CN107145399A

CN107145399A - A kind of shared-memory management method and shared-memory management equipment

Info

Publication number: CN107145399A
Application number: CN201710268053.5A
Authority: CN
Inventors: 张伟; 李清清
Original assignee: Huawei Technologies Co Ltd
Current assignee: XFusion Digital Technologies Co Ltd
Priority date: 2017-04-22
Filing date: 2017-04-22
Publication date: 2017-09-08
Anticipated expiration: 2037-04-22
Also published as: CN107145399B

Abstract

The embodiment of the invention discloses a kind of shared-memory management method and shared-memory management equipment, this method is applied in message zero-copy system, the message zero-copy system operation is in operating system, the message zero-copy system includes at least two application APP s, the shared first bag caching pbuf internal memories of APPs, this method includes：Any application APP in APPs sets the APP to protect attribute to be read-only at current time to the read-write operation of the first pbuf internal memories；If the APP receives the pbuf internal memory maloperation notification messages that the operating system is sent, the pbuf internal memory maloperation field datas in the pbuf internal memories maloperation notification message are obtained；Internal memory maloperation scene is determined based on the pbuf internal memories maloperation field data.So as to which when APP carries out maloperation to shared pbuf internal memories, the problem of being accurately positioned generation maloperation is live, safeguards system reliability.

Description

A kind of shared-memory management method and shared-memory management equipment

Technical field

The present invention relates to the communications field, and in particular to a kind of shared-memory management method and shared-memory management equipment.

Background technology

Network function virtualization (English：Network Function Virtualization, NFV) system pass through it is soft or hard Part is decoupled and function modeling, network equipment function is no longer dependent on specialized hardware, resource fully can flexibly be shared, and is realized new The quick exploitation and deployment of business.The basis of NFV technologies includes cloud computing technology and virtualization technology, general calculating, storage, The hardware devices such as network can be decomposed into a variety of virtual resources by virtualization technology, so that the various application programs in upper strata are used.

It is (English when running multiple application programs in NFV systems：Application, APP) when, in order to realize in difference Application process between carry out message forwarding, the zero duplication technology of shared drive mechanism, namely different application journey can be used Sequence shares same bag caching (English：Packet buffer, pbuf) internal memory, so as to avoid between different application Carry out copy function, the performance of lifting system.But when using shared drive mechanism simultaneously, because each Application sharing is same Pbuf internal memories, so the isolation between destroying application program, causes when there are pbuf internal memory maloperation problems, it is impossible to accurate Position memory problem scene.

The content of the invention

The embodiments of the invention provide a kind of shared-memory management method and shared-memory management equipment, with APP to altogether When enjoying the progress maloperation of pbuf internal memories, the problem of being accurately positioned generation maloperation scene.

In a first aspect, the embodiment of the present invention provides a kind of shared-memory management method, this method is applied to message zero-copy In system, the message zero-copy system operation is in operating system, and the message zero-copy system includes at least two application programs APPs, the APPs pbuf internal memories of shared drive the first, this method carry out shared-memory management by following steps.APPs first In any application APP set the APP to protect attribute to be read-only at current time to the read-write operations of the first pbuf internal memories； Then when APP receives the pbuf internal memory maloperation notification messages of operating system transmission, obtain the maloperation of pbuf internal memories and notify Pbuf internal memory maloperation field datas in message, the pbuf internal memory maloperations notification message is used to remind APP to the first pbuf Internal memory carries out write operation and the particular location of the write operation；Last APP determines APP based on pbuf internal memory maloperation field datas The internal memory maloperation scene of write operation is carried out to the first pbuf internal memories.So as to realize same pbuf internal memories are shared in multiple APP In the case of, when internal memory maloperation occur in the shared pbuf internal memories, it can quickly and accurately position and produce pbuf internal memories behaviour by mistake The APP and particular location of work, that is, determine pbuf internal memories maloperation scene, safeguards system reliability.

In a possible design, the operating system includes (SuSE) Linux OS, and the (SuSE) Linux OS includes Mprotect memory management interfaces, the mprotect memory management interfaces are used for the called read-write to set the first pbuf internal memories Operational attribute；So as to which APP is by calling the mprotect memory management interfaces to set APP to the read-write operations of the first pbuf internal memories Attribute is read-only at current time.So as to by calling the mprotect memory management interfaces of (SuSE) Linux OS to facilitate Ground is configured to the read-write operation attribute of the first pbuf internal memories under a linux operating system.

In a possible design, mprotect memory management interfaces are used to be called with based on one or more by APP Read-write operation attributes of the page table setting APP to the first pbuf internal memories.So that the mprotect internal memories based on page table management are read Write operation administrative mechanism independently can be controlled to the first pbuf internal memories.

In a possible design, the message zero-copy system also includes the 2nd pbuf internal memories, APPs shared drive Also include the 2nd pbuf internal memories, wherein, saved as in the first pbuf in small page internal memory, the 2nd pbuf and save as big page internal memory, First pbuf internal memories and one or more page table existence anduniquess corresponding relations.APP determines the first pbuf internal memories in predetermined time For APPs shared drive, APP determines to save as APPs shared drive in the 2nd pbuf outside predetermined time.When this is default The frequency at quarter is got over hour, then the frequency of explanation progress pbuf internal memory monitorings is smaller, so now system reliability is relatively small, and The relative degree being now filled using the pbuf internal memories of big page internal memory the 2nd is more, and the performance boost of system is larger；And when default The frequency at quarter is bigger, then the frequency of explanation progress pbuf internal memory monitorings is bigger, so now the reliability of system is relatively large, and The relative degree being now filled using the pbuf internal memories of big page internal memory the 2nd is small, and the performance boost of system is smaller.Therefore can root According to demand, the performance boost and reliability for weighing system set the predetermined time.

In a possible design, the first pbuf internal memories are additionally operable to cache track identification, and APP is also marked by the track Memorize records the tracing point information of the first pbuf internal memories, and the tracing point information includes institute's application program identification, application program when advance Journey is identified and current time.By the first pbuf determined in the first pbuf tracing point information track, come further really The upstream APP that fixed first pbuf is passed through, so that it is determined that APP carries out the internal memory maloperation of write operation to the first pbuf internal memories Behind scene, pbuf memory problem investigations can be further carried out by the upstream APP of the APP, it is reliable further to improve system Property.

In a first aspect, the embodiment of the present invention provides a kind of shared-memory management method, this method is applied to operating system, should Operation has message zero-copy system in operating system, and the message zero-copy system includes at least two application APP s, APPs Including the first packet buffer pbuf internal memories, this method realizes shared-memory management by following steps.First operating system according to Any application APP in APPs is set to the read-write operation attribute of the first pbuf internal memories indicates record APP to the first pbuf The read-write operation attribute of internal memory is read-only；So as to carry out write operation, operation to the first pbuf internal memories when operating system monitors APP System sends pbuf internal memory maloperation notification messages to APP, and the pbuf internal memory maloperations notification message is used to remind APP to first Pbuf internal memories carry out write operation and the particular location of the write operation.So as to realize same pbuf internal memories are shared in multiple APP In the case of, when internal memory maloperation occur in the shared pbuf internal memories, it can quickly and accurately position and produce pbuf internal memories behaviour by mistake The APP and particular location of work, that is, determine pbuf internal memories maloperation scene, systematicness can be improved while safeguards system reliability Energy.

In a possible design, the operating system includes (SuSE) Linux OS, and the (SuSE) Linux OS includes Mprotect memory management interfaces, the mprotect memory management interfaces are used to be called to set the APP to the first pbuf by APP The read-write operation attribute of internal memory；In APP by calling mprotect memory management interfaces to set readings of the APP to the first pbuf internal memories Write operation attribute is when current time is read-only, and (SuSE) Linux OS records APP and the read-write operation of the first pbuf internal memories is belonged to Property.So as to by calling the mprotect memory management interfaces of (SuSE) Linux OS can easily under a linux operating system The read-write operation attribute of first pbuf internal memories is configured.

In a possible design, the mprotect memory management interfaces be used to be called with based on one by APP or Read-write operation attributes of multiple page tables setting APP to the first pbuf internal memories.So that in the mprotect based on page table management Deposit read-write operation administrative mechanism independently can be controlled to the first pbuf internal memories.

In a possible design, the page table includes read-write operation attribute-bit, and operating system passes through in the page table It is read-only to the read-write operation attribute of the first pbuf internal memories that read-write operation attribute-bit, which records APP,.So that based on page table pipe The mprotect memory read-write operational administratives mechanism of reason independently can be controlled to the first pbuf internal memories.

The third aspect, the embodiment of the present invention provides a kind of application APP, and the application APP, which has, realizes above-mentioned the The function of the method for one side, the function can be realized by hardware, and corresponding software can also be performed by hardware and is realized. The hardware or software include one or more modules corresponding with above-mentioned functions.

Fourth aspect, the embodiment of the present invention provides a kind of computer-readable storage medium, above-mentioned for third party for saving as The computer software instructions used in application APP described in face, it includes the journey for the method for being used to perform above-mentioned first aspect Sequence.

5th aspect, the embodiment of the present invention provides a kind of operating system, and the operating system, which has, realizes above-mentioned second aspect Method function, the function can be realized by hardware, and corresponding software can also be performed by hardware and is realized.It is described hard Part or software include one or more modules corresponding with above-mentioned functions.

6th aspect, the embodiment of the present invention provides a kind of computer-readable storage medium, for saving as above-mentioned being used for the 5th side The computer software instructions used in operating system described in face, it includes the program for the method for being used to perform above-mentioned second aspect.

7th aspect, the embodiment of the present invention provides a kind of shared-memory management device, and the shared-memory management device includes Be stored with the memory of executable program code, and the processor coupled with the memory；There is operation on the processor Operation has message zero-copy system, the message zero-copy system bag in application APP and operating system, the operating system The shared first bag caching pbuf internal memories of at least two application APP s, the APPs are included, the APP is times in the APPs One APP, the APP are used for the method flow for performing first aspect present invention.

Eighth aspect, the embodiment of the present invention provides a kind of shared-memory management device, and the shared-memory management device includes Be stored with the memory of executable program code, and the processor coupled with the memory；There is operation on the processor Operation has message zero-copy system, the message zero-copy system bag in application APP and operating system, the operating system Include at least two application APP s, the APPs and share the first packet buffer pbuf internal memories, the APP is in the APPs Any APP, the operating system is used for the method flow for performing second aspect of the present invention.

9th aspect, the embodiment of the present invention provides a kind of shared-memory management equipment, and the shared-memory management equipment includes Processor and memory.The application APP of the third aspect and the operating system of the 5th aspect, the processing are run on the processor Device is configured as supporting application APP and operating system to perform the shared-memory management side that the embodiment of the present invention is provided Method.The memory is used to couple with processor, for storing the necessary program code of application APP and data, and is used for The necessary program code of storage program area and data.

Compared to prior art, in the scheme of the embodiment of the present invention in the case where multiple APP share same pbuf internal memories, By setting APP to be read-only to the read-write operation attribute of the first pbuf internal memories, so as to monitor APP to first in operating system When pbuf carries out write operation, operating system sends pbuf internal memory maloperation notification messages to APP, so that APP is based in the pbuf Deposit maloperation notification message and quickly and accurately position the APP and particular location for producing the pbuf internal memory maloperations, that is, determine Pbuf internal memories maloperation scene, safeguards system reliability.

Brief description of the drawings

In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing There is the accompanying drawing used required in technology description to be briefly described.

Fig. 1 is a kind of configuration diagram of network function virtualization system provided in an embodiment of the present invention；

Fig. 2 is the configuration diagram of another NFV systems provided in an embodiment of the present invention；

Fig. 3 is the configuration diagram of another NFV system provided in an embodiment of the present invention；

Fig. 4 is a kind of configuration diagram of message zero-copy system provided in an embodiment of the present invention；

Fig. 5 is a kind of schematic flow sheet of shared-memory management method provided in an embodiment of the present invention；

Fig. 6 is the first pbuf memory management schematic flow sheets between a kind of APP processes provided in an embodiment of the present invention；

Fig. 7 is a kind of message zero-copy flow provided in an embodiment of the present invention and the signal of the first pbuf memory managements flow Figure；

Fig. 8-a are the configuration diagrams of another message zero-copy system provided in an embodiment of the present invention；

Fig. 8-b are another message zero-copy flow provided in an embodiment of the present invention and the signal of pbuf memory managements flow Figure；

Fig. 9 is a kind of structural representation of application APP provided in an embodiment of the present invention；

Figure 10 is a kind of structural representation of operating system provided in an embodiment of the present invention；

Figure 11 is a kind of structural representation of shared-memory management device provided in an embodiment of the present invention；

Figure 12 is the structural representation of another shared-memory management device provided in an embodiment of the present invention；

Figure 13 is a kind of structural representation of shared-memory management equipment provided in an embodiment of the present invention.

Embodiment

In order that those skilled in the art more fully understand the present invention program, below in conjunction with the embodiment of the present invention Accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described.

The application scenarios that the embodiment of the present invention is based on are entered by technical scheme first below for a better understanding of the present invention Row explanation.

Referring first to Fig. 1, Fig. 1 is a kind of network function virtualization system NFV100 provided in an embodiment of the present invention framework Schematic diagram.As shown in figure 1, the NFV systems include hardware layer 110, management program (English：Hyperviser) layer 120, operation system System (English：OperSystem, OS) layer 130, application program (English：Application, APP) layer 140, wherein, operating system Layer 130 and each application program constitute virtual machine (English：Virtual Machine, VM).

Wherein, hardware layer 110 is the hardware platform that virtualized environment is run, and the hardware layer 110 can be by one or more Server group into；Management program 120 is directly installed on hardware computing resource, for managing and calling hardware resource, it is not necessary to Underlying operating system, namely it is appreciated that management program is a very thin operating system；Operating system layer 130 include one or Multiple operating systems, are the access operating system (English operated in management program：Guest OS), in the embodiment of the present invention In, the operating system can be (SuSE) Linux OS or other operating systems；Application layer 140 includes one or more answer With program module, it is to operate in the application accessed in operating system, is offering customers service.

Referring to Fig. 2, Fig. 2 is the configuration diagram of another NFV systems provided in an embodiment of the present invention.As shown in Fig. 2 should NFV systems include hardware layer 210, host (English：Host) layer 220, management program layer 230 accesses operating system (English： Guest) 240, APP layers 250.Wherein, the NFV system architectures shown in Fig. 2 and the network function virtualization system framework shown in Fig. 1 Difference is, in the network function virtualization system framework shown in Fig. 2, and by increasing host layers 220, the host layers can For installation foundation operating system, so that management program layer 230 is operated on underlying Operating System, a whole set of is constructed virtual Hardware platform, user installs new operating system and application software, host layers and guest layers on guest layers as needed On operating system can be with completely irrelevant.In embodiments of the present invention, the host layers of operating system can be Windos operations, Operating system on the guest layers can be (SuSE) Linux OS.

Referring to Fig. 3, Fig. 3 is the configuration diagram of another NFV system provided in an embodiment of the present invention.Shown in Fig. 3 In network function virtualization system, including network function virtualized infrastructure layer (English：NFV Infrastructure, NFVI) 310, virtual network function (English：Virtual Network Functions, VNFS) 320, operation/business support system System (English：Operation Support System/Business Support System, OSS/BSS) 330 and network Virtualization of function management and arranging system (English：NFV Management and Orchestration,NFV MANO)340.

Wherein, virtualized infrastructure layer 310 includes hardware resource layer, virtualization layer and virtual resource layer, VNFS320 Including multiple VNF and multiple NE Explorer (English：Element Manager, EM), different void can be provided the user Intend network function.Relative to the framework shown in Fig. 1 or Fig. 2, it is believed that the APP in Fig. 1 or Fig. 2 is deployed in the VNF in Fig. 3.

Wherein, NFV MANO include network function virtualization composer (English：NFV Orchestrator, NFVO), one Individual or multiple virtualization network function manager (English：VNF Manager, VNFM) and virtualized infrastructure manager (English Text：Virtualized Infrastructure Manager, VIM), Web Services directory (English：Network Service Catalogue, NS Catalogue), for being responsible for management and layout to whole NFVI resources, it is responsible for business network and NFVI The mapping and association of resource, are responsible for implementation of OSS service resources flows etc., MANO is by parsing the virtual network work(that product is provided Can descriptor (English：VNF Descriptor, VNFD) file realizes the establishment of virtual machine, the installation of network element, affine and anti-parent The operations such as the configuration of sum, the scalable appearance of system.

Referring to Fig. 4, Fig. 4 is a kind of configuration diagram of message zero-copy system 400 provided in an embodiment of the present invention, is such as schemed Shown in 4, the message zero-copy system 400 includes N number of APP, namely including APPs, respectively APP1, APP2, APPn, First network interface card and the first network interface card queue, the second network interface card and the second network interface card queue, and a packet buffer pbuf internal memory.Wherein, Different APP belongs to different business process, and the pbuf internal memories are shared by N number of APP.

In embodiments of the present invention, N number of APP can be to run on same VM APP in above-mentioned Fig. 1 and Fig. 2, also may be used Think the VNF in Fig. 3, or the different application run in physical computer.

In embodiments of the present invention, the message zero-copy system can run in the operating system on physical computer, It can run in the operating system in NFV systems.

What deserves to be explained is, the APP can be an independent functional module, the function for performing feature, for example, can Think the APP for the performing transport layer process function or APP for performing application layer process function, or can also be one The functional module that independent function is write.

Alternatively, in one embodiment of the invention, the plurality of APP can be different process；Can also be same Different threads in individual process.No matter the plurality of APP is in different processes, it is in not collinear in same process Journey, as long as the plurality of APP shares same pbuf internal memories, and carries out message forwarding between the plurality of APP using existing in the pbuf Scene be application scenarios that the embodiment of the present invention is provided.

In embodiments of the present invention, above-mentioned first network interface card and the second network interface card can be Microsoft Loopback Adapter (English：virtual Network card, VNIC), or physical network card.

In embodiments of the present invention, when carrying out the forwarding of zero-copy message using the message zero-copy system, APP1 first Direct memory storage (English of the process toward filling pbuf internal memories in the first network interface card queue：Direct Memory Access, DMA) Behind address, the first network interface card is toward message is filled in the pbuf memory address, then APP1 packet receiving messages from the first network interface card queue To pbuf internal memories, and the message is sequentially passed through into multiple APP processes handled, after last APPn is disposed, the second network interface card Again toward the pbuf memory address is filled in the second network interface card queue, APPn fills the message to the pbuf internal memories of the second network interface card queue Address, then the second network interface card reclaim the message from the second network interface card queue and send.After the second network interface card sends message, APPn withdraws the pbuf internal memories to memory pool.As can be seen that the message zero-copy system is by pbuf memory sharing internal memories, realize Message forwarding between different APP, so as to realize message zero-copy.

But meanwhile, based on the message zero-copy system, because different APP shares same pbuf internal memories, so destruction Isolation between different APP processes, when occurring pbuf memory problems when striding course is operated, it is impossible to find and occur in pbuf Send one's regards to where topic APP and particular location (for example, when pbuf internal memories go wrong, it is impossible to it is determined that be in n APP it is specific where Maloperation problem occurs for individual APP), namely can not orientation problem primary scene, cause the business recovery cycle long.

In order to solve the above problems, the embodiment of the present invention proposes a kind of shared-memory management method, by setting pbuf The read-write operation attribute of internal memory, when maloperation occurs for pbuf internal memories, to get the internal memory for producing the pbuf internal memory maloperations Maloperation scene, strengthens message zero-copy system reliability.

The embodiment of the present invention is described in detail below.Referring to Fig. 5, Fig. 5 is provided in an embodiment of the present invention a kind of common The schematic flow sheet of EMS memory management process is enjoyed, the shared-memory management method is applied to the zero-copy system shown in Fig. 4, the message Zero-copy system operation is in operating system, as shown in figure 5, this method may comprise steps of：

Any application APP in S501, APPs sets the APP to belong to the read-write operation of the first pbuf internal memories Property at current time to be read-only, the operating system records read-write operation attributes of the APP to the first pbuf internal memories.

In embodiments of the present invention, at least two application program is the APPs shown in Fig. 4, at least two application program In any application APP can for APP1 to any application program between APPn, for example, APP1, APP2, APPn.

Wherein, the read-write operation attribute of the first pbuf internal memories refers to be managed the pbuf of shared drive the first, prevents Only multiple application programs carry out write operation to the first pbuf simultaneously and produce memory problem, as the APP set by each APP to the The read-write operation that one pbuf should be performed at certain moment, so that when APP performs read-write operation attribute to the first pbuf internal memories During operation in addition, operating system is then considered internal memory maloperation.

In embodiments of the present invention, the read-write operation attribute includes read-only, read-write and only write.When setting APP is to first When the read-write operation attribute of pbuf internal memories is read-only, and now APP performs write operation to the first pbuf internal memories, now operation system System thinks APP to the first pbuf internal memory maloperations；When setting APP to saving as read-write in the first pbuf, namely now APP to the One pbuf internal memories perform read operation or write operation, and operating system will not be monitored as internal memory maloperation；When setting APP is to first Saved as in pbuf when only writing, so that now when APP performs read operation to the first pbuf internal memories, operating system thinks APP to One pbuf internal memory maloperations.What deserves to be explained is, also can be further in APP to first during for read-write operation attribute for read-write In the case that pbuf internal memories perform read operation, it is determined that read-write operation attribute now is read-only, in APP to counter foil in the first pbuf In the case that row write is operated, it is determined that read-write operation attribute now is only to write.

Alternatively, in one embodiment of the invention, when APP receives message by the first pbuf internal memories and is located During reason, it is read-write that now the APP, which is set to the read-write operation attributes of the first pbuf internal memories, and other moment APP to this first The read-write operation attribute of pbuf internal memories is read-only, for example, when APP does not receive message, or when APP finishes Message processing And send to other APP, APP is read-only to the read-write operation attribute of the first pbuf internal memories.

Specifically, in one embodiment of the invention, when APP receives default message by the first pbuf internal memories and locates During reason, now the APP by the read-write operation attribute to the first pbuf internal memories be set to read-write, and other APP to this first The read-write operation attribute of pbuf internal memories is read-only, after the default message is sent to other APP by APP, will now withdraw this and writes Operational attribute, namely now APP read-write operation attribute of the default message will be reseted be set to it is read-only.When each APP is set pair After the read-write operation attribute of first pbuf internal memories, the operating system that the follow-up zero-copy system is run, which is monitored, has APP pairs During the maloperation of the first pbuf internal memories, pbuf internal memories misuse notification message will be sent, to point out the pbuf internal memories of user the first to occur Problem, so that the management to the first pbuf internal memories can be realized by this way.

For example, in the example of the present invention, referring specifically to Fig. 6, Fig. 6 is one kind provided in an embodiment of the present invention First pbuf memory management schematic flow sheets between APP processes.The feelings forwarded in message between APP1, APP2 and APP3 Under condition, when APP1 receives message at the moment 1 by the first pbuf internal memories and is handled, now APP1 is to the first pbuf internal memories Read-write operation attribute be set to read-write；After APP1 sends message to APP2 processes, APP2 receives the report at the moment 2 Text, now APP2 will be set to read-write to the read-write operation attribute of the first pbuf internal memories, and now APP1 will withdraw the first pbuf The write operation of internal memory, namely now APP1 will be read-only to the read-write operation attribute of the first pbuf internal memories.At the two moment, APP3 is not handled message, thus APP3 the read-write operation attribute of the first pbuf internal memories is set to always it is read-only.

Further, referring to Fig. 7, Fig. 7 is a kind of message zero-copy flow provided in an embodiment of the present invention and first Pbuf memory management schematic flow sheets.When carrying out the forwarding of zero-copy message using the message zero-copy system, APP1 first, Read-write operation attribute to the first pbuf internal memories is disposed as read-only by APP2 and APP3.APP1 is filled toward the first network interface card queue The message will be write the first network interface card queue by DMA address, namely the first pbuf memory address, the first network interface card after message packet receiving In, then APP1 uses the first pbuf internal memories packet receiving message from the first network interface card queue, and now APP1 will be to the first pbuf The read-write operation attribute of internal memory is set to read-write；After APP1 is finished to the Message processing in the first pbuf internal memories, by the message Send to APP2, now APP1 withdraws the write operation attribute of the first pbuf internal memories, namely now APP1 processes to this first The read-write operation attribute of pbuf internal memories is read-only, and now setting is by APP2 to the read-write operation attribute of the first pbuf internal memories Read-write；After APP2 is finished to the Message processing, then next APP, the read-write behaviour of each APP the first pbuf internal memories are sent to successively Make attribute set-up mode ibid step.After message is forwarded to APPn, APPn is to the read-write operation attribute of the first pbuf internal memories Read-write, after APPn is handled message, the message is sent to APP1 by APPn, and withdraws write operation attribute.Now APP1 Receive after the message, it is read-write that APP1, which is reset to the read-write operation attributes of the first pbuf internal memories, last APP1 is by message Send to the second network interface card queue, then the second network interface card will taking-up in message the second network interface card queue, the whole network interface card transmitting-receiving packet stream of completion Journey.It should be noted that in the flow shown in Fig. 7, the send and receive packets of network interface card are managed collectively by APP1.

Specifically, in one embodiment of the invention, when each APP operates in (SuSE) Linux OS, Linux operations System includes mprotect memory management interfaces, and the mprotect memory management interfaces are used to set the first pbuf internal memories Read-write operation attribute, namely when APP is handling stored messages in the first pbuf, APP is by calling mprotect memory managements to connect Mouth sets APP to be read-write to the read-write operation attribute of the first pbuf internal memories, and other APP are then by calling mprotect internal memory pipes It is read-only to manage interface and set to the first pbuf read-write operation attribute, and now (SuSE) Linux OS will record APP to first The read-write operation attribute of pbuf internal memories.When read-write operation attribute is that read-only APP carries out write operation to the first pbuf internal memories, Linux system can listen to the internal memory maloperation immediately, and then (SuSE) Linux OS produces core dumped signal segment Fault, for pointing out APP to carry out maloperation to the first pbuf internal memory, then the APP can be carried by the error message Show and print call stack information to determine the context primary scene of exception.It is appreciated that by calling (SuSE) Linux OS Mprotect memory management interfaces easily can be entered to the read-write operation attribute of the first pbuf internal memories under a linux operating system Row is set.

Alternatively, in one embodiment of the invention, because mprotect memory management interfaces are carried out based on page table Management, namely the read-write operation attribute of corresponding internal memory will be configured simultaneously in same page table, specifically, each page table includes Read-write operation attribute-bit, thus operating system record APP to the read-write operation attributes of the first pbuf internal memories for it is read-only when, can be with It is read-only to the read-write operation attribute of the first pbuf internal memories by recording APP on the read-write operation attribute-bit in the page table. Because each page table one read-write operation attribute-bit of correspondence, namely each page table are uniformly set by the read-write operation attribute-bit The read-write operation attribute of pbuf internal memories corresponding to the page table.Therefore it is interior when being carried out based on mprotect memory read-write operational administratives mechanism When depositing read-write operation management, in order that being grasped by mprotect memory read-write operational administrative mechanism to the read-write of the first pbuf internal memories When being configured as attribute, the read-write operation attribute for the other internal memories for being unlikely to interfere with the page table, so needing to make the page The corresponding relation of table and the first pbuf internal memory existence anduniquess, namely the corresponding page table of the first pbuf internal memories or multiple page tables Independently it is controlled by mprotect memory read-write operational administrative mechanism, now one or many corresponding with the first pbuf internal memories The read-write operation attribute-bit of individual page table will be configured as read-only or read-write simultaneously.

Alternatively, in one embodiment of the invention, the page table can be the page table of common 4K bits, such as first When pbuf memory sizes are size within 4K bits, the first pbuf internal memories one page table of correspondence, now mprotect internal memories reading Write operation administrative mechanism controls the page table with to the independent control of the read-write operation attribute of the first pbuf internal memories progress by being independently arranged System, specifically, can set read-writes of each APP to the first pbuf to grasp by setting the read-write operation attribute-bit of the page table Make attribute, or if when the first pbuf memory sizes are 8K bits, the first pbuf internal memories correspond to continuous two page tables, now Mprotect memory read-write operational administratives mechanism can be grasped by controlling two page tables simultaneously to the read-write of the first pbuf internal memories It is independently controlled as attribute, specifically, can be by setting the read-write operation attribute-bit of 2 page tables to be identical simultaneously (while be read-only or while be read-write) carrys out the read-write operation attribute of the pbuf internal memories of independent control the first.It is appreciated that passing through So that the first pbuf internal memories are uniquely corresponding with page table independently of other pbuf internal memories so that in the mprotect based on page table management Deposit read-write operation administrative mechanism independently can be controlled to the first pbuf internal memories.

Alternatively, in one embodiment of the invention, the message zero-copy system also includes the 2nd pbuf internal memories, institute Stating APPs shared drive also includes the 2nd pbuf internal memories, message zero-copy system when carrying out zero-copy using shared drive, APP can determine to save as the shared drive of the APPs in the first pbuf in predetermined time, and the APP is in predetermined time Outside determine to save as in the 2nd pbuf in the shared drive of the APPs, the first pbuf and save as small page internal memory, it is described Big page internal memory, the first pbuf internal memories and one or more of page table existence anduniquess corresponding relations are saved as in 2nd pbuf.

Wherein, big page internal memory refers to reduce page table quantity by increasing page table size, systematic function is improved, in big page Leave, each page table size can be with 1G bits or bigger, and if APPs shared drive size only 4K bits when, now each Page table will the multiple pbuf internal memories of correspondence.Small page internal memory is left in the concept relative with big page internal memory, small page, each page table it is big I is the page table of common 4K bits size or the page table of 8K bit sizes.

Alternatively, in one embodiment of the invention, predetermined time refers to what APP was determined based on preset rules, for example, Can be determined based on the default sampling period, so that APP uses the first pbuf in corresponding predetermined time of default sampling period Internal memory, the 2nd pbuf internal memories are used at the time of outside the default sampling period；In another example or in message zero-copy system When the type for the message that will be handled is preset kind, preserves the message using the first pbuf internal memories and (determine that current time is Predetermined time), when the type for the message that will be handled in message zero-copy system is other types, protected using the 2nd pbuf internal memories The message (determining that current time is outside predetermined time) is deposited, so as to based on the current report by processing of message zero-copy system Literary type determines predetermined time.

Specifically, in one embodiment of the invention, due under 4K bit page tables, systematic function is than relatively low, and such as The pbuf memory settings are big page internal memory by fruit, the situation that now there will be the same page table of multiple pbuf internal memories correspondences, if When being managed to the read-write operations of current pbuf internal memories, mistakenly management to other pbuf internal memories of same page table, it will make It is abnormal into systemic-function.So, in the case where ensureing systematic function and preventing systemic-function exception, increase by second can be passed through Pbuf internal memories, set in the 2nd pbuf and save as big page internal memory, so that APP can pass through the first pbuf internal memories or the 2nd pbuf internal memories Receive and handle default message, lifting system performance.Fig. 8-a are see, for example, Fig. 8-a are provided in an embodiment of the present invention another Plant the configuration diagram of message zero-copy system.But due to saving as big page internal memory in the 2nd pbuf, (SuSE) Linux OS is not right 2nd pbuf internal memories are written and read operational administrative, are only written and read operational administrative to the first pbuf internal memories.Settable first In the default sampling period of pbuf internal memories, in the default sampling period, the first pbuf internal memories are filled in corresponding sampling instant to APP processes, so Afterwards by being written and read operational administrative to the first pbuf internal memories, it is ensured that system reliability.For example, setting the first pbuf internal memories The default sampling period be 1000 in the case of, often fill 999 the 2nd pbuf internal memories, fill a first pbuf internal memory, Linux is written and read operational administrative to ensure system reliability to the first pbuf internal memories, simultaneously because in 1000 pbuf There are 999 to use the 2nd pbuf internal memories of big page internal memory in depositing, so the high-performance of system can be ensured.

What deserves to be explained is, by setting APP to use the predetermined time of the first pbuf internal memories, when the frequency of the predetermined time More hour, then the frequency of explanation progress pbuf internal memory monitorings is smaller, so now system reliability is relatively small, and now uses The relative degree that the pbuf internal memories of big page internal memory the 2nd are filled is more, and the performance boost of system is larger；And the frequency of predetermined time Bigger, then the frequency of explanation progress pbuf internal memory monitorings is bigger, so now the reliability of system is relatively large, and now uses The relative degree that the pbuf internal memories of big page internal memory the 2nd are filled is small, and the performance boost of system is smaller.Therefore can weigh according to demand The performance boost of balance system sets the predetermined time with reliability.

Further, referring to Fig. 8-b, Fig. 8-b be another message zero-copy flow provided in an embodiment of the present invention with And pbuf memory management schematic flow sheets.Flow shown in Fig. 8-b is to be performed based on zero message copy system shown in Fig. 8-a , namely including two pbuf internal memories：First pbuf internal memories and the 2nd pbuf internal memories.Due to the first pbuf internal memories and page table one by one Big page internal memory is saved as in correspondence, the 2nd pbuf, so set the default sampling period to carry out sampling filling to the first pbuf internal memories, and Operational administrative is written and read to the first pbuf internal memories using only mprotect memory read-write operational administrative mechanism, for the 2nd pbuf Internal memory is then without memory read-write operational administrative.And the flow shown in Fig. 8-b is managed collectively the send and receive packets of network interface card by APP1.Such as Shown in Fig. 8-b, the read-write operation attribute of the first pbuf internal memories is disposed as read-only by APP1, APP2 and APP3 first.APP1 is past When filling DAM addresses in the first network interface card queue, when not for the default sampling period, filled out using the DMA address of the 2nd pbuf internal memories The first network interface card queue is filled, so that follow-up APP1 reads message from the first network interface card queue by the 2nd pbuf internal memories and handled；When for During the default sampling period, the first network interface card queue is filled using the DMA address of the first pbuf internal memories, so that when APP1 passes through first Pbuf internal memories are read after message from the first network interface card queue, need to be written and read operational attribute setting due to saving as in the first pbuf Sample pbuf internal memories, so APP1 sets the read-write operation attribute of the first pbuf internal memories to be read-write, and handle message, when having handled After message, APP1 withdraws the write operation attribute of the first pbuf internal memories, and the message is sent into APP2.

When APP2 receives the message of APP1 transmissions, whether be sampling first pbuf internal memory, judging if judging the pbuf internal memories In the case of no, set and directly handle without read-write operation attribute, in the case where being judged as YES, the first pbuf is set The read-write operation attribute of internal memory is read-write, and handles message, after message handle, and APP2 the first pbuf internal memories of withdrawal write behaviour Make attribute, and the message is sent to APP3.

When APP3 receives the message of APP2 transmissions, whether be sampling first pbuf internal memory, judging if judging the pbuf internal memories In the case of no, set and directly handle without read-write operation attribute, in the case where being judged as YES, the first pbuf is set The read-write operation attribute of internal memory is read-write, and handles message, after message has been handled, and APP3 withdraws the reading behaviour of the first pbuf internal memories Make attribute, and the message is sent to APP1.

APP1 is received after the message, in the case where being judged as the 2nd pbuf internal memories, now without read-write operation attribute Set, handled and message is sent to the second network interface card queue after message and the 2nd pbuf internal memories are reclaimed to pbuf memory pools. In the case of being judged as the first pbuf internal memories, it is read-write now to set the read-write operation attribute of the first pbuf internal memories, and is handled Message, is then sent to the second network interface card queue by message again, and reclaim from the second network interface card queue the first pbuf internal memories to Pbuf memory pools, while withdrawing the write operation attribute of the first pbuf internal memories.So as to complete whole message zero-copy process.

S502, operating system monitor APP to the first pbuf internal memories carry out write operation in the case of, operating system to APP send pbuf internal memory maloperation notification messages, now APP receive operating system transmission the maloperation of pbuf internal memories notify disappear Breath, and the pbuf internal memory maloperation field datas in the pbuf internal memories maloperation notification message are obtained, the pbuf internal memories are missed Operational notification message is used for the specific position for indicating that the APP carries out write operation and the write operation to the first pbuf internal memories Put.

Alternatively, in one embodiment of the invention, the pbuf internal memory maloperations field data includes call stack information.

Specifically, in one embodiment of the invention, when operating system monitors to save as read-only in the first pbuf, APP has carried out write operation to the first pbuf internal memories, and then operating system will send the maloperation of pbuf internal memories to the APP and notify to disappear Breath, for example, the pbuf internal memory maloperations notification message can be core dumped signal segment fault, then APP passes through this section of mistake Error signal segmentfault prints call stack information, so that further, the APP can be true according to the call stack information Surely the problem of causing the particular location of EMS memory error, and allow APP to cause EMS memory error according to the particular location reparation.

Alternatively, in one embodiment of the invention, the first pbuf internal memories are additionally operable to cache track identification, described APP is received after pbuf internal memory maloperation notification messages, and APP can also pass through the track identifications of the first pbuf internal memories record the The tracing point information of one pbuf internal memories, the tracing point information include application APP mark, application program current process identification with And current time, wherein, APP is identified for indicating sometime using the APP of the first pbuf internal memories, and application program is current Process identification (PID) indicates sometime using the APP processes of the first pbuf internal memories.

In embodiments of the present invention, tracing point information refers to the relevant information of the APP using the first pbuf internal memories, so that Relevant information using the first pbuf APP is connected to the tracing point information for then constituting the first pbuf internal memories in chronological order.

Specifically, in embodiments of the present invention, can by track identification come recording track point information, wherein, the track Mark can refer to be used to preserve the memory headroom of message in the first pbuf internal memories, so as to using the free memory space come Record the first pbuf tracing point information.

For example, in the example of the present invention, referring to Fig. 4.When APP1 processes are in the first reception and handle In first pbuf during stored messages, APP1 processes record tracing point letter now in the free memory space of the first pbuf internal memories Breath, namely APP1 marks, APP2 process identification (PID)s and the first moment；Then when in APP1 is by first stored messages send to APP2 Afterwards, APP2 processes the second reception and handle the first pbuf in stored messages when, APP2 processes again a more step this first Tracing point information now is recorded in the remaining space of pbuf internal memories, namely when APP2 marks, APP2 process identification (PID)s and second Carve；All tracing point information that the first pbuf internal memories of record are passed through successively, so that the first pbuf will be included among APPn All tracing point information according to time order and function are available first by all tracing point information of the internal memory from APP1 to APPn The track that pbuf internal memories are passed through.Further, the first pbuf internal memories that can also be recorded in zero-copy system pass through first Network interface card queue and the track of the second network interface card queue.

It is appreciated that after recorded the first pbuf tracing point information, then can work as and write behaviour in some APP generation misuses When making, can by the first pbuf tracing point information determine the first pbuf track, come further determine that this first The upstream APP that pbuf is passed through, further to improve system reliability.

S503, APP determine that the APP enters row write to the pbuf internal memories based on the pbuf internal memories maloperation field data The internal memory maloperation scene of operation.

Further, in one embodiment of the invention, can be extensive based on the live carry out business of the internal memory maloperation It is multiple.

For example, in the example of the present invention, when message zero-copy system works, and the first pbuf internal memories Send to APP3, now APP2 is by calling mprotect memory management interfaces to set read-writes of the APP2 to the first pbuf internal memories to grasp Make attribute to be read-only, APP2 will receive the pbuf internal memory maloperation notification messages of system transmission, such as core dumped signal, then APP2 is based on the core dumped signal and prints call stack information, and determines to produce to the first pbuf based on the call stack information by APP2 The position of write operation is carried out, namely produces the code position of write operation by mistake, so as to may be such that APP repairs this based on the position and asked Topic, prevents follow-up APP2 from carrying out write operation by mistake to the first pbuf internal memories.By setting read-writes of each APP to the first pbuf internal memories to grasp Make attribute, so that when the first pbuf internal memories are run to APP3, the feelings of write operation have been carried out to the first pbuf internal memories in APP2 Under condition, system can remind APP2 by pbuf internal memory maloperations notification message, to allow APP2 to position in time and change this First pbuf memory problems, it is ensured that system reliability, and also cause the APP3 of the first pbuf internal memories of positive operation to be this moment unlikely to Waste of resource goes to carry out pbuf memory problem investigations, also further increases systematic function.

Further, APP can record APP in the first pbuf when carrying out write operation by mistake to the first pbuf internal memories The numerical value that the first pbuf internal memories relevant position is stored before depositing into row write operation, so that when APP in the first pbuf to depositing into Row is write by mistake when, the numerical value of the first pbuf internal memories relevant position can be recovered to the numerical value to before write by mistake, be with further improve Reliability of uniting and systematic function, also improve system robustness.

Further, by recording the tracing point information of the first pbuf internal memories, it is determined that APP in the first pbuf to depositing into Behind the internal memory maloperation scene of row write operation, pbuf memory problem investigations can be further carried out by the upstream APP of the APP, Further improve system reliability.

What deserves to be explained is, above method embodiment cannot be only used for when APP belongs to the read-write operation of the first pbuf internal memories Property when being set to read-only, the monitoring of internal memory maloperation is carried out to the first pbuf internal memories.When APP is grasped to the read-write of the first pbuf internal memories Be set to only write as attribute, and when APP to the read-write operation attribute of the first pbuf internal memories be set to read-write when read states with And during write state, same usable aforesaid way carries out the monitoring of internal memory maloperation to the first pbuf internal memories, specifically, can basis Use of the opereating specification to the above method corresponding to read-write operation attribute carries out adaptation adjustment.

What deserves to be explained is, above-mentioned steps S501, step S502 and step S503 can also synchronously simultaneously when actually performing Row is performed, and the embodiment of the present invention is not limited strictly its execution sequence.

As can be seen that in the scheme of the embodiment of the present invention, being total to when in the zero-copy system for running on operating system by APPs When enjoying the first pbuf internal memories, any APP in the APPs sets the APP to the read-write operation attribute of the first pbuf internal memories current When moment is read-only, operating system records read-write operation attributes of the APP to the first pbuf internal memories；Then monitored in operating system When carrying out write operation to the first pbuf internal memories to the APP, operating system sends pbuf internal memory maloperation notification messages to the APP, The APP is received after the pbuf internal memory maloperation notification messages of operating system transmission, is obtained the pbuf internal memories maloperation and is notified to disappear Pbuf internal memory maloperation field datas in breath, the pbuf internal memory maloperations notification message is used to remind the APP to described the One pbuf internal memories carry out the particular location of write operation and the write operation and the write operation；Finally the APP is based on being somebody's turn to do again Pbuf internal memory maloperation field datas determine that APP carries out the internal memory maloperation scene of write operation to the first pbuf internal memories.So as to real , can quickly simultaneously when internal memory maloperation occur in the shared pbuf internal memories in the case that present multiple APP share same pbuf internal memories Positioning produces the APP and particular location of the pbuf internal memory maloperations exactly, that is, determines pbuf internal memories maloperation scene, ensure Systematic function can be improved while system reliability.

Further, business recovery can be carried out based on pbuf internal memories maloperation scene, more a step safeguards system can By property.

Referring to Fig. 9, Fig. 9 is a kind of structural representation of application APP provided in an embodiment of the present invention, for realizing this Shared-memory management method disclosed in inventive embodiments.Wherein, the APP is applied in message zero-copy system, the message zero Copy system is run in operating system, and the message zero-copy system includes at least two application APP s, the APPs Shared first bag caching pbuf internal memories, the APP is any APP in the APPs.As shown in figure 9, the embodiment of the present invention is carried A kind of APP supplied includes：

Setup module 910, acquisition module 920 and determining module 930.

Setup module 910, for setting the APP to the read-write operation attribute of the first pbuf internal memories at current time To be read-only.

Acquisition module 920, for receiving the feelings for the pbuf internal memory maloperation notification messages that the operating system is sent Under condition, the pbuf internal memory maloperation field datas in the pbuf internal memories maloperation notification message are obtained, the pbuf internal memories are missed Operational notification message is used for the specific position for reminding the APP to carry out write operation and the write operation to the first pbuf internal memories Put.

Determining module 930, for determining the APP to described first based on the pbuf internal memories maloperation field data Pbuf internal memories carry out the internal memory maloperation scene of write operation.

Alternatively, in one embodiment of the invention, the operating system includes (SuSE) Linux OS, the Linux Operating system includes mprotect memory management interfaces, and the mprotect memory management interfaces are used for called described to set The read-write operation attribute of first pbuf internal memories；The setup module 910 specifically for：

By calling the mprotect memory management interfaces to set read-writes of the APP to the first pbuf internal memories to grasp It is read-only to make attribute at current time.

Alternatively, in one embodiment of the invention, the mprotect memory management interfaces are used to be adjusted by the APP To based on read-write operation attributes of one or more page tables setting APP to the first pbuf internal memories.

Alternatively, in one embodiment of the invention, the message zero-copy system also includes the 2nd pbuf internal memories, institute State APPs and share the 2nd pbuf internal memories, the APP is configured to described in the corresponding sampling instant sampling of default sampling period First pbuf internal memories, the APP is configured to the 2nd pbuf internal memories of being sampled outside the sampling instant.

Alternatively, in one embodiment of the invention, the message zero-copy system also includes the 2nd pbuf internal memories, institute State APPs shared drive include the 2nd pbuf internal memories, the determining module 930 be additionally operable in the APPs it is any should Set the APP to the read-write operation attribute of the first pbuf internal memories before current time is read-only breath with program APP, Predetermined time determines to save as the shared drive of the APPs in the first pbuf, the APP outside predetermined time really Saved as in fixed 2nd pbuf in the shared drive of the APPs, the first pbuf and save as small page internal memory, described second Big page internal memory, the first pbuf internal memories and one or more of page table existence anduniquess corresponding relations are saved as in pbuf.

Alternatively, in one embodiment of the invention, APP900 also includes logging modle 940, for passing through the rail The tracing point information of first pbuf internal memories described in mark identification record, the tracing point information includes the application program identification, institute State application program current process identification and the current time.

In the present embodiment, application APP 900 is to present in modular form.Here " module " can refer to spy Determine application integrated circuit (application-specific integrated circuit, ASIC), perform one or more soft The processor and memory of part or firmware program, integrated logic circuit, and/or other can provide the device of above-mentioned functions.

It is understood that the function of each functional unit of the application APP 900 of the present embodiment can be according to the above method Method in embodiment is implemented, and it implements the associated description that process is referred to above method embodiment, herein not Repeat again.

Referring to Figure 10, Figure 10 is a kind of structural representation of operating system provided in an embodiment of the present invention, for realizing this Shared-memory management method disclosed in inventive embodiments.Wherein, operation has message zero-copy system in the operating system, described Message zero-copy system shares the first packet buffer pbuf internal memories including at least two application APP s, the APPs, such as schemes Shown in 10, a kind of operating system 1000 provided in an embodiment of the present invention includes：

Logging modle 1010 and sending module 1020.

Wherein, logging modle 1010, for any application APP in the APPs in the first pbuf The read-write operation attribute deposited, which is set, indicates that it is read-only to the read-write operation attribute of the first pbuf internal memories to record the APP；

Sending module 1020, row write is entered for monitoring the APP in the operating system to the first pbuf internal memories In the case of operation, the operating system sends pbuf internal memory maloperation notification messages to the APP, and the pbuf internal memories are grasped by mistake Make the specific position that notification message is used to remind the APP to carry out the first pbuf internal memories write operation and the write operation Put.

Alternatively, in one embodiment of the invention, the operating system includes (SuSE) Linux OS, the Linux Operating system include mprotect memory management interfaces, the mprotect memory management interfaces be used for by the APP call with The APP is set to the read-write operation attributes of the first pbuf internal memories.

Alternatively, in one embodiment of the invention, the page table includes read-write operation attribute-bit, the record mould Block 1010 record the APP to the read-write operation attributes of the first pbuf internal memories for it is read-only including：

Read-write operation attributes of the APP to the first pbuf internal memories is recorded by the read-write operation attribute-bit Read.

In the present embodiment, operating system 1000 is to present in modular form.Here " module " can refer to specific Application integrated circuit (application-specific integrated circuit, ASIC), performs one or more softwares Or the processor and memory of firmware program, integrated logic circuit, and/or other can provide the device of above-mentioned functions.

It is understood that the function of each functional unit of the operating system 1000 of the present embodiment can be real according to the above method The method applied in example is implemented, and it implements the associated description that process is referred to above method embodiment, herein no longer Repeat.

Referring to Figure 11, Figure 11 is a kind of structural representation of shared-memory management device provided in an embodiment of the present invention, such as Shown in Figure 11, the shared-memory management equipment 1100 includes：

Be stored with the processor 1102 that the memory 1101 of executable program code and the memory couple.The memory 1101 can be individually present, and be connected by bus with processor 1102.Memory 1101 can also be integrated with processor 1102 Together.Wherein, the processor 1102 can be central processing unit (Central Processing Unit, CPU), described Operation has operation in application APP and operating system, the operating system to have message zero-copy system, institute on processor 1102 Stating message zero-copy system includes the shared first bag caching pbuf internal memories of at least two application APP s, the APPs, described APP is any APP in the APPs.

The APP is used for：It is only at current time to the read-write operation attribute of the first pbuf internal memories to set the APP Read；In the case where the APP receives the pbuf internal memory maloperation notification messages that the operating system is sent, obtain described Pbuf internal memory maloperation field datas in pbuf internal memory maloperation notification messages, the pbuf internal memories maloperation notification message is used In the particular location for reminding the APP to carry out write operation and the write operation to the first pbuf internal memories；Based on described Pbuf internal memory maloperation field datas determine that internal memory maloperations of the APP to the first pbuf internal memories progress write operation shows .

Alternatively, in one embodiment of the invention, the operating system includes (SuSE) Linux OS, the Linux Operating system includes mprotect memory management interfaces, and the mprotect memory management interfaces are used for called described to set The read-write operation attribute of first pbuf internal memories；

It is described to set the APP to be read-only at current time to the read-write operation attribute of the first pbuf internal memories, including：

Alternatively, in one embodiment of the invention, in the reading for setting the APP to the first pbuf internal memories Write operation attribute is before current time is read-only, and the APP is additionally operable to：

Determine to save as the shared drive of the APPs in the first pbuf in predetermined time, outside predetermined time Determine to save as in the 2nd pbuf in the shared drive of the APPs, the first pbuf and save as small page internal memory, described second Big page internal memory, the first pbuf internal memories and one or more of page table existence anduniquess corresponding relations are saved as in pbuf.

Alternatively, in one embodiment of the invention, the first pbuf internal memories are additionally operable to cache track identification, described APP is additionally operable to：The tracing point information of the first pbuf internal memories is recorded by the track identification, the tracing point information includes The application program identification, the application program current process identification and the current time.

What deserves to be explained is, Figure 11 illustrate only simplifying for shared-memory management equipment 1100 and design.In practical application In, shared-memory management equipment 1100 can be comprising any amount processor and memory etc., and all can realize the present invention The shared-memory management equipment of embodiment is all within the protection domain of the embodiment of the present invention.

Referring to Figure 12, Figure 12 is the structural representation of another shared-memory management device provided in an embodiment of the present invention, As shown in figure 12, the shared-memory management equipment 1200 includes：

Be stored with the processor 1202 that the memory 1201 of executable program code and the memory couple.The memory 1201 can be individually present, and be connected by bus with processor 1202.Memory 1201 can also be integrated with processor 1202 Together.Wherein, the processor 1202 can be that operation has application program on central processor CPU, the processor 1202 Operation has message zero-copy system in APP and operating system, the operating system, and the message zero-copy system includes at least two The shared first bag caching pbuf internal memories of individual application APP s, the APPs, the APP is any APP in the APPs.

The operating system is used for：Any application APP in the APPs is to the first pbuf internal memories Read-write operation attribute, which is set, indicates that it is read-only to the read-write operation attribute of the first pbuf internal memories to record the APP；Described In the case that operating system monitors the APP to the first pbuf internal memories progress write operation, sent to the APP in pbuf Maloperation notification message is deposited, the pbuf internal memories maloperation notification message is used to remind the APP to the first pbuf internal memories Carry out the particular location of write operation and the write operation.

Alternatively, in one embodiment of the invention, the mprotect memory management interfaces are used to be adjusted by the APP Read-write operation attribute to set the first pbuf internal memories based on one or more page tables.

Alternatively, in one embodiment of the invention, the page table includes read-write operation attribute-bit, the record mould Block record the APP to the read-write operation attributes of the first pbuf internal memories for it is read-only including：

Record the APP by the read-write operation attribute-bit is to the read-write operation attribute of the first pbuf internal memories It is read-only.

What deserves to be explained is, Figure 12 illustrate only simplifying for shared-memory management equipment 1200 and design.In practical application In, shared-memory management equipment 1200 can be comprising any amount processor and memory etc., and all can realize the present invention The shared-memory management equipment of embodiment is all within the protection domain of the embodiment of the present invention.

Referring to Figure 13, Figure 13 is a kind of structural representation of shared-memory management equipment provided in an embodiment of the present invention, such as Shown in Figure 13, the shared-memory management equipment 1300 includes：

Be stored with the processor 1302 that the memory 1301 of executable program code and the memory couple.Wherein, locate It can be that can be run such as Fig. 5 on central processing unit (Central Processing Unit, CPU), the processor to manage device 1101 The APP and dress as shown in Figure 10 in APP and operating system or device embodiment as shown in Figure 9 in shown embodiment of the method Put in APP the and Figure 12 shown device embodiments in the operating system or device embodiment as shown in figure 11 in embodiment Operating system, for performing the shared-memory management method that the embodiment of the present invention is provided.Wherein, transported in the operating system Row has a message zero-copy system, and the message zero-copy system includes at least two application APP s, the APPs shared the One bag caching pbuf internal memories, the APP is any APP in the APPs.The memory 1301 is used for the program generation for storing APP Code and data, and program code and data for storage program area.The memory 1301 can be individually present, by total Line is connected with processor 1302.Memory 1301 can also be integrated with processor 1302.

Designed it is understood that Figure 13 illustrate only simplifying for shared-memory management equipment 1300.In practical application In, shared-memory management equipment 1300 can be comprising any amount processor and memory etc., and all can realize the present invention The shared-memory management equipment of embodiment is all within the protection domain of the embodiment of the present invention.

The embodiment of the present invention also provides a kind of computer-readable storage medium, wherein, the computer-readable storage medium can be stored with journey Sequence, the program can be realized when performing part including any shared-memory management method described in above method embodiment or Overall Steps.Foregoing storage medium includes：USB flash disk, read-only storage (English：Read-only memory), arbitrary access deposits Reservoir (English：Random access memory, RAM), mobile hard disk, magnetic disc or CD etc. are various can be with storage program generation The medium of code.

It should be noted that for foregoing each method embodiment, in order to be briefly described, therefore it is all expressed as a series of Combination of actions, but those skilled in the art should know, the present invention is not limited by described sequence of movement because According to the present invention, some steps can be carried out sequentially or simultaneously using other.Secondly, those skilled in the art should also know Know, embodiment described in this description belongs to preferred embodiment, involved action and module is not necessarily of the invention It is necessary.

In the above-described embodiments, the description to each embodiment all emphasizes particularly on different fields, and does not have the portion being described in detail in some embodiment Point, it may refer to the associated description of other embodiment.

, can be by another way in several embodiments provided herein, it should be understood that disclosed device Realize.For example, device embodiment described above is only schematical, such as the division of described unit is only one kind Division of logic function, can there is other dividing mode when actually realizing, such as multiple units or component can combine or can To be integrated into another system, or some features can be ignored, or not perform.It is another, it is shown or discussed each other Coupling direct-coupling or communication connection can be by some interfaces, the INDIRECT COUPLING or communication connection of device or unit, Can be electrical or other forms.

The unit illustrated as separating component can be or may not be it is physically separate, it is aobvious as unit The part shown can be or may not be physical location, you can with positioned at a place, or can also be distributed to multiple On NE.Some or all of unit therein can be selected to realize the mesh of this embodiment scheme according to the actual needs 's.

Described above, the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although with reference to before Embodiment is stated the present invention is described in detail, it will be understood by those within the art that：It still can be to preceding State the technical scheme described in each embodiment to modify, or equivalent substitution is carried out to which part technical characteristic；And these Modification is replaced, and the essence of appropriate technical solution is departed from the scope of various embodiments of the present invention technical scheme.

Claims

1. a kind of shared-memory management method, it is characterised in that methods described is applied in message zero-copy system, the message Zero-copy system operation is in operating system, and the message zero-copy system includes at least two application APP s, described APPs shared drive includes the first bag caching pbuf internal memories, and methods described includes：

Any application APP in the APPs sets the APP to exist the read-write operation attribute of the first pbuf internal memories Current time is read-only；

In the case where the APP receives the pbuf internal memory maloperation notification messages that the operating system is sent, obtain described Pbuf internal memory maloperation field datas in pbuf internal memory maloperation notification messages, the pbuf internal memories maloperation notification message is used In the particular location for reminding the APP to carry out write operation and the write operation to the first pbuf internal memories；

Determine that the APP is carried out in write operation to the first pbuf internal memories based on the pbuf internal memories maloperation field data Deposit maloperation scene.

2. according to the method described in claim 1, it is characterised in that the operating system includes (SuSE) Linux OS, described (SuSE) Linux OS includes mprotect memory management interfaces, and the mprotect memory management interfaces are used to be called to set Put the read-write operation attribute of the first pbuf internal memories；

Any application APP in the APPs sets the APP to exist the read-write operation attribute of the first pbuf internal memories Current time is read-only, including：

The APP is by calling the mprotect memory management interfaces to set readings of the APP to the first pbuf internal memories Write operation attribute is read-only at current time.

3. method according to claim 2, it is characterised in that the mprotect memory management interfaces are used for described APP is called with based on read-write operation attributes of one or more page tables setting APP to the first pbuf internal memories.

4. method according to claim 3, it is characterised in that the message zero-copy system also includes in the 2nd pbuf Deposit, the shared drive of the APPs includes the 2nd pbuf internal memories, and any application APP in the APPs sets institute APP is stated to the read-write operation attribute of the first pbuf internal memories before current time is read-only, methods described also includes：

The APP determines to save as the shared drive of the APPs in the first pbuf in predetermined time, and the APP is pre- Saved as if determining to save as in the 2nd pbuf outside the moment in the shared drive of the APPs, the first pbuf in small page Deposit, big page internal memory, the first pbuf internal memories and one or more of page table existence anduniquess pair are saved as in the 2nd pbuf It should be related to.

5. the method according to any one of Claims 1-4, it is characterised in that the first pbuf internal memories are additionally operable to caching Track identification, methods described also includes：

The APP records the tracing point information of the first pbuf internal memories, the tracing point packet by the track identification Include the application program identification, the application program current process identification and the current time.

6. a kind of shared-memory management method, it is characterised in that methods described is applied to transport in operating system, the operating system Row has message zero-copy system, and the message zero-copy system includes the shared of at least two application APP s, the APPs Internal memory includes the first bag caching pbuf internal memories, and methods described includes：

Any application APP of the operating system in the APPs belongs to the read-write operation of the first pbuf internal memories Property set and indicate that it is read-only to the read-write operation attributes of the first pbuf internal memories to record the APP；

In the case where the operating system monitors the APP to the first pbuf internal memories progress write operation, the operation System sends pbuf internal memory maloperation notification messages to the APP, and the pbuf internal memories maloperation notification message is used to remind institute State the particular location that APP carries out write operation and the write operation to the first pbuf internal memories.

7. method according to claim 6, it is characterised in that the operating system includes (SuSE) Linux OS, described (SuSE) Linux OS includes mprotect memory management interfaces, and the mprotect memory management interfaces are used to be adjusted by the APP To set the APP to the read-write operation attributes of the first pbuf internal memories.

8. method according to claim 7, it is characterised in that the mprotect memory management interfaces are used for described APP is called with based on read-write operation attributes of one or more page tables setting APP to the first pbuf internal memories.

9. method according to claim 8, it is characterised in that the page table includes read-write operation attribute-bit, the behaviour As system record the APP to the read-write operation attributes of the first pbuf internal memories for it is read-only including：

The operating system records read-writes of the APP to the first pbuf internal memories by the read-write operation attribute-bit and grasped It is read-only to make attribute.

10. a kind of shared-memory management device, it is characterised in that the equipment includes：

Be stored with the memory of executable program code；

The processor coupled with the memory；

Operation has operation in application APP and operating system, the operating system to have message zero-copy system on the processor System, the message zero-copy system includes the shared first bag caching pbuf internal memories of at least two application APP s, the APPs, The APP is any APP in the APPs, and the APP is used for：

It is read-only at current time to the read-write operation attribute of the first pbuf internal memories to set the APP；

11. shared-memory management equipment according to claim 10, it is characterised in that the operating system includes Linux Operating system, the (SuSE) Linux OS includes mprotect memory management interfaces, and the mprotect memory management interfaces are used In called to set the read-write operation attribute of the first pbuf internal memories；

By calling the mprotect memory management interfaces to set the APP to belong to the read-write operation of the first pbuf internal memories Property at current time to be read-only.

12. shared-memory management equipment according to claim 11, it is characterised in that the mprotect memory managements connect Mouth is used to call to set read-writes of the APP to the first pbuf internal memories to grasp based on one or more page tables by the APP Make attribute.

13. shared-memory management equipment according to claim 12, it is characterised in that the APP is set to institute described The read-write operation attribute of the first pbuf internal memories is stated before current time is read-only, the APP is additionally operable to：

Determine to save as the shared drive of the APPs in the first pbuf in predetermined time, determined outside predetermined time Saved as in 2nd pbuf and small page internal memory, the 2nd pbuf are saved as in the shared drive of the APPs, the first pbuf Inside save as big page internal memory, the first pbuf internal memories and one or more of page table existence anduniquess corresponding relations.

14. the shared-memory management equipment according to any one of claim 10 to 13, it is characterised in that the first pbuf Internal memory is additionally operable to cache track identification, and the APP is additionally operable to：

The tracing point information of the first pbuf internal memories is recorded by the track identification, the tracing point information includes described answer With program identification, the application program current process identification and the current time.

15. a kind of shared-memory management device, it is characterised in that the equipment includes：

Be stored with the memory of executable program code；

The processor coupled with the memory；

Operation has operation in application APP and operating system, the operating system to have message zero-copy system on the processor System, the message zero-copy system is included at least two shared first packet buffer pbuf of application APP s, the APPs Deposit, the APP is any APP in the APPs, the operating system is used for：

Any application APP in the APPs is set to the read-write operation attribute of the first pbuf internal memories indicates note It is read-only to the read-write operation attribute of the first pbuf internal memories to record the APP；

In the case where the operating system monitors the APP to the first pbuf internal memories progress write operation, to the APP Pbuf internal memory maloperation notification messages are sent, the pbuf internal memories maloperation notification message is used to remind the APP to described the One pbuf internal memories carry out the particular location of write operation and the write operation.

16. shared-memory management equipment according to claim 15, it is characterised in that the operating system includes Linux Operating system, the (SuSE) Linux OS includes mprotect memory management interfaces, and the mprotect memory management interfaces are used In being called by the APP to set the APP to the read-write operation attributes of the first pbuf internal memories.

17. shared-memory management equipment according to claim 16, it is characterised in that the mprotect memory managements connect Mouth is used to call to set the read-write operation attribute of the first pbuf internal memories based on one or more page tables by the APP.

18. shared-memory management equipment according to claim 17, it is characterised in that the page table belongs to including read-write operation Property mark, the logging modle record the APP to the read-write operation attributes of the first pbuf internal memories for it is read-only including：

The APP is recorded to the read-write operation attribute of the first pbuf internal memories for only by the read-write operation attribute-bit Read.

19. a kind of shared-memory management equipment, it is characterised in that the shared-memory management equipment includes：

Be stored with the memory of executable program code；

The processor coupled with the memory；

Operation has operation in application APP and operating system, the operating system to have message zero-copy system on the processor System, the message zero-copy system is included at least two shared first packet buffer pbuf of application APP s, the APPs Deposit, the APP is any APP in the APPs；

The APP is used for any one of perform claim requirement 1 to the 5 shared-memory management method；

The operating system is used for the shared-memory management method described in perform claim 6 to 9 any one of requirement.