FR2610157A1 - Installation for exchanging computer data, in particular for real-time on-board system - Google Patents

Abstract

In a real-time on-board system there is provided a server 10 linked by couplers such as 12 and 14 to digital buses such as 13 and 15. The controller 10 is linked to a mass memory 11 which is directly addressable, such as a backed-up random-access memory, or else an electrically erasable and programmable memory. The data exchanges between each of the items of equipment and the server are carried out via logic links such that each item of equipment feels itself to be linked with a virtual server specific to it.

Description

Installation for the exchange of computer data, in particular for real-time embedded system.

The invention relates to computer systems, more particularly on-board systems intended to operate in real time.

The on-board systems are made up of equipment linked together via digital or bus links, such as the Applicant's Digibus, or the 1553 or AMINCI buses known to those skilled in the art. These on-board systems are subject to two essential constraints: - the volume they occupy because, in an on-board system on an aircraft, space is very limited, and generally defined once and for all; - operating speed: in real time, the system must be able to respond within a very short time (typically a few tens of microseconds) to requests or wstimuli "from outside.

These two constraints have the effect of limiting the evolution of existing systems: on the one hand, the size of software is becoming larger; on the other hand, they have to process an increasingly large amount of data.

However, this need for evolution over time must be satisfied without modifying the physical architecture of the individual devices that make up the system.

To be able to increase the working capacity of all the equipment, it therefore becomes necessary to have a device, and in particular a memory capable of storing, restoring and modifying the information useful for the proper functioning of the system. This information can be parts of software, tables of constants, or records of anomalies, for example.

The systems which currently exist, based on a memory of the hard disk or magnetic tape type, are not satisfactory with regard to the constraints mentioned above. Indeed, their response times are of the order of ten milliseconds, therefore very far from the few microseconds or tens of microseconds required for the proper functioning of a real time system on board an aircraft.

On the other hand, these current systems are operating "single-user". That is to say, they are unable to inform several pieces of equipment simultaneously, on the scale of the required response times.

The present invention provides a solution to this problem.

An object of the invention is to provide a computer system which is very fast, so that it can be applied as an on-board real-time system.

Another object of the invention is to provide such a system which can jointly satisfy the requests of several users.

The invention also aims to provide a system capable of working with a directly addressable memory.

It also aims to allow the mass memory to be accessible in autonomous mode by the various devices, everything happening as if each device had a mass memory of its own.

Of course, this must be done while keeping the performance, whatever the number of requesting devices, within a limit fixed in advance.

The invention also aims to provide a solution to these problems while observing as wide compatibility as possible with currently existing standards, and in particular the so-called ISO standards as well as GAM-T-103.

The installation proposed for the exchange of computer data, in particular for an on-board real-time system, of the type comprising a server or controller member, provided with mass memory means, is suitable for communicating via at least one digital bus. with a multiplicity of user equipment.

Very generally, this installation is distinguished according to the invention by the following combination of means - the mass memory means consist of fully addressable memories, with rapid and direct access, - there is an initialization or negotiation phase allowing 'establish a logical link between the server and any of the equipment, and - data exchanges between this equipment and the server can then be carried out as part of a communication or transmission phase, using said logical link, until the rupture of this one, this phase of communication relating to records, which are each defined as a set of information accessible using a key.

According to another aspect of the invention, said phases are carried out using a predetermined set of logic messages, comprising, at least - a message requesting transmission by equipment, - an acknowledgment message of a request for transmission by the server, - a message from the server indicating that it is ready to transmit data, - a message for requesting resumption of transmission, and - a message for requesting interruption of transmission.

Advantageously, these messages have in common the fact that they include a definition part, then a code part identifying the addressing mode, the operation to be performed and the type of access required, a part indicating the resource number. allocated, this resource defining the logical link, and a part forming an access key to a record.

In a more particular embodiment, it is expected that the transmission request message also includes the logical address of the receiver, and the name of the file concerned.

It is advantageously the same for the acknowledgment messages and the message whose meaning is Ready to transmit.

For their part, these acknowledgment and ready-to-transmit messages also include a response part to the initial transmission request, in a particular embodiment of the invention.

On the other hand, it is advantageous that the messages for requests to resume transmission and requests to interrupt transmission also include a part of the transmission report.

According to another aspect of the invention, the data messages comprise a control part, followed by the data proper, which are advantageously limited to a predetermined number of words, for each data message.

Preferably, the control part of a data message contains at least the resource number and the registration number being transmitted.

According to yet another aspect of the invention, all the messages can take, in the predetermined position, a chosen code, rendering them inactive, which makes it possible to avoid taking erroneous information into account.

 In practice, the exchange of information on the bus or buses is carried out by "sectors" which form "buffers", each buffer consisting of a sector accompanied by its identification and control words.

Finally, and according to another important aspect of the invention, the mass memory is advantageously of the electrically erasable programmable memory type (EEPROM). It can also be of the saved RAM type.

Those skilled in the art will understand that the invention offers a new mode of exchanging computer data, a new mode which can be implemented with an installation as defined above.

Other characteristics and advantages of the invention will appear on examining the detailed description below, and the appended drawings, in which - Figure 1 is the block diagram of an installation according to the invention; - Figures 2A and 2B are two other diagrams showing the advantageous interconnection between the server and the equipment, according to the invention; - Figure 3 is a flow diagram of the various functions implemented by the device according to the invention; - Figures 4, 5, 6, 7 and their subsets show in more detail the structure and use of messages in the device according to the invention; - Figures 8 to 12 show different basic modes for data exchange in the installation according to the invention; - Figures 13 to 16 are timing diagrams illustrating different modes of application of the invention; and - Figures 17 to 22 are diagrams of automata useful for the implementation of the invention.

The accompanying drawings contain elements of a certain character in many respects. Consequently, they can not only serve to better understand the description below, but also contribute to the definition of the invention, if necessary.

In FIG. 1, the server of the installation according to the invention is designated by the reference 1.

This server 1 comprises as processing unit a controller 10, advantageously consisting of a microprocessor of the 68000 family sold by MOTOROLA. This controller 10 communicates with a mass memory 11, by an internal bus.

The mass memory ll can be a saved random access memory. It will be assumed hereinafter that it is an EEPROM type memory, that is to say electrically erasable and programmable. Memory structures of this kind have already been described, for a particular application, in Patent Application NO 85 00 671 filed by the Applicant on January 17, 1985 and published under ND 2,576,131.

By using a mass memory of this kind, it happens that all the words of the mass memory are directly addressable, and this with a constant access time. This is a significant improvement compared to memories of the tape or hard disk type, for which the access time to a word also depends on the time of movement of the read head on the tape or the disk.

The structure of the information recorded in the memory can be defined as follows: a "sector" is the smallest set of information serving as a basis for exchanges; a "buffer" is a unit of data information exchanged on the bus. It is therefore a sector accompanied by its identification and control words, a "block" is the smallest entity that can be written separately, a "page" is a set of words which constitutes a sub-multiple of size total mass memory, a "record" is a set of information accessible using a key, a file n is a set of records grouped logically.

As previously indicated, the controller 10 is a microprocessor of the 68000 family, for example the model 68020, from MOTOROLA. It is provided with a working memory or random access memory of 16 kilobytes, and with a program memory which is for example of the REPROM type erasable by ultraviolet, with a capacity of 64 kilobytes.

The computing power of the controller and the access time to the mass memory can then be made compatible with the performance of a bus whose speed is 1 Megabits per second.

In FIG. 1, the controller 10 is provided with a coupler 12 to a first digital bus 13 and a second coupler 14 to a second digital bus 15.

The bus 13 is for example a Digibus type link, associated with a coupler of the MAB (Autonomous Bus Module) type sold by the Applicant.

The second digital link of bus 15 can be made through a serial coupler 14, this link then being of the RS 232 type.

As a variant, any type of digital link can be used, such as the ARINC or 1553 buses, already mentioned through commercial couplers or developed by the Applicant.

Thus, the controller 10 is able to provide the dialogue functions with the digital bus (s) via the input / output couplers. It is also able to acquire the information by checking the validity of the blocks.

The controller 10 also provides writing / reading in the mass memory 11, via a specialized memory interface, adapted to the characteristics of the memory component used.

For the management of the mass memory, the controller 10 is provided with software suitable for performing the following functions - physical reading of the mass memory, per page or per block, - reading of a file, either whole or by packets of records, either by record, - physical writing in the mass memory, by page or by block, - writing in a file, whole, by packets of records or by record.

The files in mass memory can be one of the following three types: sequential file; indexed file, sequential-indexed file. In addition, the records making up these files are of variable size.

 Figure 3 summarizes the functions performed by the controller.

It manages the digital link 30 via the coupler 31 and the coupler interface 32. These functions can be carried out in a conventional manner.

Then come the functions 33 and 34 consisting in managing a so-called transmission protocol, and a so-called negotiation protocol, respectively, which are essential aspects of the invention.

Then, at 35, the controller 10 analyzes the commands originating from these protocols. In 36, it will consequently ensure the logical management of the files, from which it follows in 37 a physical management of the files.

Functions 35 to 37 are of a conventional nature, and can easily be perfected by a person skilled in the art, after having read the elements 33 and 34 described below.

Before describing in detail the structure of the various messages used by these functions, reference is made to FIGS. 2A and 2B.

FIG. 2A shows the physical connection of the server 1 by the bus 13 (for example) with equipment 2-1, 2-2, 2-3.

The object of the invention is to transform this link so that it becomes a logical link, as illustrated in FIG. 2B, such that for each of the devices 2-1, 2-2, 2-3, everything happens as if this one had a respective virtual server 1-1, 1-2, 1-3, which corresponded only with him.

FIG. 4 illustrates a transmission request message, denoted DEM. This message, designated by the reference numeral 40, comprises ten words of two bytes each.

The first word 41 of the message comprises 9 reserved bits, followed by 7 bits constituting the definition of the message.

It will be specified as of now that all the messages comprise 7 bits of definition, which, when all its bits are at 0, makes the message inactive. According to an advantageous characteristic of the invention, this makes it possible to avoid taking erroneous information into account.

A true DEM request message will therefore be characterized by the fact that its 7 definition bits have zeros everywhere, except at its least significant bit.

The second word 42 of the message 40 comprises an initial bit, the most significant, which indicates whether the addressing is physical or logical. In normal operating mode of the installation according to the invention, the addressing is logical. Physical addresses will only be used for maintenance, with special equipment. The first bit is 0 for physical addressing and 1 for logical addressing.

The other two bits indicate that the addressing level is performed by recording, by recording block or by file, depending on whether they are at 00, 01 or 10.

The rest of the first byte is reserved.

The second byte of word 42 indicates an eight-bit operation code. It can take the following values 00000001 for opening for reading, 00000010 for opening for modification, 00000011 for random reading, 00000100 for sequential reading, 000000101 for random writing, 00000110 for sequential writing, 00000111 for random erasing, 00001000 for sequential erasure, 00001001 for random positioning, 00001010 for sequential positioning, 00001011 for logical link closure, 00001111 for requesting transmission parameters.

The third word of message 40 is an assigned resource number, coded in binary on a byte, which authorizes the resource number, therefore of logical link going from 0 to 255.

The following word 44 can have two types of structures, which define the access key to a record in a different way.

If it is a recording of a sequential file, the structure is that of FIG. 44A. The first byte is the record number, and the second byte is the number of records in this batch file.

If, on the contrary, it is a direct access or sequential indexed file, the first byte is an access key to the current record, while the second byte is an access key to the last record. of the file.

The following word 45 defines the logical address of the receiver.

Its left byte is reserved. The first two bits of the byte on the right define a sub-bus coupler number, the code 00 being for example reserved for the main bus.

The next bit indicates whether it is the front bus (value 0) or the rear bus (value 1). The last five bits of the second byte indicate the bus address.

Then come four words which are used to code the name of the file concerned by the message 40. The name of the file is thus thus coded in ASCII on four words, which represents significant names being able to have eight characters. The last word of message 40 is reserved.

We will now describe, with reference to FIG. 5, the messages acknowledging a request for transmission by the server, noted in short DAC, or else the messages sent by the server to indicate that it is ready to transmit the data. , noted in brief PAT. These two messages have a common structure.

Such a message 50 begins with a first word defining the message, illustrated at 51. The general structure is the same as for message 40 in FIG. 4. The inactive state is represented by 0000000. The acknowledgment of a request is defined by 0000010, while the ready to transmit is defined by 0000011 (respective decimal values 2 and 3).

The left byte of operation code 52 is defined as for message 40. The same is true for the right byte.

The next word in message 50 is an assigned resource number, also defined as above for message 40.

This also applies to the access key to a record, which can take the two variants 54A and 54B.

The logical address of the receiver is defined by the word 55, again as for the message 40. Then comes the name of the file, expressed in four words, as before.

The essential new element of message 50 is the response to a request, which comes after the name of the file, and which can take different values such as request accepted 0000 request refused 0001 for unknown file name
0010 for a file which is not of the type
request
0011 when the equipment does not have access
to the requested file,
0100 when there are no more resources
available
Those skilled in the art will understand that such a message is capable of covering the various possible situations in the first exchange between a requesting device and the server, as will be seen below.

The message in Figure 5 has 13 words.

A message 60 which comprises ten words will now be described with reference to FIG. 6. This structure allows the definition of messages for requests to resume transmission, as well as requests to interrupt transmission, which are in principle at the initiative of a device.

The first word 61 is used to define the message as above, with the values 0000000 for inactive, 0000100 for request to resume transmission, and 0000101 for a request to interrupt transmission.

Word 62 relating to the operation code is the same as above. This also applies to the following word which designates the resource number.

 We then find the two variants 64A and 64B for the access key to a record.

Then comes the peculiarity of messages 60, which is to include a transmission report, coded on a word of two bytes. This transmission report will reflect, for example, error states regarding the self-test of the mass memory, or even the indication of a power failure of this mass memory.

The other words of message 60 are reserved.

FIG. 7 illustrates the case of a data message, the reference of which is 70.

First, it includes two control words. The first control word 71 has in its second byte a most significant bit which is representative of an end of transmission (EOT). This bit is worth 1 if the buffer which will be transmitted is the last of the transmission to be carried out.

The remaining 7 bits of the second byte of word 71 include the definition of the message. The message is inactive if all the bits are 0. It is a data message if the content of this second byte of word 71 is 0000110.

The first byte of word 71 concerns the number of the resource allocated to the transmission.

In the second control word 72, the first byte is the number or the access key to the record being transmitted. The second byte is the number of the last word making up the record being transmitted in the buffer. This byte is 0 until the last word of the record has been transmitted.

The other words of message 70 are data words, the maximum number of 126.

We will describe below how the above messages will be decoded and interpreted, on the one hand at the server level, on the other hand at the equipment level.

We will now describe with reference to Figures 8 to 12 different modes of communication by means of the installation according to the invention.

A logical link is established as soon as one of the devices sends a DEM message.

This message may include only a simple opening order, its value being 1 or 2 in the operation code of word 42 (FIG. 4). We then speak of an "explicit opening".

But the DEM message can also include in its operation code the request for an action, which can be a read, or a write, or even an erasure or a positioning.

This is called implicit opening. As we will see later, the server will send or receive the data at the end of the negotiation phase.

As previously indicated, the first phase of communication in the installation according to the invention is a "negotiation" phase. The purpose of this task is to establish the transmission parameters, and to allocate a resource to this transmission, so that this resource number defines the logical connection between the equipment and the server, and is the only number necessary to data transmission.

When a device wishes to receive or transmit data, it sends a DEM request message.

In the case of Figure 8, it is assumed that the server can respond to the request. It therefore responds with a DAC acknowledgment message having in its definition part, at word 51, the code 2 (decimal counter-value of the binary code already exposed). This DAC message also includes the indication of the resource number allocated to define the logical link between the equipment concerned and the server. Upon receipt of the DAC message, the equipment becomes inactive again.

The server then continues, looking for the required data, if any, and in any case creating the resource to which it has just given a number. When this is finished, the server sends a ready to transmit message, having the structure 50 (FIG. 5), but this time with the code 3 in the definition part of the message at word 51.

Of course, the resource number is indicated each time.

The logical connection is then made, and the data can be exchanged buffer by buffer, using data messages having the structure 70 of FIG. 7.

FIG. 9 illustrates the case where the server cannot respond to the request made by a DEM message. In this case, the DAC response does not contain an allocated resource number, but has the closing value (decimal value 11 in the operation code of word 52). Communication is immediately terminated on the server side and on receipt of such a DAC message, the equipment becomes inactive again. These are the two cases that can arise for the negotiation protocol, the most frequent case being naturally that where the logical connection could be established.

We will now examine the transmission or communication phase between the equipment and the server.

During this phase, the equipment and the server no longer use physical addressing, but only the resource number assigned during the negotiation phase.

In fact, for the equipment, everything now happens as if its only interlocutor was the resource which has been allocated to it for transmission, and with which it communicates via the communication protocol.

We have previously described with reference to FIG. 8 the case of an implicit opening, in which the initial request for the equipment included the indication of a desired action, for example a read or a write.

If the transmission phase does not follow a negotiation phase in which the equipment has made such a read or write request, the equipment must then send a new request in which it specifies the desired records. But the DEM message which it then sends now includes the number of the resource, the other indications on the required records.

FIG. 10 supposes that it was initially an explicit opening, and the transmission proper therefore begins with a new message DEM.

In response to this message, the server sends the record
N01 in the form of a data message which the equipment analyzes. As long as it does not receive DRT or DIT messages, the server continues to send records. This happens each time in the context of data messages as defined with reference to FIG. 7. For its part, the equipment performs checks which can be synchronous or asynchronous with the arrival of the records.

It is now assumed that the equipment has detected an error in the ith record. He then sends a message
DRT to request the resumption of transmission from this ith recording. This is done by the server, up to the last block.

 As we have already seen, this last block includes not only the data, but also the binary indication that it is an end of transmission (value 1 of the first bit of the second byte of word 71 of the message of data 70 in Figure 7).

The equipment will of course have continued to analyze the data.

If everything has gone well, the equipment closes the session, and sends a final DEM message to close the transmission. The value of the operation code of the message DEM is then 11, in its operation code, namely the second byte of word 42 of the message 40 (FIG. 4).

On the server side, the resource is then deallocated, and the file is closed.

This concerns a relatively simple transmission. It is of course possible that the equipment, instead of requesting the closing of the file, issues a new read or write request concerning the file already open. We then resume the entire mechanism described with reference to Figure re 8. Of course, the resource is then kept.

Advantageously, it is also possible to implement according to the invention a control phase. As soon as the equipment wishes to check the consistency of the transmission parameters, it issues a special request to this effect, in response to which the equipment returns the parameters in the form of a message.
DAC, which allows the equipment to control the consistency of information, as shown in Figure 11.

The structure of the request message is then the following decimal value 15 (00001111) in the operation code of word 42.

The server's DAC response has the format of the message 50 as before. The special nature of this DAC message, intended here for checking the consistency of the information, is expressed as follows decimal value 2 (0000010) in the message definition field.

It can also happen that the equipment wishes to interrupt the transmission, which will now be described with reference to FIG. 12.

The equipment then sends a DIT transmission interrupt request message having the structure 60 of FIG. 6. The interruption is represented by the existence of a code 5 in decimals in the definition part of the message, that is ie the right byte of the word 61. The server saves this interrupt request, but will not actually end the transmission except in response to a repetition of the DIT message from the equipment, and this in order to avoid undue interruptions.

The Applicant has conducted experiments with an installation according to the invention based on a digital bus with a speed of 1 Megabits per second, of the Digibus type.

Figures 13 to 16 are timing diagrams illustrating on two adjacent lines on the one hand the procedure and on the other hand the data transfers on the Digibus.

FIG. 13 corresponds to the case of an explicit opening, that is to say consisting of a first request without action, followed by the DAC then PAT messages, after which the equipment must send a second request so that finally make a data exchange, by a first and a second data buffer.

The DT1 delay between the first request and the DAC / PAT message is approximately 60 microseconds, including 20 microseconds due to the Digibus. The following delay DT12, counted in the same way, is at least 20 microseconds due to the Digibus, but this delay can be modified on the initiative of the equipment, which issues its second request when it sees fit. The DT13 delay is, for a read operation, 40 microseconds including 20 microseconds due to the Digibus. For a write operation, it is 20 microseconds due to the Digibus.

The DT14 delay is 20 microseconds due to the Digibus being read.

If it is a write operation in a saved RAM, the delay is also 20 microseconds due to the Digibus.

If, on the contrary, it is an operation of writing to an EEPROM memory, this delay is 2 milliseconds.

Overall, it elapses between the moment when the second request circulates on the bus and the appearance of the first data buffer - 280 microseconds which are entirely due to the write transmission time, - 300 microseconds of which 280 are due to the read transmission.

The case of an implicit opening will now be studied with reference to FIG. 14.

Counted as before, the DT21 time interval is 60 microseconds, including 20 microseconds due to the Digibus.

The DT22 delay is 20 microseconds due to the Digibus. The DT23 delay is 20 microseconds due to the Digibus in reading, 20 microseconds due to the Digibus in writing on a saved RAM, and 2 milliseconds in writing on an EEPROM memory.

In this example, there are 680 microseconds between the time when the request flows on the bus and the time when the first data buffer is available on this same bus.

With reference to FIG. 15, we will now examine the case of two simultaneous implicit openings emanating from two pieces of equipment, individualized by EQ1 and EQ2.

The DT31 and DT32 delays, equal, are both worth 60 microseconds including 20 microseconds due to the Digibus. The deadline
DT 33 is worth 20 microseconds due to Digibus, and the delay
DT34 is also worth 20 microseconds due to the Digibus.

Those skilled in the art will understand that the performances given in these examples depend essentially on the speed of the bus used. Indeed, the server is capable, as far as it is concerned, of responding simultaneously to the two devices with the same delays as for a user motorcycle transmission.

Finally, in the case of FIG. 16, a request for resumption of transmission is examined. The DT41 delay is again 60 microseconds, including 20 microseconds due to the Digibus.

The foregoing examples clearly show that the installation according to the invention completely solves the various components of the problem posed, namely: great speed in order to be able to operate as an on-board real-time system; ability to jointly meet the demand of multiple users; directly addressable memory; treatment of different equipment as if they were individually connected to the mass memory; independence of the number of equipment requesting.

 We will now describe how the messages described above will be decoded and interpreted, on the one hand at the server level, on the other hand at the equipment level.

With reference to FIGS. 17 to 22, the sequential blocks of the processing operations will be described in the form of an automaton. In the graphical representation, the nodes of the graph represent the states and the relationships correspond to the transitions - a state corresponds to the monitoring of an event and to an action, - a transition corresponds to a referral and to the reception or the emission of a message.

It is estimated that a person skilled in the art can define software for interpreting messages conforming to these automata.

Server-level message interpretation includes negotiation, interrupt handling, as well as reading and writing.

The negotiation task is still active. It is she who manages the negotiation protocol for the real-time embedded system server.

The states of the negotiation task (figure 17) are (170) Waiting for a request
This state corresponds to a passive phase of the task,
it is associated with an inactive type message.

(171) Verification of the conformity of the request
Upon receipt of the request, the task should verify
the consistency of the information in the request with that of
residing in the server directory.

1) It checks the existence of the requested file.

2) It checks that the access mode to the file is correct.

3) It checks that the equipment
- has access to the file
- can use the requested mode
- can use the requested operation.

4) She checks if the file is not already open in mo
dification.

5) It checks that the access key is suitable for the access mode
file and the record (s) exist.

6) It checks that the acknowledgment mode is authorized.

7) Finally, it checks that there are still resources available.

After these various checks, the task sends a DAC to the equipment where it indicates the response and specifies a resource number if communication can be established.

(172) Creation and allocation of a resource
The task dynamically creates from the defined number
In the previous state a resource which is affected
to communication.

The task then sends a PAT to the equipment to
signal that the logical connection is established, in
repeating the parameters of the request and specifying
the resource number.

The task of handling interrupt requests (Figure 18) is always active. It is she who manages the DIT.

The states of this task are (180) Waiting for an interrupt request
At any time, a DIT can occur. In that case,
the task waits for the next message.

(181) Waiting for a second interrupt request
If the message is again a DIT, the task asks
deactivation of the resource concerned, otherwise
transmission continues.

(182) Deactivation of the resource
The task issues a request to the task
mistress to tell her the resource to deactivate.

The read resource (Figure 19) is created dynamically by the task of managing the negotiation protocol. It is allocated to a communication.

Its states are (190) Segmentation of records
The resource reads and cuts the records into
depending on the size of the buffers that can circulate
on the bus. She adds the words at the head of the buffer
necessary for the reconstitution and con
monitoring of records.

(191) Deactivation of the resource
The resource issues a request for the task
mistress to let her know she can be turned off
vee.

The write resource (FIG. 20) is created dynamically by the task of managing the negotiation protocol. It is allocated to a communication.

Its states are (200) Reconstitution of records
The resource receives the different buffers and recons
titrate recordings using control words
located at the head of the buffer.

(201) Control of records
Once the recording has been restored
checks the validity of the transmission of the
registration. In the event of an error, the resource issues a
DRT towards the transmitting equipment by giving back
the transmission parameters as well as the key
access to the wrong record.

(202) Updating the registration
The resource updates the record using the
server file management system.

(203) Deactivation of the resource
The resource issues a request for the task
mistress to let her know she can be turned off.

The interpretation at the equipment level, in reading, is illustrated in FIG. 21.

Its states are (210) Inactive
The equipment has not initialized a transmission phase.

(211) Waiting for an acquittal
The equipment awaits the acknowledgment of its request by
mass memory.

(212) Waiting for a loan to be transmitted
The device expects a PAT from the mass memory.

(213) Waiting for a tampon
The equipment is waiting for a buffer.

(214) Reconstitution of the recording.

 The equipment reconstructs the buffers into records.

(215) Checking the validity of the registration
The equipment checks that the recording it
has just been reconstituted is valid (by means of a "sum
control "for example).

(216) Copy of the recording
The device writes the recording to its memory.

(217) End of transmission.

In writing, interpretation at the level of the equipment is carried out as illustrated in FIG. 22.

Its states are (220) Inactive
The equipment has not initialized a trans phase
mission.

(221) Waiting for an acquittal
The equipment awaits the acknowledgment of its request by
mass memory.

(222) Waiting for a loan to transmit
The device expects a PAT from the mass memory.

(223) Playing a recording
The device reads a record from its memory.

(224) Segmentation of registration
The equipment segments the recording into buffers.

(225) Issuing a buffer
The equipment issues a buffer.

(226) End of recording test:
The equipment tests whether the current recording is
finished.

(227) End of transmission.

Of course, the present invention is not limited to the embodiment described, but extends to any variant in accordance with its spirit, and covered by the claims below.

Claims (12)

Claims. 1. Installation for the exchange of computer data, in particular for real-time embedded systems, of the type comprising a server or controller unit (1), provided with mass memory means (11), and suitable for communicating by the intermediary of at least one digital bus (13, 15) with a multiplicity of user equipment (2), characterized in combination - in that the mass memory means (11) consist of fully addressable memories with fast access and direct, - in that there is an initialization or negotiation phase enabling a logical connection to be established between the server (1) and any one of the devices (2), and - in that the data exchanges between this equipment and the server can then be carried out as part of a communication or transmission phase, using said logical link, until the latter breaks; this communication phase relating to records, each of which is a set of information accessible using a key. 2. Installation according to claim 1, characterized in that said phases are carried out using a predetermined set of logic messages, comprising, at least - a message requesting transmission by equipment (DEM); - a message acknowledging a request for transmission by the server (DAC); - a message from the server indicating that it is ready to transmit data (PAT); - a transmission restart request message (DRT): and - a transmission interruption request message (DIT). 3. Installation according to claim 2, characterized in that said messages include a definition part, then a code part identifying the addressing mode, the operation to be performed and the type of access required, a part indicating the assigned resource number, defining the logical link, and a part forming the access key to a record. 4. Installation according to one of claims 2 and 3, characterized in that the transmission request message (DEM) also includes the logical address of the receiver, and the name of the file concerned, as well as the acknowledgment messages (DAC) and ready to transmit (PAT). 5. Installation according to claim 4, characterized in that the acknowledgment messages (DÂC) and ready to transmit (PAT) further comprise a response part to the initial transmission request (DEM). 6. Installation according to one of claims 2 to 5, characterized in that the transmission resumption request (DRT) and transmission interruption request (DIT) messages further comprise a part of transmission count. 7. Installation according to one of claims 2 to 6, characterized in that the data messages include a control part, followed by the data itself. 8. Installation according to claim 7, characterized in that the data proper is limited to a predetermined number of words, for each data message. 9. Installation according to one of claims 7 and 8, characterized in that the control part contains at least the resource number and the registration number during transmission.  10. Installation according to one of claims 2 to 9, characterized in that all the messages can take, in the predetermined position, a chosen code, making them inactive, which makes it possible to avoid taking erroneous information into account. 11. Installation according to one of the preceding claims, characterized in that the exchange of information on the bus or buses is carried out by "sectors", which form "buffers" accompanied by their identification and control words. 12. Installation according to one of the preceding claims, characterized in that the mass memory (11) is of the programmable and electrically erasable memory type, or else of the saved random access memory type.