WO2024113996A1

WO2024113996A1 - Optimization method and apparatus for host io processing, device, and nonvolatile readable storage medium

Info

Publication number: WO2024113996A1
Application number: PCT/CN2023/115975
Authority: WO
Inventors: 崔健; 王江; 李树青; 李幸远; 孙华锦
Original assignee: 苏州元脑智能科技有限公司
Priority date: 2022-11-29
Filing date: 2023-08-30
Publication date: 2024-06-06
Also published as: CN115543219B; CN115543219A

Abstract

An optimization method and apparatus for host IO processing, a device, and a nonvolatile readable storage medium, relating to the field of storage. The method comprises: acquiring space sizes and an execution sequence of control blocks required by a host IO, and according to the space sizes and the execution sequence, establishing a plurality of control block rings; according to the execution sequence, filling in batches the plurality of control block rings with the control blocks required by the host IO; and in response to completing the first-batch filling of the control blocks required by the host IO, starting the execution of the control blocks required by the host IO. The present application can improve use efficiency of storage spaces, allow IOs to strike the balance between the performance and efficiency, and reduce the delay of IO processing.

Description

A method, device, equipment and non-volatile readable storage medium for optimizing host IO processing

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims priority to a Chinese patent application filed with the China Patent Office on November 29, 2022, with application number 202211508344.4 and application name “A method, device, equipment and medium for optimizing host IO processing”, all contents of which are incorporated by reference in this application.

Technical Field

The present application relates to the field of storage, and in particular to a method, device, equipment and non-volatile readable storage medium for optimizing host IO processing.

Background technique

With the rise of computational storage technology, computational storage architecture reduces the corresponding data movement operations by migrating data calculations from the host CPU to the data processing acceleration unit close to the storage unit, thereby achieving the greatest degree of system performance.

In the related technology, in a general computing acceleration chip architecture based on microcode drive, UAA (Unified Acceleration Architecture) is connected to the host through a PCIe (Peripheral Component Interconnect express, a high-speed serial computer expansion bus standard) interface. UAA is divided into two parts: a control plane and a data plane. The control plane is based on a microcode drive architecture to implement acceleration tasks, and flows between various acceleration engine modules in steps. There are problems with work efficiency and performance in the main modes of CP (Control Page) and CB (Control Block) under the UAA architecture. Optionally, as storage functions become more and more complex, the number of CBs required for an IO (Input/Output) has a clear growth trend, and the demand for storage space for CP or CP chain has increased accordingly. The number of CPs and CP chains that can be accommodated at the same time by a given size of storage space on the chip has a decreasing trend, resulting in a reduction in the ability to process parallel IO; the processing time of CB processing varies greatly. A considerable number of CBs in the CP page table and CP chain corresponding to an IO need to perform DDR (Double Data Rate), hard disk access or complex calculations, which take a long time to complete. During the execution process, the occupation of the memory by all the previous CBs that have been executed is meaningless. On the other hand, for the subsequent CBs waiting to be executed, the closer to the tail CB, the longer the waiting time, which reduces the efficiency of memory use; from the perspective of IO processing delay, according to the method of creating all CBs and then handing the first CB to UAA for execution, the delay will increase due to the increase in the number of CBs that need to be created.

Summary of the invention

In view of this, the present application proposes a method, device, equipment and non-volatile readable storage medium for optimizing host IO processing. Among them, in order to improve the utilization efficiency of CP storage space and reduce IO processing delay, the present application proposes a method for optimizing host IO processing. On the basis of the definition of a single CP and a chained CP, a ring control page table (ring CP) mode is defined. When the CP or CP chain uses the ring control page table mode, all CB storage spaces of a host IO are organized in a ring manner, namely, a control block ring (CB Ring). In the CB Ring mode, all CBs required for an IO are dynamically generated in batches according to the execution order and filled into the CB Ring. After the execution is completed, they are dynamically recycled until a complete IO process is processed. The storage space of the CB Ring can be smaller than the total space required for a complete CB sequence corresponding to an IO, thereby reducing space occupancy; the space can be recycled after the CB is executed, which improves the efficiency of storage space utilization; the use and generation of CBs are dynamic, and only part of the CB needs to be generated to start execution, which reduces the delay of the host IO.

First, based on the above purpose, an embodiment of the present application provides a method for optimizing host IO processing, the method comprising the following steps: obtaining the space size and execution order of the control blocks required by the host IO, and establishing multiple control block rings according to the space size and the execution order; filling the control blocks required by the host IO into the multiple control block rings in batches according to the execution order; in response to the completion of the first batch of filling of the control blocks required by the host IO, starting the execution of the control blocks required by the host IO.

In some embodiments, the method further includes: adding a determination of the ring mode to the definition of the control page table and generating parameter information required for the corresponding control block ring from the control page table of the ring mode to obtain an updated definition of the control page table.

In some embodiments, the determination of the ring mode is added to the definition of the control page table, and the parameter information required for generating the corresponding control block ring from the control page table of the ring mode is obtained to obtain the definition of the updated control page table, including:

The definition of the additional parameter table is added to the definition of the control page table to obtain the definition of the updated control page table, wherein the definition of the additional parameter table is used to indicate the parameters required for generating the CB in the ring mode.

In some embodiments, the updated control page table includes: a control page table header area, a control block ring control header, a control block storage area, a data cache pointer area, and an original NVMe management and IO instruction backup area.

In some embodiments, the judgment of the annular mode and the parameter information required for generating the corresponding control block ring from the control page table of the annular mode are added to the definition of the control page table, and the definition of the updated control page table includes: updating the definition of the header area of the control page table based on judging whether the current control page table is in annular mode and obtaining the position of the current control page table in the control page linked list; in response to the current control page table being a control page table of annular mode, obtaining the parameter information required for generating the corresponding control block ring from the control page table of the annular mode, and setting the corresponding parameter information in the additional parameter table based on the parameter information, and setting the additional parameter table offset address in the control page table of the annular mode to obtain the starting position of the additional parameter table of the control page table of the annular mode.

In some embodiments, the definition of updating the header area of the control page table based on determining whether the current control page table is in a ring mode and obtaining the position of the current control page table in the control page linked list includes: updating the attribute determination of the control page table based on determining whether the current control page table is in a ring mode; in response to the current control page table being a control page table in a ring mode, obtaining the position of the control page table in the ring mode in the control page linked list, and setting a pointer to the address of the first control block of the next control page table in the control page linked list in the control page table in the ring mode.

In some embodiments, obtaining the space size and execution order of the control block required by the host IO, and establishing multiple control block rings according to the space size and the execution order include: establishing multiple control page tables in a ring mode based on the definition of the updated control page table and the obtained space size and execution order of the control block required by the host IO, and generating corresponding multiple control block rings through the multiple control page tables in a ring mode.

In some embodiments, obtaining the space size and execution order of the control block required by the host IO, and establishing multiple control block rings according to the space size and the execution order also includes: setting an additional parameter table in the control block storage area of the last control page table in the control page linked list in a serial order, for storing the parameter information required for the corresponding control block ring generated by the control page table in the ring mode.

In some embodiments, an additional parameter table is set in the control block storage area of the last control page table of the control page chain list in a serial order to store parameter information required for the corresponding control block ring generated by the control page table in a ring mode, including: in response to the parameter volume of the additional parameter table being larger than the control block storage area of the last control page table, the control block storage area of the previous control page table is occupied in reverse order from the serial order.

In some embodiments, obtaining the space size and execution order of the control block required by the host IO, and establishing multiple control block rings according to the space size and the execution order also includes: setting the control block type corresponding to the mode of the control page table; establishing a control block status mark in the control block header of the control block to execute different processing logic according to different control block status marks.

In some embodiments, a status mark of a control block is established in a control block header of the control block to execute different processing logic according to the status marks of different control blocks, including: setting a control block valid flag, a completion flag, and an end flag, and in response to the valid flag being valid, the completion flag and the end flag exist.

In some embodiments, when the valid flag is 1, it indicates that the control block is a legal control block; when the valid flag is 0, it indicates that the control block is an illegal control block.

In some embodiments, when the completion flag is 1, it indicates that the control block is a control block that has been executed; when the valid flag is 0, it indicates that the control block is a control block that has not been executed.

In some embodiments, when the end flag is 1, it indicates that the control block is the control page table for processing an IO process or the last control block of the control page linked list for processing an IO process; when the end flag is 0, it indicates that the control block is not the control page table for processing an IO process or the last control block of the control page linked list for processing an IO process.

In some embodiments, setting the control block type corresponding to the mode of controlling the page table includes: setting a loopback control block flag for the table Indicates that the current control block is located at the end of the linear address of the storage space of the control block ring; sets an active generation control block flag to indicate that the control block behind the current control block has not been generated. In response to calling a control block with an active generation control flag, a generation engine is set for the control block with the active generation control flag to generate the subsequent control block and fill it into the corresponding control block ring.

In some embodiments, filling the control blocks required by the host IO into multiple control block rings in batches according to the execution order includes: creating a first batch of control page tables for the host IO according to the firmware's analysis of the host IO commands, and filling the control blocks into the control block rings corresponding to the first batch of control page tables in sequence according to the execution order; setting the firmware to pass the address of the first filled control block to the work queue of the corresponding engine for queuing and waiting.

In some embodiments, in response to the completion of the first batch of filling of control blocks required by the host IO, starting the execution of the control blocks required by the host IO includes: in response to detecting an active generation of a control block flag, sending a notification to the firmware through the work queue management engine to complete the first batch of filling of the control blocks, and reclaiming the space corresponding to the first batch of filled control page tables, and starting the execution of the control blocks required by the host IO.

On the second aspect, an embodiment of the present application also provides an optimization device for host IO processing, including the following modules: a first module is configured to obtain the space size and execution order of the control blocks required by the host IO, and establish multiple control block rings according to the space size and the execution order; a second module is configured to fill the control blocks required by the host IO into batches according to the execution order onto the multiple control block rings; a third module is configured to start the execution of the control blocks required by the host IO in response to the completion of the first batch of filling of the control blocks required by the host IO.

On the third aspect, an embodiment of the present application also provides a computer device, including at least one processor; and a memory, the memory storing computer instructions that can be executed on the processor, and the instructions implement the steps of any of the above methods when executed by the processor.

In a fourth aspect, an embodiment of the present application further provides a non-volatile readable storage medium, which stores a computer program that implements any of the above method steps when executed by a processor.

The present application has at least the following beneficial effects: the present application proposes a method, device, equipment and non-volatile readable storage medium for optimizing host IO processing, wherein the method for optimizing host IO processing proposed by the present application allows the storage space of the control block ring to be smaller than the total space required for a complete control block sequence corresponding to a host IO by establishing a control block ring, thereby reducing the demand for storage space of the control page table; the control block ring generated under the control page table in the ring mode is based on the granularity of the control page table, so it is flexible, that is, it allows the host IO to balance between performance and efficiency; the control block running on the corresponding control block ring under the control page table in the ring mode reclaims the corresponding space after execution, thereby improving the efficiency of storage space utilization; and the control block is dynamically generated and recycled, so in a host IO, only part of the control block needs to be generated to start serving the IO, thereby reducing the processing delay of the IO; further, by setting the loopback control block flag and the active generation control block flag, the CB execution and generation are allowed to proceed synchronously, that is, the time consumption of CB generation is hidden in the execution process of other CBs.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other embodiments can be obtained based on these drawings without paying creative work.

FIG1 is a schematic diagram showing a general computing architecture;

FIG2 is a schematic diagram showing an embodiment of a method for processing host IO in the related art;

FIG3 is a schematic diagram showing the structure of a single control page table in the related art;

FIG4 is a schematic diagram showing the structure of a chain control page table in the related art;

FIG5 is a schematic diagram showing an embodiment of a method for optimizing host IO processing provided by the present application;

FIG6 is a schematic diagram showing another embodiment of a method for optimizing host IO processing provided by the present application;

FIG. 7 is a schematic diagram showing active triggering and triggering scenarios in a method for optimizing host IO processing provided by the present application;

FIG8 is a schematic diagram showing an embodiment of a device for optimizing host IO processing provided by the present application;

FIG9 is a schematic diagram showing an embodiment of a computer device provided by the present application;

FIG. 10 is a schematic diagram showing an embodiment of a non-volatile readable storage medium provided in the present application.

Detailed ways

The following describes embodiments of the present application. However, it should be understood that the disclosed embodiments are merely examples and other embodiments may take various alternative forms.

In addition, it should be noted that all expressions using "first" and "second" in the embodiments of the present application are intended to distinguish two entities with the same name but not the same or non-identical parameters. It can be seen that "first" and "second" are only for the convenience of expression and should not be understood as limitations on the embodiments of the present application. The subsequent embodiments will not explain this one by one. The terms "include", "comprise" or any other variation thereof are intended to cover non-exclusive inclusions, so that a process, method, article or device containing a series of elements includes not only those elements, but may also include elements that are not explicitly listed or inherent to these processes, methods, articles or devices.

One or more embodiments of the present application will be described below with reference to the accompanying drawings.

Referring to FIG. 1 and FIG. 2, FIG. 1 shows a schematic diagram of a universal computing architecture, and FIG. 2 shows a schematic diagram of an embodiment of a method for processing host IO in the related art. In the universal computing architecture (UAA), the size of the control page table (CP) can be 512 bytes, 1024 bytes or 2048 bytes. Except for the CP header, data cache and original host IO instruction backup, the remaining space is used to store the control block CB. All CPs are placed in a continuous memory space, thereby forming a CP resource pool. In addition, in order to facilitate the management of the CP resource pool, the CP granularity in a single resource pool needs to be consistent, that is, only one CP size can be selected. The size of CB can be 16 bytes, 32 bytes, 64 bytes and 128 bytes according to different application engine types. Therefore, the number of CBs that can be carried in a single CP is jointly determined by the size of CP and the size of CB. Therefore, in some complex application scenarios, the length difference of CB chains corresponding to different types of IO is too large, and the selection of CP size will cause certain difficulties. If it is too small, it cannot carry a long CB chain; on the contrary, if the CP granularity is too large, it will cause a certain waste of CP resources.

The process of a typical host IO process in UAA is as follows: First, AEM (Acceleration Engine Manager) retrieves the original IO request in accordance with a certain interface protocol, and notifies the firmware through the hardware event queue managed by WQS (Work Queue Scheduler). After receiving the notification, the firmware parses the IO command, creates a CP for this IO operation, and fills the CB of the step-by-step operation into the CP as required. After completing the above steps, the firmware passes the address of the first CB1 (4B wide) to WQS, and WQS puts the CB1 address into the work queue of the corresponding engine for waiting. When the corresponding engine starts to execute the CB1, it performs the corresponding acceleration operation according to the configuration information of CB1. After completion, the execution status is returned to WQS (optionally, the firmware can be notified), and the entry address of the next CB2 is prepared, which is also returned to WQS to enter the work queue of the corresponding engine and wait, and so on. The entire transfer process is under the control of WQS hardware, and the firmware does not need to be involved unless necessary. When the last CBX is completed, it is necessary to respond to the host and notify the firmware to reclaim the CP space.

On the basis of FIG. 1 and FIG. 2, refer to FIG. 3 and FIG. 4. FIG. 3 shows a schematic diagram of the structure of a single control page table in the related art; FIG. 4 shows a schematic diagram of the structure of a chain control page table in the related art. As shown in FIG. 3 and FIG. 4, CP is generally divided into ordinary CP and chain CP. Chain CP is introduced on the basis of single CP. Ordinary CP adds an 8-byte address pointer pointing to the next chain CP on the basis of the definition of single CP; the size of chain CP is consistent with that of ordinary CP. In chain CP, except for the CP header, the remaining space is used to store CB. Three 8-byte chain pointers are added in the chain CP header, one pointing to the next chain CP (if any), one pointing to the previous CP, and the last one directly pointing to the first ordinary CP (for convenient search of data cache, etc.), and each CB identifies its position in the entire CP chain, so that the engine can quickly locate the required information when processing. However, with the development of UAA architecture, single CP and chain CP have encountered work efficiency and performance problems. Alternatively, as storage functions become more and more complex, the number of CBs required for an IO has a clear growth trend, and with it the demand for storage space for CPs or CP chains. The number of CPs and CP chains that can be accommodated at the same time by a given size of storage space on the chip is decreasing, resulting in a reduction in the ability to process parallel IO. The processing time of CB processing varies greatly. A considerable number of CBs in the CP page table and CP chain corresponding to an IO require DDR (Double Data Rate), hard disk access or complex calculations, which take a long time to complete. During the execution process, the occupation of memory by all the preceding CBs that have been executed is meaningless. On the other hand, for the subsequent CBs waiting to be executed, the closer to the end of the CB, the longer the waiting time, which reduces the efficiency of memory use. From the perspective of IO processing delay, according to the method of creating all CBs and then handing over the first CB to UAA for execution, the delay will increase due to the increase in the number of CBs that need to be created.

Based on the above objectives, the first aspect of the embodiment of the present application proposes an embodiment of a method for optimizing host IO processing. FIG5 shows a schematic diagram of an embodiment of a method for optimizing host IO processing provided by the present application. As shown in FIG5, an optimization method for host IO processing in an embodiment of the present application includes the following steps:

S1, obtain the space size and execution order of the control block required by the host IO, and establish multiple control block rings according to the space size and execution order;

S2, filling the control blocks required by the host IO into multiple control block rings in batches according to the execution order;

S3. In response to the completion of the first batch of filling of the control blocks required by the host IO, the execution of the control blocks required by the host IO is started.

Based on the above purposes, the first aspect of the embodiment of the present application proposes an embodiment of a method for optimizing host IO processing. FIG6 shows a schematic diagram of another embodiment of a method for optimizing host IO processing provided by the present application. As shown in FIG6, on the basis of ordinary CP and chained CP, a ring CP mode is proposed. To this end, a definition of the ring CP mode is added on the basis of the original CP and CB definitions. Optionally, in order to indicate whether the CP has a ring CP mode, a mark of the ring CP mode is added to the CP attributes; secondly, in order to dynamically generate other parameters that may be required for CB in the ring CP mode, the definition of an additional parameter table is added to the definition of the CP page in the ring CP; the size of the modified single CP page table can be 512 bytes, 1K bytes or 2K bytes, and it mainly consists of five parts.

1. 128 bytes of CP header area (control page header area), including:

(1) CP attributes: the type of CP, and whether the CP is in a CP linked list, whether it is a ring CP page mode, and its position in the linked list;

(2) The serial number of the current CP, which is specified when the CP is created;

(3) Information related to the NVMe (Non-Volatile Memory Express) queue, including a 2-byte completion queue ID, a 2-byte submission queue ID, and a 4-byte training queue header information;

(4) If a CP linked list is formed, the 8 bytes include the address pointer to the first CB of the next linked CP;

(5) Four 64-bit timestamps, used to record key time nodes during the execution of the CP;

(6) 64 bytes of space are reserved for firmware.

2. When the CP attribute is the ring CP page mode, it contains 32 bytes of CB Ring control header (control block ring control header):

(1) Additional parameter table offset address 4 bytes, indicating the starting address of the additional parameter table in a single CP, that is, the relative offset address starting from the address of the first CP;

(2) 28 bytes of reserved space.

3. A CB storage area (control block storage area) of several bytes corresponds to a CP length of 512B/1KB/2KB. When the CP is a circular CP page table, the length of the CB storage area is 192 bytes/704 bytes/1728 bytes respectively; when the CP is in other modes, the length of the CB storage area is 224 bytes/738 bytes/1760 bytes respectively.

4. 64-byte data cache pointer area, used to point to the common data cache area.

5. 64-byte raw NVMe management and IO instruction backup area, which facilitates firmware intervention and error recovery when an exception occurs.

Optional, as shown in Table 1 below:

Table 1

In the chained CP, the additional parameter table is located in the CB storage area starting from the last CP. If the additional parameter table itself is large, it will occupy the CB storage area of the previous CP in reverse order from the last CP, and the remaining CB storage areas in the CP chain will form a CB ring.

The modification of CB includes adding CB header flag definition and two new CB types. Optional, including:

CB validity flag, 1 indicates a legal CB, 0 indicates an illegal CB;

CB completion flag, 1 indicates that the CB has been executed, and 0 indicates that the CB has not been executed;

Last CB flag, 1 indicates that it is the last CB in the CP or CP chain that processes an IO process, and 0 indicates that it is not the last CB.

The above CB completion flag and Last CB flag are only meaningful when the CB valid flag is 1.

Taking the above modification of CB definition as an example, the ring CP mode is supported. In addition to the original CB types, two new CB types are added in the ring:

Loopback CB, i.e. CB Return (CB_R), means that the current CB is located at the end of the linear address of the CB Ring storage space, and the next CB needs to be obtained from the starting position of the CB storage space of the first CP. For the ring CP mode, CB_R must be used.

The CB that actively generates CB is CB Builder (CB_B). When this CB appears in the CB Ring, it means that the subsequent CB of this CB has not been generated yet. When WQS (Work Queue Scheduler) schedules to this CB, it needs to give the CB generation engine (implemented by software or hardware) to process this CB, generate subsequent CBs and fill them into the CB Ring. For the ring CP mode, CB_B is optional.

Among them, FIG7 shows a schematic diagram of active triggering and triggering scenarios in an optimization method for host IO processing provided by the present application. Under the UAA framework, WQS is responsible for distributing CBs, and a dedicated software or hardware engine processes the corresponding CBs, and returns the processing results and the address of the next CB to WQS. In CB Ring mode, when the hardware engine that processes the CB submits the next CB, the different CB valid flags, CB completion flags, and Last CB flags in the next CB header are detected to complete different logics such as CB filling and CB processing. The optional processing logic is as follows:

1. WQS receives the processing result of the current CB and first determines whether the current CB is the last CB, that is, whether the Last CB flag of the current CB is 1. If the Last CB flag of the current CB is 1, the IO process ends and the CP is recycled; otherwise, it determines whether the next CB of the current CB is valid.

2. Determine whether the next CB of the current CB is valid, that is, determine whether the valid flag of the next CB of the current CB is 1. If the valid flag of the next CB of the current CB is not 1 (is 0), although the IO process is not completed at this time, due to the lack of subsequent CBs, the firmware is notified to perform subsequent filling processing. The firmware fills the subsequent CBs, updates the CB Ring, and hands the next CB to WQS scheduling. This process is the construction of a passively triggered CB. and recycling; if the valid flag of the next CB of the current CB is 1, determine whether the next CB of the current CB has been completed.

3. Determine whether the next CB of the current CB has been completed, that is, determine whether the completion flag of the next CB of the current CB is 1. If the completion flag of the next CB of the current CB is 1, although the IO process is not completed at this time, due to the lack of subsequent CBs, the firmware is notified to perform subsequent filling processing. The firmware fills the subsequent CBs, updates the CB Ring, and hands the next CB to the WQS scheduling. This process is the construction and recycling of passively triggered CBs; if the completion flag of the next CB of the current CB is not 1, determine whether the next CB of the current CB is CB_B.

4. Determine whether the next CB of the current CB is CB_B. If the next CB of the current CB is CB_B, WQS distributes CB_B to the CB generation engine, and the CB generation engine (SW/HW) fills the subsequent CB. After the filling is completed, it replies to the WQS completion status and hands the subsequent CB to the WQS for scheduling. This process is the actively triggered CB construction and recovery process; if the next CB of the current CB is not CB_B, it is scheduled to the corresponding engine according to the content of the CB.

When the CB engine finds that the next CB is CB_R, it needs to perform additional processing, that is, relocate to the first CB of the first CP and submit this CB as the next CB to WQS. The dynamic generation of CB can be achieved by the firmware in the original UAA architecture. It only needs to change the one-time generation of all CBs to batch generation. In order to improve real-time performance and reduce the impact of time uncertainty during software execution, a dedicated hardware engine can be designed and implemented for dynamic creation and recycling of CBs.

A typical IO command processing flow working in the ring CP mode includes: First, AEM will follow a certain interface protocol to retrieve the original IO request and notify the firmware through the hardware event queue managed by WQS. After the firmware is notified, it will parse the IO command, create the first batch of CPs for this IO operation, and fill the CBs of the step-by-step operation into the CP as required. After completing the above steps, the firmware will pass the address of the first CB1 (4B wide) to WQS, and WQS will put the CB1 address into the work queue of the corresponding engine for waiting. When an engine processing CB detects CB_B, it will be submitted to the CB generation engine for processing through WQS. When an engine processing CB detects a passive CB filling trigger scenario, it sends a message to the firmware through WQS, and the firmware completes the CB filling. After the firmware fills, it is still handed over to WQS for distribution. When the last CBX is completed, it is necessary to respond to the host and notify the firmware to reclaim the CP space. Active trigger scenarios and passive scenarios are allowed to appear one after another in an IO process, or only one can appear.

The above method allows the storage space of the control block ring to be smaller than the total space required for the complete control block sequence corresponding to a host IO, thereby reducing the demand for storage space for the control page table; the control block ring generated under the control page table in the ring mode is based on the granularity of the control page table, so it is flexible, that is, it allows the host IO to balance between performance and efficiency; the control block running on the corresponding control block ring under the control page table in the ring mode reclaims the corresponding space after execution, thereby improving the efficiency of storage space utilization; and the control block is dynamically generated and recycled, so in a host IO, only part of the control block needs to be generated to start serving the IO, thereby reducing the processing delay of the IO; by setting the loopback control block flag and the active generation control block flag, the CB execution and generation are allowed to proceed synchronously, that is, the time consumption of CB generation is hidden in the execution process of other CBs.

The second aspect of the embodiment of the present application proposes an optimization device for host IO processing. FIG8 shows a schematic diagram of an embodiment of an optimization device for host IO processing provided by the present application. As shown in FIG8, an optimization device for host IO processing provided by the present application includes: a first module 011, which is configured to obtain the space size and execution order of the control block required by the host IO, and establish multiple control block rings according to the space size and execution order; a second module 012, which is configured to fill the control blocks required by the host IO into multiple control block rings in batches according to the execution order; a third module 013, which is configured to complete the first batch of filling in response to the control blocks required by the host IO, and start the execution of the control blocks required by the host IO.

In some embodiments, the optimization device configured for host IO processing also includes being configured to: add a judgment of a ring mode to the definition of a control page table and generate parameter information required for a corresponding control block ring from the control page table of the ring mode, to obtain an updated definition of the control page table.

In some embodiments, the optimization device configured to process host IO also includes being configured to: update the definition of the header area of the control page table based on determining whether the current control page table is in a ring mode and obtaining the position of the current control page table in the control page linked list; in response to the current control page table being a control page table in a ring mode, obtain the parameter information required for the control page table in the ring mode to generate the corresponding control block ring, and set the corresponding parameter information in the additional parameter table based on the parameter information, and set the additional parameter table offset address in the control page table in the ring mode to obtain the starting position of the additional parameter table of the control page table in the ring mode.

In some embodiments, the optimization device configured to process host IO also includes being configured to: update the attribute judgment of the control page table based on whether the current control page table is in a ring mode; in response to the current control page table being a control page table in a ring mode, obtain the position of the control page table in the ring mode in the control page linked list, and set a pointer to the address of the first control block of the next control page table in the control page linked list in the control page table in the ring mode.

In some embodiments, the first module 011 is further configured to: establish multiple ring-mode control page tables based on the definition of the updated control page table and the space size and execution order of the control block required by the acquired host IO, and generate corresponding multiple control block rings through the multiple ring-mode control page tables.

In some embodiments, the first module 011 is further configured to: set an additional parameter table in the control block storage area of the last control page table in the control page chain list in a serial order to store parameter information required for the corresponding control block ring generated by the control page table in the ring mode.

In some embodiments, the first module 011 is further configured to: in response to the parameter volume of the additional parameter table being larger than the control block storage area of the last control page table, occupy the control block storage area of the previous control page table in reverse order of the serial connection sequence.

In some embodiments, the first module 011 is further configured to: set a control block type corresponding to the mode of the control page table; establish a control block status mark in a control block header of the control block to execute different processing logics according to different control block status marks.

In some embodiments, the first module 011 is further configured to: set a control block valid flag, a completion flag, and an end flag, and in response to the valid flag being valid, the completion flag and the end flag being present.

In some embodiments, the first module 011 is further configured to: set a loopback control block flag to indicate that the current control block is located at the end of the linear address of the storage space of the control block ring; set an active generation control block flag to indicate that the control block located after the current control block has not been generated, and in response to calling a control block with an active generation control flag, set a generation engine for the control block with the active generation control flag to generate the subsequent control block and fill it into the corresponding control block ring.

In some embodiments, the second module 012 is further configured to: create a first batch of control page tables for the host IO according to the firmware's analysis of the host IO command, and fill the control blocks into the control block rings corresponding to the first batch of control page tables in sequence according to the execution order; set the firmware to pass the address of the first filled control block to the work queue of the corresponding engine for queuing and waiting.

In some embodiments, the third module 013 is further configured to: in response to detecting an active generation of a control block flag, send a notification to the firmware through the work queue management engine to complete the first batch of filling of the control block, and reclaim the space corresponding to the first batch of filled control page tables, and start the execution of the control block required for the host IO.

Based on the above objectives, the third aspect of the embodiment of the present application proposes a computer device, and FIG9 shows a schematic diagram of an embodiment of a computer device provided by the present application. As shown in FIG9, an embodiment of a computer device provided by the present application includes the following modules: at least one processor 021; and a memory 022, the memory 022 stores computer instructions 023 that can be run on the processor 021, and the computer instructions 023 implement the steps of the above method when executed by the processor 021.

The present application also provides a non-volatile readable storage medium. Figure 10 shows a schematic diagram of an embodiment of a non-volatile readable storage medium provided by the present application. As shown in Figure 10, the non-volatile readable storage medium 031 stores a computer program 032 that performs the above method when executed by a processor.

Finally, it should be noted that a person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program of the method for setting system parameters can be stored in a non-volatile readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the non-volatile readable storage medium of the program can be a magnetic disk, an optical disk, a read-only storage memory (ROM) or a random access memory (RAM), etc. The above-mentioned computer program embodiments can achieve the same or similar effects as the corresponding above-mentioned arbitrary method embodiments.

In addition, the method disclosed in the embodiment of the present application can also be implemented as a computer program executed by a processor, and the computer program can be stored in a non-volatile readable storage medium. When the computer program is executed by the processor, the above functions defined in the method disclosed in the embodiment of the present application are performed.

In addition, the above method steps and system units may also be implemented using a controller and a non-volatile readable storage medium configured to store a computer program that enables the controller to implement the above steps or unit functions.

It will also be appreciated by those skilled in the art that various exemplary logic blocks, modules, circuits and algorithmic steps described in conjunction with the disclosure herein can be implemented as electronic hardware, computer software or a combination of the two. In order to clearly illustrate this interchangeability of hardware and software, a general description has been given to the functions of various schematic components, blocks, modules, circuits and steps. Whether this function is implemented as software or implemented as hardware depends on application and the design constraints imposed on the entire system. Those skilled in the art can implement the function in various ways for each application, but this implementation decision should not be interpreted as causing a departure from the disclosed scope of the present application embodiment.

In one or more exemplary designs, the functions may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored as one or more instructions or codes on or transmitted through a non-volatile readable storage medium. Non-volatile readable storage media include computer storage media and communication media, which include any media that facilitate the transfer of computer programs from one location to another. Non-volatile readable storage media may be any available media that can be accessed by a general or special purpose computer. As an example and not limitation, the non-volatile readable storage medium may include RAM, ROM, EEPROM, CD ROM or other optical disk storage devices, magnetic disk storage devices or other magnetic storage devices, or any other medium that can be configured to carry or store the required program code in the form of instructions or data structures and can be accessed by a general or special purpose computer or a general or special purpose processor. In addition, any connection may be appropriately referred to as a non-volatile readable storage medium. For example, if a coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwaves are used to transmit software from a website, server, or other remote source, then the coaxial cable, fiber optic cable, twisted pair, DOL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of medium. As used herein, disks and optical disks include compact disks (CDs), laser disks, optical disks, digital versatile disks (DVDs), floppy disks, and Blu-ray disks, where disks typically reproduce data magnetically, while optical disks reproduce data optically using lasers. Combinations of the above should also be included within the scope of non-volatile readable storage media.

The above are exemplary embodiments disclosed in the present application, but it should be noted that various changes and modifications may be made without departing from the scope disclosed in the embodiments of the present application as defined in the claims. The functions, steps and/or actions of the method claims according to the disclosed embodiments described herein do not need to be performed in any particular order. In addition, although the elements disclosed in the embodiments of the present application may be described or required in individual form, they may also be understood as multiple unless explicitly limited to the singular.

It should be understood that, as used herein, the singular forms "a", "an" are intended to include the plural forms as well, unless the context clearly supports an exception. It should also be understood that, as used herein, "and/or" refers to any and all possible combinations including one or more of the associated listed items.

The serial numbers of the embodiments disclosed in the above-mentioned embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

A person of ordinary skill in the art will appreciate that all or part of the steps to implement the above embodiments may be accomplished by hardware, or may be accomplished by instructing related hardware through a program, and the program may be stored in a non-volatile readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk, or an optical disk, etc.

Those skilled in the art should understand that the discussion of any of the above embodiments is only exemplary and is not intended to imply that the scope of the disclosure of the embodiments of the present application (including the claims) is limited to these examples; under the idea of the embodiments of the present application, the technical features in the above embodiments or different embodiments can also be combined, and there are many other changes in different aspects of the embodiments of the present application as above, which are not provided in detail for the sake of simplicity. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the protection scope of the embodiments of the present application.

Claims

A method for optimizing host IO processing, characterized by comprising:

Obtaining the space size and execution order of the control block required by the host IO, and establishing multiple control block rings according to the space size and execution order;

Filling the control blocks required by the host IO into the multiple control block rings in batches according to the execution order;

In response to the first batch of filling of the control blocks required by the host IO is completed, the execution of the control blocks required by the host IO is started.
The method according to claim 1, characterized in that the method further comprises:

The determination of the ring mode is added to the definition of the control page table, and parameter information required for generating a corresponding control block ring from the control page table of the ring mode is used to obtain the definition of the updated control page table.
The method according to claim 2 is characterized in that the step of adding the judgment of the ring mode to the definition of the control page table and generating the parameter information required for the corresponding control block ring from the control page table of the ring mode to obtain the definition of the updated control page table comprises:

The definition of the additional parameter table is added to the definition of the control page table to obtain the definition of the updated control page table, wherein the definition of the additional parameter table is used to indicate the parameters required for generating the CB in the ring mode.
The method according to claim 3 is characterized in that the updated control page table includes: a control page table header area, a control block ring control header, a control block storage area, a data cache pointer area, and an original NVMe management and IO instruction backup area.
The method according to claim 2 is characterized in that the adding of the judgment of the ring mode in the definition of the control page table and the parameter information required for generating the corresponding control block ring from the control page table of the ring mode to obtain the definition of the updated control page table comprises:

Based on determining whether the current control page table is in a ring mode and obtaining the position of the current control page table in the control page linked list, a definition of a header area of the control page table is updated;

In response to the fact that the current control page table is a control page table in an annular mode, parameter information required for the control page table in annular mode to generate a corresponding control block ring is obtained, and corresponding parameter information in an additional parameter table is set based on the parameter information, and an additional parameter table offset address is set in the control page table in annular mode to obtain the starting position of the additional parameter table of the control page table in annular mode.
The method according to claim 5 is characterized in that the updating of the definition of the header area of the control page table based on determining whether the current control page table is in a ring mode and obtaining the position of the current control page table in the control page linked list comprises:

Update the attribute judgment of the control page table based on whether the current control page table is in a ring mode;

In response to the current control page table being a ring-mode control page table, the position of the ring-mode control page table in the control page chain is obtained, and a pointer to the address of the first control block of the next control page table in the control page chain is set in the ring-mode control page table.
The method according to claim 2, characterized in that the obtaining of the space size and execution order of the control block required by the host IO, and establishing multiple control block rings according to the space size and execution order comprises:

Based on the definition of the updated control page table and the acquired space size and execution order of the control block required by the host IO, multiple control page tables in annular modes are established, and corresponding multiple control block rings are generated through the multiple control page tables in annular modes.
The method according to claim 7, characterized in that the obtaining of the space size and execution order of the control blocks required by the host IO, and establishing multiple control block rings according to the space size and execution order further comprises:

An additional parameter table is set in the control block storage area of the last control page table of the control page chain list in a serial connection order to store parameter information required by the corresponding control block ring generated by the control page table in the ring mode.
The method according to claim 8 is characterized in that the additional parameter table is set in the control block storage area of the last control page table of the control page chain table in the serial order to store the parameter information required for the corresponding control block ring generated by the control page table in the ring mode, including:

In response to the parameter volume of the additional parameter table being larger than the control block storage area of the last control page table, the control block storage area of the previous control page table is occupied in reverse order according to the serial connection sequence.
The method according to claim 2 is characterized in that the space size of the control block required for obtaining the host IO and the execution sequence Sequence, establishing multiple control block rings according to the space size and the execution order also includes:

Set the control block type corresponding to the mode of controlling the page table;

A status flag of the control block is established in the control block header of the control block, so as to execute different processing logics according to different status flags of the control block.
The method according to claim 10, characterized in that the step of establishing a status flag of the control block in the control block header of the control block so as to execute different processing logics according to different status flags of the control block comprises:

A control block valid flag, a completion flag, and an end flag are set, and in response to the valid flag being valid, the completion flag and the end flag exist.
The method according to claim 11 is characterized in that, when the valid flag is 1, it indicates that the control block is a legal control block; when the valid flag is 0, it indicates that the control block is an illegal control block.
The method according to claim 11 is characterized in that, when the completion flag is 1, it indicates that the control block is a control block that has been executed; when the valid flag is 0, it indicates that the control block is a control block that has not been executed.
The method according to claim 11 is characterized in that, when the end flag is 1, it indicates that the control block is a control page table for processing an IO process or the last control block of the control page linked list for processing an IO process; when the end flag is 0, it indicates that the control block is not a control page table for processing an IO process or the last control block of the control page linked list for processing an IO process.
The method according to claim 11, characterized in that the setting of the control block type corresponding to the mode of controlling the page table comprises:

The loopback control block flag is set to indicate that the current control block is located at the end of the linear address of the storage space of the control block ring;

An active generation control block flag is set to indicate that the control block behind the current control block has not been generated. In response to calling a control block with an active generation control flag, a generation engine is set for the control block with the active generation control flag to generate the subsequent control blocks and fill them into the corresponding control block ring.
The method according to claim 15, characterized in that the step of filling the control blocks required by the host IO into the multiple control block rings in batches according to the execution order comprises:

Creating a first batch of control page tables for the host IO according to the firmware's analysis of the host IO command, and sequentially filling control blocks into control block rings corresponding to the first batch of control page tables according to the execution order;

The firmware is configured to pass the address of the first filled control block to the work queue of the corresponding engine for queuing.
The method according to claim 16, wherein in response to the control blocks required by the host IO completing the first batch of filling, starting the execution of the control blocks required by the host IO comprises:

In response to detecting the active generation of the control block flag, a notification is sent to the firmware through the work queue management engine to complete the first filling of the control block, and the space corresponding to the first filled control page table is reclaimed to start the execution of the control block required by the host IO.
A device for optimizing host IO processing, characterized by comprising:

The first module is configured to obtain the space size and execution order of the control block required by the host IO, and establish multiple control block rings according to the space size and execution order;

A second module is configured to fill the control blocks required by the host IO into the multiple control block rings in batches according to the execution order;

The third module is configured to start the execution of the control blocks required by the host IO in response to the completion of the first batch of filling of the control blocks required by the host IO.
A computer device, comprising:

at least one processor; and

A memory storing computer instructions executable on the processor, wherein the instructions, when executed by the processor, implement the steps of the method according to any one of claims 1 to 17.
A non-volatile readable storage medium storing a computer program, wherein the computer program implements the steps of the method according to any one of claims 1 to 17 when executed by a processor.