Process for transferring data between nodes in a multinodal information system

• US 6,009,472 A
• Filed: 07/22/1996
• Issued: 12/28/1999
• Est. Priority Date: 07/21/1995
• Status: Expired due to Term

• Current Assignee: Bull Sas (Atos SE)
• Original Assignee: Bull Sas (Atos SE)
• Inventors: Derne, Henriette, Boudou, Alain
• Primary Examiner(s): Geckil, Mehmet B.
• Application Number:
  US08/684,996
• Time in Patent Office: 1,254 Days
• Field of Search: 395/200.31, 395/200.47, 395/200.62, 395/200.6, 395/200.63, 395/200.64
• US Class Current: 709/232
• CPC Class Codes: G06F 15/17 using an input/output type connection, e.g. channel, I/O port

Abstract

In a multinode information system (SYS), a processor (Pxm) of a first node (Nx) emits a message of predetermined size (MESS) in the form of a write operation in the communication module (ISLx) of the first node and transmits it to the communication module (ISLy) of a second node (Ny), where it is stored in a queue of messages received (FIFO-MGT), in the order of their reception, then dequeued by a processor (Pyn) of the second node (Ny) in the form of a read operation in the communication module (ISLy) of this node so that the information contained in the message can be processed.

Claims

1. A process for transmitting a message (MESS) between data processors (Pxm, Pyn) in an information system (SYS) distributed in at least a first node (Nx) and a second node (NY), each of which has a module for communication (ISL) with another node by means of a transmission link (Lxy), comprising emitting a message (MESS) of predetermined size, by a first processor of the first node in the form of a write operation (PIO Store) in a first communication module (ISLx) of the first node and transmitting the message through the transmission link (Lxy) to a second communication module (ISLy) of the second node, storing the message in the second communication module in a queue of messages received (Qhp, Qlp) in the order of reception, and retrieving the message from the queue in the form of a read operation (PIO Load) in the second communication module of said second node by a second processor of the second node which incorporates the queue so that the information (SWM) contained in the message can be processed, said storing and retrieving operations being carried out sequentially and signalizing the arrival and the overflow status of a message at the second processor of the second node, the queue being constituted by a bank of registers (FIFO-MESS) having a first address (FIFO-IN) for executing message engineering operations and a second address (FIFO-OUT) for executing message dequeuing operations, the first and second addresses being respectively contained in a global space (GSP) visible for all nodes, and in a local space (LSP) only visible for the node where the message is to be received, of an input-output address space (ISLO) related to the registers which contain the modules of said nodes.

2. The process according to claim 1, characterized in that the message emitted by the processor of the first node is intended for said first node, where it is stored in the queue of messages received, in the order of their reception.

3. The process according to claim 1, characterized in that since the size of the queue is limited, the communication module incorporating the queue sends the communication module of the first node a response which indicates whether or not the message has been stored in the queue.

4. The process according to claim 1, characterized in that the current status of the queue is obtained from a field (MTAG) of the message by the second processor which retrieves the message from the queue.

5. The process according to claim 4, characterized in that the field includes at least one bit (E) which represents an empty status of the queue and indicates that the contents of the message read are invalid.

6. The process according to claim 4, characterized in that the field includes at least one bit (F) which represents an overflow status of the queue, indicating that since the queue is full, at least one message received has been rejected.

7. The process according to claim 4, characterized in that the field includes at least one bit (L) which indicates that the dequeued message is the last message in the queue.

8. The process according to claim 1, characterized in that a processor of the node incorporating the queue is only notified of the arrival of a message by an interruption if the queue is empty when it receives a message.

9. An information processing system (SYS) for transmitting messages comprising data processors (Pxm, Pyn) distributed In at least two nodes (Nx) (Ny), each node having a module for communication (ISL) with another node, a transmission link (Lxy) connecting said nodes to implement a message transmission process, a first node including a first processor adapted to emit a message having a predetermined size in the form of a write operation (PIO Store) in a first communication module (ISLx) of the first node and is transmitted, through the transmission link (Lxy), to a second node having a second communication module (ISLy) for storing the message in a queue of messages received (Qhp, Qlp) in the order of reception, and a read operation (PIO Load) retrieving the message from the queue in the second communication module of said second node by a second processor of the second node which incorporates the queue so that the information (SWM) contained in the message can be processed, said storing and retrieving being carried out sequentially in the second node and means for signalizing the arrival and the overflow status of a message to the second processor of the second node, the queue being constituted by a bank of registers (FIFO-MESS) having a first address (FIFO-IN) for executing message enqueuing operations and a second address (FIFO-OUT) for executing message dequeuing operations, the first and second addresses being respectively contained in a global space (GSP) visible for all nodes, and in a local space (LSP) only visible for the node where the message is to be received, of an input-output address space (ISLO) related to the registers which contain the modules of said nodes.

10. The system according to claim 9, characterized in that each of said nodes contains several receiving queues which correspond to different priorities, all of which are specified by said first and second addresses.

11. The system according to claim 9, characterized in that each communication module includes an overflow register (FIFO-OVF) which indicates the identity a the node in which at least one message has been rejected because the queue was full.

12. The system according to claim 9, characterized in that the arrival of a message is indicated by an interruption of a processor of the second node, which interruption is controlled by a bit associated with the queue in an interruption register (FIFO-ITR) included in the second communication module, which bit has a first value when a message is stored in an empty queue and has a second value through a write operation at the address of this register.

13. The system according to claim 9, characterized in that each node has its own operating system.

14. An integrated circuit, characterized in that it incorporates a communication module (ISL) for the implementation of a process for transmitting a message (MESS) between data processors of an information system (SYS), said data processors (Pxm), Pyn) being distributed in at least a first node (Nx) and a second node (Ny), each of which has a module for communication (ISL) with another node, a transmission link (Lxy) for effecting communications between nodes, a first processor for the first node for emitting a message, of predetermined size, in the form of a write operation (PIO Store) in a first communication module (ISLx) of the first node and the transmission link Lxy for transmitting the message to a communication module (ISLy) of the second node, means for storing the message in a queue of messages received (Qhp, Qlp) in the order of reception and a read operation for retrieving messages from the queue in a second communication module of said second node by a second processor of the second node which incorporates the queue so that the information (SWM) contained in the message can be processed, said storing and retrieving operations being carried out sequentially, means for signalizing the arrival and the overflow status of a message to the second processor of the second node, and wherein at least said first and second nodes implement the message transmission process, the queue being constituted by a bank of registers (FIFO-MESS) having a first address (FIFO-IN) for executing message enqueuing operations and a second address (FIFO-OUT) for executing message dequeuing operations, the first and second addresses being respectively contained in a global space (GSP) visible for all nodes, and in a local space (LSP) only visible for thenode where the message is to be received, of an input-output address space (ISLO) related to the registers which contain the modules of said nodes.