WO2020135504A1

WO2020135504A1 - Virtualization method and apparatus

Info

Publication number: WO2020135504A1
Application number: PCT/CN2019/128310
Authority: WO
Inventors: 李翌; 彭浩
Original assignee: 中兴通讯股份有限公司
Priority date: 2018-12-27
Filing date: 2019-12-25
Publication date: 2020-07-02
Also published as: CN111381926A

Abstract

Provided are a virtualization method and apparatus. The method comprises: in the case where an application of a virtual machine sends a pre-determined request, an NVMe-Blk drive of the virtual machine allocating, from a virtual NVMe direct memory access (DMA) memory management area, a first DMA memory, wherein the first DMA memory comprises a second DMA memory of a submission queue and a third DMA memory of a completion queue of an input/output (IO) queue (200); the NVMe-Blk drive constructing a submission queue item according to an HPA of the second DMA memory, and notifying an NVMe device of a host of processing the pre-determined request (201); and the NVMe-Blk drive reading response information in the completion queue (202).

Description

Virtualization method and device

This application requires the priority of the Chinese patent application with the application number 201811612722.7 filed by the China Patent Office on December 27, 2018. The entire content of this application is incorporated by reference in this application.

Technical field

The embodiments of the present application relate to but are not limited to the computer field, for example, a virtualization method and device.

Background technique

Solid State Drives (SSD) are hard drives made with solid-state electronic storage chip arrays, and most of them are currently based on Flash chips as storage media. Compared with conventional mechanical hard drives, SSDs are fast to read and write, light in weight, low in energy consumption, shock-proof and drop-resistant, and small in size. It has been used more and more in the fields of server and personal computing.

The non-volatile memory host controller interface specification (Non-Volatile Memory Express (NVMe) is a storage device interface specification that can take full advantage of the low latency of the external device high-speed interconnect standard (Peripheral Component Interconnect express (PCI-E) channel) And the characteristics of parallelism simplify the input and output (IO) access path of the solid state disk. Compared with the serial connection of small computer system interface (Small Computer System Interface (SCSI) (Serial Attached SCSI, SAS) or serial advanced connection technology (Serial Advanced Technology Attachment, SATA) interface specification, NVMe can provide lower latency, Higher transmission performance and lower power consumption control. NVMe adapts to the high-speed IO characteristics of SSD, which greatly improves the read and write performance of solid-state disks, and is widely used in high-end SSD devices.

At present, many cloud hosts provide virtual machine storage support based on NVMe SSD. The general virtualization method of NVMe SSD is shown in Figure 1. The virtual machine uses the front-end driver to make read and write requests with the back-end driver in the Quick Emulator (Qemu). The back-end driver runs in the context of the host and interfaces with the virtual file system (Virtual File System (VFS)). , Read and write requests to the NVMe driver in the host. In the NVMe interface specification, NVMe devices provide two types of queues: management and IO. There is only one management queue and multiple groups of IO queues. Each group of queues consists of two circular lists of submission and response, which are used by the NVMe device to receive commands and place execution information.

As shown in Figure 1, when the virtual machine performs IO read and write, the front-end driver sends the read-write request to the back-end driver; after the back-end driver receives the read-write request, the read-write request is converted to a host (Host) (also (Referred to as physical machine) VFS read and write operations; Host operating system (Operating System, OS) kernel calls NVMe SSD drive read and write commands according to VFS read and write operations, so that the NVMe driver generates IO read and write instructions into the IO of the NVMe device In the submission queue of the queue; after the NVMe device executes the IO read and write instructions, the result is placed in the response queue of the IO queue, and a message interrupt (Message Signaled Interrupts, MSI) is generated to notify the host; the host then processes through the NVMe driver and is encapsulated by VFS , Return the result to the back-end driver, and return to the front-end through the virtualization mechanism to complete the IO read and write operations of the virtual machine or guest (Guest).

It can be seen that in the ordinary NVMe virtualization method, one IO operation needs to involve two front-end and back-end interactions, and involves multiple copies of data in the front-end and back-end, and the long IO stack from VFS in the Host to NVMe, efficiency Low, is not conducive to the guests to take full advantage of the high IO characteristics of NVMe SSD devices.

Summary of the invention

The embodiments of the present application provide a virtualization method and device, which can improve IO efficiency.

An embodiment of the present application provides a virtualization method, including: in the case where a predetermined request is issued by an application of a virtual machine, the interface specification of the non-volatile memory host controller of the virtual machine—the block NVMe-Blk driver directly stores memory from the virtual NVMe The first DMA memory is allocated in the DMA memory management area; wherein, the first DMA memory includes the second DMA memory of the commit queue of the input/output IO queue and the third DMA memory of the completion queue; the NVMe-Blk driver is based on the first The physical address HPA of the physical machine in the DMA memory constructs a submission queue item, and notifies the NVMe device of the host to process the predetermined request; the NVMe-Blk driver reads the response information in the completion queue.

An embodiment of the present application also provides a virtualization method, which includes: the host's non-volatile memory host controller interface specification NVMe device learns that a predetermined request needs to be processed, performs an operation corresponding to the predetermined request, and places the response information Into the completion queue.

An embodiment of the present application provides a virtualization device, including: a non-volatile memory host controller interface specification—a block NVMe-Blk driver, which is configured to directly store memory from a virtual NVMe when the virtual machine application issues a predetermined request The first DMA memory is allocated in the DMA memory management area; wherein, the first DMA memory includes a second DMA memory of the commit queue of the input and output IO queue and a third DMA memory of the completion queue; according to the second DMA memory The physical machine physical address HPA constructs a submission queue item, notifies the NVMe device to process the predetermined request; reads the response information in the completion queue.

An embodiment of the present application provides a virtualization device, including: a non-volatile memory host controller interface specification NVMe device, which is configured to perform an operation corresponding to the predetermined request when it is informed that a predetermined request needs to be processed, and place the response information Into the completion queue.

An embodiment of the present application further provides a virtualization device, including a processor and a computer-readable storage medium, where the computer-readable storage medium stores instructions, and when the instructions are executed by the processor, the foregoing A method of virtualization.

An embodiment of the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, any one of the foregoing virtualization methods is implemented.

BRIEF DESCRIPTION

Figure 1 is a schematic diagram of a common virtualization method of related technology NVMe SSDs;

2 is a flowchart of a virtualization method proposed by an embodiment of this application;

FIG. 3 is a schematic diagram of a doorbell (DoorBell) area mapping with an IO queue in an embodiment of this application;

4 is a schematic diagram of a virtualization method in an embodiment of this application;

5 is a flowchart of a virtualization method proposed by another embodiment of this application;

6 is a schematic structural diagram of a virtualization device according to an embodiment of the present application.

detailed description

The steps shown in the flowcharts of the figures can be performed in a computer system such as a set of computer-executable instructions. And, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in an order different from here.

Referring to FIG. 2, an embodiment of the present application provides a virtualization method, including steps 200 to 202.

In step 200, in the case that the application of the virtual machine issues a predetermined request, the NVMe-block (Block, Blk) driver of the virtual machine allocates the first from the virtual NVMe direct memory access (Direct Memory Access, DMA,) memory management area. A DMA memory; wherein, the first DMA memory includes the second DMA memory of the commit queue (Summit Queue, SQ) of the IO queue and the third DMA memory of the completion queue (Complete Queue, CQ).

In step 201, the NVMe-Blk driver constructs an SQ item based on the physical machine physical address (HPA) of the second DMA memory, and notifies the NVMe device of the host to process the predetermined request.

In the embodiment of the present application, the SQ item further includes information such as the sector number and priority of the NVMe device.

Among them, the second DMA memory is the source address, and the sector number is the destination address, that is, the NVMe device is notified through the SQ item to write data from the second DMA memory to the sector number.

In the embodiment of the present application, the NVMe device may be an SSD.

In the embodiments of the present application, any one of the following methods may be used to notify the NVMe device of the host to process the reservation request.

Method 1: The NVMe-Blk driver sends a write doorbell (DoorBell) request to the host; wherein the write DoorBell request includes submission queue item information, that is, submission queue item subscript; the host writes the submission queue item information Into the field corresponding to the IO queue identifier in the DoorBell area of the register of the NVMe device.

In this method, the NVMe-Blk driver can be based on a generalized virtual memory address trapping mechanism, or the HyperCall instruction of the Central Processing Unit (CPU) sends a write DoorBell request to the host hypervisor.

In this method, the DoorBell area is arranged according to the IO queue ID, and the host can know the field corresponding to the IO queue ID based on the IO queue ID and the start address of the DoorBell area.

Method 2: The NVMe-Blk driver pre-stores the second correspondence between the IO queue identifier and the field of the doorbell DoorBell area in the register of the NVMe device; the NVMe-Blk driver writes the submission queue item information to the NVMe In the doorbell DoorBell area of the device's register, in the field corresponding to the IO queue identifier.

In this method, as shown in Figure 3, the DoorBell area of the NVMe device is a continuous area, which can be mapped into a virtual machine, and each IO queue corresponds to a 16-bit field in it, which can be initialized by the NVMe-Blk driver. When the mapping operation to the DoorBell area is completed, the second corresponding relationship is saved.

In the second way, the NVMe-Blk driver is used to directly write the submission queue item information into the field corresponding to the IO queue identifier in the DoorBell area of the NVMe device register, without writing through the host, which improves the IO efficiency.

In step 202, the NVMe-Blk driver reads the response information in the completion queue.

In the embodiment of the present application, the NVMe-Blk driver may use any of the following methods to read the response information in the CQ.

Method 1: The NVMe-Blk driver polls the completion queue to obtain response information.

Method 2: Before the host sends the first correspondence between the IO queue identifier and the HPA of the first physical virtual machine memory to the NVMe-Blk driver, the host allocates message interrupts (MSI) to the IO queue Interrupt number, the MSI interrupt number is set to be processed directly by the virtual machine; when the host sends the first correspondence between the IO queue identifier and the HPA of the first physical virtual machine memory to the NVMe-Blk driver, Send the MSI interrupt number to the NVMe-Blk driver; after the host performs the operation corresponding to the predetermined request, the host triggers an interrupt according to the MSI interrupt number; the NVMe-Blk driver reads the interrupt when the interrupt corresponding to the MSI interrupt number is triggered Describe the response information in the completion queue.

In this method, the NVMe-Blk driver can set the corresponding interrupt handler in the virtual machine. When the interrupt is triggered, the top handler of the interrupt reads the completion queue information and releases the CQ queue to trigger the bottom handler of the interrupt; when the predetermined request is In the case of a read request, the interrupt bottom handler obtains the location of the third DMA memory from the completion queue information, completes the copy of the read data, and releases the DMA memory.

In Linux, the interrupt handler will trigger the IO protocol stack of the virtual machine OS to complete the action, and return the corresponding operation return information to the read and write requester.

In this method, the MSI interrupt number is set to be processed directly by the virtual machine. In the X86 processor, it can be implemented based on the mechanism of Direct IO Virtualization Technology (Virtualization for Directed I/O, VT-D) PostInterrupt. That is, the MSI interrupt number is set to The VT-D PostInterrupt virtual universal interrupt controller (Virtual Generic Interrupt, VGI) feature area of the X86 processor of the virtual machine; it can be completed based on the vGSI mechanism in the ARM processor; in the case of the virtual machine monopolizing the host CPU, it can be Based on the X86 processor, a virtual machine control structure (Virtual-Machine Control Structure, VMCS) is provided with an interrupt (non-exit) method.

In the embodiment of the present application, the predetermined request includes at least one of the following: a read request and a write request.

In the case where the predetermined request is a write request, step 200 and step 201 further include: the NVMe-Blk driver copies the written data to the second DMA memory.

In the case where the predetermined request is a read request, the response information includes the HPA of the third DMA memory, and the NVMe-Blk driver reads data from the third DMA memory according to the response information, and copies the read data to the application's read cache, Free the third DMA memory.

In another embodiment of the present application, before step 200, the method further includes: the NVMe-Blk driver applies to the virtual machine for a continuous first physical virtual machine memory, and the virtual machine physical of the first physical virtual machine memory The address (Guest Physical) (GPA) is transmitted to the host; among them, the GPA of the starting position of the first physical virtual machine memory can be transmitted to the host; the GPA of the first physical virtual machine memory can be transmitted to the host Qemu; NVMe- The Blk driver receives the first correspondence between the IO queue identifier and the HPA of the first physical virtual machine memory; the NVMe-Blk driver applies to the virtual machine for a continuous second physical virtual machine memory, and the second The physical virtual machine memory is set to the DMA access method, and the virtual machine physical address GPA of the second physical virtual machine memory is transmitted to the host; wherein, the GPA of the starting position of the second physical virtual machine memory can be transmitted to the host; Transfer the GPA of the memory of the second physical virtual machine to Qemu of the host; the NVMe-Blk driver creates the virtual NVMe DMA memory management area according to the received HPA of the memory of the second physical virtual machine.

In an embodiment of the present application, as shown in FIG. 4, when the NVMe-Blk driver is created, a continuous first physical virtual machine memory and a second physical virtual machine memory may be applied to the virtual machine.

Among them, the NVMe-Blk driver can be created when the virtual machine operating system (Operating System) is initialized, and it is registered as a block (Block, Blk) driver of the virtual machine to complete random read and write operations based on Blk.

In an embodiment of the present application, the transfer method of the GPA address from the virtual machine to the host includes any one of the following: under a kernel virtual machine (Kernel-based Virtual Machine, KVM), it can be transferred based on a virtual input output (virtIO) channel; based on Qemu shared memory area transfer; HyperCall instruction parameter transfer based on X86vt technology.

Among them, the virtIO channel may be a virtIO control channel. When the NVMe-Blk driver is created, the virtIO channel may be established by the NVMe-Blk driver and Qemu of the host.

In this embodiment of the present application, the NVMe-Blk driver is used to achieve direct access to the NVMe device, which reduces the participation of the host, thereby achieving efficient IO operations of the virtual machine, and does not need to occupy the CPU, reduces the consumption of virtualization, and does not need to support specific NVMe device hardware has good versatility.

Referring to FIG. 5, another embodiment of the present application provides a virtualization method, including step 500.

In step 500, when the host's NVMe device learns that a predetermined request needs to be processed, it performs an operation corresponding to the predetermined request and puts the response information into the completion queue.

In the embodiment of the present application, in the case that the field corresponding to the IO queue identifier in the DoorBell area of the register of the NVMe device has commit queue item information written, the NVMe device learns that a predetermined request needs to be processed.

Wherein, the host or the NVMe-Blk driver may write the submission queue item information into the field corresponding to the IO queue identifier in the DoorBell area of the register of the NVMe device.

In the case where the NVMe-Blk driver writes the submission queue item information into the field corresponding to the IO queue identifier in the DoorBell area of the NVMe device's register, the NVMe-Blk driver needs to save the IO queue identifier and the NVME device's The second correspondence between the fields of the DoorBell area in the register.

In the case where the host writes the submission queue item information into the field corresponding to the IO queue identifier in the DoorBell area of the NVMe device's register, the NVMe-Blk driver sends a write DoorBell request to the host; where the write DoorBell request includes Submission queue information; the host writes the submission queue item information into the field corresponding to the IO queue identifier in the DoorBell area of the register of the NVMe device.

In an embodiment of the present application, the reservation request is a write request or a read request; in the case where the reservation request is a write request, performing operations corresponding to the reservation request includes: the NVMe device stores the second DMA memory according to the submission queue item information. The data is written to the corresponding sector of the NVMe device.

In the case where the predetermined request is a read request, the operation corresponding to the predetermined request includes: the NVMe device reads data from the corresponding sector of the NVMe device according to the submission queue item information, and copies the read data to the third DMA In memory.

In another embodiment of the present application, before performing the operation corresponding to the predetermined request in step 500, before placing the response information in the completion queue, the method further includes: the host converts the received GPA of the first physical virtual machine memory into HPA, wherein the device command includes the HPA of the first physical virtual machine memory; wherein, Qemu of the host calls the linux kernel interface to convert GPA to HPA, and sets the noCache attribute, Qemu sends the HPA to the host, and the host sends the HPA Send to the NVMe-Blk driver; the host creates an IO queue based on the HPA of the first physical virtual machine memory, and sends the first correspondence between the IO queue identifier and the HPA of the first physical virtual machine memory to NVMe -Blk driver; wherein, Qemu of the host creates an IO queue by calling NVMe device control commands; the host converts the received GPA of the second physical virtual machine memory into HPA, and converts the second physical virtual machine memory Set to host DMA access mode, send the HPA of the second physical virtual machine memory to the NVMe-Blk driver; where, Qemu of the host calls the Linux kernel interface to convert GPA to HPA, and set the noCache attribute, Qemu will HPA Send to the host, the host sends HPA to the NVMe-Blk driver.

In another embodiment of the present application, before the host sends the first correspondence between the IO queue identifier and the HPA of the memory of the first physical virtual machine to the NVMe-Blk driver, the method further includes: the host is the IO The queue allocation message interrupts the MSI interrupt number, and sets the MSI interrupt number to be directly processed by the virtual machine; the host sends the first correspondence between the IO queue identifier and the HPA of the first physical virtual machine memory to When driving the NVMe-Blk, the MSI interrupt number is also sent to the NVMe-Blk driver; after the host performs the operation corresponding to the predetermined request, the method further includes: the host triggering an interrupt according to the MSI interrupt number .

In the embodiment of the present application, the host may transfer the HPA or IO queue identification (ID) or MSI interrupt number to the virtual machine based on the shared memory area of Qemu or the virtIO channel.

Referring to FIG. 6, another embodiment of the present application provides a virtualization device, including an NVMe-Blk driver 601.

The NVMe-Blk driver 601 is configured to allocate the first DMA memory from the virtual NVMe DMA memory management area when the virtual machine application issues a predetermined request; the first DMA memory includes the input queue of the input and output IO queue. The second DMA memory and the third DMA memory of the completion queue; construct the submission queue item according to the HPA of the second DMA memory, notify the NVMe device of the host to process the predetermined request; read the response information in the completion queue.

In the embodiment of the present application, the predetermined request is a write request; the NVMe-Blk driver 601 is further configured to copy the written data to the second DMA memory.

In the embodiment of the present application, the predetermined request is a read request; the NVMe-Blk driver 601 is further configured to read the read data from the third DMA memory according to the completion queue information, and read the read data The data is copied to the read buffer of the application, and the third DMA memory is released.

In the embodiment of the present application, the NVMe-Blk driver 601 is further configured to: apply to the virtual machine for a continuous first physical virtual machine memory, and transmit the virtual machine physical address GPA of the first physical virtual machine memory to The host; receiving the first correspondence between the IO queue identifier and the HPA of the first physical virtual machine memory; applying to the virtual machine for a continuous second physical virtual machine memory, and setting the second physical virtual machine memory to The DMA access method transmits the virtual machine physical address GPA of the second physical virtual machine memory to the host; the virtual NVMe DMA memory management area is created according to the received HPA of the second physical virtual machine memory.

In the embodiment of the present application, the NVMe-Blk driver 601 is further configured to: save the second correspondence between the IO queue identifier and the field of the doorbell DoorBell area in the register of the NVMe device; the queue item will be submitted The information is written in the doorbell DoorBell area of the NVMe device's register, in the corresponding field of the IO queue identifier.

In the embodiment of the present application, the NVMe-Blk driver 601 is configured to notify the NVMe device of the host to process a predetermined request in the following manner: send a doorbell write DoorBell request to the host; wherein, the write DoorBell request includes submission queue item information, That is to submit the subscript of the queue item.

In the embodiment of the present application, the NVMe-Blk driver 601 is further configured to: receive the MSI interrupt number; and read the response information in the completion queue when the interrupt corresponding to the MSI interrupt number is triggered.

In the embodiment of the present application, the NVMe-Blk driver 501 is configured to read the response information in the completion queue in the following manner: poll the completion queue to obtain the response information.

The implementation of the foregoing virtualization device is the same as the virtualization method of the foregoing embodiment, and details are not described here.

Another embodiment of the present application provides a virtualization apparatus (such as a host), including an NVMe device 602.

The NVMe device 602 is set to perform an operation corresponding to the reservation request when it learns that a reservation request needs to be processed, and put the response information into the completion queue.

In the embodiment of the present application, the reservation request is a write request, and the NVMe device 602 is set to implement the operation corresponding to the reservation request in the following manner:

Write the data in the second DMA memory to the corresponding sector of the NVMe device according to the submission queue item information.

In the embodiment of the present application, the reservation request is a read request, and the NVMe device 602 is configured to implement the operation corresponding to the reservation request in the following manner:

Read data from the corresponding sector of the NVMe device according to the submission queue item information, and copy the read data to the third DMA memory.

In the embodiment of the present application, it further includes: a controller 603 configured to convert the received GPA of the first physical virtual machine memory into a physical machine physical address HPA; create an IO queue according to the HPA of the first physical virtual machine memory , Sending the first correspondence between the IO queue identifier and the HPA of the first physical virtual machine memory to the NVMe-Blk driver; converting the received GPA of the second physical virtual machine memory into the physical machine physical address HPA, Setting the memory of the second physical virtual machine to a host DMA access mode, and sending the HPA of the memory of the second physical virtual machine to the NVMe-Blk driver.

In the embodiment of the present application, it further includes: a controller 603 configured to allocate a message interrupt MSI interrupt number to the IO queue, set the MSI interrupt number to be directly processed by the virtual machine; send the MSI interrupt number Drive the NVMe-Blk; after the NVMe device 602 performs an operation corresponding to the predetermined request, the NVMe device 602 triggers an interrupt according to the MSI interrupt number; in the embodiment of the present application, the controller 603 is further configured to place the submission queue The item information is written into the field corresponding to the IO queue identifier in the DoorBell area of the register of the NVMe device.

Another embodiment of the present application provides a virtualization device, including a processor and a computer-readable storage medium, where the computer-readable storage medium stores instructions, which are implemented when the instructions are executed by the processor Any kind of virtualization method.

Another embodiment of the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of any one of the foregoing virtualization methods are implemented.

Those of ordinary skill in the art may understand that all or some of the steps, systems, and functional modules/units in the method disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware implementation, the division between the functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be composed of multiple The physical components are executed in cooperation. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer-readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). As is well known to those of ordinary skill in the art, the term computer storage medium includes both volatile and nonvolatile implemented in any method or technology for storing information such as computer readable instructions, data structures, program modules, or other data Sex, removable and non-removable media. Computer storage media include but are not limited to Random Access Memory (RAM), Read-Only Memory (ROM), Erasable Programmable Read-Only Memory (Electrically Programmable Read Only Memory (EEPROM), Flash memory or other memory technology, compact disc (Read-Only Memory, CD-ROM), digital versatile disc (Digital Video Disc, DVD) or other optical disc storage, magnetic box, magnetic tape, magnetic disk storage or other magnetic storage devices Or any other medium that can be used to store desired information and can be accessed by a computer. In addition, it is well known to those of ordinary skill in the art that the communication medium generally contains computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and may include any information delivery medium .

Claims

A virtualization method, including:

In the case that the application of the virtual machine issues a predetermined request, the non-volatile memory host controller interface specification of the virtual machine—the block NVMe-Blk driver allocates the first DMA memory from the virtual NVMe direct memory access DMA memory management area; where, The first DMA memory includes the second DMA memory of the input queue of the input/output IO queue and the third DMA memory of the completion queue;

The NVMe-Blk driver constructs a submission queue item according to the physical machine physical address HPA of the second DMA memory, and notifies the NVMe device of the host to process the predetermined request;

The NVMe-Blk driver reads the response information in the completion queue.
The virtualization method according to claim 1, wherein the predetermined request is a write request; before the NVMe-Blk driver constructs a submission queue item according to the HPA of the second DMA memory, further comprising:

The NVMe-Blk driver copies the written data to the second DMA memory.
The virtualization method according to claim 1, wherein the predetermined request is a read request; the virtualization method further comprises:

The NVMe-Blk driver reads data from the third DMA memory according to the response information, copies the read data to an application read buffer, and releases the third DMA memory.
According to the virtualization method according to any one of claims 1 to 3, in the case where a predetermined request is issued by the application of the virtual machine, the interface specification of the non-volatile memory host controller of the virtual machine—the block NVMe-Blk driver directly from the virtual NVMe Before allocating the first DMA memory in the memory access DMA memory management area, it also includes:

The NVMe-Blk driver applies to the virtual machine for a continuous first physical virtual machine memory, and transmits the virtual machine physical address GPA of the first physical virtual machine memory to the host;

The NVMe-Blk driver receives the first correspondence between the input and output IO queue identifier and the physical machine physical address HPA in the first physical virtual machine memory;

The NVMe-Blk driver applies to the virtual machine for a continuous second physical virtual machine memory, sets the second physical virtual machine memory as a DMA access method, and sets the virtual machine physical address of the second physical virtual machine memory GPA transmission to the host;

The NVMe-Blk driver creates the virtual NVMe DMA memory management area according to the received HPA of the memory of the second physical virtual machine.
The virtualization method according to claim 4, further comprising:

The NVMe-Blk driver saves the second correspondence between the IO queue identifier and the field of the doorbell DoorBell area in the register of the NVMe device;

The processing of the predetermined request by the NVMe device informing the host includes:

The NVMe-Blk driver writes the submission queue item information into the field corresponding to the IO queue identifier in the doorbell DoorBell area of the register of the NVMe device.
The virtualization method according to claim 4, further comprising:

The NVMe-Blk driver receives the message interrupt MSI interrupt number;

The response information in the read completion queue of the NVMe-Blk driver includes:

The NVMe-Blk driver reads the response information in the completion queue when the interrupt corresponding to the MSI interrupt number is triggered.
The virtualization method according to any one of claims 1 to 3, wherein the processing of the predetermined request by the NVMe device of the notification host includes:

The NVMe-Blk driver sends a doorbell write DoorBell request to the host; wherein the write DoorBell request includes submission queue item information.
The virtualization method according to any one of claims 1 to 3, wherein the response information in the read completion queue of the NVMe-Blk driver includes:

The NVMe-Blk driver polls the completion queue to obtain the response information.
A virtualization method, including:

When the NVMe device of the host's non-volatile memory host controller interface specification learns that a predetermined request needs to be processed, it performs an operation corresponding to the predetermined request and puts the response information into the completion queue.
The virtualization method according to claim 9, wherein the predetermined request is a write request, and performing the operation corresponding to the predetermined request includes:

The NVMe device writes the data in the second direct memory access DMA memory to the corresponding sector of the NVMe device according to the submission queue item information.
The virtualization method according to claim 9, wherein the predetermined request is a read request, and performing the operation corresponding to the predetermined request includes:

The NVMe device reads data from the corresponding sector of the NVMe device according to the submission queue item information, and copies the read data to the third DMA memory.
The virtualization method according to any one of claims 9 to 11, before performing the operation corresponding to the predetermined request and placing the response information in a completion queue, further comprising:

The host converts the received virtual machine physical address GPA of the first physical virtual machine memory into a physical machine physical address HPA;

The host creates an input and output IO queue according to the HPA of the first physical virtual machine memory, and sends the first correspondence between the IO queue identifier and the HPA of the first physical virtual machine memory to the NVMe-block Blk drive;

The host converts the GPA of the memory of the second physical virtual machine into a physical address HPA of the physical machine, sets the memory of the second physical virtual machine to a host DMA access mode, and sends the HPA of the memory of the second physical virtual machine Give the NVMe-Blk driver.
The virtual machine method according to claim 12, before the host sends the first correspondence between the IO queue identifier and the HPA of the memory of the first physical virtual machine to the NVMe-Blk driver, further comprising: The host allocates a message interrupt MSI interrupt number to the IO queue, and sets the MSI interrupt number to be processed directly by the virtual machine;

When the host sends the first correspondence between the IO queue identifier and the HPA in the memory of the first physical virtual machine to the NVMe-Blk driver, the host also sends the MSI interrupt number to the NVMe-Blk driver;

After performing the operation corresponding to the predetermined request, the method further includes:

The host triggers an interrupt according to the MSI interrupt number.
The virtual machine method according to claim 12, further comprising:

The host receives the doorbell write doorbell request; wherein the doorbell write request includes information about the submission queue item;

The host writes the submission queue item information into the field corresponding to the IO queue identifier in the DoorBell area of the register of the NVMe device.
A virtualization device, including:

Non-volatile memory host controller interface specification—block NVMe-Blk driver, set to allocate the first DMA memory from the virtual NVMe direct memory access DMA memory management area when the virtual machine application issues a predetermined request; where, The first DMA memory includes a second DMA memory of the submission queue of the input/output IO queue and a third DMA memory of the completion queue; a submission queue item is constructed based on the physical machine physical address HPA of the second DMA memory, and the NVMe device is notified to process The predetermined request; read the response information in the completion queue.
A virtualization device, including:

The non-volatile memory host controller interface specification NVMe device is set to perform an operation corresponding to the predetermined request when it learns that a predetermined request needs to be processed, and put the response information into a completion queue.
A virtualization device includes a processor and a computer-readable storage medium. The computer-readable storage medium stores instructions. When the instructions are executed by the processor, any one of claims 1 to 14 is implemented. The described virtualization method.
A computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, implements the virtualization method according to any one of claims 1 to 14.