TWI399637B - Fast switch machine method - Google Patents

Description

Fast switching machine method

The invention relates to a computer system switching machine method, in particular to a method for a computer system to quickly switch on and off after entering a power saving mode.

In general, the working state of the computer system, in addition to the mode of executing the operating system and various application programs, includes the power saving mode, and the computer system enters the dormant state on the one hand to save power consumption, and on the other hand, to reduce system work. Noise.

According to the design classification of Advanced Configuration & Power Interface (ACPI), the most common types of sleep will be two types, one is Suspend To RAM (STR), called S3 state, and the S3 state is only for the opposite part. The volatile memory is continuously powered, including the Frame Buffer and Main Memory, and the rest is powered off. The S3 state has two advantages. One is that the computer system returns to full speed for a faster time; the other is better for security. When the user's application and operation are private and private data is not expected to be stored in the hard disk, Adopt the S3 state. The other is the Send To Disk (STD), which is called the S4 state. The S4 state is to write all the working data to the non-volatile memory for storage, and then the system will all be powered off. The advantage of S4 is more provincial. Electricity.

The two sleep S3 states and the S4 state are not perfect. After entering the S3 state, it is still necessary to provide power to the volatile memory in the computer system to maintain the state of the system storage; Compared to the S3 state, the speed of returning to the computer system at full speed is much slower than that of the S3 state.

In view of the above, the present invention provides a method for quickly switching on and off in response to the above-mentioned deficiencies, improving the sleep described above, which can reduce the current memory usage when entering sleep, thereby improving the restart of the computer system to full speed operation. Reaction efficiency.

The main object of the present invention is to provide a method for quickly switching on and off, which is to reduce the amount of memory used in the current system when the computer system enters sleep, so as to reduce the amount of data stored and stored in the second storage device by the main memory. , thereby increasing the speed at which the computer system returns to full speed.

Another object of the present invention is to provide a method of quickly switching the machine that prevents the loss and damage of data when the computer system accidentally loses power.

A further object of the present invention is to provide a method for quickly switching on and off. The data writing or loading back adopts a random access processing technology, which will greatly increase the speed of data access when the computer system enters sleep or restarts each other. .

In order to achieve the above object, the method for quickly switching on and off according to the present invention is to divide at least one memory into a plurality of innocent storage pages and a plurality of non-innocent storage pages when a computer system enters sleep, and invent the storage page. When the storage is abandoned, the non-innocent storage page is stored in the sleep file and stored in one of the switching spaces of at least one of the second storage devices or in a file system. When the computer system is restarted, the computer system is returned to the memory by the switching space and the file system, and the data can be read from the second storage device and loaded back to the memory. Thus, the amount of data stored in the second storage device is reduced, and the rate at which the computer system is restarted and resumes to full speed is accelerated.

The purpose, technical contents, features and effects achieved by the present invention will be more readily understood by the detailed description of the embodiments and the accompanying drawings.

The invention provides a method for quickly switching on and off, reducing the number of use of the memory device when the computer system enters sleep, and reducing the amount of data written into the second storage device, so as to increase the speed at which the computer system enters sleep or restarts to resume full speed operation, underneath. The technical features of the present invention will be described in detail with reference to preferred embodiments. As shown in the first figure, a computer system 10 includes a computing processor (CPU) 12, a main memory 14 and a second storage device 16. The second storage device 16 can be a high-speed random access memory device, such as a flash memory, and is mainly used to store a file system 18, a swap-space 20, and a hibernation file. .

The second figure shows a flow chart of the method of the fast switching machine of the present invention. Please refer to the schematic diagram of the first figure. First, in step S30, the operating system (OS) of the computer system 10 enters the sleep mode, and the main memory 14 is distinguished by the area or page as the innocent storage page and the non-innocuous storage page, due to the second storage. The device 16 stores the backup with the inconsistent storage page, so the swapped innocent storage page will abandon the storage, and the main memory 14 will be configured through a core program to control the insulative storage page to be exchanged, and the core program can calculate the basis. The main memory 14 is configured by the number of innocent storage pages to be exchanged, and after the insulative storage page is replaced, the memory is controlled to be released. The non-innocent storage page, which is the machine status, is stored in the sleep file 22 and stored in the switch space 20 of the second storage device 16 or in the file system 18. The non-innocent storage pages may be sorted according to the address to merge the non-innocent storage pages of the consecutive addresses into a single write command. Thereafter, in step S32, when the computer system 10 is restarted, the computer system 10 reads the sleep file 22 from the switching space 20 and the file system 18, and restores the stored sleep file to the main memory 14 to restore the system state. Finally, in step S34, the computer system 10 loads the data from the second storage device 16 back to the main memory 14.

The speed at which the execution of the second storage device 16 (Suspend-to-flash) to the flash memory is mainly determined by the amount of data that needs to be written to the nonvolatile memory. Swapping-before-hibernating uses the existing memory management technology in the operating system (OS) to discard most memory pages without any writing. Into the action, so you can speed up the suspend. When replying, the data will be retrieved from three places: hibernation file 22, swap-space 20 and file system 18. The system belongs to the core ( The main data (data) and code (code) of the kernel part are retrieved immediately after rebooting, and the remaining data is retrieved by the user's current paging stage using the paging-on-demand method. section. Since the amount of data required by the user is usually much smaller than the amount of data in the memory before the system is suspended, the speed of the resume can be accelerated.

The memory page can be divided into three types: free page, anonymous page, and named page. Free page is the memory that is not used by the system at the moment. The content in the free page is meaningless to the system. Anonymous page is the memory dynamically allocated by the program during the execution period. This part of the memory mainly includes stack and heap. The designated page is a copy of the file in the main memory 14. Formally it behaves like a cache of the second storage device 16. The designated page mainly contains the executable file, the correspondence of the dynamic-linke library in the memory, or the program uses a memory-mapped-file to store a file. Map to memory.

Both an anonymous page and a named page may be a clean page or a dirty page. The non-invisible anonymous page must be written to the swap-space 20 when it is swapped out. The dirty named page is swap-out to the file system 18. When the operating system (OS) needs to swap-out an innocent storage page, since the page must have a completely identical copy in the second storage device 16, the operating system (OS) can directly discard the page. .

When the system starts executing suspend-to-flash, most of the paging does not need to do any write back to the second storage device 16, since most of the paging is innocent storage pages. Part of the paging needs to be written back to the switching space 20 or the file system 18. The remaining pages are non-swappable memory. These pages are mainly the kernel of the operating system (OS), or the program sets some pages to be non-swappable for performance reasons (non- Swapable). The non-swapped memory will eventually be written to the hibernation file 22.

When rebooting, after the system performs the routine hardware initialization and loading the OS loader, the system will determine whether to resume or perform a general boot action from the hibernation state. The data in the hibernation file 22 will first be loaded into the main memory 14, and this action is completed, and the operating system (OS) completes the basic booting action. The previous system state is then loaded back from the file system 18 and the switch space 20 in accordance with the needs of the user. If swapping-before-hibernating is executed again, since some non-innocent storage pages have been written to the switching space 20 when the last execution of swapping-before-hibernating, this paging becomes an innocent storage page, and then the computing processor (CPU) 12 does not write the page for this page, so this page does not need to be written to the switching space 20 again. Therefore, the swapping-before-hibernating execution speed will be faster after the second time.

Using the flash memory as the second storage device 16 through the optimization technique, there is a performance similar to sequential access on random access. Through write-combining, the small random write generated by the memory page at the time of swapping out is queued before actually writing to the flash memory. These should be written out of the physical address of the page to sort, after sorting, you can know whether there is a continuous range of writes in the pages to be written, and then you can write a larger write (write Request) replaces a few smaller writes. This speeds up the swapping-before-hibernating.

Writing the swappable pages back to the second storage device 16 can require a large amount of memory to the operating system (OS), in order to avoid the optimization of the operating system (OS) when the memory is allocated to affect the actual configuration. The amount of memory. Each configuration to a page will then write a byte (byte) of data in this page to force the OS to immediately configure the physical memory to the program. This forces the operating system (OS) to swap out most of the memory. A series of free actions are then performed to return the entire amount of memory previously requested by the system to the operating system (OS). After these two steps, most of the paging in the operating system (OS) becomes a free page, and the free page does not need to be written back to the hibernation file 22.

In addition, configuration memory can be requested directly in the kernel of the operating system (OS), and the required paging will be configured immediately through the kernel function, and no additional writes will be required to force the operating system (OS). Real distribution memory. The release action is performed as soon as the memory is configured, and the system generates a large number of free pages. Since the memory is configured by the kernel function, no additional write action is required. Therefore, the number of page breaks can be calculated in advance, thereby estimating how much memory needs to be configured.

In addition, the swapper mechanism in the operating system (OS) can be directly manipulated. Take Linux as an example. In the kernel, there is a memory management program, which is a shrink_all_memory program. This program will recycle the memory pages and pass the number of pages to be recycled when calling. It can affect the system without affecting the system. The memory page is released when the user is stable and the program is used. The mechanism of the shrink_all_memory program is that there are two recent least used directories in the kernel (least_recently_used list, LRU list) are active list and inactive list, and the active list contains the most recently used directory. The accessed page, the inactive list, is a page that has not been accessed for a while. The shrink_all_memory program will start with the inactive list and then recycle the active list. With the shirink_all_memory program, you can swap out as much as possible without affecting the operation of the system core and the user's program.

In the kernel of Linux, the second storage device 16 is treated as a general I/O device, and when we swap out the memory page to the switching space 20, the call will be made. The parameters accepted by the submit_bio program include the number of sectors to be written (sector number), read/write command (r/w command), memory address (memory address), memory length (memory length), etc. News. The page to be swapped out can be processed again by intercepting the submit_bio program.

The above-mentioned page breaks to be swapped out are determined by the shirink_all_memory program, and the decision is based on the use of the page break instead of the physical address of the page in the main memory 14. We will first arrange the write requirements generated by the shirink_all_memory program. The physical memory addresses of the paging are then sorted and combined to reduce the number of access/output (I/O) accesses and increase the input/output (I/O) request size. The larger the request size, the better the write performance of the second storage device 16 for flash memory.

When a memory page is swapped out, the kernel needs to record the position where the page is placed in the switching space 20. The kernel is enabled to reload this page from the second storage device 20. Since the write-combining will rewrite the shirink_all_memory program, after each execution of the shirink_all_memory program, it must be determined that each page is written to the location of the switching space 20, and then the information is filled into the paging table ( The swapped-out page identifier in page table).

The embodiments described above are merely illustrative of the technical spirit and the features of the present invention, and the objects of the present invention can be understood by those skilled in the art, and the scope of the present invention cannot be limited thereto. That is, the equivalent variations or modifications made by the spirit of the present invention should still be included in the scope of the present invention.

10. . . computer system

12. . . Computing processor

14. . . Main memory

16. . . Second storage device

18. . . File system

20. . . Switch space

twenty two. . . Dormant file

The first figure is a schematic diagram of the architecture of the present invention.

The second figure is a flow chart of the processing steps of the present invention.

10. . . computer system

12. . . Computing processor

14. . . Main memory

16. . . Second storage device

18. . . File system

20. . . Switch space

twenty two. . . Dormant file

Claims (9)

A method for quickly switching on and off, comprising the steps of: entering a sleep in a computer system, wherein the computer system distinguishes at least one memory into a plurality of innocent storage pages and a plurality of non-innocent storage pages, and the innocuous storage page is replaced When the storage is abandoned, the non-innocent storage page is stored in the sleep file and stored in one of the switching spaces of at least one of the second storage devices or in a file system. Rebooting the computer system, the computer system reads the sleep file from the switching space and the file system, and restores the sleep file to the memory; and the computer system reads the data from the second storage device. And load the data into the memory. The method of the fast switching machine of claim 1, wherein in the step of the computer system entering the sleep state, the memory is released after the white storage page is replaced. The method of the fast switching machine of claim 2, wherein in the step of the computer system going to sleep, the memory is configured by a core program, and the innocent storage page is controlled to be swapped out, and the innocence is performed. After the storage page is swapped out, the control releases the memory. The method of the fast switching machine of claim 3, wherein the core program calculates the number of the invented pages to be swapped out and configures the memory according to the number. The method of the fast switching machine of claim 1, wherein the second storage device is a flash memory. The method of the fast switching machine of claim 1, wherein the second storage device is a high speed random access memory device. The method of the fast switching machine of claim 1, wherein the computer system can be used as a unit for distinguishing the memory in the step of the computer system going to sleep. The method of the fast switching machine according to claim 1, wherein the second storage The device has a backup storage consistent with the insufficient storage page. The method of the fast switching machine according to claim 1, wherein in the step of the computer system entering the sleep, the non-innocent storage page to be swapped out is sorted according to the address, and the address is consecutively non-innocent. The storage pages are merged into a single write command, and the second storage device is combined combing.