CN102830965A - Method for caching and multiplexing position-independent code in Java virtual machine - Google Patents
Method for caching and multiplexing position-independent code in Java virtual machine Download PDFInfo
- Publication number
- CN102830965A CN102830965A CN2012102602800A CN201210260280A CN102830965A CN 102830965 A CN102830965 A CN 102830965A CN 2012102602800 A CN2012102602800 A CN 2012102602800A CN 201210260280 A CN201210260280 A CN 201210260280A CN 102830965 A CN102830965 A CN 102830965A
- Authority
- CN
- China
- Prior art keywords
- code
- file
- buffer memory
- cache file
- cache
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Devices For Executing Special Programs (AREA)
- Memory System Of A Hierarchy Structure (AREA)
Abstract
The invention belongs to the field of design techniques of a Java virtual machine, and particularly relates to caching and multiplexing management of a position-independent code generated by a just-in-time compiler. The invention designs an efficient method for managing the position-independent code, and realizes the method on a virtual machine Dalvik in an Android operating system. According to the method, a favorable cache file format and a cache file organization manner are designed for the position-independent code, and an efficient multiplexing detecting and loading strategy is designed. Through the management method, the positon-independent code generated by the complier of the virtual machine is effectively multiplexed, thereby improving the executing efficiency of the virtual machine.
Description
Technical field
The invention belongs to the Java Virtual Machine technical field, be specifically related to the buffer memory and the multiplexing method of position independent code.
Background technology
Java Virtual Machine is as a kind of important runtime system, because of being easy to expansion, being convenient to advantages such as trace routine execution and program portable by broad research.In recent years, by methods such as instant compiling, executed in parallel, the virtual machine of desktop platform and server platform is carried out efficient and has been obtained very big lifting.Instant technique of compiling is by dynamic profile information; Make virtual machine only pay close attention to by the use of thermal means; Dwindled the compiling scope of compiler; Reduce the compilation time expense on the whole, reduced the required storage space of code that compiling generates, therefore be widely used in PC platform, server platform even embedded platform.Yet instant compiling system is also carried out to degree and is brought overhead, as detecting time-delay and compiling time-delay.
The executable code buffer memory is multiplexing to be the technology that an important reduction detects and compiling is delayed time.But this technological core is to the design of the executing location independent code of virtual machine generation and the buffer memory and the reuse management of position independent code.Code cache is about to the executable code of compiling generation with document form, presses certain format to secondary storage device, like hard disk, flash memory etc.The purpose of buffer memory code to file be make other programs or this program after operational process in can the identical part or all of code of multiplexing function.The typical application of code cache multiplex technique is the SB of Unix or (SuSE) Linux OS.Function in the SB has only a physical copy, carries out and is independent of each other but can be loaded into different memory address space by different processes, thereby not only save code memory space but also save compilation time.
The common cache of position independent code and multiplexing method use symbol table and dynamic linker to realize jointly.In compilation process (normally static compiling), the position of compiler from code is relevant extracts the symbolic information that is relied on, and calculates the address offset of each symbol with respect to a certain overall fixed position in the space, program address; Then, symbol and off-set value thereof are stored in the unified symbol table; At last, symbol table and descriptor thereof are stored to file system with code.After handling like this, the positional information that is relied in the code is replaced by points to a certain index in the symbol table, has eliminated the specific address positional information, thus the location independent of code.Therefore, use symbol table, realized the location independentization of code on the one hand, accomplished cache management on the other hand again position independent code.The essence of this method is that directly quoting of address location replaced with indirect referencing, accomplishes when operation is postponed till in the address binding operation of position reference point in the code.Therefore, dynamic linker and link process when this method also must cooperate special operation, it is multiplexing just the buffer memory code to be able to.During multiplexing code, operating system needs the query tokens table, obtains the current location information of required symbol, and the filling symbol table is in order to follow-up use, and this process is called the resolving of symbol.If symbol is not in the present procedure mirror image and in SB, and dynamic linker need load this SB itself and the SB that relies on and accomplish essential symbol resolution process so, and this process is called the chain type of symbol and resolves.After resolving end, dynamic linker need load symbol and correlative code, needs the many information of checking in this process, and is very complicated.This makes this method be suitable for static compiler and but is inappropriate for instant compiling system, because as far as instant compiler, the expense of maintain symbol table, compute sign skew belong to the part of program execution time, and static compiler need not to consider these expenses; Special linker can increase the realization difficulty of instant compiler, the stand-by period that special Symbolic Links resolving can the increase program be carried out.
Summary of the invention
The objective of the invention is to design a kind of buffer memory of position independent code efficiently and multiplexing method.This method is used less cache file and multiplexing file overhead, does not need complex data structures and algorithm simultaneously, can simplify the code administration of instant compiler, helps to promote the execution efficient of virtual machine.
The position independent code buffer memory and the multiplexing method of the present invention's design comprise: design cache file storage format and necessary metadata information, so that can realize the buffer memory of code efficiently; Design cache file organizational form is so that can multiplexing efficiently code; Design realizes the buffer memory detection algorithm, in operational process, detects the existence and the legitimacy of cache file, and upgrades the buffer status of each Java method; According to buffer status and multiplex strategy, the code how decision loads in the cache file is multiplexing to realize.
The position independent code buffer memory and the multiplexing method of the present invention's design can be divided into the two large divisions: executable code caching method and cache file multiplexing method.Aspect the executable code caching method, virtual machine is by cache file layout memory location independent code shown in Figure 1, and press the also code of multiplexing buffer memory of Fig. 3 and flow process buffer memory shown in Figure 5.Fig. 2 has then showed the cache file method for organizing that the present invention designed.
First, the caching method of executable code.The caching method of executable code mainly comprises: the layout type of cache file, the organizational politics of cache file, position independent code buffer memory.
Described cache file layout, as shown in Figure 1.From figure, can know; Cache file is made up of head and content two parts: top of file comprises the magic number (Magic Number) of sign this document type; Total size, the constant pool space size of the verification that except that magic number, remains file content and, this document, the code area is big or small and the size of abnormal information table, and it is to be about to the assignment of code space that loads that the integrality that these information are mainly used in inspection file before loading code and guides memory allocator; File content comprises constant pool, executable code itself and the abnormal information table that code uses.The present invention uses the ccf postfix notation to use the code cache file of Fig. 1 layout.Placement strategy that the present invention designs has following advantage:
1. when integrality and safety examination, can carry out by read-only metadata information, thereby can accomplish checkout procedure fast;
2. before loading the CCF content, the size information in the metadata can instruct required all storage spaces of executable code of distributing buffer memory with code manager in advance, thereby makes loading procedure become simple file read and internal memory write operation.
Described cache file organizational politics sees shown in Figure 2.Find that through investigation each class file comprises a plurality of Java methods, each Java method is by different index signs; And different class files is by unique identity signature sign of class file.Therefore, the present invention uses the directory management cache file, and a catalogue is represented a class file, and the code cache of Java method in this class file of the representation of file in the catalogue.Can know that from Fig. 2 the present invention is with a type unique identification structure directory name, with the call number structure cache file name of method in class.This buffer organization strategy has following advantage:
1. the structure nomenclature is directly used existing identification information and is made up filename and directory name, thereby saves complicated unique key assignments process, has reduced the realization difficulty of method;
2. cache file detected when the formation rule of type catalogue and cache file name can be used for moving, and is convenient efficiently multiplexing;
3. need not information such as independent maintenance documentation table with the log file position, also needn't realize the ff algorithm, these work all by accomplishing by file system, make buffer memory and multiplex process all become simple.
Said position independent code buffer memory is the basis with cache file layout and organizational form thereof, and on this basis, the present invention has designed the buffer memory flow process, specifically sees shown in Figure 3.Bold box is represented the caching that the present invention adds among the figure.The present invention is saved in the binary code that generates in the buffer memory according to the translation result of instant compiler in the Java Virtual Machine and goes.Detailed process is: after instant compiler generates the irrelevant binary code of good position the verification of Accounting Legend Code with, prepare the metamessage of this method, deposit in the buffer memory according to preamble described cache file organizational politics and layout then.
Second portion, the multiplexing method of cache file.The multiplexing method that the present invention designed is in two stages: the Detection of Existence stage of cache file and cache file load phase.The former is responsible for detecting the code whether method of being called exists buffer memory, and the legitimacy of detection of code (if having buffer memory really), and its efficiency of algorithm directly influences the expense of virtual machine; The latter is responsible for loading cache file and carrying out the code that loads on suitable opportunity.
Method of application buffer status of the present invention intercoms between two stages mutually, shown in Fig. 4 left side.The present invention detects the cached version that whether has code earlier, and upgrades the zone bit of testing result to method.Press Fig. 4 right side designed method buffer status transformation rule and conversion condition then: during virtual machine activation, each method all is designated as original state---UNCHECKED, and whether expression the unknown exists buffer memory; CACHED representes that this method detects through buffer memory, and has reusable code cache version that the buffer memory load operation can load the buffer memory code of this method on appropriate opportunity; NOCACHE representes that this method is tested, but does not have reusable code cache file at present.Each method initially all is the UNCHECKED state; When implementing this method, carrying out buffer memory detects; If do not detect this method then change into the NOCACHE state; If cache file has existed and then changed into the CACHED state, the binary code of method can be changed into the NOCACHE state again when losing efficacy or unloading.
In the cache file Detection of Existence stage, the present invention designs following buffer memory detection algorithm: 1. the identification information that obtains class file from the class file of method; 2. construct the complete trails of cache file with class file identification information and method index in the method table in class file; 3. use the file system built-in function, whether the file pointed like this path of detection such as access, stat exists, and the buffer status sign of the testing result update method returned of foundation.Why this method efficiently is: the position of file and index information are all by file system maintenance; Algorithm and process that whether the detection file exists are also accomplished by the system call function, and this method that makes is saved when realizing and safeguarded cache file information and the task to the access interface of these information is provided; Information type of being merely sign and method index that the structure path is required, they can directly obtain from the data structure body of method when operation, need not extra storage space and virtual machine function call; Path configuration is the character string concatenation, and itself is directly simple.
In the cache file load phase, multiplex strategy that the present invention designs is that method run time MET (Method Executing Time) is multiplexing, promptly before method is carried out, checks buffer status, and determines whether to load earlier to carry out according to buffer status again.This ultimate principle with instant compiling is similar, so MET is multiplexing is also referred to as instant loading (Just-In-Time Load).The instant method that is performed that only loads that loads, multiplexing than all, this strategy has reduced the memory headroom demand of buffer memory code on the whole, and resource is rationally utilized.Fig. 5 is the multiplexing operating process of buffer memory that the present invention designs.Bold box is designed content by the present invention among the figure.Compare with traditional instant compilation process, position independent code buffer memory of the present invention and reuse management method are that virtual machine provides a new executable code source, and this source and compilation process immediately all are online completion.Detailed process is following: at first adopt the execution of explaining the mode access method of carrying out, carry out the detection of cache file this moment, if there is the buffer memory of this method in the buffer memory, then the legitimacy of cache file carried out verification.If through verification then directly load and directly carry out binary code like internal memory; If the verification existing problems then use instant compiler to recompilate, the result is preserved the meeting buffer memory, directly carry out binary code simultaneously.If in buffer memory, there is not the record of this method; The mode that then still adopts explanation to carry out is carried out; Call counting simultaneously; After counter surpasses certain threshold value, call the compiling that instant compiler carries out method, and the tissue of the cache file of describing according to preamble and layout type carry out code to two of this method and carry out buffer memory.
The invention has the beneficial effects as follows: 1) the present invention designs and has realized that a kind of buffer memory and reuse management method to the dynamic position independent code that generates of Java instant compiler, its lightweight characteristic and less expense can be used it when operation.2) the present invention has designed texture's method of the cache file of lightweight, by the identification information and the file system functions interface of existing java class and method, can manage the buffer memory code efficiently, and can detect, safeguard and load the multiplexing fine support that provides for buffer memory; This method can be used for other management application fields based on file.3) the present invention has designed the cache file detection method based on state, can detect the existence of cache file efficiently; Status Flag has separated detection and loading procedure, makes loader more flexible multiplexing opportunity.4) the present invention has designed the instant loading strategy of code reuse, makes virtual machine realize correctly multiplexing to position independent code with less expense.
Description of drawings
Fig. 1 is designed code cache file content layout by the present invention.
Fig. 2 is designed the cache file organizational form by the present invention.
Fig. 3 is a code cache operating process of the present invention.
Fig. 4 is designed buffer status conversion condition and rule by the present invention.
Fig. 5 loads multiplexing operating process for code of the present invention.
Fig. 6 is a cache file detection algorithm of the present invention.
Embodiment
Practical implementation process of the present invention divides two steps to carry out: cache location independent code to file and from the multiplexing executable code of file.When implementing, the code that the former relates generally to instant compiler generates and management, needs in appropriate, design the buffer memory form by the present invention and organizational form is carried out buffer memory to the executable code that on-the-flier compiler generates; The latter relates generally to interpreter and code manager, and need in appropriate, design the cache file detection method by the present invention and multiplex strategy loads the buffer memory code.Carefully state respectively from two steps below:
At first, virtual machine need the irrelevant verification of calculating location and with metadata in each size information; Then, create cache file and catalogue (if not having the catalogue of corresponding class file) with the structure nomenclature; At last, with binary mode ready metadata is write in the newly-built cache file with source code.Among the present invention, these operations occur in after the instant compiling end, before code is installed and carried out, shown in bold box among Fig. 4.
Virtual machine hopes that usually the code that is buffered is loaded more early well more, thereby can make virtual machine more early carry out the higher binary code of efficient.Be that method is when being performed for the first time the instant period the earliest that loads.By selectivity compiling flow process, the engine of execution first of Java method all is an interpreter.Therefore, the present invention is arranged in the multiplexing loading procedure of buffer memory in the interpreter and carries out.Owing to before instant the loading, must detect the existence and the legitimacy of cache file earlier, thereby in the implementation process, the present invention also is arranged in the buffer memory testing process in the interpreter.Fig. 5 left side thick line flow process has been showed this process.
Buffer memory existence testing process among Fig. 5 is realized by algorithm shown in Fig. 6 false code.Use Status Flag and transformation rule thereof shown in Figure 4, can guarantee that the cache file testing process only takes place once; Detection then can directly realize through comparing the buffer status sign simply.When method is performed first; Interpreter must call that dvmCodeCacheCheck detects whether cache file exists and according to the buffer status (the 6th is capable) of testing result update method; Make redirect at last and determine (the 9th row): for the CACHED state outcome, virtual machine will be from explaining that execution switches to loading buffer memory code and carries out this locality execution; For the NOCACHE state outcome, virtual machine is still carried out according to the normal processes that selectivity compiles, and is compiled execution up to the method heating, perhaps because of heat extremely end of execution in explanation inadequately.In this algorithm, dvmCodeCacheCheck accomplishes cache file Detection of Existence process with structure nomenclature and file system call interface function.
Claims (4)
1. position independent code buffer memory and multiplexing method in the Java Virtual Machine is characterized in that being divided into the two large divisions: executable code caching method and cache file multiplexing method, wherein:
First, the caching method of executable code comprises: the layout type of cache file, the organizational politics of cache file and position independent code buffer memory;
Described cache file layout, cache file is made up of top of file and file content two parts; Said top of file comprises the magic number of sign this document type; Total size, the constant pool space size of the verification that except that magic number, remains file content and, this document, the code area is big or small and the size of abnormal information table, and it is to be about to the assignment of code space that loads that the integrality that these information are used for inspection file before loading code and guides memory allocator; Said file content comprises constant pool, executable code itself and the abnormal information table that code uses;
Described cache file organizational politics uses the directory management cache file, the corresponding catalogue of each java class, and all methods in this java class correspond to a file in this catalogue, preserve the code cache of this Java method in this document; Directory name is named according to the unique identification of java class, and filename carries out unique sign and identification according to the call number of each Java method in class;
Said position independent code buffer memory is the basis with cache file layout and organizational form thereof, according to the translation result of instant compiler in the Java Virtual Machine binary code that generates is saved in the buffer memory and goes; Detailed process is: after instant compiler generates the irrelevant binary code of good position the verification of Accounting Legend Code with, prepare the metamessage of this method, deposit in the buffer memory according to cache file organizational politics recited above and layout then;
Second portion, the multiplexing method of cache file was divided into for two stages: the Detection of Existence stage of cache file and cache file load phase; The former is responsible for detecting the code whether method of being called exists buffer memory, and the legitimacy of detection of code; The latter is responsible for loading cache file and carrying out the code that loads on suitable opportunity;
The state of said cache file intercoms between two stages mutually: detect the cached version that whether has code earlier, and upgrade the zone bit of testing result to method; Undertaken multiplexing by the buffer status transformation rule and the conversion condition of designed method then: during virtual machine activation, each method all is designated as original state---UNCHECKED, and whether expression the unknown exists buffer memory; CACHED representes that this method detects through buffer memory, and has reusable code cache version that the buffer memory load operation can load the buffer memory code of this method on appropriate opportunity; NOCACHE representes that this method is tested, but does not have reusable code cache file at present; Each method initially all is the UNCHECKED state; When implementing this method, carrying out buffer memory detects; If do not detect this method then change into the NOCACHE state; If cache file has existed and then changed into the CACHED state, the binary code of method can be changed into the NOCACHE state again when losing efficacy or unloading.
2. position independent code buffer memory and multiplexing method in the Java Virtual Machine according to claim 1 is characterized in that: in the cache file Detection of Existence stage, adopt the buffer memory detection algorithm: (1) obtains the identification information of class file from the class file of method; (2) construct the complete trails of cache file with class file identification information and method index in the method table in class file; (3) according to the complete trails of this cache file, use the file system built-in function, detects file pointed and whether exist, and the buffer status sign of the testing result update method returned of foundation.
3. position independent code buffer memory and multiplexing method in the Java Virtual Machine according to claim 2; It is characterized in that: in the cache file load phase; Employing method run time MET multiplex strategy; Promptly before method is carried out, check buffer status, and determine whether to load again earlier to carry out according to buffer status.
4. position independent code buffer memory and multiplexing method in the Java Virtual Machine according to claim 3; It is characterized in that: the detailed process of said cache file multiplexing method is following: at first adopt the execution of explaining the mode access method of carrying out; Carry out the detection of cache file this moment; If have the buffer memory of this method in the buffer memory, then the legitimacy of cache file carried out verification; If through verification then directly load and directly carry out binary code like internal memory; If the verification existing problems then use instant compiler to recompilate, the result is preserved the meeting buffer memory, directly carry out binary code simultaneously; If in buffer memory, there is not the record of this method; The mode that then still adopts explanation to carry out is carried out; Call counting simultaneously; After counter surpasses certain threshold value, call the compiling that instant compiler carries out method, and the tissue of the cache file of describing according to preamble and layout type carry out code to two of this method and carry out buffer memory.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210260280.0A CN102830965B (en) | 2012-07-26 | 2012-07-26 | A kind of buffer memory of position-independentcode code in Java virtual machine and multiplexing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210260280.0A CN102830965B (en) | 2012-07-26 | 2012-07-26 | A kind of buffer memory of position-independentcode code in Java virtual machine and multiplexing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102830965A true CN102830965A (en) | 2012-12-19 |
| CN102830965B CN102830965B (en) | 2015-12-09 |
Family
ID=47334115
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210260280.0A Expired - Fee Related CN102830965B (en) | 2012-07-26 | 2012-07-26 | A kind of buffer memory of position-independentcode code in Java virtual machine and multiplexing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102830965B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103488517A (en) * | 2013-09-10 | 2014-01-01 | 新浪网技术(中国)有限公司 | PHP code compiling method and device and PHP code running method and device |
| CN104850423A (en) * | 2015-01-16 | 2015-08-19 | 中国矿业大学 | Method for identifying startup stage of application program under android system |
| CN115309405A (en) * | 2022-08-24 | 2022-11-08 | 北京熵核科技有限公司 | Code link optimization method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20070092559A (en) * | 2006-03-10 | 2007-09-13 | 삼성전자주식회사 | Apparatus and method for executing thread scheduling in virtual machine |
| CN102375731A (en) * | 2010-08-23 | 2012-03-14 | 融建信息技术(厦门)有限公司 | Coding-free integrated application platform system |
-
2012
- 2012-07-26 CN CN201210260280.0A patent/CN102830965B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20070092559A (en) * | 2006-03-10 | 2007-09-13 | 삼성전자주식회사 | Apparatus and method for executing thread scheduling in virtual machine |
| CN102375731A (en) * | 2010-08-23 | 2012-03-14 | 融建信息技术(厦门)有限公司 | Coding-free integrated application platform system |
Non-Patent Citations (2)
| Title |
|---|
| QIANG LI等: "VM-based Architecture for Network Monitoring and Analysis", 《THE 9TH INTERNATIONAL CONFERENCE FOR YOUNG COMPUTER SCIENTISTS》, 21 November 2008 (2008-11-21) * |
| 陈微等: "降低协同设计虚拟机启动开销的译码后指令缓存技术", 《计算机研究与发展》, vol. 48, no. 1, 31 January 2011 (2011-01-31) * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103488517A (en) * | 2013-09-10 | 2014-01-01 | 新浪网技术(中国)有限公司 | PHP code compiling method and device and PHP code running method and device |
| CN103488517B (en) * | 2013-09-10 | 2016-08-10 | 新浪网技术(中国)有限公司 | PHP code Compilation Method, PHP code operation method and device |
| CN104850423A (en) * | 2015-01-16 | 2015-08-19 | 中国矿业大学 | Method for identifying startup stage of application program under android system |
| CN104850423B (en) * | 2015-01-16 | 2017-12-26 | 中国矿业大学 | The method for identifying the application program launching stage under Android system |
| CN115309405A (en) * | 2022-08-24 | 2022-11-08 | 北京熵核科技有限公司 | Code link optimization method |
| CN115309405B (en) * | 2022-08-24 | 2023-08-04 | 北京熵核科技有限公司 | Code link optimization method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102830965B (en) | 2015-12-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11106475B2 (en) | Correlating class loader objects across execution environments | |
| US8914780B2 (en) | Incremental generation of managed assemblies | |
| US8677331B2 (en) | Lock-clustering compilation for software transactional memory | |
| US9298448B2 (en) | System and method for run time dependency resolution | |
| US7962959B1 (en) | Computer resource optimization during malware detection using antivirus cache | |
| US8341615B2 (en) | Single instruction multiple data (SIMD) code generation for parallel loops using versioning and scheduling | |
| CN105453054B (en) | The system and method for content dynamic select garbage collection algorithm based on heap region | |
| US8839215B2 (en) | String cache file for optimizing memory usage in a java virtual machine | |
| US20130282676A1 (en) | Garbage collection-driven block thinning | |
| CN110955431A (en) | Processing method and device of compiling environment | |
| CN102135904A (en) | Multi-core target system oriented mapping method and device | |
| KR20120082295A (en) | Method and apparatus of web browsing through code caching and optimization for javascript | |
| JP5966509B2 (en) | Program, code generation method, and information processing apparatus | |
| CN103473059B (en) | A kind of general-purpose operating system supporting multiple architectural framework | |
| CN102455971A (en) | Application-level random instruction testing method, system and device | |
| CN104133733A (en) | Memory error detection method | |
| CN100362474C (en) | Time-multiplexed speculative multi-threading to support single-threaded applications | |
| CN105224309A (en) | A kind of method and apparatus transferring function | |
| CN102830965A (en) | Method for caching and multiplexing position-independent code in Java virtual machine | |
| CN103631589B (en) | Method and device for recognizing application | |
| US20240231864A9 (en) | Hybrid just in time load module compiler with performance optimizations | |
| Trotman et al. | Micro‐and macro‐optimizations of SaaT search | |
| Gal | Efficient bytecode verification and compilation in a virtual machine | |
| KR102456017B1 (en) | Apparatus and method for file sharing between applications | |
| Prasad et al. | Improving the performance of processing for small files in Hadoop: A case study of weather data analytics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20190704 Address after: Room 1103, Building 21, 39 Jibang Road, Zhongming Town, Shanghai 202163 Patentee after: SHANGHAI JILIAN NETWORK TECHNOLOGY Co.,Ltd. Address before: 200433 No. 220, Handan Road, Shanghai, Yangpu District Patentee before: Fudan University |
|
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20151209 |