CN118069222A - Instruction execution method and device - Google Patents
Instruction execution method and device Download PDFInfo
- Publication number
- CN118069222A CN118069222A CN202410236005.8A CN202410236005A CN118069222A CN 118069222 A CN118069222 A CN 118069222A CN 202410236005 A CN202410236005 A CN 202410236005A CN 118069222 A CN118069222 A CN 118069222A
- Authority
- CN
- China
- Prior art keywords
- instruction
- cpu
- coprocessor
- illegal
- kernel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 126
- 238000004088 simulation Methods 0.000 claims abstract description 132
- 230000008569 process Effects 0.000 claims abstract description 46
- 238000012545 processing Methods 0.000 claims description 109
- 238000004458 analytical method Methods 0.000 claims description 54
- 238000004590 computer program Methods 0.000 claims description 16
- 238000013519 translation Methods 0.000 abstract description 22
- 230000006870 function Effects 0.000 description 10
- 238000010586 diagram Methods 0.000 description 9
- 239000002699 waste material Substances 0.000 description 7
- 238000012360 testing method Methods 0.000 description 5
- 230000001133 acceleration Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 229910002056 binary alloy Inorganic materials 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000012634 fragment Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000008054 signal transmission Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
- G06F9/30098—Register arrangements
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Advance Control (AREA)
Abstract
The embodiment of the invention provides an instruction execution method and device, comprising the following steps: in the process of executing the instruction in the user mode, if the CPU determines that the current instruction is an illegal instruction, switching to the kernel mode; the CPU sends illegal instructions to the coprocessor in a kernel state; the coprocessor performs simulation operation on the illegal instruction to obtain a simulation result; after the CPU obtains the simulation result in the kernel state, the CPU is switched to the user state to continuously execute the subsequent instruction. The efficiency of emulating an illegal instruction using a coprocessor may be higher than the efficiency of emulating an illegal instruction using a CPU. Through the cooperative simulation of the kernel and the coprocessor, the kernel is responsible for taking out instructions to be simulated, and then the coprocessor simulates the instructions, so that the efficiency of instruction translation/instruction simulation is improved.
Description
The present invention is a divisional application of the invention application with the application date of 2023, 10 month and 24 days, the application number of 202311378936.3 and the name of a method and a device for executing an instruction.
Technical Field
The present invention relates to the field of coprocessor technologies, and in particular, to a method and apparatus for executing instructions.
Background
The instruction set is one of the basic components of a computer for directing a processor to perform various operations. It includes information such as command, register and address coding mode. However, the binary program can only be executed on the CPU having the instruction set matched with the binary program, and if the instruction set adopted by the executed binary program does not match the instruction set used by the CPU, the CPU cannot execute the binary program.
Currently, a binary system with an unmatched instruction set is translated by a software translation mode to obtain a binary system or an instruction sequence matched with a CPU architecture, and then the translated binary system or the instruction sequence is input into the CPU for execution. However, software translation itself requires a relatively high time cost and the execution efficiency of the translated binary or instruction fragment is typically quite low, which results in quite low binary performance of executing the different instruction set by software translation.
Therefore, how to improve the efficiency of instruction translation/instruction simulation is a technical problem that needs to be solved currently.
Disclosure of Invention
The embodiment of the invention provides an instruction execution method and an instruction execution device, which are used for solving the problem of lower efficiency of instruction translation/instruction simulation in the prior art.
In a first aspect, an embodiment of the present invention provides an instruction execution method, including: in the process of executing the instruction in the user mode, if the CPU determines that the current instruction is an illegal instruction, switching to the kernel mode; the CPU sends illegal instructions to the coprocessor in a kernel state; the coprocessor performs simulation operation on the illegal instruction to obtain a simulation result; after the CPU obtains the simulation result in the kernel state, the CPU is switched to the user state to continuously execute the subsequent instruction.
In the above technical solution, since the CPU is a general-purpose circuit and the coprocessor is a special-purpose circuit that is specially used for emulating an illegal instruction, the efficiency of emulating an illegal instruction using the coprocessor is higher than the efficiency of emulating an illegal instruction using the CPU. Through the cooperative simulation of the kernel and the coprocessor, the kernel is responsible for taking out instructions to be simulated, and then the coprocessor simulates the instructions, so that the efficiency of instruction simulation is improved.
Optionally, the coprocessor performs simulation operation on the illegal instruction to obtain a simulation result, including: the coprocessor analyzes the illegal instruction to obtain an analysis result; the CPU obtains the register data corresponding to the analysis result in the kernel state and sends the register data to the coprocessor; and the coprocessor performs simulation operation of illegal instructions through the register data to obtain simulation results.
Optionally, the method further comprises: in the process of analyzing illegal instructions by the coprocessor, the CPU is switched to a user state and executes a first instruction; the first instruction is any one or more instructions except the current instruction; after the analysis result is obtained, before the CPU obtains the register data corresponding to the analysis result in the kernel state, the method further comprises the following steps: the coprocessor sends a first interrupt to the CPU; the CPU switches to kernel mode based on the first interrupt.
Optionally, the method further comprises: in the process of analyzing illegal instructions by the coprocessor, the CPU executes idle waiting in a kernel state; after the analysis result is obtained, before the CPU obtains the register data corresponding to the analysis result in the kernel state, the method further comprises the following steps: the coprocessor sends a first interrupt to the CPU.
Optionally, the method further comprises: in the process of analyzing illegal instructions by the coprocessor, the CPU executes first busy-wait in a kernel state, wherein the first busy-wait is used for obtaining analysis results from the coprocessor.
Optionally, the method further comprises: in the process of simulating illegal instructions by the coprocessor, the CPU is switched to a user state and executes a second instruction; the second instruction is any one or more instructions except the current instruction; after the coprocessor obtains the simulation result, before the CPU obtains the simulation result in the kernel state, the CPU further comprises: the coprocessor sends a second interrupt to the CPU; the CPU switches to kernel mode based on the second interrupt.
Optionally, the method further comprises: in the process of simulating illegal instructions by the coprocessor, the CPU executes idle waiting in a kernel state; after the coprocessor obtains the simulation result, before the CPU obtains the simulation result in the kernel state, the CPU further comprises: the coprocessor sends a second interrupt to the CPU.
Optionally, the method further comprises: in the process of simulating illegal instructions by the coprocessor, the CPU executes second busy-wait in the kernel state, and the second busy-wait is used for obtaining simulation results from the coprocessor.
Alternatively, the first interrupt and the second interrupt are implemented by the same vectorized interrupt, or by different vectorized interrupts.
Optionally, the coprocessor performs simulation operation on the illegal instruction to obtain a simulation result, including: the coprocessor analyzes the illegal instruction to obtain an analysis result; the coprocessor acquires shadow register data corresponding to the analysis result according to the analysis result; and the coprocessor performs simulation operation of illegal instructions through the shadow register data to obtain a simulation result.
Optionally, the coprocessor performs simulation operation of the illegal instruction through the shadow register data, and after obtaining a simulation result, the method further includes: if the illegal instruction has an association relation with the shadow register, the simulation result is written into the shadow register.
In a second aspect, an embodiment of the present invention provides an instruction execution apparatus, including: the first processing unit is switched to a kernel mode if determining that the current instruction is an illegal instruction in the process of executing the instruction in the user mode; the first processing unit sends illegal instructions to the second processing unit in a kernel state; the second processing unit performs simulation operation on the illegal instruction to obtain a simulation result; after the first processing unit obtains the simulation result in the kernel mode, the first processing unit is switched to the user mode to continuously execute the subsequent instruction.
Optionally, the second processing unit analyzes the illegal instruction to obtain an analysis result; the first processing unit obtains register data corresponding to an analysis result in a kernel mode and sends the register data to the second processing unit; and the second processing unit performs simulation operation of illegal instructions through the register data to obtain a simulation result.
Optionally, in the process of resolving the illegal instruction by the second processing unit, the first processing unit switches to a user state and executes the first instruction; the first instruction is any one or more instructions except the current instruction; the second processing unit sends a first interrupt to the first processing unit; the first processing unit switches to a kernel mode based on the first interrupt.
Optionally, in the process that the second processing unit analyzes the illegal instruction, the first processing unit executes idle waiting in a kernel mode; the second processing unit sends a first interrupt to the first processing unit.
Optionally, in the process of resolving the illegal instruction by the second processing unit, the first processing unit executes a first busy-wait in a kernel state, where the first busy-wait is used to obtain a resolving result from the co-processing.
Optionally, in the process of simulating the illegal instruction by the second processing unit, the first processing unit switches to the user state and executes the second instruction; the second instruction is any one or more instructions except the current instruction; the second processing unit sends a second interrupt to the first processing unit; the first processing unit switches to a kernel mode based on the second interrupt.
Optionally, in the process of simulating the illegal instruction by the second processing unit, the first processing unit executes idle waiting in a kernel mode; the second processing unit sends a second interrupt to the first processing unit.
Optionally, in the process of simulating the illegal instruction by the second processing unit, the first processing unit performs a second busy-wait in a kernel state, where the second busy-wait is used to obtain a simulation result from the second processing unit.
Alternatively, the first interrupt and the second interrupt are implemented by the same vectorized interrupt, or by different vectorized interrupts.
Optionally, the second processing unit analyzes the illegal instruction to obtain an analysis result; the second processing unit acquires shadow register data corresponding to the analysis result according to the analysis result; and the second processing unit performs simulation operation of illegal instructions through the shadow register data to obtain a simulation result.
Optionally, if the illegal instruction has an association relationship with the shadow register, writing the simulation result into the shadow register.
In a third aspect, an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the method for executing the instructions of any of the first aspects.
In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing a computer program executable by a computer device, which when run on the computer device causes the computer device to perform the method of executing instructions of any of the first aspects described above.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments will be briefly described below, it will be apparent that the drawings in the following description are only some embodiments of the present invention, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a method for kernel trap simulation according to an embodiment of the present invention;
FIG. 2 is a flowchart of an instruction execution method according to an embodiment of the present invention;
FIG. 3 is a flowchart of a method for determining simulation results according to an embodiment of the present invention;
FIG. 4 is a flow chart of a method for executing instructions according to an embodiment of the present invention;
FIG. 5 is a flowchart of another method for executing instructions according to an embodiment of the present invention;
FIG. 6 is a flowchart of another method for instruction execution according to an embodiment of the present invention;
FIG. 7 is a flowchart of a method for executing another instruction according to an embodiment of the present invention;
FIG. 8 is a flow chart of a method for executing instructions according to an embodiment of the present invention;
FIG. 9 is a flow chart of a method for executing instructions according to an embodiment of the present invention;
FIG. 10 is a flow chart of a method for executing instructions according to an embodiment of the present invention;
FIG. 11 is a schematic diagram of an instruction execution device according to an embodiment of the present invention;
fig. 12 is a schematic structural diagram of a computer device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the following, the present application is described in part for general explanation, so as to be understood by those skilled in the art, but not limited thereto.
1. User mode: the CPU may implement a number of privilege states, including a user state, which is typically the lowest privilege state where a user program is executed without the system crashing due to an error.
2. Kernel mode: a CPU may implement a number of privilege states, including a kernel state, which is typically the most privileged privilege state (and may sometimes extend the more privileged privilege states), may execute certain privileged instructions, may crash the system once mishandled, and may only execute there by the operating system kernel.
3. Register: the registers are small storage areas used for storing data inside the CPU and used for temporarily storing the data and operation results participating in the operation.
4. Shadow register: by introducing shadow registers, the instruction may reuse the same register encodings, but in different modes, these encodings correspond to different physical registers.
5. A coprocessor: a coprocessor is a processor that assists a central processing unit in developing and applying processing tasks that it cannot perform, or that is inefficient in execution. The central processing unit cannot execute a lot of work, such as signal transmission among devices, management of access devices and the like; and the execution efficiency and the effect are low, such as graphic processing, audio processing and the like.
6. An internal coprocessor: the coprocessor may reside on the same die as the central processor, and such a coprocessor is conventionally referred to as an internal coprocessor. The CPU that is common in the market includes some internal coprocessors and more peripheral circuits in addition to a central processing unit (i.e., CPU in a narrow sense). Such coprocessors are typically used to accomplish tasks that the CPU cannot do. Because of their close relationship and most of them have a very important role in the operation of the whole system, it is customary not to refer to this whole as SoC but to refer to it directly as CPU.
7. An external coprocessor: some central processors expose interfaces outward for access to optional coprocessors, which are traditionally referred to as external coprocessors. The cpu is usually coupled to an external coprocessor with a low degree of coupling and works well without having to access the external coprocessor. In general, the external coprocessor is presented as a stand-alone chip or as an expansion card.
The CPU can only execute instruction sets of the same architecture as the CPU. For example, if the architecture of the CPU is MIPS, the CPU of the MIPS architecture can only execute MIPS instructions. As another example, if the CPU architecture is RISC-V instructions, then the RISC-V architecture CPU can only execute RISC-V instructions.
However, in one possible scenario, a binary program can only be executed on a CPU having an instruction set matching it, and if the instruction set employed by the binary program being executed does not match the instruction set used by the CPU, the CPU cannot execute the binary. The case in which the instruction set of the binary program being executed does not match the instruction set executed by the CPU can be divided into two cases, and the two cases are described below, respectively:
First case: the binary uses the same architecture as the CPU, however, the latter does not implement some of the instruction set extensions used by the former, or the instruction set version of the latter is not compatible with the former. For example, the CPU supports the MIPS instruction set, wherein the extended instruction of the MIPS instruction set is instruction a, but since some CPUs can process the a instruction, some CPUs cannot process the instruction a, if the CPU cannot process the instruction a executes the instruction a, illegal instruction exception may be caused. As another example, during the evolution of an instruction set, a new version of the instruction set may add instructions, resulting in a binary compiled for the new version of the instruction set not being able to run on a CPU that uses the old version of the instruction set. The new version of the instruction set may also delete some instructions, resulting in a binary compiled for the old version of the instruction set not being able to run on the CPU using the new version of the instruction set.
Second case: the target architecture of the binary program is different from the CPU architecture. For example, a CPU supporting the MIPS instruction set cannot execute the binary program of the RISC-V architecture.
The execution of binary programs incompatible therewith on the CPU may be implemented in a software translation manner, where possible. In particular, software translation, i.e., the translation of unsupported instructions (instructions of other architectures/instruction extensions not implemented by the native CPU) by software into a series of equivalent supported instruction sequences that, after execution, are externally equivalent to the simulated original instructions. The most representative instruction translation software is QEMU, which almost performs instruction translation and some additional work when running user state simulations; it also requires a virtualization of the hardware when running system-level simulations, which is not a software translation as discussed herein. The compiler can be divided into AOT (Ahead-Of-Time) which compiles binary before execution and JIT (Just-In-Time) which compiles while executing and software translation. QEMU is used almost exclusively for JIT translation, where an untranslated instruction block is translated and then executed, the translated instruction block is cached in a policy and need not be translated again by the time it is executed. While the AOT translation is represented by Apple's Rosetta2, after the x86 software is installed on ARM Mac, rosetta2 scans and translates all binaries in the software immediately, translating them completely into ARM64; when running these software, it is ARM64 binary that is executed without the need for re-translation. Problems with software translation are: 1. software translation itself requires a relatively high time cost and the translated binary or instruction fragment is typically quite inefficient to execute, which results in quite poor binary performance of executing the different instruction sets through software translation. 2. For the first case described above (instruction set extensions/versions are different), not every instruction needs to be translated, for example, if most instructions do not need to be translated, only a small portion of the instructions need to be translated, the translator translates all instructions, which is time consuming and resource consuming, and there may be instructions that do not need to be translated that may no longer be the original instructions after translation, other instructions or a combination of other instructions.
In another possible scenario, to solve the inefficiency of software translation without introducing complex hardware costs, an instruction set extension may be designed that abstracts the commonality of those instructions in the target instruction set that are inefficient to execute after translation, providing acceleration in hardware. It should be noted that the translation acceleration instruction set extension itself is neither a re-implementation of the target instruction set nor a hardware-implemented translator, which needs to be used with a software translator specifically adapted to the instruction set extension. The instruction set extension can effectively solve the problem of 2, the instruction set extension has the problems of requiring the support of a CPU, occupying valuable coding space of the instruction set, having higher coupling degree with the CPU and poorer flexibility, and when the instruction set extension is formally determined, continuing iteration is easy to introduce compatibility, and the instruction set extension needs to be matched with a special software translator for use.
In one possible application scenario, a large number of instructions in the binary program can be directly executed by the CPU, and these instructions do not need to be emulated/translated, but when the CPU executes to an unsupported instruction, the CPU cannot execute the unsupported instruction, so that the CPU falls into an illegal instruction exception, which is the case 1 mentioned above. Since software translation and instruction set extension cannot better solve the problem in case 1, in order to better solve the problem in case 1, a kernel trap simulation (trap and empty) method is introduced. Specifically, the CPU directly executes the binary program, and when illegal instruction abnormality is encountered, the kernel trap simulation method is started again, so that the instruction which can be directly executed by the CPU can be executed with the original efficiency, and the overall cost is greatly reduced.
As shown in fig. 1, a flowchart of a method for kernel trap simulation according to an embodiment of the present invention is provided, and the method includes the following steps:
Step 101, executing an instruction to be executed.
Step 102, the decoder decodes the instruction to be executed.
In the embodiment of the invention, the instruction consists of an operation code and an address code. The opcode indicates the nature of the operation to be performed, i.e., what operation to perform, or what to do; the address code is an address of an operation object at the time of execution of the operation code. When a computer executes a given instruction, it must first analyze what the opcode of the instruction is to determine the nature and method of operation, and then control other components of the computer to cooperatively perform the function of instruction expression. This analysis is done by the decoder.
Step 103, judging whether the instruction to be executed is a legal instruction, if yes, executing step 104, and if not, executing step 105.
Step 104, executing the target instruction normally.
In the embodiment of the invention, the step does not need to enter the subsequent steps.
Step 105, raise illegal instruction exception.
In the embodiment of the invention, if the instruction to be executed is not a legal instruction, the instruction to be executed is determined to be an illegal instruction, and the illegal instruction is caused to be abnormal.
At step 106, the CPU is trapped in kernel mode.
In the embodiment of the invention, the core trap simulation is that the result of the target instruction is calculated in the core state and then written into the target register to return, so that the CPU can sink into the core state after illegal instruction exception is caused.
In step 107, the CPU saves the register context to the kernel stack.
In the embodiment of the invention, the function of saving the register context to the kernel stack is to restore the execution state of the user state when the user state is returned later.
Step 108, entering an exception handling routine to determine that the exception is an illegal instruction exception.
Step 109, fetch illegal instruction that raised exception.
In step 110, the instruction simulation module in the kernel parses the illegal instruction.
And step 111, entering a simulation function corresponding to the illegal instruction according to the analysis result of the illegal instruction.
Step 112, the emulation function obtains the value of the register used by the illegal instruction from the kernel stack.
Step 113, the simulation function simulates the operation of the corresponding illegal instruction to obtain the operation result.
And 114, correspondingly operating according to the operation result.
In the embodiment of the invention, performing the corresponding operation may include modifying a register value stored in a kernel stack.
In step 115, the context of the register is restored from the kernel stack.
At step 116, the CPU returns to the user state.
As can be seen from fig. 1, the instruction is simulated by the CPU, so that the CPU can execute an illegal instruction. However, since the CPU belongs to a general-purpose circuit, although the CPU can simulate an instruction, the instruction simulation is not efficient.
Therefore, the embodiment of the invention provides an instruction execution method, through the cooperative simulation of the kernel and the coprocessor, the kernel is responsible for taking out the instruction to be simulated, and then the coprocessor simulates the instruction, so that the efficiency of instruction translation/instruction simulation is improved.
It should be noted that fig. 2-7 below each require that the register context be saved upon entering the kernel mode and restored before returning to the user mode.
As shown in fig. 2, a flowchart of an instruction execution method according to an embodiment of the present invention is provided, where the method includes the following steps:
in step 201, the cpu determines whether the current instruction is a legal instruction during the process of executing the instruction in the user mode, if yes, executes step 202, and if not, executes step 203.
In the embodiment of the invention, a large number of instructions in the binary program can be directly executed by the CPU under a possible condition, and only a few instructions cannot be directly executed. It should be noted that, the CPU executes the instruction when in the user state, and in the process of executing the instruction in the user state, the CPU needs to determine whether the current instruction is a legal instruction, if so, the current instruction is a legal instruction, and the CPU can directly execute the instruction. If not, illegal instruction exception is raised.
In step 202, the cpu normally executes the current instruction.
In the embodiment of the invention, if the current instruction is judged to be a legal instruction in the process of executing the instruction in the user mode, the CPU normally executes the current instruction without entering the subsequent steps.
In step 203, the cpu switches to kernel mode.
In the embodiment of the invention, because the current instruction is an illegal instruction, the illegal instruction is abnormal, and the CPU is required to be switched to the kernel mode at first, so that the illegal instruction can be simulated later.
In step 204, the cpu sends an illegal instruction to the coprocessor in kernel mode.
In the embodiment of the invention, the CPU is a general-purpose circuit, and the coprocessor is a special-purpose circuit which is specially used for simulating the illegal instruction, so that the efficiency of simulating the illegal instruction by using the coprocessor is higher than that of simulating the illegal instruction by using the CPU. In the kernel state, the CPU needs to send the illegal instruction to the coprocessor, so that the subsequent coprocessor can simulate and operate the illegal instruction conveniently.
And 205, performing simulation operation on the illegal instruction by the coprocessor to obtain a simulation result.
In the embodiment of the invention, the coprocessor performs simulation operation on the illegal instruction, so that a simulation result can be accurately obtained.
In step 206, the CPU obtains the simulation result from the coprocessor.
Step 207, switching to the user mode to continue executing the subsequent instruction.
In the embodiment of the invention, after the CPU obtains the simulation result in the kernel state, the CPU operates according to the correspondence of the simulation result, and then switches to the user state to continue executing the subsequent instruction. For example, if the simulation result obtained by the CPU in the kernel state is the read memory, the CPU reads the memory according to the simulation result. For another example, if the simulation result obtained by the CPU in the kernel state is to change the register value in the kernel stack, the CPU changes the register value in the kernel stack according to the simulation result. For another example, if the simulation result obtained by the CPU in the kernel state is a change control register, the CPU changes the control register according to the simulation result.
As can be seen from the above steps 201 to 207, since the CPU is a general-purpose circuit and the coprocessor is a special-purpose circuit dedicated to the emulation of an illegal instruction, the efficiency of the emulation of an illegal instruction using the coprocessor is higher than that of the emulation of an illegal instruction using the CPU. Through the cooperative simulation of the kernel and the coprocessor, the kernel is responsible for taking out instructions to be simulated, and then the coprocessor simulates the instructions, so that the efficiency of instruction translation/instruction simulation is improved.
In order to improve the efficiency of instruction translation/instruction emulation through the co-emulation of the kernel and the coprocessor, the CPU is required to switch the state to the kernel state first, then send an illegal instruction to the coprocessor in the kernel state, and then the coprocessor performs emulation operation on the illegal instruction. Thus, the simulation result can be obtained efficiently. The following describes how the coprocessor simulates illegal instructions to obtain simulation results.
As shown in fig. 3, a flowchart of a method for determining a simulation result according to an embodiment of the present invention is provided, where the method includes the following steps:
in step 301, the coprocessor analyzes the illegal instruction to obtain an analysis result.
In the embodiment of the invention, the coprocessor analyzes the illegal instruction to obtain an analysis result, wherein the analysis result comprises a register identifier corresponding to the illegal instruction.
In step 302, the cpu obtains the parsing result from the coprocessor.
In step 303, the cpu obtains register data corresponding to the analysis result.
In step 304, the CPU sends the register data to the coprocessor.
In step 305, the coprocessor performs simulation operation of illegal instructions through register data to obtain simulation results.
In the embodiment of the invention, the coprocessor performs simulation operation of illegal instructions through register data, thereby realizing obtaining simulation results.
As can be seen from steps 301 to 305, the coprocessor replaces the CPU to perform the emulation operation on the illegal instruction, so as to improve the efficiency of instruction translation/instruction emulation.
As shown in fig. 4, a flowchart of a method for executing an instruction according to an embodiment of the present invention includes the following steps:
in step 401, the cpu determines that the current instruction is an illegal instruction in the user mode, and switches to the kernel mode.
In the embodiment of the invention, when the CPU executes the instruction in the user mode, if the current instruction is determined to be an illegal instruction, the illegal instruction needs to be simulated, and the CPU needs to be switched from the user mode to the kernel mode first.
In step 402, the cpu sends an illegal instruction to the coprocessor in kernel mode.
In the embodiment of the invention, when the CPU is in the kernel state, the kernel state is the privilege state with the highest authority, and illegal instructions can be sent to the coprocessor.
In step 403, the coprocessor parses the illegal instruction, and the CPU switches to the user state and executes the first instruction.
In the embodiment of the invention, after the coprocessor receives the illegal instruction sent by the CPU, the illegal instruction is analyzed, and the CPU is switched to the user state and executes the first instruction when the coprocessor analyzes the illegal instruction, wherein the first instruction is any instruction or a plurality of instructions except the current instruction.
At step 404, the coprocessor sends a first interrupt to the CPU.
In the embodiment of the invention, after the coprocessor analyzes the illegal instruction, a first interrupt is sent to the CPU.
In step 405, the cpu switches to the kernel mode based on the first interrupt and enters the first interrupt handling routine.
In step 406, the cpu obtains the parsing result from the coprocessor.
In step 407, the cpu obtains register data corresponding to the analysis result according to the analysis result.
In step 408, the CPU sends the register data to the coprocessor.
In the embodiment of the invention, the CPU is switched from the user mode to the kernel mode based on the first interrupt, the register identification corresponding to the illegal instruction in the analysis result is obtained through the first interrupt processing routine, then the CPU obtains the register data corresponding to the illegal instruction according to the register identification corresponding to the illegal instruction, and the register data is sent to the coprocessor, so that the subsequent coprocessor can simulate the illegal instruction according to the register data.
In step 409, the coprocessor simulates illegal instructions, and the CPU switches to user mode and executes the second instructions.
In the embodiment of the invention, after the coprocessor receives the register data sent by the CPU, the illegal instruction is simulated, and the CPU is switched to the user state and executes the second instruction when the coprocessor simulates the illegal instruction, wherein the second instruction is any instruction or a plurality of instructions except the current instruction.
In step 410, the coprocessor sends a second interrupt to the CPU after obtaining the simulation result.
In step 411, the cpu switches to kernel mode based on the second interrupt, entering the second interrupt handling routine.
In step 412, the CPU obtains the simulation results from the coprocessor.
In step 413, after the cpu obtains the simulation result in the kernel mode, the cpu switches to the user mode to continue executing the subsequent instruction.
As can be seen from steps 401 to 413, since the CPU is a general-purpose circuit and the coprocessor is a circuit specially used for simulating instructions, the efficiency of instruction simulation can be improved by taking out illegal instructions from the CPU in the kernel state and then simulating the illegal instructions by the coprocessor.
In the embodiment of the present invention, in steps 403 and 409, the CPU is switched to the user mode to execute other instructions, and in fact, in the process of waiting for the coprocessor to parse/simulate the instructions, the computing resources of the CPU are allocated to other processes, so that the system can be prevented from being stopped for a long time. It can be seen that the CPUs in steps 401, 403, 405, 409, 411, 413 all perform context switching, i.e. from the kernel mode to the user mode or vice versa, in order to be able to give up the computing resources to other processes while waiting for the coprocessor to emulate an instruction, and have to introduce context switching. Since the computational resources are only available to other processes when the coprocessor is expected to parse/emulate instructions with a relatively long time, the overhead of context switching is higher than that incurred by the additional context switching, otherwise, the efficiency of instruction emulation is significantly reduced.
Optionally, because the speed of the coprocessor simulating the illegal instruction is high, if the method of fig. 4 is used, there is a problem of operation resource waste, so the embodiment of the invention provides another instruction execution method, and the method reduces the number of times of context switching and avoids operation resource waste by making the CPU in a state of waiting empty.
As shown in fig. 5, another method for executing an instruction according to an embodiment of the present invention includes the following steps:
step 501, determining that the current instruction is an illegal instruction, and switching to a kernel mode.
In step 502, the cpu sends an illegal instruction to the coprocessor in kernel mode.
In step 503, the coprocessor parses the illegal instruction, and the CPU performs idle waiting in the kernel state.
In the embodiment of the invention, because the speed of simulating the illegal instruction by the coprocessor is high, when the coprocessor analyzes the illegal instruction, the CPU does not switch from the kernel mode to the user mode, but performs idle waiting, does not perform context switching, and remains in the kernel mode to wait for the analysis result of the coprocessor.
At step 504, the coprocessor sends a first interrupt to the CPU.
In step 505, the cpu enters a first interrupt handling routine.
In the embodiment of the invention, the CPU is left in the kernel mode to execute the idle waiting, so that the CPU does not need to enter the kernel mode again before processing the interrupt; also, since no meaningful instructions are executed on the CPU during the idle waiting, the current state does not need to be saved before the interrupt is processed. Thus, no context switch is required before processing the interrupt.
In step 506, the cpu obtains the parsing result from the coprocessor.
In step 507, the cpu obtains register data corresponding to the analysis result according to the analysis result.
In step 508, the CPU sends the register data to the coprocessor.
In step 509, the coprocessor simulates an illegal instruction, and the CPU performs a wait empty in the kernel state.
In the embodiment of the invention, because the speed of simulating the illegal instruction by the coprocessor is high, when the coprocessor simulates the illegal instruction, the CPU does not switch from the kernel mode to the user mode, but performs idle waiting, does not perform context switching, and remains in the kernel mode to wait for the simulation result of the coprocessor.
Step 510, after obtaining the simulation result, the coprocessor sends a second interrupt to the CPU.
In step 511, the cpu enters a second interrupt processing routine.
In the embodiment of the invention, the CPU is left in the kernel mode to execute the idle waiting, so that the CPU does not need to enter the kernel mode again before processing the interrupt; also, since no meaningful instructions are executed on the CPU during the idle waiting, the current state does not need to be saved before the interrupt is processed. Thus, no context switch is required before processing the interrupt.
In step 512, the CPU obtains the simulation result from the coprocessor.
In step 513, after the cpu acquires the simulation result in the kernel mode, the cpu switches to the user mode to continue executing the subsequent instruction.
As can be seen from the above steps 501 to 513, by letting the CPU execute idle waiting when the coprocessor parses the illegal instruction and when the coprocessor simulates the illegal instruction, the CPU does not perform context switching, but remains in the kernel mode to wait for the processing result of the coprocessor, thereby avoiding the waste of operation resources caused by the context switching of the CPU, and further improving the efficiency of instruction translation/instruction simulation.
Optionally, since the time for the coprocessor to parse the illegal instruction is shorter in step 503 and the time for the coprocessor to simulate the illegal instruction is longer in step 509, in order to more fully utilize the operation resources, the embodiment of the present invention provides another method for executing the instruction, by making the CPU still wait for in step 503, but allowing for context switching in step 509 to allow the operation resources to be given to other programs, so that the operation resources can be more fully utilized.
As shown in fig. 6, another method for executing an instruction according to an embodiment of the present invention includes the following steps:
In step 601, the cpu switches to kernel mode if it is determined that the current instruction is an illegal instruction during the process of executing the instruction in user mode.
In step 602, the cpu sends an illegal instruction to the coprocessor in kernel mode.
In step 603, the coprocessor parses the illegal instruction, and the CPU executes a wait empty in the kernel state.
In the embodiment of the invention, the time for analyzing the illegal instruction by the coprocessor is short and the speed is high, so that the CPU is enabled to execute idle waiting in the kernel state, not execute the context switching and remain in the kernel state in order to avoid the problem of operation resource waste caused by the context switching of the CPU.
In step 604, the coprocessor sends a first interrupt to the CPU.
In step 605, the cpu enters a first interrupt handling routine.
In the embodiment of the invention, the CPU is left in the kernel mode to execute the idle waiting, so that the CPU does not need to enter the kernel mode again before processing the interrupt; also, since no meaningful instructions are executed on the CPU during the idle waiting, the current state does not need to be saved before the interrupt is processed. Thus, no context switch is required before processing the interrupt.
In step 606, the CPU obtains the parsing result from the coprocessor.
In step 607, the cpu obtains the register data corresponding to the analysis result according to the analysis result.
At step 608, the CPU sends the register data to the coprocessor.
In step 609, the coprocessor simulates an illegal instruction, and the CPU switches to the user mode and executes a third instruction.
In the embodiment of the invention, because the time for simulating the illegal instruction by the coprocessor is longer, in order to fully utilize the operation resources, when the CPU simulates the illegal instruction by the coprocessor, the CPU is switched to a user state and instructs a third instruction, wherein the third instruction is any one or more than one instruction except the current instruction, that is, the CPU performs context switching to enable the operation resources to be given to other programs, thereby fully utilizing the operation resources.
In step 610, the coprocessor sends a second interrupt to the CPU after obtaining the simulation result.
In step 611, the cpu switches to kernel mode based on the second interrupt, entering the second interrupt handling routine.
In the embodiment of the invention, the CPU is left in the kernel mode to execute the idle waiting, so that the CPU does not need to enter the kernel mode again before processing the interrupt; also, since no meaningful instructions are executed on the CPU during the idle waiting, the current state does not need to be saved before the interrupt is processed. Thus, no context switch is required before processing the interrupt.
In step 612, the CPU obtains the simulation results from the coprocessor.
In step 613, after the cpu obtains the simulation result in the kernel mode, the cpu switches to the user mode to continue executing the subsequent instruction.
As can be seen from the above steps 601 to 613, the CPU is kept waiting when the coprocessor parses the illegal instruction, so as to avoid the waste of operation resources caused by the context switch of the CPU, and further improve the efficiency of instruction translation/instruction simulation. When the coprocessor simulates illegal instructions, context switching is performed so as to obtain the operation resources to other programs, thereby realizing full utilization of the operation resources.
Optionally, the internal register of the coprocessor can be directly accessed by the CPU, so that the CPU can actively access the internal register of the coprocessor to judge whether the coprocessor finishes processing when waiting for the processing result of the coprocessor, and the coprocessor is not required to send an interrupt to the CPU.
As shown in fig. 7, a flowchart of a method for executing another instruction according to an embodiment of the present invention includes the following steps:
step 701, determining that the current instruction is an illegal instruction, and switching to a kernel mode.
In step 702, the cpu sends an illegal instruction to the coprocessor in kernel mode.
In step 703, the coprocessor parses the illegal instruction, and the CPU performs a first busy-wait in the kernel state.
In the embodiment of the invention, in the process of analyzing illegal instructions by the coprocessor, the CPU executes first busy-wait in a kernel mode, wherein the first busy-wait is used for the CPU to repeatedly inquire whether an analysis result exists in a register of the coprocessor.
In step 704, the cpu obtains the analysis result from the coprocessor based on the first busy-wait.
In the embodiment of the invention, the CPU is based on the first busy waiting, so that the analysis result is obtained from the coprocessor, then the CPU obtains the register data corresponding to the illegal instruction according to the register identification corresponding to the illegal instruction in the analysis result, and then the register data is sent to the coprocessor, thereby facilitating the subsequent coprocessor to simulate the illegal instruction according to the register data.
Step 705, the cpu obtains register data corresponding to the analysis result according to the analysis result.
In step 706, the CPU sends the register data to the coprocessor.
In step 707, the coprocessor emulates an illegal instruction and the CPU performs a second busy-wait.
In the embodiment of the invention, in the process of simulating illegal instructions by the coprocessor, the CPU executes a second busy-wait in a kernel mode, wherein the second busy-wait is used for the CPU to repeatedly inquire whether a simulation result exists in a register of the coprocessor.
At step 708, the CPU obtains the simulation result from the coprocessor through the second busy-wait.
In step 709, after the cpu obtains the simulation result in the kernel mode, the cpu switches to the user mode to continue executing the subsequent instruction.
As can be seen from the foregoing steps 701 to 709, by executing the first busy wait by the CPU, the CPU may obtain the analysis result from the coprocessor by repeatedly querying the register of the coprocessor, thereby sending an interrupt to the CPU instead of the coprocessor to obtain the analysis result, and reducing the number of interrupt times. The CPU can acquire the simulation result from the coprocessor by repeatedly inquiring the register of the coprocessor by executing the second busy waiting, so that the CPU is replaced by the coprocessor to acquire the simulation result and then send an interrupt to the CPU, and the interrupt times are reduced.
In one possible case, the CPU realizes the shadow register, so that when the CPU enters the kernel mode, the kernel can directly switch to another group of shadow registers without saving the register context, thereby avoiding the overhigh cost of context switching and improving the efficiency of instruction translation/instruction simulation. It should be noted that using shadow registers does not eliminate the need for context switching, but only reduces overhead. In this case, when the coprocessor can complete instruction emulation by directly reading and writing the shadow register, the efficiency of instruction emulation can be further improved. The coprocessor may be an internal coprocessor, an external coprocessor, or other coprocessors, and is not limited herein.
As shown in fig. 8, a flowchart of a method for executing an instruction according to an embodiment of the present invention includes the following steps:
In step 801, when the cpu determines that the current instruction is an illegal instruction during the process of executing the instruction in the user mode, the cpu switches to the kernel mode.
Step 802, a CPU sends illegal instructions to a coprocessor in a kernel mode;
in step 803, the coprocessor parses the illegal instruction, and the CPU switches to the user mode and executes the third instruction.
In the embodiment of the invention, after the coprocessor receives the illegal instruction sent by the CPU, the illegal instruction is analyzed, and the CPU is switched to the user state and executes the third instruction when the coprocessor analyzes the illegal instruction, wherein the third instruction is any instruction or a plurality of instructions except the current instruction.
In step 804, the coprocessor determines the identity of the shadow register according to the analysis result, and reads the corresponding shadow register data.
In step 805, the coprocessor emulates an illegal instruction.
In the embodiment of the invention, if the illegal instruction has an association relation with the shadow register, the coprocessor can read and write the register, and then the simulation result is written into the shadow register. If there is a partial simulation result that requires kernel processing, it is provided to the kernel in a subsequent step.
At step 806, the coprocessor sends a third interrupt to the CPU.
In step 807, the cpu switches to the kernel mode based on the third interrupt, and enters a third interrupt processing routine.
In step 808, the CPU obtains the simulation results from the coprocessor.
Step 809, after the cpu obtains the simulation result in the kernel mode, it switches to the user mode to continue executing the subsequent instruction.
Through the steps 801 to 809, it can be seen that, when the CPU is in idle waiting state and is combined with the shadow register, not only can context switch be performed during simulation of an illegal instruction by the coprocessor to obtain the operational resource for other programs, thereby realizing full utilization of the operational resource, but also further reducing the workload of the kernel and improving the efficiency of instruction translation/instruction simulation.
As shown in fig. 9, a flowchart of a method for executing an instruction according to an embodiment of the present invention includes the following steps:
In step 901, when the cpu executes the instruction in the user mode, if it is determined that the current instruction is an illegal instruction, the cpu switches to the kernel mode.
Step 902, the CPU sends illegal instructions to the coprocessor in a kernel mode;
In step 903, the coprocessor parses the illegal instruction, and the CPU performs idle waiting in the kernel state.
And step 904, the coprocessor determines the identification of the shadow register according to the analysis result and reads the corresponding shadow register data.
In step 905, the coprocessor emulates an illegal instruction.
Step 906, after obtaining the simulation result, sending a third interrupt to the CPU.
In step 907, the cpu enters a third interrupt processing routine.
In step 908, the CPU obtains the simulation results from the coprocessor.
In step 909, after the cpu acquires the simulation result in the kernel mode, the cpu switches to the user mode to continue executing the subsequent instruction.
As can be seen from the above steps 901 to 909, if the CPU is in a state of waiting for a shadow register, and if the coprocessor simulates an illegal instruction faster, if the method of fig. 8 is used, the benefit of letting the computing resource to other processes is lower by context switching, there is a problem of computing resource waste, so that by making the CPU in a state of waiting for a CPU, the number of times of context switching is reduced, and the computing resource waste is avoided.
As shown in fig. 10, a flowchart of a method for executing an instruction according to an embodiment of the present invention includes the following steps:
In step 1001, the cpu switches to the kernel mode if it is determined that the current instruction is an illegal instruction during the process of executing the instruction in the user mode.
Step 1002, a CPU sends an illegal instruction to a coprocessor in a kernel mode;
In step 1003, the coprocessor parses the illegal instruction, and the CPU executes the third busy-wait in the kernel state.
In step 1004, the coprocessor determines the identity of the shadow register according to the analysis result, and reads the corresponding shadow register data.
In step 1005, the coprocessor emulates an illegal instruction.
At step 1006, the cpu obtains the simulation result from the coprocessor based on the third busy-wait.
In step 1007, after the cpu obtains the simulation result in the kernel mode, the cpu switches to the user mode to continue executing the subsequent instruction.
As can be seen from the above steps 1001 to 1007, the CPU is in busy-wait and shadow registers are combined, so that the CPU can obtain the simulation result from the coprocessor by repeatedly querying the registers of the coprocessor, thereby sending an interrupt to the CPU after obtaining the simulation result instead of the coprocessor. Further reduction of kernel workload can be achieved, and efficiency of instruction translation/instruction simulation is improved.
In the embodiment of the invention, the instruction execution method can be divided into two stages, wherein the first stage is that the CPU sends the illegal instruction to the coprocessor, the coprocessor receives the illegal instruction and analyzes the illegal instruction, the analysis result of the illegal instruction is sent to the CPU, and the CPU sends the register data corresponding to the analysis result to the coprocessor. The second stage is that the coprocessor simulates illegal instructions according to register data, and sends simulation results of the illegal instructions to the CPU, then the CPU obtains the simulation results, and then the CPU switches to a user state to continuously execute subsequent instructions after obtaining the simulation results in a kernel state.
Referring to fig. 4, 5 and 7, steps 401 to 408 in fig. 4 correspond to the first stage of the instruction execution method. Steps 409 to 413 in fig. 4 correspond to the second stage of the instruction execution method. Steps 501 to 508 in fig. 5 correspond to the first stage of the instruction execution method. Steps 509 through 513 in fig. 5 correspond to the second stage of the instruction execution method. Steps 701 to 706 in fig. 7 correspond to the first stage of the instruction execution method. Steps 706 to 709 correspond to the second stage of the instruction execution method.
Alternatively, the first stage of the instruction execution method corresponding to fig. 4 may be arbitrarily combined with the second stage of the instruction execution method corresponding to fig. 5 or fig. 7. Similarly, the first stage of the instruction execution method of fig. 5 may be arbitrarily combined with the second stage of the corresponding instruction execution method of fig. 4 or 7. Similarly, the first stage of the instruction execution method of fig. 7 may be arbitrarily combined with the second stage of the corresponding instruction execution method of fig. 4 or 5.
Alternatively, the coprocessor may also be connected to other CPU components for other types of acceleration. For example, if the coprocessor is connected to the MMU, the virtual address may be translated directly via the MMU, eliminating the need for a page table or MMU to be walked by the kernel (note that for software-managed TLB architectures, such as MIPS, the MMU itself can only walk the TLB, whether the TLB is to be filled by the kernel upon a TLB miss). This feature allows the instruction emulation coprocessor to also be used for the translation acceleration instruction set extensions described above, as follows: it is first necessary to define an instruction set extension whose purpose is to accelerate virtual address translation using the coprocessor, then the CPU itself cannot decode the instructions, and when executing the instructions, it will enter the core trap flow in fig. 2-7, then execute the core trap flow, then the coprocessor queries the MMU, then the coprocessor sends the translated instructions to the CPU, then the CPU obtains the translated instructions from the core state, and then the CPU switches to the user state.
Alternatively, early streaming may have been performed during the CPU development process. After this, it is sometimes necessary to add new instructions, but before the new instructions are streamed again, a lot of testing is still required for the new instructions. However, if the FPGA is fully used for software local testing, its clock frequency can greatly impact efficiency. If old instructions can be dispatched to the CPU of the streamed slice, a significant efficiency improvement can be achieved by dispatching only new instructions to the FPGA.
Optionally, a module required by a new instruction which is stripped from a new version of an unfused slice can be programmed into the FPGA or directly programmed in whole, and a program which can be used as a coprocessor is run on the module. In this way, the old instruction still runs on the CPU of the streamed chip, and as the clock frequency can be several times to tens of times faster than the FPGA, quite high overall performance can be obtained; the new instruction is transferred to the CPU programmed on the FPGA to run, so that the debugging is very convenient. In this case, the execution of the new instruction on the FPGA is not necessarily faster than the kernel simulation directly. However, since this simulation is not a speed concern, but rather the correctness of the new circuit design under software local testing, the overall speed of the entire test platform is faster than that of a pure FPGA test platform.
Based on the same technical concept, an embodiment of the present application provides a schematic structural diagram of an instruction execution device, as shown in fig. 11, the device 1100 includes: the first processing unit 1101 switches to a kernel mode if it is determined that the current instruction is an illegal instruction in the process of executing the instruction in the user mode; the first processing unit 1101 sends an illegal instruction to the second processing unit 1102 in the kernel mode; the second processing unit 1102 performs simulation operation on the illegal instruction to obtain a simulation result; after the first processing unit 1101 obtains the simulation result in the kernel mode, it switches to the user mode to continue executing the subsequent instruction.
Optionally, the second processing unit 1102 analyzes the illegal instruction to obtain an analysis result; the first processing unit 1101 obtains register data corresponding to the analysis result in the kernel mode, and sends the register data to the second processing unit 1102; the second processing unit 1102 performs simulation operation of illegal instructions through the register data, and obtains a simulation result.
Optionally, in the process of the second processing unit 1102 analyzing the illegal instruction, the first processing unit 1101 switches to the user state and executes the first instruction; the first instruction is any one or more instructions except the current instruction; the second processing unit 1102 sends a first interrupt to the first processing unit 1101; the first processing unit 1101 switches to the kernel mode based on the first interrupt.
Optionally, during the process of resolving the illegal instruction by the second processing unit 1102, the first processing unit 1101 performs a wait in kernel mode; the second processing unit 1102 sends a first interrupt to the first processing unit 1101.
Optionally, during the process of resolving the illegal instruction by the second processing unit 1102, the first processing unit 1101 performs a first busy-wait in the kernel state, where the first busy-wait is used to obtain the resolving result from the co-processing.
Optionally, during the process of simulating the illegal instruction by the second processing unit 1102, the first processing unit 1101 switches to the user state and executes the second instruction; the second instruction is any one or more instructions except the current instruction; the second processing unit 1102 sends a second interrupt to the first processing unit 1101; the first processing unit 1101 switches to the kernel mode based on the second interrupt.
Optionally, during the simulation of the illegal instruction by the second processing unit 1102, the first processing unit 1101 performs a wait in kernel mode; the second processing unit 1102 sends a second interrupt to the first processing unit 1101.
Optionally, during the process of the second processing unit 1102 simulating the illegal instruction, the first processing unit 1101 performs a second busy-wait in the kernel mode, and the second busy-wait is used to obtain a simulation result from the second processing unit 1102.
Alternatively, the first interrupt and the second interrupt are implemented by the same vectorized interrupt, or by different vectorized interrupts.
Optionally, the second processing unit 1102 analyzes the illegal instruction to obtain an analysis result; the second processing unit 1102 obtains shadow register data corresponding to the analysis result according to the analysis result; the second processing unit 1102 performs simulation operation of illegal instructions through the shadow register data to obtain a simulation result.
Optionally, if the illegal instruction has an association relationship with the shadow register, writing the simulation result into the shadow register.
Based on the same technical concept, the embodiment of the present application further provides a computer device, as shown in fig. 12, where the computer device 1200 includes at least one processor 1201 and a memory 1202 connected to the at least one processor, and in the embodiment of the present application, a specific connection medium between the processor 1201 and the memory 1202 is not limited, and in fig. 12, the processor 1201 and the memory 1202 are connected by a bus, for example. The buses may be divided into address buses, data buses, control buses, etc.
In the embodiment of the present application, the memory 1202 stores instructions executable by the at least one processor 1201, and the at least one processor 1201 can execute the steps included in the foregoing instruction execution method by executing the instructions stored in the memory 1202.
Where the processor 1201 is a control center of a computing device, various interfaces and lines may be used to connect various portions of the computing device, through execution or execution of instructions stored in the memory 1202, and invocation of data stored in the memory 1202, to effect data processing. Alternatively, the processor 1201 may include one or more processing units, and the processor 1201 may integrate an application processor and a modem processor, wherein the application processor primarily processes operating systems, user interfaces, application programs, etc., and the modem processor primarily processes issuing instructions. It will be appreciated that the modem processor described above may not be integrated into the processor 1201. In some embodiments, processor 1201 and memory 1202 may be implemented on the same chip, or they may be implemented separately on separate chips in some embodiments.
The processor 1201 may be a general purpose processor such as a Central Processing Unit (CPU), digital signal processor, application SPECIFIC INTEGRATED Circuit (ASIC), field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or may implement or perform the methods, steps, and logic blocks disclosed in embodiments of the application. The general purpose processor may be a microprocessor or any conventional processor or the like. The steps of a method disclosed in connection with an embodiment of a method of obtaining data may be embodied directly in a hardware processor for execution, or in a combination of hardware and software modules in a processor for execution.
Memory 1202 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. The Memory 1202 may include at least one type of storage medium, which may include, for example, flash Memory, hard disk, multimedia card, card Memory, random access Memory (Random Access Memory, RAM), static random access Memory (Static Random Access Memory, SRAM), programmable Read-Only Memory (Programmable Read Only Memory, PROM), read-Only Memory (ROM), charged erasable programmable Read-Only Memory (ELECTRICALLY ERASABLE PROGRAMMABLE READ-Only Memory, EEPROM), magnetic Memory, magnetic disk, optical disk, and the like. Memory 1202 is any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited to such. The memory 1202 in embodiments of the present application may also be circuitry or any other device capable of performing storage functions for storing program instructions and/or data.
Based on the same technical idea, the embodiments of the present application further provide a computer-readable storage medium storing a computer program executable by a computing device, which when run on an electronic device causes the electronic device to execute the steps of the above-described instruction execution method.
Based on the same technical idea, an embodiment of the present application also provides a computer program product, which comprises a computer program/instruction, which when executed by a processor, implements the steps of the instruction execution method.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (10)
1. A method of executing instructions, comprising:
In the process of executing the instruction in the user mode, if the CPU determines that the current instruction is an illegal instruction, switching to the kernel mode;
The CPU sends the illegal instruction to a coprocessor in a kernel state;
The coprocessor analyzes the illegal instruction to obtain an analysis result;
The coprocessor acquires shadow register data corresponding to the analysis result according to the analysis result;
The coprocessor performs simulation operation of the illegal instruction through the shadow register data to obtain a simulation result;
And after the CPU acquires the simulation result in the kernel mode, switching to the user mode to continuously execute the subsequent instruction.
2. The method as recited in claim 1, further comprising:
in the process of analyzing and simulating the illegal instruction by the coprocessor, the CPU is switched to a user state and executes a third instruction; the third instruction is any one or more instructions other than the current instruction.
3. The method as recited in claim 1, further comprising:
and in the process of analyzing and simulating the illegal instruction by the coprocessor, the CPU executes idle waiting in a kernel state.
4. The method as recited in claim 1, further comprising:
And in the process of analyzing and simulating the illegal instruction by the coprocessor, the CPU executes a third busy-wait in a kernel state, wherein the third busy-wait is used for obtaining an analysis result from the coprocessor and a simulation result from the coprocessor.
5. A method according to claim 2 or 3, wherein after said obtaining the simulation result, said CPU further comprises, before obtaining the simulation result in a kernel state:
the coprocessor sends a third interrupt to the CPU;
The CPU switches to kernel mode based on a third interrupt.
6. The method of claim 1, wherein the coprocessor performs the simulated operation of the illegal instruction through the shadow register data, and further comprises, after obtaining a simulation result:
and if the illegal instruction has an association relation with the shadow register, writing the simulation result into the shadow register.
7. An instruction execution device, comprising:
the first processing unit is switched to a kernel mode if determining that the current instruction is an illegal instruction in the process of executing the instruction in the user mode;
The first processing unit sends the illegal instruction to a second processing unit in a kernel mode; the second processing unit analyzes the illegal instruction to obtain an analysis result; the second processing unit obtains shadow register data corresponding to the analysis result according to the analysis result; the second processing unit performs simulation operation of the illegal instruction through the shadow register data to obtain a simulation result; and after the first processing unit acquires the simulation result in the kernel mode, switching to the user mode to continuously execute the subsequent instruction.
8. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method of any of claims 1-6 when the program is executed.
9. A computer readable storage medium, characterized in that it stores a computer program executable by a computer device, which program, when run on the computer device, causes the computer device to perform the steps of the method according to any one of claims 1-6.
10. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the steps of the method of claim 1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410236005.8A CN118069222A (en) | 2023-10-24 | 2023-10-24 | Instruction execution method and device |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410236005.8A CN118069222A (en) | 2023-10-24 | 2023-10-24 | Instruction execution method and device |
CN202311378936.3A CN117112029B (en) | 2023-10-24 | 2023-10-24 | Instruction execution method and device |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311378936.3A Division CN117112029B (en) | 2023-10-24 | 2023-10-24 | Instruction execution method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN118069222A true CN118069222A (en) | 2024-05-24 |
Family
ID=88800585
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410236005.8A Pending CN118069222A (en) | 2023-10-24 | 2023-10-24 | Instruction execution method and device |
CN202311378936.3A Active CN117112029B (en) | 2023-10-24 | 2023-10-24 | Instruction execution method and device |
Family Applications After (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311378936.3A Active CN117112029B (en) | 2023-10-24 | 2023-10-24 | Instruction execution method and device |
Country Status (1)
Country | Link |
---|---|
CN (2) | CN118069222A (en) |
Family Cites Families (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7234144B2 (en) * | 2002-01-04 | 2007-06-19 | Microsoft Corporation | Methods and system for managing computational resources of a coprocessor in a computing system |
CN101387969B (en) * | 2008-10-16 | 2011-04-13 | 上海交通大学 | Dynamic binary translation method for cooperation design of software and hardware |
US8239620B2 (en) * | 2010-09-27 | 2012-08-07 | Mips Technologies, Inc. | Microprocessor with dual-level address translation |
US8789042B2 (en) * | 2010-09-27 | 2014-07-22 | Mips Technologies, Inc. | Microprocessor system for virtual machine execution |
US9405549B2 (en) * | 2011-03-04 | 2016-08-02 | Nec Corporation | Deadlock avoidance method and deadlock avoidance mechanism |
CN106325819B (en) * | 2015-06-17 | 2019-08-02 | 华为技术有限公司 | Computer instruction processing method, coprocessor and system |
US11055401B2 (en) * | 2017-09-29 | 2021-07-06 | Intel Corporation | Technologies for untrusted code execution with processor sandbox support |
WO2021217529A1 (en) * | 2020-04-29 | 2021-11-04 | 华为技术有限公司 | Method and system for inter-process communication |
CN114077379B (en) * | 2020-08-19 | 2024-03-26 | 华为技术有限公司 | Computer equipment, exception handling method and interrupt handling method |
CN114721719B (en) * | 2022-04-20 | 2023-08-01 | 上海道客网络科技有限公司 | Method and system for containerized deployment of heterogeneous applications in cluster |
CN116302549A (en) * | 2023-03-24 | 2023-06-23 | 郑州云海信息技术有限公司 | Hardware resource acquisition method, device, equipment and medium of AI video accelerator |
CN116866449A (en) * | 2023-07-08 | 2023-10-10 | 浙江大学 | TOE acceleration system for improving application network performance |
-
2023
- 2023-10-24 CN CN202410236005.8A patent/CN118069222A/en active Pending
- 2023-10-24 CN CN202311378936.3A patent/CN117112029B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN117112029B (en) | 2024-03-12 |
CN117112029A (en) | 2023-11-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US5838978A (en) | System and method of using annotations to optimize dynamically translated code in the presence of signals | |
JP3753743B2 (en) | Method and apparatus for memory data aliasing in advanced processors | |
US11604643B2 (en) | System for executing new instructions and method for executing new instructions | |
US7809547B2 (en) | Host computer system emulating target system legacy software and providing for incorporating more powerful application program elements into the flow of the legacy software | |
US11803387B2 (en) | System for executing new instructions and method for executing new instructions | |
US9785506B2 (en) | Reducing likelihood of concurrency error in virtualized computing environment | |
US7823140B2 (en) | Java bytecode translation method and Java interpreter performing the same | |
JPH0782441B2 (en) | Simulation method | |
JP3776132B2 (en) | Microprocessor improvements | |
US11625247B2 (en) | System for executing new instructions and method for executing new instructions | |
US20210240467A1 (en) | Method for Implementing Function Jump, Apparatus, and Computer Storage Medium | |
KR20110050230A (en) | Virtual system and instruction executing method thereof | |
WO2022237098A1 (en) | Heterogeneous program execution method and apparatus, and computing device and readable storage medium | |
US10949207B2 (en) | Processor core supporting a heterogeneous system instruction set architecture | |
CN115543547B (en) | Migration method and system for virtual machine in heterogeneous virtualization platform | |
CN107368739B (en) | Kernel drive monitoring method and device | |
CN117573419B (en) | Page exception handling method and device | |
US11188361B1 (en) | Cross-architecture software distribution using a virtual instruction set architecture | |
CN117112029B (en) | Instruction execution method and device | |
Wang et al. | Dynamic translation optimization method based on static pre-translation | |
US20050091022A1 (en) | Ultra fast multi-processor system simulation using dedicated virtual machines | |
US10133655B1 (en) | Emulation of target system using JIT compiler and bypassing translation of selected target code blocks | |
CN113849397A (en) | Execution engine, virtual machine, related apparatus and related methods | |
Hindborg | Accelerating Instruction Set Emulation using Reconfigurable Hardware and Trace Based Optimization | |
Kurikka | Testing embedded software in a simulated environment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |