DE102004006566A1 - A method of reading / writing information regarding a destination network subscriber station in a network of distributed stations, and request network subscriber station for performing the method - Google Patents

Abstract

In an IEEE 1394 home network, after a bus reset operation, the network stations (10, 11, 12) read at least portions of the configuration ROM of all or at least some network stations of the local network using asynchronous read transactions since the configuration ROM contains important identity information and the capabilities of the respective network station. If such read transactions are carried out strictly serially with asynchronous read transactions, the waiting times for each read transaction continue to accumulate, so that the phase of acquiring the station-specific information in the network can take a very long time. DOLLAR A In order to increase the efficiency in this phase, the invention proposes to perform the read transactions quasi-parallel, that is, so that a further read request to another network station (12, 10) can already be sent if the response to a previously sent Reading transaction has not yet received. Thus, read requests can be interleaved with the responses, so that the individual waiting times can overlap with each other and do not add up completely. The phase of the detection of the station-specific information can thereby be substantially shortened. For the quasi-parallel request method, the concept of so-called "split transactions" contained in the IEEE 1394 standard is exploited. DOLLAR A The invention can also be used when writing data ...

Description

The The invention relates to the technical field of data communication in a network of distributed stations, in particular a home network.

Background of the invention

By a home network, the various devices are interconnected. Such devices can be out come in the field of consumer electronics, such. As a TV, VCR, DVD player, Satellite receiver, CD player, MD player, amplifier, Radio, camcorder, etc. In this context, there is also a personal computer mentioned, today also as a device the consumer electronics can be viewed. meanwhile But home networks are also known in which other devices are integrated. As an example, devices the white one Goods mentioned, such as washing machine, electric stove, microwave oven, refrigerator, Dishwasher, etc.

For networking of devices from the consumer electronics sector were from the industry developed appropriate communication systems. Thought is there primarily to the wired networking of devices with the help the so-called IEEE 1394 bus system with which it is possible to Data with a very high data rate between the individual network subscriber stations, hereafter referred to as network station, to exchange. By doing original IEEE 1394-1995 standard are the data transfer speeds S100, S200 and S400 specified. These correspond to data transfer rates of 100 Mbps, 200 Mbps and 400 Mbps. Meanwhile, the original standard extended and in the current version IEEE-1394b are even higher data transfer speeds S800, S1200 and S1600 specified.

Such High data rates are needed in particular for transmissions between consumer electronics devices. This This is because the typical application of the data exchange between consumer electronic devices therein is that a title is played on a video or audio source, either video or music, and the associated one Data stream to another consumer electronics device or more other consumer electronic devices is transmitted. For this application is between the devices in question, the data together exchange, set up a logical data connection. About these Data connection are then regularly transmitted data packets. This form of data transfer is in the IEEE 1394-1995 standard as isochronous data transmission designated at the regular, in certain Intervals, Transfer data packets from the data source to the data sink or data sinks become.

Furthermore also finds asynchronous data transfer in the IEEE 1394 network instead of. Here, data packets are sent as needed. As many such asynchronous data packets are sent over the bus, depends on the accumulated data volume. Asynchronous data transmission becomes predominantly for the detection and control of a device in the network by a used by other devices in the network.

Of the IEEE 1394-1995 standard There are only a few limitations regarding the topology of the IEEE 1394 network. The approved bus topology corresponds to a tree structure. Depending on In the case of use, however, the tree structure can be different. The network can do this be designed very variable.

One Another outstanding feature of the IEEE 1394 bus system is that the connection and disconnection of devices during operation of the network is possible. This feature is Also known in the jargon as the "Life Insertion Feature." Every time after connecting or disconnecting a device becomes a bus reset triggered. Following each bus reset There is a self-configuration phase of the network. In this sends each IEEE 1394 network station Self-ID information (called "Self-ID Packet" by default) to the other network stations. This ensures that each network station informs about it is which other network stations are connected in the network are. The self-ID information is used by each network station facing yourself identify the other network stations of the network. With Help received from the other network stations Self-ID information, every network station is capable of one create so-called node list and in one of the network station to store associated memory device.

These Stored information is then stored in a driver program The network station uses even more detailed information about each request further network station. This more detailed information can the driver program z. B. in a so-called "Node Information Table" compile. This is prescribed in the IEEE 1394 standard, each network station in a specially reserved memory area called "Configuration ROM "station-specific information about the characteristics and abilities the network station must hold.

invention

The Invention busy mainly with the problem of station-specific information can be detected efficiently, without an application program running on the polling network station blocked for too long. For certain Operations must be done the application program on the assembled station-specific Access information in the "Node-Information-Table". If However, if this table is not yet fully available, longer delay times may result for the application program. what for the user as annoying can prove if so long reaction times arise in the program. Although a read transaction according to IEEE 1394-1995 standard as a block read access carried out be so that the desired Posts but can not be detected in the configuration ROM with a read request all available today IEEE 1394 network stations support such block read accesses. In particular, such devices, the IEEE 1394 circuits newer types (so-called OHCIICs), often support only word read accesses. With such network stations, the Information through several consecutive word read accesses be queried, so that multiple responses each queried Network station would have to wait. As already described before Network topology corresponds to a tree structure can become for different transmission paths as well different transmission times result. In addition, in the IEEE 1394 bus system, a variable Limit for the maximum allowed response time after a request from to allowed to 8 seconds. Even if the application program designed to do other tasks in the meantime, can the delay possibly make it very noticeable to the user. This is especially the case when the read requests strictly serial, d. H. that first the response of a previously sent Read request is awaited before the next read request to the next Network station is sent. This strictly serial method of Although the granting of read requests offers the advantage of being programmatically can be realized very easily, but has the disadvantage that in this case, the different response times are each sum up and thus the collection of station-specific information network stations can take an extremely long time. It is also too Bear in mind that the phase of querying the station-specific information in principle, must be traversed by each network station, d. h. that each network station each station-specific information of all other network stations must query.

to Increase the efficiency of querying the station-specific Information is proposed according to the invention from the strictly serial form of the query of the station-specific To move away from information and instead apply a quasi-parallel query method. In this case, a polling network station sends the read request to another network station, without first the reply to wait for the previous read request. In this query method Not all response times add up on and the response time of the application program can be drastically reduced be reduced. The invention assumes in the purchase that for the administration the quasi parallel sent read requests more effort arises.

The Read request and its response are performed on the IEEE 1394 bus with asynchronous transactions. For the quasi The parallel method of issuing read requests is the IEEE 1394-1995 standard concept of so-called "split transactions". After this concept can by an application several requests are sent one after the other and the associated ones responses will be reassigned to the corresponding request as soon as they arrive at the requesting station. The assignment takes place over here the transaction code, the so-called "Transaction Label", which complies with the IEEE 1394-1995 standard from the requesting network station in the read request for the destination network station be clearly set and in the associated response of the destination network station must be identical.

The The same concept can according to the invention for the transmission of writing requests the network stations are used. It can happen that after a bus reset a network station with special bus management expertise, such as z. B. "Bus Manager "or" Isochronous Resource Manager, "determined Posts must update at all network stations. These are write requests the respective stations sent. This can also according to the invention according to the Quasi-parallel method done, leaving another write request to another Network station is then sent before the previously mentioned Network station feedback has delivered. In this case, writing data, in particular for updating setting parameters after a bus reset, accelerates, so that an application program with a corresponding shorter Reaction time manages.

The inventive measure is in addition to the improvement of the reaction time also better exploiting the available bandwidth of the data bus, as the application program can request further data transfers pending the arrival of outstanding responses if the transactions are managed using the quasi-parallel method.

At the Reading out the configuration ROMs of the network stations after the strictly serial poll method, the operation is determined by the sum of the Response times of all network stations determined. In addition, will possibly with long response times Give away bandwidth on the bus that uses the application program for itself could. When reading the station-specific information according to the quasi-parallel method will be the cheapest Case the readout time only by the sum of the response times of the am determined the slowest reacting node. Even in the most unfavorable In this case, the process takes exactly as long as after the strictly serial readout method. The possible Time advantage justifies the higher one Administrative overhead within the application program, all manage both active and asynchronous transactions simultaneously respective replies must allocate while in the strictly serial read only one asynchronous transaction is active at a time.

The inventive measures are in the independent claims 1 and 10 called.

By those in the dependent Claims listed measures Further improvements are possible.

The Collection of station-specific information can be done with individual Word read accesses piecewise respectively. Instead of word read accesses, block read accesses can also be used used to capture the information in sections can be. With a block read access, a variable number of data words to be read.

The same thing Concept leaves also use in the write accesses.

In the case, that the station-specific information of a network station by multiple word reading requests piece by piece or by several Block read access queried sections, it may be advantageous when between the word read requests / block read requests for a- and the same network station first the answer the former Word read request / block read request awaited to this network station becomes. It can happen already from the first reply show that the data word to be read is faulty. In one In such case, the remaining word read requests / block read requests not to be sent at all. In addition, this measure is the necessary Administrative burden for the quasi-parallel query method limited. This is a good compromise between desired short reaction time on the one hand and on the other hand not ausuferndem Administrative burden for the programming possible.

Vice versa can it for a further optimization of the reaction time of the application program advantageous be if the station-specific information of a network station by multiple word read requests piecewise or by multiple block read requests be queried in sections without the respective reply to the previously sent word read request / block read request to wait for this station. It depends on the specific network if that measure An additional Time advantage brings. When the network is fully developed, are on an IEEE 1394 bus 63 network stations connected. In this Case, it may make sense to dispense with the said measure, because Alone due to the large number of network stations already very many Transactions for determining the station specific information about the Bus connection run and the number of transactions through the multiple word read requests / block read requests per requested network station drastically increased again becomes. The described measure but can be advantageous if a lower stage of the Network exists and / or the station-specific information with a few words or words Block read accesses can be determined.

Of the Administrative burden for This measure increases accordingly, as more and more active asynchronous Transactions must be managed and the respective responses must be assigned.

drawings

The The invention will be explained in more detail with reference to drawings. These show in:

1 the structure of an exemplary IEEE 1394 network;

2 the general format of the "Configuration ROM";

3 the structure of the "Bus Info Block" within the "Configuration ROM";

4 the so-called protocol architecture of an IEEE 1394 node according to IEEE 1394-1995 standard;

5 the structure of a transmission cycle of the IEEE 1394 bus system;

6 an example of the flow of an asynchronous transaction according to IEEE 1394-1995 standard;

7 a schematic representation of the transmission method with so-called "split transactions" according to IEEE 1394-1995 standard;

8th the query of station-specific information after a bus reset according to the strictly serial polling method;

9 the query of the station-specific information according to the quasi-parallel polling method according to a first embodiment of the invention; and

10 the query of the station-specific information according to the quasi-parallel query method according to a second embodiment of the invention.

embodiments the invention

1 shows an exemplary IEEE 1394 network with three network stations. These are interconnected in a tree structure. One node is connected to another via a separate IEEE 1394 network cable. The number of cable ports ("ports") can vary from node to node In the example shown, the station is 12 equipped with a so-called 3-port physical layer module, the network station 10 with a 2-port physical layer device and the station 11 with a 7-port physical layer device.

To a bus reset Businitialization takes place in two phases. These are "Tree Identification" and "Self Identification". During the Phase "Tree Identification "will the current bus structure by analyzing the assignment of the ports determined the network stations. At this stage it is also determined which network station takes over the root function. During the phase "Self Identification" identifies itself each network station itself and at the same time also provides the information about which transmission speed she supports. This happens by sending self-ID information packets in one order determined by the bus structure.

In 1 is indicated in which order the network nodes send their self-ID information packets during the bus initialization phase. The station with reference number 11 gets assigned the bus due to the tree structure shown first. It is the node with the lowest node number (node ID, # 0). Due to the bus topology shown next, the bus would be the network station 12 be assigned and last of the network station 10 , After all network stations have sent out their self-ID information packets, the node list can be established in all network stations by analyzing all self-ID information packets. Both phases of bus initialization are described in detail in the IEEE 1394-1995 standard.

To In the bus initialization, the structure of the "Node Information Table" takes place by providing more detailed information about the individual network stations are queried. This phase will be below described in more detail. In this phase, all network stations send Read requests to the "Configuration ROM "of each Network station over the Bus. This phase is complete when each network station receives the station-specific information of all other network stations has queried and the "Node Information Table" with the station-specific Information was created in each network station.

2 shows the general format of the "Configuration ROM" according to the IEEE 1394-1995 standard As shown, the "Configuration ROM" is divided into data words of 32 bits corresponding to 4 bytes Each 32-bit wide data word, also called Quadlet, can It is also specified in the IEEE 1394-1995 standard under which address the "Configuration ROM" must be set up.This is usually done so that in the freely assignable RAM memory of the network station the corresponding memory area for the " The first quadlet of the "Configuration ROM" contains the information about the length of the "bus_info_blocks", the CRC length as well as the actual CRC check code for the area of the "Configuration" covered by the CRC check code ROMs. "This is followed by the so-called" bus_info_block ". Further entries in the "Configuration ROM" concern the "root_directory", "unit_directory", "root & unit leaves" and the information about "vendor_dependent_information". All intended areas in the "Configuration ROM" are defined in the IEEE 1394-1995 standard. The station-specific information required for the construction of the "Node Information Table" varies according to the desired content of the "Node Information Table". For example, to capture all the information from the "bus_info_block" in the "Node-Information-Table", the first 5 quadlets are read in the "Configuration ROM".

3 shows the format of the "bus_info_blocks" In the first quadlet is only the Printout "1394" stored in ASCII format The second quadlet of the "bus_info_blocks" contains entries about the capabilities of the network station. The entry irmc indicates whether the network station has the ability to operate as an "isochronous resource manager." The cmc entry specifies whether the network station has the ability to operate as a "cycle master." The isc entry indicates whether the network station supports isochronous data transfer. The information bmc indicates whether the network station has the ability to work as a "bus manager." Important are the entries "node_vendor_id" and "chip_id_hi" in the third quadlet and "chip_id_lo" in the fourth quadlet. Namely, this information defines a unique global identification number for the network station. This identification number is given once and is also referred to as GUID, where GUID stands for "Global Unique Identifier." Since network transactions are partially GUID-addressed in some home networks, this information is of great importance, for example the HAVi network HAVi stands for "Home Audio Video Interoperability".

Hereinafter, the protocol architecture of an IEEE 1394 interface will be described with reference to FIG 4 explained. The lowest communication layer called physical layer 20 is always realized in hardware. For this purpose, integrated circuits have been available for some time. The next higher communication layer called Data Link Layer 21 is usually also implemented by hardware. Separate link ICs are also available in the market. The other layers shown, namely "Transaction Layer" 22 , "Serial Bus Management" 23 and "Application Layer" 24 are usually implemented by software, which is then executed on a powerful microcontroller in the network station.

Within the shift for "Serial Bus Management" 23 are the components "node controller" 27 , "Isochronous Resource Manager" 26 and "Bus Manager" 25 highlighted. In an IEEE 1394 network, a maximum of one "Bus Manager" 25 and at most one "Isochronous Resource Manager" 26 active at a time, even if several network nodes offer the respective functionality. Which network node performs the respective function is determined after each bus reset operation according to a predetermined method in the IEEE 1394-1995 standard. If the "root" node can perform the respective function, it is quite probable by the said methods that the respective function of the "root" is activated. It can also happen that no "Bus Manager" 25 is present, so that then the "Isochronous Resource Manager" 26 (if available) some tasks of the "Bus Manager" 25 must take over.

For the invention described here, the components "Transaction Layer" 22 and "Application Layer" 24 essential, which is why these components are discussed in more detail below.

The component "Transaction Layer" 22 consists of parts of the network layer and the transport layer according to the OSI / ISO reference model of data communication. This component is described in detail in the IEEE 1394-1995 standard. It is therefore expressly referred to this standard for the purpose of disclosing the invention.

The component "Transaction Layer" 22 offers it as a service to write data to a specific specified address of another network station, read data from a specified network station at a specified address and send data to another network station so as to perform a function with return of the result. Essential for the invention is firstly the service with which a read transaction can be carried out and secondly the service with which a write transaction can be carried out.

In the embodiment according to the invention, the application layer 24 a read request of type READ at the "Transaction Layer" 22 Request. The "Transaction Layer" 22 converts this request into a data packet request, whereby, in addition to the node number of the network station to be interrogated, the address under which the data is to be found at the network station to be interrogated is also specified. If appropriate, the number of data to be read from the specified memory address may also be included in the request. The requested data is processed via the component "Transaction Layer" 22 to the requesting application layer 24 forwarded. If an error code has been received as an answer due to the requested data or data transmission interruptions, the "Transaction Layer" component reports 22 this error code also to the application layer 24 ,

For the exact procedure of READ and WRITE Transactions, refer to the detailed description of the "Transaction Layer". 22 in the IEEE 1394-1995 standard under chapter 7. Transaction Layer Specification.

According to the IEEE 1394 standard, the data traffic over the bus takes place in transfer cycles. Such a data transfer cycle is in the 5 shown. In this variant of the data transfer via the bus, a cycle generator, called "cycle master", is determined in the network. This is a network station that contains a relatively accurate clock generator. This cycle generator sets the data transfer cycle. For this purpose, the cycle generator periodically transmits a so-called "cycle start packet" over the bus connection to all other network stations in an interval of 125 μs. All other network stations then have to set their cycle time register ("cycle timer register") with the received value in the "cycle start The standard specifies that the network stations must wait at least for the time period of a so-called "subaction gap" in the case of asynchronous data transmission, before they can apply for the bus again. After receiving the "Cycle Start Packet," in isochronous data transmission, the network stations only need to wait for the shorter time period of an "isochronous gap" before they can issue their own isochronous transmission requests. The data transfer in the 125 μs cycle is as in 5 illustrated, practically divided into two parts. In the first phase after receiving the "Cycle Start Packet", the isochronous data packets are transmitted and wait for the short waiting time of the "isochronous (short) gap" between these packets. After the isochronous data traffic has been transmitted, there follows a longer waiting time of the so-called "subaction (long) gap." Thereafter, asynchronous data traffic takes place via the bus "waited.

6 illustrates the principle of asynchronous data transfer by means of so-called "subactions." According to the IEEE 1394 standard, the term "subaction" is defined as a process of delivering a single data packet over the bus 6 A transaction over the bus is divided into two "Subactions." The first part refers to the request itself (eg, READ or WRITE Request) .The receipt of the request is acknowledged with a confirmation message, which then ends the end of the request first "Subactions" marked. This is followed by the specified waiting time of a "Subaction Gap", after which the requested network station returns its response, and a confirmation message is returned directly afterwards, which is referred to as "split transaction" in the standard and under point 3 , 3.6.2.2 Split Transactions. The peculiarity of this method is that a transaction in two parts in the request and the response does not necessarily have to be completed before a new transaction can take place. Thus, multiple transactions may be requested before the return must be returned relative to the first transaction. An example of such a "split transaction" for a READ transaction is in the 7 shown. The transaction is issued with a READ request as a transaction request "TR_DATA.request" from the application layer 24 started. The "Transaction Layer" 22 then returns a data packet request "LK_Data.regest" to the backup layer 21 , The backup layer 21 transmits a corresponding data packet to the requested network station via the bus. The backup layer 21 in the destination network station delivers a data packet input display "LK_data.indication" to the "Transaction Layer" 22 in the destination network station. The "Transaction Layer" 22 then gives a corresponding read request as a transaction notification "TR_DATA.indication" to the application layer 24 the destination network station on. Then the "Transaction Layer" 22 the destination network station information "LK_DATA.response" to the backup layer 21 the way that the response is in progress. The backup layer 21 the destination network station then returns an acknowledgment to the requesting network station. If this confirmation message in the backup layer 21 the requesting network station arrives, it also gives a notification "LK_DATA.confirmation" to the "Transaction Layer" 22 , Thus, the request subaction is done and it can be sent from this point further requests on the bus without having to wait for the actual response to the first request. This is in the 7 each represented by interruption of the time axis. Then, when the data is ready for the response, the application layer takes over 24 the destination network station has a READ response as a transaction response "TR_DATA.response" to the "Transaction Layer" 22 delivered. The "Transaction Layer" 22 the destination network station issues a data packet request to the data protection layer 21 the destination network station again via LK_DATA.request. The corresponding data packet is taken from the data protection layer 21 sent over the bus to the requesting network station. After receipt of this data packet in the data protection layer 21 the requesting network station delivers this data packet to the "Transaction Layer" 22 via LK_DATA.indication. The data is sent to the application layer as part of the transaction acknowledgment "TR_DATA.confirmation" 24 forwarded. The "Transaction Layer" 22 additionally gives the notification to the backup layer 21 about the successful receipt of the requested data via LK_DATA.response. The backup layer 21 sends the confirmation message to the backup layer 21 the destination network station. This in turn notifies the "Transaction Layer" 22 in the destination network station via LK_DATA.confirmation. There with the entire reading transaction is completed.

Of the Administrative burden is for the transmission principle with "Split Transactions "increases as the requirements and answers of several transactions can be nested within each other. To this purpose is in the IEEE 1394-1995 standard provided that one waiting period is monitored for each split transaction within which is the transaction response at the requesting network station must be received. The duration of all these waiting times can be in Split timeout registers can be set in the request network station. The default entry in this register is 100 ms as a split timeout value. If this time exceeded was, the corresponding applies Transaction ended unsuccessfully.

The Driver program in the network station is according to IEEE 1394 standard as an application and used asynchronous transactions for querying the station-specific information additional network stations. While thus the previously described handling of the transactions from the transaction request of the driver program until the transaction confirmation to the driver program in the IEEE 1394 standard is required, it is up to the driver program, such as sending the requests and receiving the returns in conjunction with the rest of the IEEE 1394 protocol software the network station is organized internally. In the simplest case sends the driver program only one request per time and waits after receiving the reply. In this case, only one set of management information, e.g. B. the node number of the destination network station, the destination address within the destination network station, the transmission speed, possible error conditions etc., for the one requirement and the assignment of requirement and response is trivial. To speed up the query of station-specific Information is it z. Possible, Send one request per time per destination network station, without to the receipt of the answers to wait from other destination network stations. These are, however so many sentences Administration information needed, how target network stations are on the IEEE 1394 bus. moreover have to the management information Means for assigning the requirements to the respective answers included, for. B. a transaction identifier o. Ä. If even more requirements per Target network station used per time, this increases first once again the number of needed Sets management information corresponding. Furthermore have to all administrative information for a respective destination network station are compared with each other, including additional, parent Administrative information, e.g. B. a marking already performed requirements, possible parent error conditions etc., needed per destination network station.

8th shows the timing of the query of the station-specific information in the pattern network according to 1 according to the conventional, strictly serial polling method. In the case shown, the network station asks 11 the entries of the first five quadlets of the "configuration ROM" in succession at the two other network stations 10 and 12 from. This station-specific information polling phase takes place after a bus reset phase that started at time BRS and ended at time BRE. In the upper part of the 8th in each case the time is marked at which the polling network station 11 directs a word read request to one of the other two network stations. Each read request is designated by a pair of digits. The first digit indicates the node number of the network station to be queried, and the second digit separated by a comma indicates which quadlet is to be read in the configuration ROM. In the lower part of the 8th the time at which the response to the read request from the destination network station is returned is marked. Here, too, is indicated by the pair of digits, which specific read request is affected. In addition, it is still highlighted by an arcuate connection, which read request belongs to which response. As shown, the next next quadlet read request in the destination network station's Configuration ROM is not generated until the previous read request's response has already arrived, as shown, resulting in the total time taken to poll station specific information from both network stations The individual read transactions may take a different amount of time, and if necessary, a read transaction may also exceed the set timeout value, so that a repeated request is necessary 8th not shown. The total time can therefore be much longer than in 8th shown. The station-specific information retrieval phase is ended when the response to the last read request, identified by the two digits 2, 4, arrives.

9 shows a first embodiment of performing the query of the station-specific information using the quasi-parallel query method. As shown, the read request is made on the first quadlet of the "Configuration ROM" of the network station 12 sent first. This is followed immediately afterwards by the read request for the first quadlet in the "Configuration-Rom" of the network station 10 without first the reply on the read request, directed to network station 12 to wait. After that, however, no further read requests are initially sent. Only after the response for the read request on quadlet 0 of the network station 10 has arrived, the next read request is on Quadlet 1 of the network station 10 sent. At this time, the response to the read request is at quadlet 0 of the network station 12 not yet received. After this response has arrived, then the further read request on Quadlet 1 of the network station 12 sent. Thus, the individual read requests are nested, as in 9 shown. The result of this is that in total the phase of the query of the station-specific information is dealt with much faster than in the strictly serial polling method according to FIG 8th , Although the individual read requests require exactly the same time as previously described in 8th However, as the read requests have been nested within each other and thus the individual requests are sent significantly earlier, the responses will arrive earlier and the entire phase of the station-specific information query is completed earlier. As shown, in the case shown, the query phase is shortened by about half compared to the standard query method.

The 10 shows still a second embodiment of the invention. Here, the individual reading requirements are interlaced even more closely. Namely, it does not wait until the previous read request to the same destination network station has been answered before the next read request is sent to this destination network station. In the case shown results in this method, a further shortening of the phase of the query of station-specific information. On the other hand, this advantage is offset by a further increase in the administrative burden, since significantly more transactions may be pending in a given time interval. It is therefore only advisable to use this solution when relatively few network stations are present in the network and / or the query station-specific information as many network stations with a block read access or at least with less block read accesses can take place as word read accesses.

As explained in the introduction, instead of read requests, write requests can also be sent using the quasi-parallel method. The write transactions can also be interleaved with each other as a result of the transfer mode with "split transactions." In the IEEE 1394-1995 standard, this is one of the 7 corresponding figure included. This will be on 3 - 12 of the standard under point "3.6.2.2. Refer to "Split Transactions".

As an example of a standard-compliant application in the case of write requests, mention is made of updating the "Broadcast_Channel_Register" of the network stations after a bus reset, which is done by the "Isochronous Resource Manager". 26 performed after a bus reset. However, these registers exist only on those network nodes that are designed according to the extended IEEE 1394a standard. This register is optionally available in the network stations except for all Isochronous Resource Manager Capable network stations where it must be provided The Isochronous Resource Manager 26 accesses this register if it wants to notify the other network stations that the IEEE 1394a default channel number 31 has been reserved for asynchronous broadcast stream packets.

Claims (18)

Method for reading / writing information relating to a destination network subscriber station ( 10 . 12 ) in a network of distributed stations, wherein a request network subscriber station ( 11 ) each have a read request / write request to the destination network subscriber station ( 10 . 12 ), and the destination network subscriber station ( 10 . 12 ) returns a response within a certain or indefinite time, characterized in that the request network subscriber station ( 11 ) another read / write request directed to another network subscriber station ( 12 . 10 ) without waiting for the reply to the previously sent read / write request. The method of claim 1, wherein the queried ones Information station-specific information, in particular configuration data affect. The method of claim 2, wherein the station-specific information of a network subscriber station is interrogated by a plurality of word-reading requests, piecewise or by a plurality of block-reading requests, and between the word-reading requests or block-reading requests for one and the same destination network subscriber station ( 10 . 12 ) first the response of the previous word read request or block read request is awaited. Method according to claim 2, wherein the station-specific information of a destination network subscriber station ( 10 . 12 ) are queried piece by piece or by a plurality of read block requests by a plurality of word read requests are sent to the same destination network subscriber station without the reply to the previously sent word read request or block read request ( 10 . 12 ) to wait. Method according to one of the preceding claims, characterized characterized in that the network of distributed stations is an IEEE 1394 network is. Method according to claim 5, characterized in that that as station-specific information one or more entries in one Requested configuration ROM of an IEEE 1394 network subscriber station become. Method according to Claim 6, characterized that with a word reading request one quadlet of data, this means four bytes each, in the configuration ROM of an IEEE 1394 network subscriber station be queried. Method according to one of the preceding claims, characterized characterized in that the write request is in multiple word-writing requests for the piecemeal Write or multiple block write requests for piecewise writing is decomposed. Method according to one of the preceding claims, characterized in that the updating of a Broadcast_Channel_Register to the destination network subscriber station ( 10 . 12 ) is sent. Request network subscriber station for carrying out the method according to one of the preceding claims, having means for generating and sending a read / write request to a destination network subscriber station ( 10 . 12 ), with means for receiving and processing a response of the destination network subscriber station ( 10 . 12 ) in response to the read / write request, characterized in that the create / send means are arranged to issue another read / write request directed to another destination network subscriber station ( 12 . 10 ), without waiting for the reply to the previously sent read / write request. Request network subscriber station according to claim 10, characterized in that the requested information station-specific Information, in particular configuration data concern. A request network subscriber station as claimed in claim 11, characterized in that the station specific information is requested by a plurality of word read requests piecewise or by a plurality of read block requests and the create / send means is adapted to first return the previous word read request or block read request to the same destination network subscriber station ( 10 . 12 ) before sending the next word read request or block read request to the same destination network subscriber station ( 10 . 12 ). A request network subscriber station according to claim 11, characterized in that the station-specific information is queried by a plurality of word-reading requests, piecewise or by a plurality of block-reading requests, and the generating / sending means are adapted to pass the next word-reading request or block-reading request to the same target network subscriber station (12). 10 . 12 ) without the reply to the previously sent word read request or block read request to the same destination network subscriber station ( 10 . 12 ) to wait. Request network subscriber station after a the claims 10-13, having an IEEE 1394 interface. Request network subscriber station according to claim 14, characterized in that as station-specific information one or more entries in a configuration ROM of an IEEE 1394 network subscriber station. A request network subscriber station according to claim 15, characterized in that said one-read request generation / dispatching means each carry one quadlet of data, that is, four bytes, in the configuration ROM of a destination network subscriber station (15). 10 . 12 ) Interrogate. Request network subscriber station after a the claims 10 to 16, characterized in that the creation / dispatch means so are designed to handle the write request in multiple word-writing requests for the piecemeal Write or in multiple block write requests for the sections Disassemble the letter. A request network subscriber station according to any one of claims 10 to 17, characterized in that the creation / dispatch means are arranged to act as a write request updating the entry in a Broadcast_Channel_Register to a destination network subscriber station ( 10 . 12 ).