relate to electronic devices and data protection methods using the same.

Recently, as portable electronic devices such as smart phones and tablet PCs have become popular, the hardware and software of electronic devices have also developed dramatically, and the use environment of portable electronic devices is becoming similar to that of a PC. In addition, convenient functions desired by users are provided by downloading various applications from the Internet, app stores, etc.

These electronic devices can run various applications based on machine learning model data learned through deep learning or machine learning based on artificial neural networks.

However, as various applications are downloaded, malicious applications, including malware and spyware, are introduced into portable electronic devices, causing damage from cyber attacks such as network traffic, system performance degradation, file deletion, and personal information leakage.

In particular, cyber attacks on learned machine learning model data files can cause application execution errors, which can lead to fatal consequences such as safety threats.

An electronic device according to various embodiments of the present disclosure includes a memory and a processor operatively connected to the memory, and when the memory is executed, the processor starts an operation according to the execution of an application based on a model data file. Then, the model data file is loaded from the memory, a hash value for the model data file is calculated in the hypervisor space, stored together with model-related information for the model data file, and based on a timer generated in the kernel space. Calculate the hash value of the loaded model data file in the hypervisor space, compare the calculated hash value with the stored hash value to confirm identity, and if the hash values are not identical according to the comparison, the model data You can store instructions that determine whether a file has been modified.

A method of an electronic device according to various embodiments of the present disclosure includes the operation of loading the model data file from the memory when an operation according to execution of an application based on the model data file begins, and the operation of loading the model data file in the hypervisor space. An operation of calculating a hash value and storing it together with model-related information about the model data file, an operation of calculating a hash value of the loaded model data file in the hypervisor space based on a timer generated in the kernel space, the calculation It may include an operation of comparing the hash value with the stored hash value to confirm identity, and an operation of determining that the model data file has been modified if the hash values are not the same according to the comparison.

1 is a block diagram of an electronic device in a network environment, according to various embodiments.
FIG. 2 is a diagram illustrating the structure of an operating system running on an electronic device, according to various embodiments.
FIG. 3 is a block diagram illustrating functional modules of an operating system structure running on an electronic device, according to various embodiments.
FIG. 4 is a block diagram illustrating functional modules of an operating system structure running on an electronic device, according to various embodiments.
FIG. 5 is a diagram for explaining a method of protecting data in an electronic device, according to various embodiments.
FIG. 6 is a flowchart illustrating a method for protecting data in an electronic device, according to various embodiments.
FIG. 7 is a flowchart illustrating a method for protecting data in an electronic device, according to various embodiments.

1 is a block diagram of an electronic device 101 in a network environment 100, according to various embodiments. Referring to FIG. 1, in the network environment 100, the electronic device 101 communicates with the electronic device 102 through a first network 198 (e.g., a short-range wireless communication network) or a second network 199. It is possible to communicate with at least one of the electronic device 104 or the server 108 through (e.g., a long-distance wireless communication network). According to one embodiment, the electronic device 101 may communicate with the electronic device 104 through the server 108. According to one embodiment, the electronic device 101 includes a processor 120, a memory 130, an input module 150, an audio output module 155, a display module 160, an audio module 170, and a sensor module ( 176), interface 177, connection terminal 178, haptic module 179, camera module 180, power management module 188, battery 189, communication module 190, subscriber identification module 196 , or may include an antenna module 197. In some embodiments, at least one of these components (eg, the connection terminal 178) may be omitted or one or more other components may be added to the electronic device 101. In some embodiments, some of these components (e.g., sensor module 176, camera module 180, or antenna module 197) are integrated into one component (e.g., display module 160). It can be.

The processor 120, for example, executes software (e.g., program 140) to operate at least one other component (e.g., hardware or software component) of the electronic device 101 connected to the processor 120. It can be controlled and various data processing or calculations can be performed. According to one embodiment, as at least part of data processing or computation, the processor 120 stores commands or data received from another component (e.g., sensor module 176 or communication module 190) in volatile memory 132. The commands or data stored in the volatile memory 132 can be processed, and the resulting data can be stored in the non-volatile memory 134. According to one embodiment, the processor 120 includes a main processor 121 (e.g., a central processing unit or an application processor) or an auxiliary processor 123 that can operate independently or together (e.g., a graphics processing unit, a neural network processing unit ( It may include a neural processing unit (NPU), an image signal processor, a sensor hub processor, or a communication processor). For example, if the electronic device 101 includes a main processor 121 and a secondary processor 123, the secondary processor 123 may be set to use lower power than the main processor 121 or be specialized for a designated function. You can. The auxiliary processor 123 may be implemented separately from the main processor 121 or as part of it.

The auxiliary processor 123 may, for example, act on behalf of the main processor 121 while the main processor 121 is in an inactive (e.g., sleep) state, or while the main processor 121 is in an active (e.g., application execution) state. ), together with the main processor 121, at least one of the components of the electronic device 101 (e.g., the display module 160, the sensor module 176, or the communication module 190) At least some of the functions or states related to can be controlled. According to one embodiment, co-processor 123 (e.g., image signal processor or communication processor) may be implemented as part of another functionally related component (e.g., camera module 180 or communication module 190). there is. According to one embodiment, the auxiliary processor 123 (eg, neural network processing unit) may include a hardware structure specialized for processing artificial intelligence models. Artificial intelligence models can be created through machine learning. For example, such learning may be performed in the electronic device 101 itself on which the artificial intelligence model is performed, or may be performed through a separate server (e.g., server 108). Learning algorithms may include, for example, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but It is not limited. An artificial intelligence model may include multiple artificial neural network layers. Artificial neural networks include deep neural network (DNN), convolutional neural network (CNN), recurrent neural network (RNN), restricted boltzmann machine (RBM), belief deep network (DBN), bidirectional recurrent deep neural network (BRDNN), It may be one of deep Q-networks or a combination of two or more of the above, but is not limited to the examples described above. In addition to hardware structures, artificial intelligence models may additionally or alternatively include software structures.

The memory 130 may store various data used by at least one component (eg, the processor 120 or the sensor module 176) of the electronic device 101. Data may include, for example, input data or output data for software (e.g., program 140) and instructions related thereto. Memory 130 may include volatile memory 132 or non-volatile memory 134.

The program 140 may be stored as software in the memory 130 and may include, for example, an operating system 142, middleware 144, or application 146.

The input module 150 may receive commands or data to be used in a component of the electronic device 101 (e.g., the processor 120) from outside the electronic device 101 (e.g., a user). The input module 150 may include, for example, a microphone, mouse, keyboard, keys (eg, buttons), or digital pen (eg, stylus pen).

The sound output module 155 may output sound signals to the outside of the electronic device 101. The sound output module 155 may include, for example, a speaker or a receiver. Speakers can be used for general purposes such as multimedia playback or recording playback. The receiver can be used to receive incoming calls. According to one embodiment, the receiver may be implemented separately from the speaker or as part of it.

The display module 160 can visually provide information to the outside of the electronic device 101 (eg, a user). The display module 160 may include, for example, a display, a hologram device, or a projector, and a control circuit for controlling the device. According to one embodiment, the display module 160 may include a touch sensor configured to detect a touch, or a pressure sensor configured to measure the intensity of force generated by the touch.

The audio module 170 can convert sound into an electrical signal or, conversely, convert an electrical signal into sound. According to one embodiment, the audio module 170 acquires sound through the input module 150, the sound output module 155, or an external electronic device (e.g., directly or wirelessly connected to the electronic device 101). Sound may be output through the electronic device 102 (e.g., speaker or headphone).

The sensor module 176 detects the operating state (e.g., power or temperature) of the electronic device 101 or the external environmental state (e.g., user state) and generates an electrical signal or data value corresponding to the detected state. can do. According to one embodiment, the sensor module 176 includes, for example, a gesture sensor, a gyro sensor, an air pressure sensor, a magnetic sensor, an acceleration sensor, a grip sensor, a proximity sensor, a color sensor, an IR (infrared) sensor, a biometric sensor, It may include a temperature sensor, humidity sensor, or light sensor.

The interface 177 may support one or more designated protocols that can be used to connect the electronic device 101 directly or wirelessly with an external electronic device (eg, the electronic device 102). According to one embodiment, the interface 177 may include, for example, a high definition multimedia interface (HDMI), a universal serial bus (USB) interface, an SD card interface, or an audio interface.

The connection terminal 178 may include a connector through which the electronic device 101 can be physically connected to an external electronic device (eg, the electronic device 102). According to one embodiment, the connection terminal 178 may include, for example, an HDMI connector, a USB connector, an SD card connector, or an audio connector (eg, a headphone connector).

The haptic module 179 can convert electrical signals into mechanical stimulation (e.g., vibration or movement) or electrical stimulation that the user can perceive through tactile or kinesthetic senses. According to one embodiment, the haptic module 179 may include, for example, a motor, a piezoelectric element, or an electrical stimulation device.

The camera module 180 can capture still images and moving images. According to one embodiment, the camera module 180 may include one or more lenses, image sensors, image signal processors, or flashes.

The power management module 188 can manage power supplied to the electronic device 101. According to one embodiment, the power management module 188 may be implemented as at least a part of, for example, a power management integrated circuit (PMIC).

Battery 189 may supply power to at least one component of electronic device 101. According to one embodiment, the battery 189 may include, for example, a non-rechargeable primary battery, a rechargeable secondary battery, or a fuel cell.

Communication module 190 is configured to provide a direct (e.g., wired) communication channel or wireless communication channel between electronic device 101 and an external electronic device (e.g., electronic device 102, electronic device 104, or server 108). It can support establishment and communication through established communication channels. Communication module 190 operates independently of processor 120 (e.g., an application processor) and may include one or more communication processors that support direct (e.g., wired) communication or wireless communication. According to one embodiment, the communication module 190 is a wireless communication module 192 (e.g., a cellular communication module, a short-range wireless communication module, or a global navigation satellite system (GNSS) communication module) or a wired communication module 194 (e.g., : LAN (local area network) communication module, or power line communication module) may be included. Among these communication modules, the corresponding communication module is a first network 198 (e.g., a short-range communication network such as Bluetooth, wireless fidelity (WiFi) direct, or infrared data association (IrDA)) or a second network 199 (e.g., legacy It may communicate with an external electronic device 104 through a telecommunication network such as a cellular network, a 5G network, a next-generation communication network, the Internet, or a computer network (e.g., LAN or WAN). These various types of communication modules may be integrated into one component (e.g., a single chip) or may be implemented as a plurality of separate components (e.g., multiple chips). The wireless communication module 192 uses subscriber information (e.g., International Mobile Subscriber Identifier (IMSI)) stored in the subscriber identification module 196 within a communication network such as the first network 198 or the second network 199. The electronic device 101 can be confirmed or authenticated.

The wireless communication module 192 may support 5G networks after 4G networks and next-generation communication technologies, for example, NR access technology (new radio access technology). NR access technology provides high-speed transmission of high-capacity data (eMBB (enhanced mobile broadband)), minimization of terminal power and access to multiple terminals (mMTC (massive machine type communications)), or high reliability and low latency (URLLC (ultra-reliable and low latency). -latency communications)) can be supported. The wireless communication module 192 may support high frequency bands (eg, mmWave bands), for example, to achieve high data rates. The wireless communication module 192 uses various technologies to secure performance in high frequency bands, for example, beamforming, massive array multiple-input and multiple-output (MIMO), and full-dimensional multiplexing. It can support technologies such as input/output (FD-MIMO: full dimensional MIMO), array antenna, analog beam-forming, or large scale antenna. The wireless communication module 192 may support various requirements specified in the electronic device 101, an external electronic device (e.g., electronic device 104), or a network system (e.g., second network 199). According to one embodiment, the wireless communication module 192 supports Peak data rate (e.g., 20 Gbps or more) for realizing eMBB, loss coverage (e.g., 164 dB or less) for realizing mmTC, or U-plane latency (e.g., 164 dB or less) for realizing URLLC. Example: Downlink (DL) and uplink (UL) each of 0.5 ms or less, or round trip 1 ms or less) can be supported.

The antenna module 197 may transmit or receive signals or power to or from the outside (eg, an external electronic device). According to one embodiment, the antenna module 197 may include an antenna including a radiator made of a conductor or a conductive pattern formed on a substrate (eg, PCB). According to one embodiment, the antenna module 197 may include a plurality of antennas (eg, an array antenna). In this case, at least one antenna suitable for a communication method used in a communication network such as the first network 198 or the second network 199 is connected to the plurality of antennas by, for example, the communication module 190. can be selected Signals or power may be transmitted or received between the communication module 190 and an external electronic device through the at least one selected antenna. According to some embodiments, in addition to the radiator, other components (eg, radio frequency integrated circuit (RFIC)) may be additionally formed as part of the antenna module 197.

According to various embodiments, the antenna module 197 may form a mmWave antenna module. According to one embodiment, a mmWave antenna module includes: a printed circuit board, an RFIC disposed on or adjacent to a first side (e.g., bottom side) of the printed circuit board and capable of supporting a designated high frequency band (e.g., mmWave band); And a plurality of antennas (e.g., array antennas) disposed on or adjacent to the second side (e.g., top or side) of the printed circuit board and capable of transmitting or receiving signals in the designated high frequency band. can do.

At least some of the components are connected to each other through a communication method between peripheral devices (e.g., bus, general purpose input and output (GPIO), serial peripheral interface (SPI), or mobile industry processor interface (MIPI)) and signal ( (e.g. commands or data) can be exchanged with each other.

According to one embodiment, commands or data may be transmitted or received between the electronic device 101 and the external electronic device 104 through the server 108 connected to the second network 199. Each of the external electronic devices 102 or 104 may be of the same or different type as the electronic device 101. According to one embodiment, all or part of the operations performed in the electronic device 101 may be executed in one or more of the external electronic devices 102, 104, or 108. For example, when the electronic device 101 needs to perform a certain function or service automatically or in response to a request from a user or another device, the electronic device 101 may perform the function or service instead of executing the function or service on its own. Alternatively, or additionally, one or more external electronic devices may be requested to perform at least part of the function or service. One or more external electronic devices that have received the request may execute at least part of the requested function or service, or an additional function or service related to the request, and transmit the result of the execution to the electronic device 101. The electronic device 101 may process the result as is or additionally and provide it as at least part of a response to the request. For this purpose, for example, cloud computing, distributed computing, mobile edge computing (MEC), or client-server computing technology can be used. The electronic device 101 may provide an ultra-low latency service using, for example, distributed computing or mobile edge computing. In another embodiment, the external electronic device 104 may include an Internet of Things (IoT) device. Server 108 may be an intelligent server using machine learning and/or neural networks. According to one embodiment, the external electronic device 104 or server 108 may be included in the second network 199. The electronic device 101 may be applied to intelligent services (e.g., smart home, smart city, smart car, or healthcare) based on 5G communication technology and IoT-related technology. .

Electronic devices according to various embodiments disclosed in this document may be of various types. Electronic devices may include, for example, portable communication devices (e.g., smartphones), computer devices, portable multimedia devices, portable medical devices, cameras, wearable devices, or home appliances. Electronic devices according to embodiments of this document are not limited to the above-described devices.

The various embodiments of this document and the terms used herein are not intended to limit the technical features described in this document to specific embodiments, and should be understood to include various changes, equivalents, or replacements of the embodiments. In connection with the description of the drawings, similar reference numbers may be used for similar or related components. The singular form of a noun corresponding to an item may include one or more of the above items, unless the relevant context clearly indicates otherwise. As used herein, “A or B”, “at least one of A and B”, “at least one of A or B”, “A, B or C”, “at least one of A, B and C”, and “A Each of phrases such as “at least one of , B, or C” may include any one of the items listed together in the corresponding phrase, or any possible combination thereof. Terms such as "first", "second", or "first" or "second" may be used simply to distinguish one component from another, and to refer to that component in other respects (e.g., importance or order) is not limited. One (e.g., first) component is said to be “coupled” or “connected” to another (e.g., second) component, with or without the terms “functionally” or “communicatively.” When mentioned, it means that any of the components can be connected to the other components directly (e.g. wired), wirelessly, or through a third component.

The term “module” used in various embodiments of this document may include a unit implemented in hardware, software, or firmware, and is interchangeable with terms such as logic, logic block, component, or circuit, for example. It can be used as A module may be an integrated part or a minimum unit of the parts or a part thereof that performs one or more functions. For example, according to one embodiment, the module may be implemented in the form of an application-specific integrated circuit (ASIC).

Various embodiments of the present document are one or more instructions stored in a storage medium (e.g., built-in memory 136 or external memory 138) that can be read by a machine (e.g., electronic device 101). It may be implemented as software (e.g., program 140) including these. For example, a processor (e.g., processor 120) of a device (e.g., electronic device 101) may call at least one command among one or more commands stored from a storage medium and execute it. This allows the device to be operated to perform at least one function according to the at least one instruction called. The one or more instructions may include code generated by a compiler or code that can be executed by an interpreter. A storage medium that can be read by a device may be provided in the form of a non-transitory storage medium. Here, 'non-transitory' only means that the storage medium is a tangible device and does not contain signals (e.g. electromagnetic waves), and this term refers to cases where data is semi-permanently stored in the storage medium. There is no distinction between temporary storage cases.

According to one embodiment, methods according to various embodiments disclosed in this document may be provided and included in a computer program product. Computer program products are commodities and can be traded between sellers and buyers. The computer program product may be distributed in the form of a machine-readable storage medium (e.g. compact disc read only memory (CD-ROM)) or through an application store (e.g. Play StoreTM) or on two user devices (e.g. It can be distributed (e.g. downloaded or uploaded) directly between smart phones) or online. In the case of online distribution, at least a portion of the computer program product may be at least temporarily stored or temporarily created in a machine-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.

According to various embodiments, each component (e.g., module or program) of the above-described components may include a single or plural entity, and some of the plurality of entities may be separately placed in other components. there is. According to various embodiments, one or more of the components or operations described above may be omitted, or one or more other components or operations may be added. Alternatively or additionally, multiple components (eg, modules or programs) may be integrated into a single component. In this case, the integrated component may perform one or more functions of each component of the plurality of components in the same or similar manner as those performed by the corresponding component of the plurality of components prior to the integration. . According to various embodiments, operations performed by a module, program, or other component may be executed sequentially, in parallel, iteratively, or heuristically, or one or more of the operations may be executed in a different order, or omitted. Alternatively, one or more other operations may be added.

FIG. 2 shows the structure of an operating system executed by the processor 120 (e.g., the processor 120 of FIG. 1) of an electronic device (e.g., the electronic device 101 of FIG. 1) according to various embodiments. It can be expressed.

According to one embodiment, the processor 120 is a main processor (e.g., the main processor 121 in FIG. 1) (e.g., a central processing unit or an application processor) or a secondary processor that can operate independently or together with the main processor (e.g., the main processor 121 in FIG. 1). may include an auxiliary processor 123 (e.g., a graphics processing unit, a neural processing unit (NPU), an image signal processor, a sensor hub processor, or a communication processor).

According to one embodiment, the processor 120 is operatively connected to a memory (e.g., the memory 130 of FIG. 1), and permissions or properties related to an operating system, middleware, or various applications stored in the memory 130 are set. By executing instructions including various instructions or processes, various related data can be loaded and various operations can be performed. For example, the processor 120 may load and execute various data related to the machine learning model stored in the memory 130 to perform inference using the machine learning model and execute various instructions related to various applications using the results. there is.

According to one embodiment, the processor 120 may execute an operating system stored in the memory 130, and the operating system structure accordingly includes a user space 210, a kernel space 220, and a hyper It may include a hypervisor space 230. According to the operating system structure running on the processor 120, a logical separation structure of the user space 210, kernel space 220, and hypervisor space 230 with respect to execution rights or properties of instructions stored in memory 130. You can cause commands or processes to be executed. The processor 120 can run various applications using machine learning models under this operating system structure and can block external attacks on machine learning model data to ensure the integrity of the data.

According to various embodiments, when inference begins through a machine learning model as a specific application is executed, the user space 210 may transmit information related to the machine learning model to the kernel space 220. For example, model-related information may include a model identifier.

According to various embodiments, an application may be identified by a name string of the application, such as “com.sec.android.app.sbrowser”. According to various embodiments, a learning machine model may be identified by a model name string, for example, “mobilenet”. In the user space 210, a list of learning machine models that can be used by a specific application can be managed as shown in Table 1 below, and an unused number from 0 to 100, for example, can be assigned as a model identifier for the learning machine model. You can.

ID app model 0 com.sec.android.app.sbrowser mobilenet One com.sec.android.app.camera Resnet 2 com.sec.android.app.camera Mobilenet ... ... ...

According to various embodiments, when a learning machine model is newly executed in the user space 210, the application name (app string) and model name (model string) can be searched from the above-described list. If the corresponding application name and model name are not in the list according to the search, a new identifier can be assigned to this pair.

Depending on the search, pairs of the corresponding application name and model name are in the list, and information about the corresponding model can be used depending on the execution of the application. Additionally, when the application is terminated, information about the model can be deleted from the list of items related to the application (app string). For example, when the camera app is terminated, items 1 and 2 may be deleted from the list in Table 1 described above.

According to various embodiments, a device driver or syscall may receive model-related information transmitted from the user space 210 and transmit it to the kernel space 220.

According to various embodiments, the kernel space 220 changes the properties of the machine learning model data file to read only based on the model-related information transmitted from the user space 210 and stores the model-related information in the hypervisor space. It can be delivered to (230).

According to various embodiments, the hypervisor space 230 may calculate a hash value for the machine learning model data file based on model-related information and store the calculated hash value along with the model-related information. In the hypervisor space 230, the calculated hash value can be stored in the memory 130, for example.

According to various embodiments, the hypervisor space 230 may block access of the kernel space 220 to the stored hash value.

According to various embodiments, the hypervisor space 230 may block access to the kernel space 220 by changing the entry value of the memory management table of the kernel space 220 for the internal memory 240.

According to various embodiments, the hypervisor space 230 may generate a new hash value for the machine learning model data file over a specified period of time and compare it with the stored hash value.

According to various embodiments, the hash value comparison operation as described above may be periodically performed in the hypervisor space 230.

According to various embodiments, authentication of the model data file may be performed based on whether the hash values are the same through a hash value comparison operation in the hypervisor space 230.

According to various embodiments, if the hash values do not match according to the hash value comparison operation in the hypervisor space 230, the hash value generation and comparison operations may be performed again a specified number of times (e.g., 3 times).

According to various embodiments, if authentication of the model data file fails according to the hash value comparison operation in the hypervisor space 230, the model data file is reloaded through the user space 210 or the currently running application is terminated. can do.

FIG. 3 illustrates a functional module of an operating system structure executed by a processor (e.g., processor 200 of FIG. 2) of an electronic device (e.g., electronic device 101 of FIG. 1) according to various embodiments. It is a block diagram.

According to one embodiment, the operating system architecture of Figure 3 includes user space (e.g., user space 210 of Figure 2), kernel space (e.g., kernel space 220 of Figure 2), and hypervisor space (e.g., Figure 2). It may include 2 hypervisor spaces (230).

According to various embodiments, as a program or application stored in a memory (e.g., memory 130 of FIG. 2) is executed, the processor (e.g., processor 120 of FIG. 2) uses a machine learning framework based on the program or application instructions. (211) (e.g., machine learning libraries such as tflite, caffee, pytorch, keras) can be executed and inference may be performed through the loaded machine learning model (212).

According to various embodiments, the model inference module (e.g., inference model module) 213 in the user space 210 is used in the kernel space to change the properties of the machine learning data file before inference by the machine learning model 212 begins. An executable command can be called at (220).

According to various embodiments, the model inference module 213 may read a machine learning data file and transfer model-related information to the kernel space 220.

According to various embodiments, the model inference module 213 may call the kernel space 220 to stop and organize related operations before the inference of the machine learning model 212 is terminated.

According to various embodiments, the model inference module 213 may use ioctl or implement syscall as shown in Table 2 below to perform the above-described operations.

Use ioctl syscall implementation #define CMD_CHMOD 0#define CMD_START 1
#define CMD_END 2

ioctl(CMD_CHMOD, File)
fd = open(File, O_RDONLY)
read(fd, buf, size)
ioctl(CMD_START, fd, size, model_id)
..
~inference~
...
ioctl(CMD_END, model_id)
exit()




model_chmod(File)
fd = open(File, O_RDONLY)
read(fd, buf, size)
model_start(fd, size, model_id)
..
~inference~
...
model_end(model_id)
exit()

FIG. 3 may show an example in which the model inference module 213 uses ioctl to perform the above-described operations. In this case, a device driver (e.g., /dev/protect_ml) 221 may be implemented in order for the model inference module 213 to communicate between the user space 210 and the kernel space 220. The device driver 221 may be included in the kernel space 220, for example. Meanwhile, when syscall is implemented, communication between the user space 210 and kernel space 220 can be implemented with syscall, so the device driver 221 can be omitted.

According to various embodiments, the model protection module 222 of the kernel space 220 read-only (reads) the properties of the model data file stored in memory (e.g., memory 130 in FIG. 1) based on the received model-related information. can be changed to only).

According to various embodiments, the model protection module 222 may generate a timer 223. For example, when a notification is periodically generated in the timer 223, the model protection module 222 may transmit that a notification has been created to the hash comparison module 232 in the hypervisor space 230.

According to one embodiment, the model protection module 222 may include a total of four functions: change_ro(), register_timer(), hvc_for_hash(), and clean().

According to one embodiment, when an ioctl (e.g., ioctl(CMD_CHMOD)) or syscall (e.g., model_chmod syscall) is received in the user space 210, the model protection module 222 changes the corresponding model data file by the change_ro() function. You can change the inode access permission property to read-only (r--r--r--).

According to one embodiment, the register_timer() and hvc_for_hash() functions of the model protection module 222 operate when an ioctl (e.g., ioctl(CMD_START)) or syscall (e.g., model_start syscall) is received in the user space 220. You can.

According to one embodiment, the register_timer() function can create a timer that generates a notification (e.g., interrupt) at a specified time interval (e.g., 5 minutes).

According to one embodiment, the hvc_for_hash() function may transmit the notification that a notification has occurred through a hypervisor call (hvc) to the hash management module 231 to request hash registration and/or management. For example, the hvc call can transmit model-related information such as fd (file descriptor), size, and model identifier to the hash management module 231.

According to one embodiment, the clean() function may terminate the timer 223 created upon receipt of an ioctl (e.g., ioctl(CMD_END)) or syscall (e.g., model_end syscall) from the user space 210. The expiration of the timer 223 may be communicated to the hypervisor space 230 through an hvc call, allowing the hash management module 231 to end management of the hash value.

According to one embodiment, the timer 223 may operate once at a designated time point (e.g., every 5 minutes). When an interrupt is received according to the operation of the timer 223, it is transmitted to the hypervisor space 230 through an hvc call, the hash management module 231 calculates the hash value, and the hash comparison module 232 calculates the calculated hash value. The integrity of the model file can be verified by comparing the hash value with the stored hash value.

According to various embodiments, the hash management module 231 in the hypervisor space 230 calculates a hash value for the machine learning model data file currently loaded in memory based on model-related information, and the calculated hash value. can be saved along with model-related information. The hash management module 231 may store the calculated hash value in, for example, the memory 130.

According to various embodiments, the hash management module 231 in the hypervisor space 230 may block access of the kernel space 220 to the stored hash value.

According to various embodiments, the hash management module 231 in the hypervisor space 230 changes the entry value of the memory management table of the kernel space 220 for the internal memory 240, so that the kernel space 220 changes the hash value. Access to this stored memory space can be blocked.

According to various embodiments, the hypervisor space 230 may generate a new hash value for the model data file currently loaded in memory when a specified time has elapsed and compare it with the stored hash value. The timer 223 can be set to operate once at a designated time point (e.g., every 5 minutes). When an interrupt is received according to the operation of the timer 223, the hash management module 231 calculates a hash value in the hypervisor space 230 through an hvc call, and the hash comparison module 232 calculates the calculated hash value. The integrity of the model file can be verified by comparing it with the stored hash value.

According to one embodiment, the hash management module 231 may include a create_hash function and a delete_hash function, and the hash comparison module 232 may include a compare_hash function.

According to one embodiment, the create_hash function uses model-related information such as file descriptor (fd) and file size (size) received through ioctl and hvc to create a hash value (e.g. SHA) of the model file being used by the user process of the application. -256) can be calculated. The calculated hash value can be stored in the form of a pair along with the model identifier (model id). The stored hash value can later be used as a value to verify the integrity of file data. Accordingly, even if the kernel space 220 is vulnerable to security and is attacked, the hash value is retrieved by accessing the memory where the hash value is stored in the kernel space 220. Access to the kernel space 220 can be blocked to prevent modification. To this end, the hash management module 231 may block access to the memory location where the hash value is stored by setting the value indicating the virtual address where the hash value is stored in the virtual address table of the kernel space 220 to invalid.

FIG. 4 is a block diagram illustrating the structure of a logical operation module executed by a processor (e.g., processor 120 of FIG. 2) of an electronic device (e.g., electronic device 101 of FIG. 1) according to various embodiments. It is also a degree.

According to one embodiment, the processor architecture of Figure 4 includes user space (e.g., user space 210 in Figure 2), kernel space (e.g., kernel space 220 in Figure 2), and hypervisor space (e.g., Figure 2). It may include a hypervisor space (230).

According to various embodiments, as a program or application stored in a memory (e.g., memory 130 of FIG. 2) is executed, the processor (e.g., processor 120 of FIG. 2) uses a machine learning framework based on the program or application instructions. 311 (e.g., machine learning libraries such as tflite, caffee, pytorch, keras) can be executed and inference may be performed through the loaded machine learning model 312.

According to various embodiments, the model inference module (e.g., inference model module) 313 of the user space 210 calls the kernel space 220 before inference by the machine learning model 312 starts to file a machine learning data file. You can change the properties of .

According to various embodiments, the model inference module 313 may read a machine learning data file and call the kernel space 220.

According to various embodiments, the model inference module 313 may call the kernel space 220 to stop and organize related operations before the application of the machine learning model 312 is terminated.

According to various embodiments, the model inference module 313 may implement a syscall as described in Table 2 above to perform the above-described operations.

According to one embodiment, Figure 4 may show an example in which the model inference module 313 uses a syscall to perform the above-described operations. In this case, the model inference module 313 may use a syscall to perform communication between the user space 210 and the kernel space 220. Accordingly, the device driver 221 as shown in FIG. 3 may be omitted. The configuration of FIG. 4 may be similar in function to the configuration of FIG. 3 except for the device driver 221, and detailed descriptions may be omitted below.

According to various embodiments, the model protection module 322 of the kernel space 220 read-only (read-only) properties of a model data file stored in memory (e.g., internal memory 240 of FIG. 2) based on the received model-related information. You can change it to read only).

According to various embodiments, the model protection module 322 may generate a timer 323. A notification periodically generated by the timer 323 may be delivered to the hash comparison module 332 in the hypervisor space 230.

According to one embodiment, the model protection module 322 may be composed of a total of four functions: change_ro(), register_timer(), hvc_for_hash(), and clean(). The functions of the four functions are similar to those described with reference to FIG. 2, and detailed descriptions thereof are omitted here.

According to one embodiment, when a syscall (e.g., model_chmod syscall) is received from the user space 210, the model protection module 322 sets the inode access permission attribute of the corresponding model data file to read-only (r--r--r). --), and model-related information such as fd(), size, and model identifier can be transmitted to the hash management module 331 through hvc call.

According to one embodiment, the timer 323 operates once at a designated point in time (e.g., every 5 minutes), and the hash management module 331 in the hypervisor space 230 compares the newly calculated hash value to the hash comparison module 332. The integrity of the model file can be verified by comparing it with the stored hash value.

According to various embodiments, the hash management module 331 in the hypervisor space 230 calculates a hash value for the machine learning model data file based on model-related information, and combines the calculated hash value with model-related information and You can save them together. The hash management module 231 may store the calculated hash value in, for example, the internal memory 240.

According to various embodiments, the hash management module 331 in the hypervisor space 230 may block access of the kernel space 220 to the stored hash value.

According to various embodiments, the hash management module 331 in the hypervisor space 230 changes the entry value of the memory management table of the kernel space 220 for the internal memory 240, so that the kernel space 220 changes the hash value. Access to this stored memory space can be blocked.

There is a need to ensure the integrity of data in response to kernel attacks on data stored by electronic devices, such as machine learning model data files learned according to various embodiments.

The technical problem to be achieved in this document is not limited to the technical problem mentioned above, and other technical problems not mentioned can be clearly understood by those skilled in the art from the description below. There will be.

An electronic device according to various embodiments includes a memory (e.g., memory 130 of FIG. 1) and a processor (e.g., processor 120 of FIG. 1) operatively connected to the memory, wherein the memory includes, When executed, the processor loads the model data file from the memory when an operation according to application execution based on the model data file begins, calculates a hash value for the model data file in the hypervisor space, and generates the model data file. Store it with model-related information, calculate the hash value of the loaded model data file in the hypervisor space based on a timer generated in the kernel space, and compare the calculated hash value with the stored hash value. Instructions may be stored to confirm identity and determine that the model data file has been modified if the hash values are not identical according to the comparison.

According to various embodiments, the memory may store instructions that, when executed, cause the processor to block access of the hypervisor space to the hash value storage location in the kernel space.

According to various embodiments, the memory may store instructions that, when executed, cause the processor to block access to the hash value storage location by changing an entry in a page table of the kernel space.

According to various embodiments, the memory may store instructions that, when executed, cause the processor to change the entry of the page table in the kernel space to invalid.

According to various embodiments, the memory, when executed, may store instructions that cause the processor to change the properties of the model data file to read-only when an operation according to execution of an application based on the model data file begins. .

According to various embodiments, when the memory is executed, the processor transfers the model-related information about the model data file from the kernel space to the hypervisor space to calculate and store the hash value in the hypervisor space. You can store instructions that allow you to perform actions.

According to various embodiments, the memory may store instructions that, when executed, cause the processor to transfer the model-related information for the model data file from the kernel space to the hypervisor space through a device driver or syscall. You can.

According to various embodiments, the memory may store instructions that, when executed, cause the processor to transmit a notification generated by the timer generated in the kernel space to the hypervisor space to perform the hash value comparison operation. You can.

FIG. 5 is a diagram for explaining a method of protecting data in an electronic device according to various embodiments.

According to various embodiments, a processor designed as an Advance RISC Machine (ARM) (e.g., processor 120 in FIG. 2) may be configured with four exception levels to handle exceptions when they occur, and each level is connected to each other. You may have other privileges.

In one embodiment, ARM's exception levels can be divided into EL0, EL1, EL2, and EL3, and an operation can be performed at one of these exception levels. The four levels EL3, EL2, EL1, and EL0 are in that order. For example, the security level of EL2 is higher than that of EL1, the privilege level is high, and the level of authority to access resources may also be higher.

According to one embodiment, the hypervisor space, which is EL2, may have higher privileges than the kernel space, which is EL1.

According to one embodiment, the hash management module of the hypervisor space (e.g., the hypervisor space 230 of Figures 2, 3, or 4 (e.g., the hash management module 231 or 331 of Figure 3 or 4)) creates_hash It may include a function, delete_hash function, and the hash comparison module 232 may include a compare_hash function.

According to one embodiment, the create_hash function calculates the hash value (e.g., SHA-256) of the model file used by the user process of the application using fd and size passed through ioctl, syscall, and/or hvc to create memory 530. ) (e.g., internal memory 240 of FIG. 2). The calculated hash value can be stored in the form of a pair along with the model identifier (model id). The stored hash value can later be used as a value to verify file data integrity, and accordingly, the memory location 531 where the hash value is stored in the kernel space (e.g., the kernel space 220 in FIG. 2, FIG. 3, or FIG. 4) Access to can be blocked.

According to one embodiment, the hash management module 231 or 331 of the hypervisor space at the EL2 level creates the last page table 511 entry 513 of the page table of the kernel space 220 at the EL1 level. By setting it to invalid, access to the memory location 531 in the kernel space can be blocked.

FIG. 6 is a flowchart illustrating a method for protecting data in an electronic device, according to various embodiments.

According to various embodiments, in operation 601, a processor (e.g., processor 120 of FIG. 2, FIG. 3, or FIG. 4) of an electronic device (e.g., electronic device 101 of FIG. 1) generates a specific learning machine model and A variety of related data can be loaded.

According to one embodiment, the processor executes a specific application in user space (e.g., user space 210 in Figures 2, 3, or 4), starts operations to perform various instructions, and runs a running program used by the specific application. Data files related to the machine model can be loaded into memory.

According to various embodiments, when inference begins through a machine learning model as a specific application is executed, the user space stores information related to the machine learning model in the kernel space (e.g., kernel space 220 of Figures 2, 3, or 4). ) can be transmitted. For example, model-related information may include a model identifier.

According to various embodiments, the processor may create a timer in kernel space. A timer created in kernel space can generate notifications (e.g. prompts) at points when the integrity of the model data file is verified to ensure the security of the model data file.

According to various embodiments, the processor changes the properties of the machine learning model data file to read only based on model-related information in kernel space and stores the model-related information in hypervisor space (e.g., Figure 2, Figure 3, or It can be delivered to the hypervisor space 230 in FIG. 4.

According to various embodiments, in operation 603, the processor calculates a hash value for the machine learning model data file in the hypervisor space based on model-related information transmitted from the kernel space, and stores the calculated hash value as model-related information. You can save it with . In the hypervisor space 230, the calculated hash value can be stored in, for example, internal memory (eg, internal memory 240 in FIG. 2).

According to various embodiments, the processor may block kernel space access to the hash value stored in the hypervisor space so that the hash value cannot be modified accordingly if the kernel space is attacked.

According to various embodiments, the processor may generate a new hash value for the machine learning model data file in the hypervisor space at operation 605 at a designated time based on a timer generated in the kernel space.

According to various embodiments, the processor may compare the generated hash value and the stored hash value in the hypervisor space in operation 607.

According to various embodiments, if the hash values do not match according to the hash value comparison operation in the hypervisor space, the processor may re-perform the hash value generation and comparison operation a specified number of times (e.g., 3 times).

According to various embodiments, the processor may determine in operation 609 that the model data file has been modified if the hash values are not the same.

According to various embodiments, the processor may periodically perform a hash value comparison operation in the hypervisor space.

According to various embodiments, authentication of a model data file can be performed based on whether the hash values are the same through a hash value comparison operation in the hypervisor space.

According to various embodiments, if authentication of the model data file fails according to a hash value comparison operation in the hypervisor space, the processor may reload the model data file through user space or terminate the currently running application.

FIG. 7 is a flowchart illustrating a method for protecting data in an electronic device, according to various embodiments.

According to various embodiments, in operation 601, a processor (e.g., processor 120 of FIG. 2, FIG. 3, or FIG. 4) of an electronic device (e.g., electronic device 101 of FIG. 1) generates a specific learning machine model and A variety of related data can be loaded.

According to one embodiment, the processor executes a specific application in user space (e.g., user space 210 in Figures 2, 3, or 4) and starts operations to perform various commands, Data files related to the machine model can be loaded into memory.

According to various embodiments, models generated from machine learning frameworks (e.g., machine learning framework 211 or 311 of Figure 3 or Figure 4) may be trained in the learning step, and the trained model may include optimized weight information.

According to various embodiments, in the model inference step, model data including the tree structure and weights of the model may be loaded. There may be differences in the language expression of commands or functions depending on the type of machine learning framework, but performing inference using a machine learning model may include the process of open() and read() the model data. According to one embodiment, a device driver (e.g., /dev/protect_ml) can be created to check the integrity of the model.

According to one embodiment, each model may have its own model identifier (model id).

According to various embodiments, in operation 701, the processor may initiate inference through a machine learning model as a specific application is executed in user space, thereby sending information related to the machine learning model to kernel space (e.g., FIGS. 2 and 3 Alternatively, it can be delivered to the kernel space 220 of FIG. 4).

According to one embodiment, when inference begins through a model, the processor sends a device driver (e.g. : /dev/protect_ml). After the file open and read is performed, CMD_ID 1 to indicate that inference of the model has started, for example via ioctl, and the model file's file descriptor, file size, or model identifier (model id). Related information can be written to the device driver.

According to various embodiments, in operation 703, the processor changes the properties of the machine learning model data file to read only based on model-related information in kernel space and stores the model-related information in hypervisor space (e.g., Figure 2, It can be delivered to the hypervisor space 230 of Figure 3 or Figure 4. For example, in the kernel space that receives CMD_ID 0 and a file indicating that model inference has started through a device driver (e.g. /dev/protect_ml), the chage_ro() function is executed to change the file's properties to change the file's properties to the corresponding model data file. You can change the inode attribute to r--r--r--. Therefore, attempts to write to the model data file from the outside can be blocked, and even when the data file is opened, it can be opened as read-only.

According to one embodiment, if the inode property of a data file is changed to read-only, then when the file is opened for model inference in user space (e.g., open(FILE, O_RDONLY)), the file is opened with write permission. When accessing , such access may fail because the file's attribute is read only.

Meanwhile, if root permission is obtained, it may be possible to write to a file if the inode is set to read only, and for this purpose, a hash comparison operation can be performed as an additional protection operation. According to various embodiments, the processor may open a model data file for read-only in operation 705 and may read a model data file from the open model data file in operation 707.

According to one embodiment, the same model file may be loaded in different applications or different processes. Different applications or user processes can have different model identifiers.

According to various embodiments, when inference begins through a machine learning model, the processor may generate a timer in kernel space in operation 709 accordingly. For example, in the kernel space that receives CMD_ID 0, which indicates that model inference has started, through the device driver (e.g. /dev/protect_ml), a timer is created based on the start point of model data file inference, and the created timer is For example, you can periodically generate notifications (e.g. prompts). In kernel space, you can confirm that model inference has started and create and register a timer that runs once every 5 minutes, for example. The kernel space can transmit model-related information such as fd, file size, and model id to the hypervisor space through the hvc call and notify that model inference has begun.

According to various embodiments, in operation 711, the processor may periodically check the hash value based on a timer generated in kernel space. The processor may periodically generate a new hash value for the machine learning model data file in the hypervisor space based on a timer generated in the kernel space and compare the hash value generated in the hypervisor space with the stored hash value. The timer in the registered kernel space can generate a timer interrupt, for example, once every 5 minutes and call a function (e.g. Compare_hash()) that compares hash values in the hypervisor space through hvc. In the hypervisor space, you can perform a hash operation on the model data file in memory accessible in the user space through compare_hash() and compare it with the stored hash value.

According to various embodiments, in operation 713, the processor may check whether the hash value mismatch (fail in operation 711) has been performed more than a specified number of times (e.g., 3 times) according to the hash value comparison operation in the hypervisor space. If the specified number of times has not been exceeded according to the hash value comparison operation, the operation returns to operation 711 and the hash value generation and comparison operation can be performed again. For example, the compare_hash() function in hypervisor space can deliver a value indicating the integrity of the file (e.g., return 0) if the hash values are the same.

According to various embodiments, if the processor confirms in operation 713 that the hash values are not the same more than a specified number of times, the processor determines that the model data file has been modified, and determines whether the application is terminated in operation 715. If it is determined that the model data file has been modified, the model data file can be reloaded or the application can be terminated. If the two hash values are different, the hash operation can be retried a specified number of times, for example, 3 times, taking into account the possibility of malfunction. Accordingly, if the hash values are different all three times, the processor can determine that the file has been modified. In this case, a new file may be loaded depending on the case, but for applications that are very sensitive to security, such as a biometric recognition system, the system may be turned off.

According to various embodiments, if the result of comparing the hash values more than a specified number of times is not the same, but this does not correspond to the case of terminating the application (operation 715 - No), the processor may recover the damaged data by reloading the model data file in operation 717. there is.

According to various embodiments, in operation 719, the processor calculates a hash value for the machine learning model data file in the hypervisor space based on model-related information transmitted from the kernel space, and stores the calculated hash value in the model-related information. You can save it with . For example, the calculated hash value can be stored along with the model identifier. The processor can verify integrity by periodically comparing the stored hash value with a file loaded in memory accessible to the user process.

According to various embodiments, in operation 721, the processor may block kernel space access to the hash value stored in the hypervisor space. In the hypervisor space, the page table value in the kernel space can be changed to make the memory where the hash value is stored inaccessible. In the hypervisor space, for example, an entry in the page table in the kernel space can be changed to invalid to unmap it. As a result, the kernel space can be attacked, making it impossible to modify the file contents and recalculate and save the hash value. Therefore, if the kernel space is attacked, the hash value managed by the hypervisor may not be modified even if the contents of the file loaded in memory are changed.

According to various embodiments, the processor may store the hash value in hypervisor space and unmap the page table in kernel space, and then start model inference through user space in operation 723.

According to various embodiments, when the model inference is terminated and the corresponding process is terminated, the processor may transmit a notification before process termination from the user space to the kernel space in operation 725. For example, the user space can use CMD_ID 2 and model id to notify termination through ioctl, and the kernel space can terminate the timer accordingly in operation 727, and deliver notification through hvc to execute operation 729 in the hypervisor. You can perform an operation to delete the stored hash value.

According to various embodiments, the processor includes a main processor and an auxiliary processor, and when a plurality of machine learning models according to the above-described embodiments operate and a model inference operation is performed by two or more processors, a plurality of machine learning models It is possible to create and operate only one timer (global timer) described above. For example, when inference of a machine learning model begins, all operations except the timer creation operation in the above-described embodiments may be performed in the same or similar manner. In some cases, for security reasons, a timer may be registered separately by adjusting the interval time for a specific model. When the global timer periodically issues interrupts for multiple models, the hash value of the currently used model file can be calculated in the hypervisor space and compared with the stored hash value.

The embodiments disclosed in this document are merely presented as examples to easily explain the technical content and aid understanding, and are not intended to limit the scope of the technology disclosed in this document. Therefore, the scope of the technology disclosed in this document should be interpreted as including all changes or modified forms derived based on the technical idea of the various embodiments disclosed in this document in addition to the embodiments disclosed herein.

Claims (20)

In electronic devices,
Memory; and
comprising a processor operatively connected to the memory,
The memory, when executed, causes the processor to:
When the calculation according to the execution of the application based on the model data file starts, the model data file is loaded from the memory,
Calculating a hash value for the model data file in the hypervisor space and storing it together with model-related information about the model data file,
Calculating a hash value of the loaded model data file in the hypervisor space based on a timer generated in the kernel space,
Compare the calculated hash value with the stored hash value to confirm identity,
An electronic device that stores instructions for determining that the model data file has been modified if the hash values are not identical according to the comparison.
According to paragraph 1,
The memory, when executed, causes the processor to:
An electronic device that stores instructions that allow the hypervisor space to block access to the hash value storage location in the kernel space.
According to paragraph 2,
The memory, when executed, causes the processor to:
An electronic device that stores instructions for blocking access to the hash value storage location by changing an entry in a page table of the kernel space.
According to paragraph 3,
The memory, when executed, causes the processor to:
An electronic device that stores instructions for changing the entry of the page table in the kernel space to invalid.
According to paragraph 1,
The memory, when executed, causes the processor to:
An electronic device that stores instructions for changing the properties of the model data file to read-only when an operation according to execution of an application based on the model data file begins.
According to paragraph 1,
The memory, when executed, causes the processor to:
If the hash values are not the same, the operation of calculating the hash value for the model data file according to the timer in the hypervisor space and comparing whether it is the same as the hash value stored with the model ID is performed a specified number of times. An electronic device that stores instructions.
According to paragraph 1,
The memory, when executed, causes the processor to:
An electronic device that stores instructions for reloading the model data file from the memory or terminating the application when it is determined that the model data file has been modified.
According to paragraph 1,
The memory, when executed, causes the processor to:
An electronic device that stores instructions for transferring the model-related information about the model data file from the kernel space to the hypervisor space to perform the hash value calculation and storage operations in the hypervisor space.
According to clause 8,
The memory, when executed, causes the processor to:
An electronic device that stores instructions for transferring the model-related information about the model data file from the kernel space to the hypervisor space through a device driver.
According to paragraph 1,
The memory, when executed, causes the processor to:
An electronic device that stores instructions for transmitting a notification generated by the timer generated in the kernel space to the hypervisor space to perform the hash value comparison operation.
In the method of the electronic device,
loading the model data file from the memory when an operation according to execution of an application based on the model data file begins;
An operation of calculating a hash value for the model data file in the hypervisor space and storing it together with model-related information for the model data file;
calculating a hash value of the loaded model data file in the hypervisor space based on a timer generated in the kernel space;
Comparing the calculated hash value with the stored hash value to confirm identity; and
A method comprising determining that the model data file has been modified if the hash values are not identical according to the comparison.
According to clause 11,
The method further includes an operation of allowing the hypervisor space to block access to the hash value storage location in the kernel space.
According to clause 12,
The method further includes an operation of blocking access to the hash value storage location by changing an entry in the page table of the kernel space.
According to clause 13,
The blocking operation changes the entry of the page table in the kernel space to invalid.
According to clause 11,
The method further includes changing the properties of the model data file to read-only when an operation according to execution of an application based on the model data file begins.
According to clause 11,
If the hash values are not the same, an operation of calculating a hash value for the model data file according to the timer in the hypervisor space and comparing whether it is the same as the hash value stored with the model ID a specified number of times. How to include more.
According to clause 11,
If it is determined that the model data file has been modified, the method further includes reloading the model data file from the memory or terminating the application.
According to clause 11,
The method further includes transferring the model-related information about the model data file from the kernel space to the hypervisor space to perform the hash value calculation and storage operation in the hypervisor space.
According to clause 18,
The method further includes transferring the model-related information about the model data file from the kernel space to the hypervisor space through a device driver.
According to clause 11,
The method further includes transmitting a notification generated by the timer generated in the kernel space to the hypervisor space to perform the hash value comparison operation.

Images