METHOD AND COMPUTER FOR EXECUTING MEMORY-STORED TIMER

    公开(公告)号:JPH10307767A

    公开(公告)日:1998-11-17

    申请号:JP2997296

    申请日:1996-01-23

    Inventor: LIN WILLIAM

    Abstract: PROBLEM TO BE SOLVED: To set a timer in a system without making long the time needed to set and reset the timer by traversing a cell in a non-free state which is related to a time storage entry and implementing a function which is related to the time storage entry. SOLUTION: Time storage entries 60 and 65 in a link list 50 are related to a timer for a specific data packet. After the time storage entries 60 and 65 which have expired are traversed, the time storage entries 60 and 65 are removed from the link list 50. After time storage entries 60 and 65 like those are all removed from cells 1 and 6 in a specific non-free state, an array entry related to the cells 1 and 6 in the non-free state is removed from a double link list 50 and an arbitrary memory which relates to an arbitrary data structure is reallocated.

    COMPUTER SYSTEM DATA I/O USING REFERENCE BETWEEN CPU AND MULTIPLEX MEMORY DEVICE

    公开(公告)号:JPH103441A

    公开(公告)日:1998-01-06

    申请号:JP34141396

    申请日:1996-12-20

    Abstract: PROBLEM TO BE SOLVED: To eliminate an unnecessary copying of data in a processor and also between processors by using a multiple DSC/S(data source/sink) to transform a DST(data stream). SOLUTION: A 1st pointer is generated by a 1st DSC/S which includes a 1st global network address designating a 1st storage position where a 1st DST is stored, and only the 1st pointer is transferred to a 2nd DSC/S. In the same way, a 2nd pointer is generated for a 2nd DST. At the same time, the 1st and 2nd pointers are chained together so that a 1st chained pointer is obtained. Then only the 1st chained pointer is transferred to a 3rd DSC/S via a 3rd node where a 3rd data fragment of a message is stored, and also the message is transferred to its destination. The 1st pointer is processed by the 3rd DSC/S and therefore transformed into plural secondary pointers. At the same time, a protocol header is chained to each part of the 1st DST.

    I/O DEVICE AND COMPUTER SYSTEM DATA I/O FOR REFERENCE BETWEEN MULTIPLEX MEMORY DEVICES

    公开(公告)号:JPH09325944A

    公开(公告)日:1997-12-16

    申请号:JP34141596

    申请日:1996-12-20

    Abstract: PROBLEM TO BE SOLVED: To exclude the unwanted copying of data by transferring only a 1st pointer generated by a 1st data source/sink to a 2nd data source/sink, generating a 2nd pointer at the 2nd data source/sink and linking the 1st and 2nd pointers. SOLUTION: When a global pointer indicates data 560 and specified data 561 of its own application exist in a memory 551, an application server process 531 dispatches the global painter indicating the data 560 and the other global pointer indicating data 562 specified for that application to an intermediate protocol process 532. The intermediate protocol process 532 does not copy the data 561 and 560 into the related memory but dispatches two global pointers indicating the data 561 and 560 to a TCP/IP process 533 together with a 3rd global pointer indicating intermediate protocol header data 562.

    METHOD FOR MAINTENANCE OF NETWORK CONNECTION BETWEEN APPLICATIONS AND DATA-PROCESSING SYSTEM

    公开(公告)号:JPH0926891A

    公开(公告)日:1997-01-28

    申请号:JP2997396

    申请日:1996-01-23

    Abstract: PROBLEM TO BE SOLVED: To make related data the checkpoint from the processor of a main node to a backup processor when a main processor is taken over and to enable a smooth shift so as not to disconnect the connection between the program of a server side and a client side when a spontaneous processing is taken over. SOLUTION: A network is provided with a server node 15 communicating with a protocol process node 20 and the protocol process node 20 communicates with a transmission network 30. The transmission network 30 is connected with a client note 25. When a main processor 26 is taken over, related data is made the checkpoint from the processor 26 of a main node 22 to a backup processor 46 so that the backup processor 46 may take over the role of the main processor 26 and a socket may be maintained to an open state. Thus, a network connection can be maintained when a spontaneous processing switch is performed.

    DISTRIBUTED DATA CACHE FOR MULTIPROCESSOR SYSTEM WITH CACHE

    公开(公告)号:JPH0922384A

    公开(公告)日:1997-01-21

    申请号:JP862596

    申请日:1996-01-22

    Abstract: PROBLEM TO BE SOLVED: To secure reliability by distributing a cache. SOLUTION: Respective processors 12 maintain the cache for discriminating the states of the caching of respective files, the blocks of the respective files to be cached and cached open files. As long as only one processor opens the file and opens it for read/write access, the processor is allowed to perform read/write caching to the file. When the processor opens the file for the read access, the processor is allowed to perform read caching as long as the other processor does not open the file for the read/write access. After the last processor which performs the read/wire access to the file closes the file, a disk system upgrades a cache state to the file.

    DATA TRANSMISSION METHOD
    100.
    发明专利

    公开(公告)号:JPH0916531A

    公开(公告)日:1997-01-17

    申请号:JP2997096

    申请日:1996-01-23

    Abstract: PROBLEM TO BE SOLVED: To provide a new system capable of efficiently transmitting/receiving data among plural processes under a fixed condition when much time is required for data copying as compared with the execution of processing and the data copying operation is not important for the security of failure tolerance. SOLUTION: In a parallel processing and fault tolerant type computer system, data are transmitted among plural processes in a single CPU 110 (142) by two methods. In the 1st method, data are copied in each transmission. In the 2nd method, data are transmitted through a shared memory queuing system without being copied. The 1st method is used for securing failure tolerance and linear extendability. The 2nd method minimizes time required for inter-process communication. Since the shared memory queuing system is used, the vertical and horizontal modules of a process executed in the same CPU 110 (142) are increased.

Patent Agency Ranking