KR20120070326A - A apparatus and a method for virtualizing memory - Google Patents

Abstract

PURPOSE: An apparatus and a method for virtualizing a memory are provided to improve the performance of a virtualizing system by rapidly mapping memories in order to change page tables of a guest operating system in the virtualizing system. CONSTITUTION: Shadow page tables(62) are generated respectively for guest operating systems. A guest page table is provided from one of the guest operating systems as the reference of a physical memory page. A processor processes mapping information which maps one of the shadow page tables and the guest page table. The mapping information is machine address information in which the guest operating system is loaded and is the virtual address information of the guest page table.

Description

Memory apparatus and method {A apparatus and a method for virtualizing memory}

The present invention relates to a memory virtualization method using a shadow page table in a virtualization system supporting a plurality of guest operating systems.

Virtualization technology is a technology that virtualizes input / output devices such as CPU, memory, network, and monitor, which are hardware resources used by an operating system (OS). In general, operating systems are designed to directly access and control hardware, but virtualized systems prevent multiple operating systems from directly accessing hardware and provide a virtual hardware interface to the operating system, allowing multiple operating systems to run on a single piece of hardware. . Virtualization technology began in the first server segment, and has been applied to desktops and, more recently, embedded hair devices such as mobile phones and handheld terminals.

Virtualization technology can be classified into CPU virtualization, memory virtualization, disk virtualization, and virtualization of I / O devices. In the case of CPUs, CPU architectures generally support more than one mode.

1 and 2 are schematic diagrams illustrating a privileged mode, an unprivileged mode, a privileged mode of a virtualization system, and a nonprivileged mode of a non-virtualized system, respectively. The system differs at the privileged level.

In the non-virtualization system shown in FIG. 1, the operating system 14 is run in privileged mode 12, and the general application 13 is run in non-privileged mode 11. In the privileged mode 12, all instructions provided by the CPU (not shown) may be used, and in the non-privileged mode 11, only the remaining instructions except the instructions that can be executed in the privileged mode 12 may be executed in the CPU. . The operating system generally operates in privileged mode, which acquires the authority to control the application program.

In contrast, in the virtualization system shown in FIG. 2, the operating system 24 operates in the non-privileged mode 21. That is, the privileged mode 22 is assigned to the virtual machine monitor (VMN) 25 because the operating system 24 can not operate the hardware directly to operate a plurality of virtual machines. In addition, the operating system 24 and the application program 23 virtualize the CPU to be executed in the non-privileged mode 21.

In the case of memory, in a non-virtualized system, the operating system manages all the physical memory and allocates it to applications. To do this, the operating system maps the physical memory to the virtual memory space used by the application and prevents memory interference by granting different access rights to the memory area.

In the virtualization system, the virtual machine monitor performs the memory mapping and memory access control that the operating system manages. The virtual machine monitor refers to the guest page table (GPT), which is memory mapping information provided by the operating system, and transforms it appropriately to generate a shadow page table (SPT) used for actual physical memory mapping. . The shadow page table contains memory mapping information and access rights options by memory area.

In full-virtualization, the virtual machine monitor cannot see information such as data structures maintained by the operating system, but only when the operating system asks the hardware to execute a particular instruction. In other words, the operating system manages all the page tables of each process that the operating system runs, but the virtual machine monitor can only see the page table currently selected by the operating system. Based on the information, a method of recreating the shadow page each time may be used.

Since the change of the guest page table occurs every time the application process is switched, the guest page table can be changed frequently. Since the shadow pages must be regenerated frequently each time, the virtual system performance is greatly reduced. There is a problem.

In order to solve this problem, an object of the present invention is to provide a memory virtualization method for improving performance of a virtualization system through efficient management of a shadow page table used for memory virtualization in a virtualization system.

Memory virtual method as an aspect of the present invention for solving the above technical problem, in the memory virtualization method in a virtualization system supporting a plurality of guest operating systems, by generating a respective shadow page table for each of the plurality of guest operating systems Caching; Mapping a guest page table for reference of a physical memory page from one of the cached shadow page tables and one of the plurality of guest operating systems using mapping information; It may include.

The mapping may include using the machine address location information loaded by the guest operating system and the virtual address location information of the guest page table as the mapping information.

In the caching operation, location information of the shadow page table may be cached as a list for each guest operating system, and caching may be maintained even when execution of the guest operating system is stopped.

The memory virtualization method may further include: determining, by the virtual machine monitor of the virtualization system, a caching frequency of the shadow page table for each guest operating system, and dynamically changing a caching space of the shadow page table according to the caching frequency. It may also be included.

The memory virtualization method may form the guest page table as a primary page table and a secondary page table, and when the secondary page table is shared, the shadow page table may be cached in consideration of the sharing relationship.

Here, the virtual machine monitor of the virtualization system may manage the list of the secondary page table shared to identify the sharing relationship between the secondary page table by referring to the list when caching the shadow page table. You can also identify the sharing relationship of secondary page tables.

On the other hand, in another aspect of the present invention, a memory virtualization apparatus using a shadow page table in a virtualization system supporting a plurality of guest operating systems, the shadow page table generated for each of the plurality of guest operating systems; A guest page table provided for reference of a physical memory page from one of the plurality of guest operating systems; And mapping information for mapping one of the cached shadow page tables and the guest page table.

The present invention has the following effects.

First, the processing overhead of the virtualization system can be minimized by caching and reusing shadow page tables.

Second, in the virtualization system operating multiple operating systems at the same time, memory mapping can be performed quickly to change the page table of the guest operating system to improve the performance of the virtualization system.

Third, the processing speed of the application can be improved by reducing the memory mapping burden due to switching of frequently occurring applications.

1 is a block diagram illustrating a privileged mode and a non-privileged mode of a general non-virtualization system.
2 is a diagram illustrating a privileged mode and a non-privileged mode of a general virtualization system.
3 is a diagram illustrating memory mapping using a page table according to an embodiment of the present invention.
4 is a diagram illustrating memory map recognition of a virtual machine monitor and a guest operating system according to an exemplary embodiment of the present invention.
5 is a diagram illustrating a memory structure of a virtualization system according to an embodiment of the present invention.
6 is a diagram illustrating shadow page table caching in the memory virtualization method according to an embodiment of the present invention.
FIG. 7 illustrates a case in which secondary page tables are shared in a memory virtualization method according to an embodiment of the present invention.
8 is a flowchart illustrating a memory virtualization method according to an embodiment of the present invention.

The following embodiments are a combination of elements and features of the present invention in a predetermined form. Each component or feature may be considered to be optional unless otherwise stated. Each component or feature may be implemented in a form that is not combined with other components or features. In addition, some of the elements and / or features may be combined to form an embodiment of the present invention. The order of the operations described in the embodiments of the present invention may be changed. Some configurations or features of certain embodiments may be included in other embodiments, or may be replaced with corresponding configurations or features of other embodiments.

Embodiments of the invention may be implemented through various means. For example, embodiments of the present invention may be implemented by hardware, firmware, software, or a combination thereof.

For a hardware implementation, the method according to embodiments of the present invention may be implemented in one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs) , Field programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, and the like.

In the case of an implementation by firmware or software, the method according to embodiments of the present invention may be implemented in the form of a module, a procedure or a function for performing the functions or operations described above. The software code may be stored in a memory unit and driven by a processor. The memory unit may be located inside or outside the processor, and may exchange data with the processor by various known means.

Throughout the specification, when a part is "connected" with another part, this includes not only a case where the part is directly connected, but a case where the part is electrically connected with another element in between. In addition, when a part includes a certain component, this means that unless otherwise stated, it may include other components other than excluding other components.

In addition, the term module described herein refers to a unit for processing a specific function or operation, which may be implemented in hardware or software, or a combination of hardware and software.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are intended to provide further explanation of the invention, and are not intended to limit the scope of the invention.

The present invention is a memory virtualization method for effectively managing a shadow page table used for memory virtualization in a virtualization system.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

3 is a diagram illustrating memory mapping using a page table according to an embodiment of the present invention.

The shadow page table used in the memory virtualization method according to the embodiment of the present invention is generated according to the page table. First, the mapping according to the page table will be described with reference to FIG. 3. 3 is for the ARM architecture, which is one of a number of CPU architectures, but the invention is not limited thereto. Other architectures have some differences, but the key role of the page table is the same. In other words, this function supports dynamic memory mapping so that relatively small physical memory can be used in a relatively large virtual memory space of a process, and gives a different access permission for each memory area to prevent memory interference between processes. The ARM architecture also provides a number of page mapping methods. FIG. 3 is an example of the most commonly used page table.

The first page table 31 or the first page table 31 and the second page table 32 are used to map page frames 35 and 36 which are physical memories.

First, when the primary page table 31 is used, the primary page table 31 having a size of 16 KB is configured. Since each entry 33 of the primary page table 31 has a size of 4 bytes, the primary page table 31 having a size of 16 KB has 4096 entries, and each entry 33 is a mapped physical memory. It points to the page frame 35 of 1MB.

One page table 31 is allocated per process, and each process can map 4 GB of virtual memory space. This is the case with 32bit machines as an example.

On the other hand, when the primary page table 31 and the secondary page table 32 are used, the entry 33 of the primary page table 31 indicates the secondary page table 32 having a size of 1 KB. The secondary page table 32 has 256 entries 34, each entry pointing to a page frame 36 of size 4KB. Since 256 4KB page frames 36 are 1MB, the primary page table entry 33, which has gone through the secondary page table 32, also points to a 1MB page frame.

When the secondary page table 32 is used, it is possible to indicate a page frame, which is distributed 4 KB physical memory even if not 1 MB of contiguous memory, and the page access authority can be set for each page frame, so that it can be set in 4 KB units. Access rights can be set differently.

4 is a diagram illustrating memory map recognition of a virtual machine monitor and a guest operating system according to an exemplary embodiment of the present invention.

Referring to this, the guest operating system 1, the guest operating system 2, and the virtual machine monitor recognize the memory map differently.

The difference occurs because in the virtualization system, the location of the physical memory known to the guest operating system at the time of compilation and the physical memory location to be used by the guest operating system are different.

The virtual machine monitor compensates for this difference by creating and using a shadow page table that changes the guest page table created by the guest operating system to the memory location where the operating system is actually loaded.

Referring to the embodiment of the present invention shown in Figure 4 as follows.

The virtual machine monitor recognizes the location of the real physical memory 41 into which the virtual machine monitor, guest operating system 1, and guest operating system 2 are loaded.

However, the guest operating system 1 recognizes the use of the physical memory 42 from 0x50000000 to 0x51000000, and the guest operating system 2 also recognizes the use of the physical memory 43 from 0x50000000 to 0x51000000.

To compensate for this difference, the virtual machine monitor provides a shadow page table that allows the operating system to map the actual physical memory 41 allocated to it.

FIG. 5 is a diagram illustrating a memory structure of a virtualization system according to an exemplary embodiment of the present invention. Referring to this, a mapping relationship between a virtual memory and a physical memory will be described.

The memory address in the virtualization system is divided into a virtual address (51), a pseudo-physical address (52), and a machine address (Machine Address, 53). The virtual address 51 is a virtual memory area recognized by the virtual machine, the pseudo virtual address 52 is a physical memory area recognized by the virtual machine, and the machine address 53 is an actual physical memory area seen by the virtual machine monitor. do.

When creating a shadow page table, the virtual machine monitor gives different access rights to each memory area as described above to prevent memory interference between processes.

In other words, when creating an entry in the shadow page table, if a guest operating system writes a memory to a specific location, the virtual machine monitor can take control and change it to read-only. Privileged mode virtual machine monitor has gained control of the non-privileged guest operating system.

In the memory virtualization method of the present invention, when generating the shadow page table, as shown in FIG. Consider.

In addition, if a change in the access control setting is required, such as when memory interprocess occurs, a corresponding entry in the shadow page table is created by changing the access control setting.

In the shadow page table generation process, the amount of creation is determined by the number of entries in the page table. In the case of a virtualization system supporting a plurality of guest operating systems, since the shadow page table must be generated each time a process of each operating system is executed, the generation process occurs very frequently, which places a heavy load on the system.

Different operating systems map physical memory to processes, and a memory virtualization method according to an embodiment of the present invention proposes a method of reducing a shadow page table generation load in consideration of such characteristics of a guest operating system.

For 32-bit Linux, if the physical memory size is less than 896MB, Linux maps all physical memory directly from the 0xC0000000 address used by the kernel in the process's 4GB virtual address space. Due to the nature of Linux, all processes have the same kernel address mapping, and the page tables of processes used in the Linux operating system all have the same entry above 0XC0000000.

Therefore, if the guest operating system is a Linux operating system, the address of the upper part of 0xC0000000 can reuse the previous shadow page table entry contents without having to interpret the guest page table and create a shadow page table. For other operating systems other than Linux described in the above embodiments, considering the memory mapping characteristic of the operating system, the shadow page table generation load can be reduced.

In a virtualization system operating multiple guest operating systems, in order to reuse a shadow page table, there must be information capable of mapping the guest page table and the cached shadow page table used in the guest operating system. The mapping information may be machine address location information loaded with a guest OS, that is, actual physical memory location information.

In addition, the virtual address location information of the guest page table may be used as the mapping information.

When the machine address and the virtual address are used as mapping information, when the guest operating system uses the page table of another process at the same virtual address location, the page table cannot be identified by the two pieces of information. In the case of para-virtualization, the virtual machine monitor can receive the operating system's information from the guest operating system. This problem can be solved. In the case of full-virtualization, additional information for identification is required. Do. For example, a guest operating system may be identified by adding an identifier for each operating system to the guest page table.

6 is a diagram illustrating shadow page table caching in the memory virtualization method according to an embodiment of the present invention.

Referring to FIG. 6, in the memory virtualization method according to an embodiment of the present invention, the shadow page table 62 may be reused by generating and caching each shadow page table 62 for each of a plurality of guest operating systems. In particular, it is characterized in that the cache by the guest operating system list 61 including the location information of the shadow page table (62).

The operating system-specific list 61 is information for mapping the guest page table (not shown) and the shadow page table 62. As a result, whenever the guest page table (not shown) is changed, the stored shadow page table 62 may be reused without regenerating the shadow page table 62.

Caching of the shadow page table 62 may be performed in various ways. One of the methods may be to maintain caching even when at least one of the plurality of guest operating systems is stopped.

Due to the nature of the virtualization system, many guest operating systems are often executed at the same time. If the shadow page tables of multiple guest operating systems share a single cache space, the cache space is finite, so the existing shadow page table is not needed to store a new shadow page table. The shadow page table of the guest operating system in which the execution of the table is stopped must exit the cache. However, the shadow page table of the guest operating system in which the execution is interrupted and then executed again becomes unusable and needs to be regenerated. Memory virtualization method according to an embodiment of the present invention, so that the shadow page table used by the guest operating system scheduled out so that such a problem does not occur is maintained without being pushed out of the cache.

In addition, the caching of the shadow page table 62 is a method in which the virtual machine monitor of the virtualization system determines the caching frequency of the shadow page table for each guest operating system, and dynamically changes the caching space of the shadow page table according to the caching frequency. Can also be used. This is to efficiently manage when a plurality of guest operating systems share a cache area.

FIG. 7 illustrates a case in which secondary page tables are shared in a memory virtualization method according to an embodiment of the present invention.

Referring to FIG. 7, in the memory virtualization method according to an embodiment of the present invention, a guest page table is divided into a primary page table and a secondary page table 71, and the second page table 71 is divided into 1. Shared with the tea page table. In this case, it is desirable to cache the shadow page table in consideration of the sharing relationship.

When the guest page table shares the secondary page table 71 as shown in FIG. 7, the page tables affect each other, and when the shadow page table is cached, the secondary page table 71 does not reflect the sharing relationship of the secondary page table 71. As a result, the shared guest page table cannot be reflected in the shadow page table, which causes the page table to be inconsistent and the virtualization system cannot operate normally. Therefore, the embodiment of the present invention solves the problem by using a technique that reflects the sharing relationship of the guest page table in the shadow page table. There are many ways to recognize a sharing relationship.

In an embodiment of the present invention, the virtual machine monitor of the virtualization system manages the list of shared secondary page tables 71 to check whether the guest page table is shared by referring to the list when creating the shadow page table.

In addition, as another embodiment of the present invention, the hashing scheme may be configured to determine the sharing relationship of the secondary page table. Linked lists are an easy way to group data structures, but searching through linked lists can be inefficient. Because you have to go through the list until you find what you want. A hash table is an array of pointers, or vectors of pointers, that can be accessed more efficiently by accessing them with an index that indicates their position in the array.

Meanwhile, a memory virtualization apparatus using a shadow page table according to another aspect of the present invention is configured to process mapping information for mapping a cached shadow page table, a guest page table, and a cached shadow page table and a guest page table. It may include a processor.

8 is a flowchart illustrating a memory virtualization method according to an embodiment of the present invention.

First, in a virtualization system supporting a plurality of operating systems, a shadow page table is generated for each operating system (S110).

Next, the generated shadow page table is stored in the cache area (S130). In this case, as described above, a caching method may use various methods. Even when execution of a particular operating system is stopped, a caching area may be maintained, or a method of dynamically allocating a caching area in consideration of the frequency of use may be used.

Next, a process of one operating system among a plurality of operating systems is performed, and a guest page table is provided accordingly (S150).

Thereafter, the cached shadow page table and the guest page table are mapped based on the mapping information (S170). The mapping information may be provided as a list for each operating system, and the list includes address information of the physical memory to allow access to the physical memory in the virtualization system.

As a result, the cached shadow page table can be reused using mapping information without going through the conventional shadow page table generation procedure, thereby improving the performance of the virtualization system.

The present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. Accordingly, the above detailed description should not be construed as limiting in all aspects and should be considered as illustrative. The scope of the invention should be determined by reasonable interpretation of the appended claims, and all changes within the equivalent scope of the invention are included in the scope of the invention. In addition, claims that do not have an explicit citation in the claims may be combined to form an embodiment or included in a new claim by amendment after the application.

The memory virtualization apparatus and method according to the present invention is applicable anywhere in the field of virtualization technology operating a plurality of operating systems.

31: Primary page table 32: Secondary page table
51: virtual address 52: doctor virtual address
53: Machine Address 62: Shadow Page Table

Claims (14)

In the memory virtualization method in a virtualization system supporting a plurality of guest operating systems,
Generating and caching a shadow page table for each of the plurality of guest operating systems; And
Mapping a guest page table for reference of a physical memory page from one of the cached shadow page tables and one of the plurality of guest operating systems using mapping information; Memory virtualization method comprising a.
The method of claim 1,
The mapping step,
And the machine address location information loaded by the guest operating system and the virtual address location information of the guest page table as the mapping information.
The method of claim 1,
The caching step,
And caching location information of the shadow page table as a list for each guest operating system.
The method of claim 1,
The caching step,
Memory caching method, characterized in that the caching is maintained even when the guest operating system is stopped.
The method of claim 1,
The caching step,
Determining, by the virtual machine monitor of the virtualization system, a caching frequency of the shadow page table for each guest operating system, and dynamically changing a caching space of the shadow page table according to the caching frequency. Memory virtualization method.
The method of claim 1,
The memory virtualization method,
And forming the guest page table as a primary page table and a secondary page table, and when the secondary page table is shared, caching the shadow page table in consideration of the sharing relationship.
The method of claim 6,
The memory virtualization method,
And a virtual machine monitor of the virtualization system manages the list of shared secondary page tables to identify a sharing relationship between the secondary page tables with reference to the list when caching the shadow page table.
The method of claim 6,
The memory virtualization method,
Memory virtualization method characterized by grasping the sharing relationship of the secondary page table by a hashing technique.
In the memory virtualization apparatus using a shadow page table in a virtualization system supporting a plurality of guest operating systems,
A shadow page table generated for each of the plurality of guest operating systems;
A guest page table provided for reference of a physical memory page from one of the plurality of guest operating systems;
And a processor for processing mapping information for mapping one of the cached shadow page tables and the guest page table.
10. The method of claim 9,
The mapping information,
And virtual machine location information of the guest address table and machine address location information loaded by the guest operating system.
10. The method of claim 9,
The shadow page table,
Memory virtualization device characterized in that the cached to the guest operating system-specific list.
10. The method of claim 9,
The shadow page table is
Memory caching is maintained even when the execution of the guest operating system is stopped.
10. The method of claim 9,
The shadow page table is
And a virtual machine monitor of the virtualization system determines the caching frequency of the shadow page table for each guest operating system, and dynamically changes the caching space of the shadow page table according to the caching frequency.
10. The method of claim 9,
The guest page table is
A virtual virtualization apparatus comprising a primary page table and a secondary page table, and when the secondary page table is shared, a virtual machine monitor of the virtualization system manages a list of shared secondary page tables. .

Images