TW201805820A - Method for transferring command from host to device controller and system using the same - Google Patents

Abstract

A method for transferring a command from a host to a device controller and a system using the method are disclosed. The method includes the steps of: A. determining a segment size; B. dividing a command into a number of sections each having a size the same as the segment size; C. sequentially distributing the sections to n groups and if the last section is smaller than a segment, then fill this section to a segment size; D. keeping the same distributing order in step C if a cycle of distribution is finished while there are sections left for distributing; E. restructuring the section(s) in each group as a sub-command after all sections are distributed; and F. providing the sub-commands to a device controller synchronously.

Description

Method of transmitting commands from a host to a device controller and using the party Legal system

The present invention relates to a method and system for transmitting commands from a host to a device controller, and more particularly to a method and system for transferring commands from a host to a device controller by dividing each command into a number of subcommands.

Traditionally, when a host transmits commands to a device, it must transmit a command through a transport interface, one transfer operation. With the development of command transmission technology, the host can transmit multiple commands, such as command queues, and the device controller can handle two or more commands. Therefore, the host can split a command into two or more subcommands and send them to the device controller, which can speed up command processing.

The process of command transmission is as shown in Figure 1. When the host is powered on or before the command is transmitted, the host driver sends an identification command to the device controller via an interface to identify a separate storage device (Hard Disk Drive (HDD), Solid State Drive (Solid State Drive). , SSD) or memory card) or device controller managed storage device information. The recognition command and other later will The commands mentioned can be adapted to a certain bus protocol. For non-volatile memory architecture storage devices, the bus protocol can be a protocol developed by the Non Volatile Memory Express standard. The non-volatile memory high-speed standard describes the scratchpad interface, command set, and feature set of the PCIe architecture SSD. The device controller then sends back device information about the storage device to the host driver. Device information can include the size of the storage device unit, such as the size of the flash memory page and block, the type of flash memory, the fastest transfer speed, and so on. According to the device information, the host driver can provide an address to prepare the command to the device controller. The device controller can read the command for the address and execute it.

As mentioned above, because the host driver can split the command into several (at least two) subcommands, so that the device controller can read the subcommands from different addresses, and on a single processor different task (task) or Multiple processors execute it. This is feasible, but there is no way to achieve it. The present invention discloses a solution to the aforementioned needs.

This paragraph of text extracts and compiles certain features of the present invention. Other features will be revealed in subsequent paragraphs. The intention is to cover various modifications and similar arrangements in the spirit and scope of the appended claims.

In order to meet the above requirements, the method provided by the present invention for transmitting commands from a host to a device controller includes the steps of: A. determining a segment size; B. dividing a command into a plurality of segments, each segment having a size and The size of the section C. sequentially distribute the segments into the n-group, and if the last segment is smaller than one segment, fill up to a segment size; D. if the distribution cycle ends and there are remaining segments to be distributed, then Continue to maintain the same distribution order in step C; E. After all the segmentation distributions, reconstruct the segmentation of each group into a subcommand; and F. Synchronize to provide the subcommands to a device controller.

According to this method, the size of the segment that is finally divided by a command must be the same size as the segment. The segment size can be the same as one or more pages of a non-volatile memory chip. n is a positive integer and may be greater than or equal to 2, n may also be the maximum number of sub-commands that can be synchronously acquired and executed by the device controller. The subcommand providing methods are obtained by sending a host neutron command address to the device controller. If the non-volatile memory high speed specification is used for command transmission, the addresses may be submitted queues.

Another aspect of the present invention is a system using the foregoing method, the system comprising: a host, a driver installed, and a memory for storing command subcommands, wherein the driver is used to determine a segment size; The split command is a plurality of segments, each segment having the same size as the segment; the segments are sequentially distributed to the n groups; if the distribution loop ends and there are remaining segments to be distributed, then If the block is smaller than a segment; after all the segments are distributed in the memory, the segments in each group are reconstructed into a sub-command; and the sub-commands are provided synchronously; and a storage The device is connected to the host remotely or proximally for storing data for access, including: several non-volatile a memory chip; and a device controller connected to the host for synchronizing the sub-commands from the command and executing the sub-commands.

In accordance with the present invention, the size of the segment that was last divided by a command must be the same size as the segment. The segment size can be the same as one or more pages of the non-volatile memory chips. N can be an integer and greater than or equal to 2, and n can also be the maximum number of subcommands that can be synchronously acquired and executed by the device controller. The subcommand providing methods are obtained by sending the subcommand address in the host to the device controller. If the non-volatile memory high-speed specification is used for command transmission, the addresses are the commit queue. The device controller executes all subcommands from a command in synchronization with the driver settings so that the device controller can feed back the host after the subcommands are executed.

The present invention utilizes the advantages of a small sub-command size and synchronous execution of an operating program or a plurality of processors in the device controller. The execution time of the sub-commands is shorter than the time at which the original commands are executed, and the overall performance of the stored access can be improved.

10‧‧‧System

100‧‧‧Host

110‧‧‧Central Processing Unit

115‧‧‧Driver

120‧‧‧RAM module

130‧‧‧PCIe controller

200‧‧‧Storage equipment

210‧‧‧Device Controller

220‧‧‧ volatile memory unit

230‧‧‧ Non-volatile memory chips

300‧‧‧PCIe bus

Figure 1 shows a related pre-transmission of a command transmission.

2 is a flow chart of a method for transmitting commands from a host to a device controller in accordance with the present invention.

Figure 3 depicts a system using this method.

Figure 4 shows the data structure in a RAM module.

Figure 5 shows how a command is split.

Figure 6 shows the two subcommands formed.

Figure 7 shows how another command is split.

Figure 8 shows the three subcommands formed.

The invention will be more specifically described by reference to the following embodiments.

Please refer to FIG. 2. FIG. 2 is a flow chart of a method for transmitting commands from a host to a device controller in accordance with the present invention. The method can be implemented by a system having a host and a device controller for executing read and write commands to a storage device managed by the device controller. It should be noted that the system can refer to a personal computer, a tablet computer, a smart phone or a stand-alone electronic device for special operations, such as a car computer for a controller car device, and a host computer and a storage device are integrated. The system can also have a host and a detachably connected storage device. For example, a central processing unit in a digital camera and a memory card indirectly connected to the central processing unit via a connector. In addition, the system referred to in the present invention may also refer to a cloud storage device (virtual or physical) of a client computer and a remote data center. Read and write commands can be transmitted and received via the Local Area Network or the Internet. The way in which command transmissions are handled is the same for all applications. An embodiment will be described below using a local application.

Figure 3 shows a system 10 having a host 100 and a storage device 200 to which the disclosed method of the present invention is applied. The host 100 and the storage device 200 are disposed on A casing (not shown) is connected via a PCI (Peripheral Component Interconnect) express (PCIe) bus 300. The host 100 has a central processing unit 110, a RAM module 120, and a PCIe controller 130. It may also have other electronic components or devices that are not within the scope of the present invention. Therefore, the electronic components or devices are not further described. . The host 100 installs a driver 115 that is activated after the central processing unit 100 is powered on. The main functions of the driver 115 include: determining a segment size; dividing a command into a plurality of segments, each segment having the same size as the segment; sequentially distributing the segments to the n group; if distributed If the loop ends and there are still remaining segments to be distributed, then the size is added to a segment if the block is smaller than a segment; after all segments are distributed in the memory, the segment of each cluster is reconstructed as a subcommand; Synchronize the subcommands to a device controller. The details of each function will be explained later. The RAM module 120 is used to store commands and subcommands. The PCIe controller 130 is responsible for communicating with a device controller 210 via the PCIe bus 300.

The storage device 200 is connected to the host 100 at the local end for storing data in the casing for access. Of course, as noted above, in other embodiments, storage device 200 can be remotely connected to host 100 by the Internet. In accordance with the present invention, storage device 200 should be a non-volatile memory type storage device such as an SSD or eMMC. Thus, storage device 200 can have a plurality of non-volatile memory chips 230 for storing data. In practice, each non-volatile memory chip 230 can be a NAND flash memory chip, a NOR flash memory chip, or a charge-defect storage flash memory chip. The storage device 200 also has a volatile record The memory unit 220 is a dynamic random access memory (DRAM) module. The volatile memory unit 220 can be used to temporarily store certain important data from the host 100 for executing commands. For example, volatile memory unit 220 can have a mapping table for mapping physical addresses and corresponding logical addresses for access to job execution. The storage device 200 has a device controller 210. The storage device 200 is connected to the host 100 via the PCIe bus 300 using the device controller 210. The device controller 210 can synchronously acquire the sub-commands generated by a command and execute the sub-commands. The following is an example of how the system 10 operates by the disclosed method.

It is emphasized that the method provided by the present invention is implemented by the driver 115 and the device controller 210. If the invention is commercialized, the driver 115 should be sold at the same time as the device controller 210, otherwise the system 10 will not work. According to the method, a first step is to determine a segment size (S01). The extent size is a commanded (read or write) split unit size, which is the same as the size of the page or pages of the non-volatile memory chip 230. The commonly used page size is 4kB, and this embodiment uses 4kB as the segment size. The extent size is set by the driver 115.

The second step is to divide a command into a plurality of segments, each segment having the same size as the segment (S02). For a better understanding, see Figure 5, which shows how a command is split. A command 1 has a size of 18 kB and occupies an address of 0 to 17 in the RAM module 120. Since the extent size is 4kB, the data in the 4 addresses divided by the command 1 form a segment: a first segment (from 0) To 3), a second segment (from 4 to 7), a third segment (from 8 to 11), a fourth segment (from 12 to 15) and a fifth segment (from 16 to 17) ). . In most cases, the size of the last segment (eg, the fifth segment) of a command may be smaller than the segment size, so the fifth segment is padded to 4kB.

Then, the segments are sequentially distributed to the n group (S03). As a result of the distribution shown in Fig. 6, n is set to 2 to form two groups. The first segment is distributed to the first group, and then the second segment is distributed to the second group. If the distribution loop ends and there is still a segment remaining to be distributed, the same distribution order in step S03 is maintained (S04). The third segment is distributed to the first group, the fourth segment is distributed to the second group, and finally the fifth segment is distributed to the first group. The first group has a first segment, a third segment and a fifth segment in sequence. The second group has a second segment and a fourth segment in sequence. The results of the distribution are shown in Figure 6.

The next step is to reconstruct the segment of each group into a subcommand after all the segments are distributed (S05). As shown in Fig. 6, the first group displayed above is reconstructed as a sub-command 1, and the second group shown below is reconstructed as a sub-command 2.

The final step of the method is to provide the subcommands (subcommand 1 and subcommand 2) to the device controller 210 (S06), which means that subcommand 1 and subcommand 2 should not be acquired separately for the device controller 210 because they The same as a command. The subcommands should be listed in the RAM module 120. Please see figure 4. FIG. 4 shows the data structure in the RAM module 120. The two commands C1 and C2 are queued according to the original plan of the central processing unit 110. After the driver 115 processes, the command C1 is divided into sub-commands SC1-1 and SC1-2. Command C2 is divided into subcommands SC2-1 and SC2-2. According to non Volatile memory high-speed specifications, sub-commands SC1-1, SC1-2, SC2-1, and SC2-2 are considered separate submission queues. Then, the sub-command is provided by the device controller 210 by the transmission address (RAM module 120) of the sub-command in the host 100. Because of the non-volatile memory high-speed specification for command transmission, these addresses are in this case a commit queue.

From step S01 to step S06, all the steps can be completed by the driver 115. That is, when the driver 115 is executed from step S01 to step S06, all necessary calculations are performed by the central processing unit 110, and the related data is stored in the RAM module 120. The device controller 210 and the driver 115 must set a subcommand to execute a command in synchronization so that the device controller 210 can feed back the host 100 after the subcommand is executed. At the same time, the segment size and the number n should be the same for the device controller 21 and the driver 115 during operation.

In accordance with the present invention, n is the maximum number of subcommands that can be synchronously retrieved and executed by the device controller 210. If device controller 210 can process 2 commands synchronously, n is 2; if device controller 210 processes 4 commands synchronously, n is 4; and so on. Finally, n should be a positive integer greater than or equal to 2. In the present embodiment, n = 2 is used. In the following embodiments, an example shows how the method is applied when n is 3.

See Figure 7 and Figure 8. Figure 8 shows a command splitting to form three subcommands, and Fig. 7 shows the results of the segmentation distribution. Command 2 has a size of 19kB and occupies an address from 0 to 18 of RAM module 120. In step S01, the extent size is determined to be 3kB instead of 4kB in the previous embodiment. Next, in step S02, the life Divide 2 into 7 segments: a first segment (from 0 to 2), a second segment (from 3 to 5), a third segment (from 6 to 8), and a fourth segment. (from 9 to 11), a fifth segment (from 12 to 14), a sixth segment (from 5 to 17) and a seventh segment (18). Since the seventh segment is the last segment and is smaller than the segment size, it is necessary to fill a segment size (from 18 to 20).

Next, in step S03, the commands are sequentially distributed to the three groups: a first group, a second group, and a third group. The first segment is distributed to the first group, the second segment is distributed to the second group, and the third segment is distributed to the third group. After the end of the distribution cycle, the same distribution order in step S03 is maintained for all distributions after the first cycle. The fourth segment is distributed to the first group, the fifth segment is distributed to the second group, and the sixth segment is distributed to the third group and the last seventh segment is distributed to the first group. The first group has a first segment, a fourth segment and a seventh segment in sequence. The second group has a second segment and a fifth segment in sequence. The third group has a third segment and a sixth segment in sequence.

After the step of reconstructing in step S05, in FIG. 8, the first group is reconstructed into a subcommand, the second group is reconstructed as a subcommand 2, and the third group is reconstructed as a subcommand 3. The three subcommands are controlled by the device controller 210. Acquired, so it may be necessary for three operating programs to simultaneously operate the three subcommands so that the original commands can be executed more quickly.

It is emphasized that while the above embodiments use a non-volatile memory high speed specification, in practice, the present invention can apply other specifications that allow multiple commands to be transceived between the host and the device controller. For example, the Universal Flash Storage specification.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the invention, and those skilled in the art can make some modifications and refinements without departing from the spirit and scope of the invention. The scope of the invention is defined by the scope of the appended claims.

Claims (15)

A method for transmitting a command from a host to a device controller, comprising the steps of: A. determining a segment size; B. dividing a command into a plurality of segments, each segment having a size equal to the segment size; C. The segments are sequentially distributed to the n-group, and if the last segment is smaller than one segment, it is added to a segment size; D. If the distribution cycle ends and there are remaining segments to be distributed, continue to maintain step C. The same distribution order; E. After all the segments are distributed, the segments of each group are reconstructed into a sub-command; and F. Synchronously provides the sub-commands to a device controller. The method of claim 1, wherein a segment of the last segment of the command is the same size as the segment. The method of claim 1, wherein the segment size is the same as one or more pages of a non-volatile memory chip. The method of claim 1, wherein n is a positive integer and greater than or equal to 2. The method of claim 4, wherein n is the maximum number of subcommands that are synchronously fetched and executed by the device controller. The method of claim 1, wherein the subcommand providing method is obtained by sending a host neutron command address to the device controller. The method of claim 6, wherein if a non-volatile memory (Non Volatile Memory Express) specification is used for command transmission, the addresses are submitted queues. A system comprising: a host, a driver installed, a memory for storing command subcommands, wherein the driver is used to determine a segment size; and a command is divided into a plurality of segments, each segment The size of the segment is the same as the size of the segment; the segments are sequentially distributed to the n group; if the distribution loop ends and there are still remaining segments to be distributed, the segment is added to a segment size if the block is smaller than a region a segment; after all the segments are distributed in the memory, reconstructing the segments in each group as a subcommand; and providing the subcommands in synchronization; and a storage device, connecting the remote or near end to the host, For storing data for access, comprising: a plurality of non-volatile memory chips; and a device controller connected to the host for synchronizing obtaining the sub-commands from a command and executing the sub-commands. The system of claim 8, wherein the size of the last segment of the command is the same as or smaller than the size of the segment. The system of claim 8 wherein the segment size is the same as a page of the non-volatile memory wafers. The system of claim 8 wherein n is an integer and greater than At 2. The system of claim 11, wherein n is the maximum number of subcommands that can be synchronously acquired and executed by the device controller. The system of claim 8, wherein the sub-command providing method is obtained by sending the sub-command address of the host to the device controller. The system of claim 13, wherein if the non-volatile memory high speed specification is used for command transmission, the addresses are submitted queues. The system of claim 8 wherein the device controller executes all subcommands from a command in synchronization with the driver settings so that the device controller can feed back the host after the subcommands are executed.