WO1997028504A1

WO1997028504A1 - Application programming interface for data transfer and bus management over a bus structure

Info

Publication number: WO1997028504A1
Application number: PCT/US1997/001742
Authority: WO
Inventors: Scott D. Smyers; Bruce Fairman
Original assignee: Sony Electronics, Inc.
Priority date: 1996-02-02
Filing date: 1997-01-29
Publication date: 1997-08-07
Also published as: ATE199987T1; DE69704344T2; US6243783B1; KR100472908B1; JP2000510659A; DE69723726D1; EP0877983B1; CA2244713C; EP1056016A2; AU1855797A; EP1056016A3; US5991520A; KR19990082226A; DE69704344D1; TW317623B; CA2244713A1; EP0952526A3; EP2284713A3; JP3993893B2; EP2284713A2

Abstract

An application programming interface implements and manages isochronous and asynchronous data transfer operations between an application and a bus structure. During an asynchronous transfer the API includes the ability to transfer any amount of data between one or more local data buffers within the application and a range of addresses over the bus structure using one or more asynchronous transactions. An automatic transaction generator may be used to automatically generate the transactions necessary to complete the data transfer. The API also includes the ability to transfer data between the application and another node on the bus structure isochronously over a dedicated channel. During an isochronous data transfer, a buffer management scheme is used to manage a linked list of data buffer descriptors. This linked descriptor list can form a circular list of buffers and include a forward pointer to the next buffer in the list and a backward pointer to the previous buffer in the list for each buffer. The linked descriptor list may also form a linear list to which the application can append additional buffers or remove existing buffers from the list. During isochronous transfers of data, the API provides implementation of a resynchronization event in the stream of data allowing for resynchronization by the application to a specific point within the data. Implementation is also provided for a callback routine for each buffer in the list which calls the application at a predetermined point during the transfer of data.

Description

APPLICATION PROGRAMMING INTERFACE

FOR DATA TRANSFER AND BUS MANAGEMENT

OVER A BUS STRUCTURE

FIELD OF THE INVENTION:

The present invention relates to the field of providing an interface for applications to communicate over a bus structure. More particularly, the present invention relates to the field of controlling bus management and data transfer operations between applications over a bus structure in both asynchronous and isochronous formats.

BACKGROUND OF THE INVENTION:

The IEEE 1394 standard, "P1394 Standard For A High Performance Serial Bus," Draft 8.0 lvl, June 16, 1995, is an international standard for implementing an inexpensive high-speed serial bus architecture which supports both asynchronous and isochronous format data transfers. Isochronous data transfers are real-time transfers which take place such that the time intervals between significant instances have the same duration at both the transmitting and receiving applications. Each packet of data transferred isochronously is transferred in its own time period. An example of an ideal application for the transfer of data isochronously would be from a video recorder to a television set. The video recorder records images and sounds and saves the data in discrete chunks or packets. The video recorder then transfers each packet, representing the image and sound recorded over a limited time period, during that time period, for display by the television set. The IEEE 1394 standard bus architecture provides multiple channels for isochronous data transfer between applications. A six bit channel number is broadcast with the data to ensure reception by the appropriate application. This allows multiple applications to simultaneously transmit isochronous data across the bus structure. Asynchronous transfers are traditional data transfer operations which take place as soon as possible and transfer an amount of data from a source to a destination.

- 1 - SUBSTITUTE SHEET The IEEE 1394 standard provides a high-speed serial bus for interconnecting digital devices thereby providing a universal I/O connection. The IEEE 1394 standard defines a digital interface for the applications thereby eliminating the need for an application to convert digital data to analog data before it is transmitted across the bus. Correspondingly, a receiving application will receive digital data from the bus, not analog data, and will therefore not be required to convert analog data to digital data. The cable required by the IEEE 1394 standard is very thin in size compared to other bulkier cables used to connect such devices. Devices can be added and removed from an IEEE 1394 bus while the bus is active. If a device is so added or removed the bus will then automatically reconfigure itself for transmitting data between the then existing nodes. A node is considered a logical entity with a unique address on the bus structure. Each node provides an identification ROM, a standardized set of control registers and its own address space.

The IEEE 1394 standard defines a protocol as illustrated in Figure 1. This protocol includes a serial bus management block 10 coupled to a transaction layer 12, a link layer 14 and a physical layer 16. The physical layer 16 provides the electrical and mechanical connection between a device or application and the IEEE 1394 cable. The physical layer 16 also provides arbitration to ensure that all devices coupled to the IEEE 1394 bus have access to the bus as well as actual data transmission and reception. The link layer 14 provides data packet delivery service for both asynchronous and isochronous data packet transport. This supports both asynchronous data transport, using an acknowledgement protocol, and isochronous data transport, providing real-time guaranteed bandwidth protocol for just-in-time data delivery. The transaction layer 12 supports the commands necessary to complete asynchronous data transfers, including read, write and lock. The serial bus management block 10 contains an isochronous resource manager for managing isochronous data transfers. The serial bus management block 10 also provides overall configuration control of the serial bus in the form of optimizing arbitration timing, guarantee of adequate electrical power for all devices on the bus, assignment of the cycle master, assignment of isochronous channel and bandwidth resources and basic notification of errors.

- 2 - - SUBSTITUTE SHEET An application programming interface (API) for applications using the IEEE 1394 standard serial bus has been developed by Skipstone for enabling the application to use the IEEE 1394 bus for data transfers. With their API, Skipstone includes a manual entitled "The SerialSoft IEEE 1394 Developer Toolkit," available from Skipstone, Inc., 3925 West Braker Lane, #425, Austin, Texas 78759. Skipstone defines their API as a collection of programming calls to be used by the application to manage data being written to and obtained from a device over an IEEE 1394 bus. To initialize an isochronous transfer, several asynchronous data transfers may be required to configure the applications and to determine the specific channel which will be used for transmission of the data. Once the channel has been determined, buffers are used at the transmitting application to store the data before it is sent and at the receiving application to store the data before it is processed. In a transmitting application, the Skipstone API actively manages the transfer of data from the appropriate portion of the appropriate buffer onto the bus structure, during the appropriate time period. In a receiving application, the Skipstone API actively manages the reception of data from the bus structure, storing the data in the appropriate portion of the appropriate buffer and the processing of the data in the appropriate time period.

During asynchronous data transfers, the Skipstone API actively manages the required transactions to complete the data transfer. During an asynchronous incoming write transaction, the application provides a buffer to the API, mapped to a certain area of the 1394 bus address space. As write transactions arrive at the API, their data is written to the buffer. During an asynchronous incoming read transaction the application is responsible for making sure that the buffer contains useful information. The 1394 bus driver then reads the data from the buffer at the requested address when the read transaction arrives. For both write and read transactions, the Skipstone API actively manages and generates each necessary transaction. For example, if a block of data is being transferred to the application, of a size requiring multiple transactions, the Skipstone API requires the application to describe each 1394 transaction necessary to complete the transfer of the block of data. This consumes significant overhead by the processor of the

- 3 - SUBSTITUTE SHEET application as well as the full attention of the API during an asynchronous data transfer operation.

The Skipstone API supports isochronous data transfer operations in a similar way. Specifically, the application must describe each isochronous packet to the Skipstone API. The Skipstone API then transmits each packet at the proper time. This requires significant processor overhead and thereby prohibits efficient processing of the isochronous data by the application.

What is needed is an API that provides automated generation of transactions necessary to complete a data transfer, without requiring supervision by the API and the processor of an application. What is further needed is an API which implements isochronous transfer features of the IEEE 1394 standard bus structure very efficiently, permitting a high degree of hardware automation, if needed by the application.

SUMMARY OF THE INVENTION: An applications programming interface implements and manages isochronous and asynchronous data transfer operations between an application and a bus structure. During an asyncronous transfer the API includes the ability to transfer any amount of data between one or more local data buffers within the application and a range of addresses over the bus structure using one or more asynchronous transactions. An automatic transaction generator may be used to automatically generate the transactions necessary to complete the data transfer without direct processor control or supervision by the applications programming interface. The API also includes the ability to transfer data between the application and another node on the bus structure isochronously over a dedicated channel. During an isochronous data transfer, a buffer management scheme is used to manage a linked list of data buffer descriptors provided by the application. The linked list of buffer descriptors is maintained by the API to ensure the uninterrupted flow of the continuous stream of isochronous data. This linked descriptor list can form a circular list of buffers and include a forward pointer to the next buffer in the list and a backward pointer to the previous buffer in the list for each buffer. The linked descriptor list may also form a linear list to which the application can append additional buffers or

- 4 - - SUBSTITUTE SHEET - remove existing buffers from the list. During isochronous transfers of data, the API provides implementation of a resynchronization event in the stream of data allowing for resynchronization by the application to a specific point within the data. Implementation is also provided for a callback routine for each buffer in the list which calls the application at a predetermined point during the transfer of data.

BRIEF DESCRIPTION OF THE DRAWINGS:

Figure 1 illustrates a protocol defined by the IEEE 1394 standard.

Figure 2 illustrates a block diagram schematic of an applications programming interface within a system including a bus structure.

Figure 3 illustrates a system including a video camera 50, a video cassette recorder 52 and a computer 54 connected together by the input/output (I/O) busses 56 and 58.

Figure 4 illustrates a block diagram of a hardware system resident in each system for implementing the applications programming interface of the present invention. Figure 5 illustrates a list of buffer descriptors corresponding to buffers assigned to an API 20 by an application.

Figure 6 illustrates a flow chart illustrating API buffer processing for isochronous send and receive operations.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT:

Figure 3 illustrates a system including a video camera 50, a video cassette recorder 52 and a computer 54 connected together by the input/output (I/O) busses 56 and 58. The I/O bus 56 couples the video camera 50 to the video cassette recorder 52, allowing the video camera 50 to send data to the video cassette recorder 52 for recording. The I/O bus 58 couples the video cassette recorder 52 to the computer 54, allowing the video cassette recorder 52 to send data to the computer 54 for display.

An applications programming interface (API) according to the present invention could be implemented within any one or all of the connected subsystems including the video camera 50, the video cassette recorder 52 or the computer 54, for controlling data transfer operations communicated across the bus structures 56 and 58. In the preferred

- 5 - - SUBSTITUTE SHEET - embodiment of the present invention the bus structures 56 and 58 are IEEE 1394 standard cables.

A block diagram of a hardware system resident in each system for implementing the applications programming interface of the present invention is illustrated in Figure 4. In the hardware system illustrated in Figure 4, a printed circuit board 60 is coupled to a user interface 70. The printed circuit board 60 includes a central processing unit (CPU) 62 coupled to system memory 64 and to an I/O bus interface 66 by the system bus 68. The user interface 70 is also coupled to the system bus 68. The user interface 70 is subsystem specific, but can include a keyboard, display or other I/O devices for communicating with a user of the subsystem.

Each of the subsystems including the video camera 50, the video cassette recorder 52 and the computer 54, in order to implement the applications programming interface of the present invention, will include a hardware system such as the system illustrated in Figure 4. The CPU 62 within each of these devices is used to execute the application program instructions. The API of the present invention will then manage both isochronous and asynchronous data transfer operations between the resident subsystem and one of the other subsystems over an appropriate one of the busses 56 or 58.

An applications programming interface according to the present invention implements isochronous and asynchronous data transfers to and from an application over a bus structure. An application as used herein will refer to either an application or a device driver. The bus structure over which the data transfer operations are completed is preferably an IEEE 1394 standard bus structure. However, as will be apparent to those skilled in the art, the applications programming interface of the present invention will also be applicable for use in managing data transfers over other types of bus structures. The applications programming interface includes the ability to transfer any amount of data between a local data buffer provided by the application and a range of addresses over the bus structure using one or more asynchronous transactions. When an asynchronous transfer of a block of data is initiated, the applications programming interface sends a command to an automatic transaction generator. The automatic transaction generator then automatically generates the read or write transactions necessary to transfer the complete

- 6 - - SUBSTITUTE SHEET - block of data asynchronously without direct processor control or requiring supervision by the applications programming interface.

The applications programming interface also includes the ability to transfer data between the application and another node on the bus structure isochronously over a dedicated channel. During an isochronous data transfer, a buffer management scheme is used to manage data buffers within the application. The application may use one, more than one or a linked list of buffers depending on the type and amount of data to be transferred. A linked list of buffer descriptors that point to the buffers is maintained by the API to ensure the uninterrupted flow of the continuous stream of isochronous data. This linked descriptor list may implement a linear or a circular list of buffers and includes a forward pointer to the descriptor for the next buffer in the list and a backward pointer to the descriptor for the previous buffer in the list for each buffer. When a linear list is implemented, the application can dynamically append buffers to the list or remove existing buffers from the list, as necessary, for the processing of the data. During an isochronous transfer of data, the applications programming interface of the present invention provides implementation of a resynchronization event in the stream of data allowing for resynchronization to a specific point within the data. Implementation is also provided for a callback routine for each buffer which calls the application at a predetermined specific point during the data transfer operation. Both the resynchronization event and the callback routine are supported by the IEEE 1394 standard.

The applications programming interface of the present invention also includes the ability to perform bus management operations, as necessary, over the bus structure. Such bus management operations include allocating and deallocating isochronous channel numbers, as necessary, and allocating and deallocating isochronous bandwidth. If the bus structure is an IEEE 1394 standard bus structure, then the applications programming interface also performs other bus management operations as required by the IEEE 1394 standard.

A block diagram schematic of an applications programming interface, according to the present invention, within a system including a bus structure is illustrated in Figure 2. The API 20 serves as an interface between the applications 22 and 24 and the bus

- 7 - - SUBSTITUTE SHEET - structure 28, managing the transfer of data between the bus structure 28 and the applications 22 and 24. As illustrated in Figure 2, a single API 20 may serve as an interface between multiple applications and the bus structure 28. For example, within the computer system 54, illustrated in Figure 3, a single API 20 could serve as an interface between one or more applications being run by the computer system 54.

A hardware and physical interface 26 is included between the API 20 and the bus structure 28. The hardware and physical interface 26 includes an automatic transaction generator 38 for automatically generating the necessary transactions for completion of an asynchronous data transfer between one of the applications 22 or 24 and another node on the bus structure 28. The hardware and physical interface 26 also includes a bus interface

40 for monitoring and managing the flow of data to and from the bus structure 28. The hardware and physical interface 26 is shown coupled to a set of memory buffers 30, as controlled by the API 20. The set of memory buffers 30 includes the memory buffers 32, 34 and 36. As will be described below, the memory buffers 32, 34 and 36 are dedicated to the API 20 by the application 22 for use in sustaining isochronous data transfers to and from the application 22.

ISOCHRONOUS DATA TRANSFERS

To initialize an isochronous data transfer operation an application first requests an isochronous channel from the API 20. The application may either request a specific channel number or any currently available channel number. The API 20 then obtains a channel for the isochronous transfer per the requirements of the IEEE 1394 standard. The IEEE 1394 standard supports a six bit channel number which is broadcast with a stream of data across the bus structure 28. Once a channel is allocated for an isochronous data transfer between an application and another node on the bus structure 28, no other nodes may use that specific channel number. After a channel is allocated, data buffers must be assigned by the application to the API 20 to be used for the data transfer. The API 20 allows the application to assign one, more than one or a list of data buffers to use for receiving or transmitting the isochronous stream of data. Each buffer assigned to the API 20 may be contiguous or fragmented and logical or physical. The list of data buffers may

- 8 - - SUBSTITUTE SHEET - be circular or linear. If a linear list of data buffers is assigned to the API 20 the application 22 can add additional buffers or remove buffers from the list as necessary to process the data.

In the system illustrated in Figure 2, the application 22 has assigned three buffers 30, including the buffers 32, 34 and 36 to the API 20 for isochronous data transfers. The application has also assigned a linked list of three buffer descriptors to the API, one for each of the buffers 32, 34 and 36. The API 20 maintains a buffer descriptor for each buffer within the linked list and manages the flow of the isochronous data between the application, the assigned buffers and the bus structure 28. Within the list of descriptors managed by the API 20, each buffer is represented by a buffer descriptor, including a forward pointer to the descriptor for the next buffer in the list and a backward pointer to the descriptor for the previous buffer in the list. A list of buffer descriptors corresponding to buffers assigned to an API 20 by an application is illustrated in Figure 5. Each of the buffer descriptors 1-n correspond to a memory buffer 1-n. Specifically, the buffer descriptor 80 corresponds to the memory buffer 84 and the buffer descriptor 82 corresponds to the memory buffer 86.

The buffer descriptors each include an address and length of the corresponding buffer. The buffer descriptor also includes a callback completion routine to call after the buffer has been filled or emptied, depending on the direction of the current data transfer operation. The buffer descriptors further include an optional synchronization event field which is programmed by the application and is how the buffer is synchronized to a specific event or time. Specifically, the buffer descriptor 80 corresponding to the memory buffer 84, includes an address 80a and a length 80b for the memory buffer 84. A completion routine 80c and a synchronization event 80d are also included, if necessary. This use of buffer descriptors and memory buffers allows great flexibility to an application using the API of the present invention, since the descriptors, buffers, completion routines and synchronization events are all set up by the application according to its specific needs. As an example, for an application that is running in a digital video camera transferring data isochronously to a digital video monitor, data is loaded in memory buffers, for which the API maintains buffer descriptors. The API then manages

- 9 - - SUBSTITUTE SHEET - the transfer of each packet of data from the buffers to the video momtor. The video camera is able to implement a 2x compression feature in the vertical dimension by having pairs of descriptors point to the same memory buffer. That is, the descriptors 1 and 2 will point to the memory buffer 1 , the descriptors 3 and 4 will point to the memory buffer 2, and so on. A completion routine in the second descriptor of each pair notifies the video monitor that data in the memory buffer is ready to be read. This means that as the video camera outputs first and second scan line data, the second scan line data overwrites the first scan line data in the memory buffer with the second scan line data. The video monitor does not read the memory buffer until after the second scan line is written so the momtor never sees the first scan line data. In this manner, every other scan line is skipped.

The descriptors allow the list to be circular in nature and thereby maintain the continuous stream of data to or from the buffers 32, 34 and 36. During an isochronous data transfer from the application 22 to another node along the bus structure 28, the application 22 fills the buffers 32, 34 and 36, in turn, with the data. The API 20 then manages the transferring of the data from the appropriate buffer to the bus structure 28 during an appropriate time period. The bus interface 40 within the hardware and physical interface 26 controls transferring the data from the buffers 32, 34 and 36 onto the bus structure 28. During an isochronous data transfer from another node along the bus structure 28 to the application 22, the API 20 manages transferring the data from the bus structure 28, through the bus interface 40, to the appropriate buffer 32, 34 and 36. As one allocated buffer is filled up, the data is stored in the next buffer in the linked list. The application 22 then reads the data from the appropriate one of the buffers 32, 34 and 36 during the appropriate time period. Once the application 22 has finished reading the data from a buffer, the buffer is provided back to the API 20 and the application 22 processes the data from the next buffer.

The buffer descriptors will also implement a linear list of buffers which allows the application to assign buffers to or remove buffers from the API 20, as necessary to complete a data transfer operation. For example, during an isochronous receive operation, as the application is finished processing each buffer it can then reassign it to the API for

- 10 - - SUBSTITUTE SHEET - receiving more data. Correspondingly, if additional buffers are necessary to complete a data transfer operation, the application can assign more buffers to the API.

The API 20 will execute a resynchronization event and/or a callback routine during the transfer of isochronous data if requested by the application 22. A resynchronization event allows for resynchronization by the application to a predetermined specific point in time within the data during the transfer. Because the data is being transferred isochronously, this resynchronization event will also synchronize the application to an appropriate point in time relative to the data flow. The transfer of video data provides an ideal example for the implementation of a resynchronization event. During the transfer of video data from an application such as a video recorder, the data is transferred in blocks representing the data necessary to display one horizontal line on a monitor or television. After the display of each horizontal line, the monitor must reset itself to be ready to display the next horizontal line. A resynchronization event could be employed by the monitor at the end of the data for each horizontal line, allowing the monitor to resynchronize itself to the beginning of the next horizontal line.

In the preferred embodiment of the API of the present invention an isochronous operation may be synchronized or scheduled to be performed immediately, at a specific bus time, when a specific value appears in the isochronous data block packet header, or when isochronous data appears on a specific channel of the bus for start operations or ends on a specific channel of the bus for stop operations.

Each buffer assigned to the API 20 can have a resynchronization event and a callback routine. A callback routine could be employed during the transfer of video data at the end of the transfer of a block of data representing a frame. A monitor or television groups horizontal lines into a frame and at the end of each frame resets itself to the top of the screen to be ready for the beginning of the next frame. A callback routine could be used at the end of the stream of data representing each frame. Such a scheme would allow a buffer to be filled with the data representing a video frame from a source coupled to the bus structure 28. After the data representing the video frame has been transferred, the callback routine can be used to notify the application that the data representing the next frame has been transferred and is available for processing. The application could

- 1 1 - - SUBSTITUTE SHEET - then process the data for this frame of data while the data for the next frame is being loaded into the next buffer.

A flow chart illustrating API buffer processing for isochronous send and receive operations is shown in Figure 6. It is assumed that at the start 102 of an isochronous receive operation that the application has set up the buffers/descriptors, completion routine calls and synchronization events. The flowchart 100 is entered at step 102 for each isochronous stream that requires processing in the bus system. The API 20 keeps track of a current descriptor for processing the incoming data. In other words, the API maintains a pointer to the next buffer, and location within the next buffer where data can be stored. At step 104 the next buffer descriptor is obtained from the linked list. At step 106 a check is made to determine if any more descriptors are included within the linked list. If there are no more descriptors in the linked list then the processing is stopped at the step 108. If there are additional descriptors then the routine moves to the step 1 12 where it waits until the synchronization event for the current buffer is reached. Once the synchronization event is reached, then at the step 1 14 the current buffer is either filled with the incoming data for a receive operation or the data from the buffer is transmitted for a send operation. After the buffer has been processed, then at the step 116 it is determined if a callback routine was included for this buffer. If a callback routine was included, then, at the step 1 18 the callback routine is called. Otherwise, the routine goes back to the step 104 and obtains the next descriptor. Whether a callback routine is provided or not, the API and hardware subsystem 26 assure that the next buffer descriptor is obtained such that no isochronous data is lost.

The steps of the flowchart 100 may be performed by a CPU and related subsystems such as found in a typical personal computer (PC), embedded processing system, etc. as discussed above in connection with Figures 3 and 4. In general, the steps of flowcharts presented in this specification may be implemented in any suitable programming language such as "C", PASCAL, FORTRAN, BASIC, assembly language, etc., or in a combination of such languages. Any suitable computer programming technique may be used for a software design to implement the steps, such as procedural or object oriented programming, parallel or distributed processing, interrupt driven or polled event

- 12 - - SUBSTITUTE SHEET - processing, etc. Steps may be modified, added to, or taken away from, those shown in the flowcharts while still achieving the method steps and apparatus elements described in this specification and recited in the claims. The processing in a single step may be broken into two or more steps. Also, in some embodiments, two or more steps may be accomplished at the same time, or their tasks interleaved. The sequencing, or routing, of the steps may also be changed. Each flowchart is but one instance of a primitive example of the logic used to achieve a function in a preferred embodiment of the present invention.

For purposes of discussion, the cumulative steps of a flowchart are referred to as constituting a single "routine," or program, although they may be implemented in two or more routines, programs, processes, etc. Flowchart steps may also be distributed among processors residing in the same or different devices.

As an example of an isochronous data transfer operation, if the application 22 is a video monitor which is receiving data isochronously from a video recorder at a node coupled to the bus structure 28, the API 20 will manage the flow of data from the bus structure to the buffers 32, 34 and 36, each represented by a buffer descriptor in the linked list. A first buffer 32 is filled with the data received from the video recorder. When the first buffer 32 is filled, it is processed and displayed by the video monitor 22 while the next buffer 34 in the linked list is filled. If the first buffer 32 included a callback routine at the end of the data for a frame, then the callback routine could be used to notify the video monitor 22 that it could process the data in the first buffer 32, representing the first frame. When the video monitor 22 is finished processing the data within the first buffer 32 it can then provide the buffer 32 back to the API 20 for storing additional data received from the bus structure 28.

If the application 22 is a video recorder transmitting isochronous data to another node coupled to the bus structure, then the application loads the buffers 32, 34 and 36, in turn, with data. The API 20 will then manage the transmission of the data from the buffers 32, 34 and 36 onto the bus structure 28 with the appropriate channel number at the appropriate time. In this manner the API 20 of the present invention manages isochronous data transfers to and from an application 22.

- 13 - - SUBSTITUTE SHEET - ASYNCHRONOUS DATA TRANSFERS

To execute an asynchronous data transfer operation between an application 24 and another node coupled to the bus structure 28, the API 20 defines essentially a direct memory access (DMA) model, utilizing a level of hardware automation to automatically generate the requests necessary to complete the transfer and allowing the application and the API 20 to perform other functions while the data transfer operation is being completed. The API 20 provides a memory-mapped interface to the application for asynchronous data transfers. To initiate an asynchronous data transfer, an application 24 transmits a descriptor to the API 20 including an address of a buffer within the application's address space, a starting address in the address space of the bus structure at which the transfer is to take place, a length of the block of data to be transferred and a code representing whether the transfer is to be a read or write operation. The API 20 provides the necessary data to the hardware automatic transfer generator 38 which then generates the one or more transactions necessary to complete the transfer of the entire block of data across the bus structure 28. The automatic transfer generator 38 then generates the necessary read or write transactions to complete the transfer of data between the buffer assigned by the application 24 and the appropriate addresses across the bus structure 28. This automation does not require the attention of the API 20 or the application 24 to complete an asynchronous data transfer operation. While in the preferred embodiment of the present invention the automatic transaction generator 38 is implemented in hardware, it should be apparent to those skilled in the art that the automatic transaction generator could also be implemented in software within the API 20. If the application does not require this level of hardware automation, the API 20 can also generate the transactions necessary to complete a data transfer operation, without using the automatic transaction generator 38. As is known to those skilled in the art each read or write transaction can only transfer a certain amount of data depending on the system and the capabilities of the bus structure 28. Therefore, to transfer a block of data it may be necessary to generate multiple read or write transactions. In contrast to the systems of the prior art, the API 20 of the present invention sends a single command to the automatic transaction generator block 38. The automatic transaction generator block 38 then generates the read or write

- 14 - - SUBSTITUTE SHEET - transactions necessary to transfer the complete block of data over the bus structure 28, without requiring further attention by the API 20. This allows the system to be more efficient, as the API 20 and the application 24 can perform other tasks while the transfer is taking place. Because the transfer is asynchronous, once the transfer of the entire block of data is complete, the API 20 will notify the application 24.

As discussed above, in the preferred embodiment of the present invention, the bus structure 28 is an IEEE 1394 standard bus structure. For asynchronous data transfers the bus structure 28 therefore provides a 64 bit address space. Within the descriptor provided to the automatic transaction generator 38, the remote address at which the data transfer is to take place is specified by a 64 bit address.

To initiate an asynchronous read operation, the application 24 transmits a descriptor to the API 20 including the address of the buffer within the application's address space to which the data is to be transferred, a 64 bit starting address in the address space of the bus structure 28 from which the data is to be read, the length of the block of data to be transferred and a code representing that the transfer is a read operation. The API 20 then transmits the required information to the automatic transaction generator 38. The automatic transaction generator 38 then generates the necessary read commands to transfer the data to the application's buffer from the proper node on the bus structure 28. The application is responsible for ensuring that the specified buffer is available before the read transactions are generated. The data is then read in response to the transactions generated by the automatic transaction generator 38, in a known manner.

To initiate an asynchronous write operation, the application 24 transmits a descriptor to the API 20 including the address of the buffer within the application's address space from which the data is to be transferred, a 64 bit starting address in the address space of the bus structure 28 to which the data is to be written, the length of the block of data to be transferred and a code representing that the transfer is a write operation. The API 20 then transmits the required information to the automatic transaction generator 38. The automatic transaction generator 38 then generates the necessary write commands to transfer the data to the proper node on the bus structure 28 from the application's buffer. The data is then transferred from the application's buffer in response

- 15 - - SUBSTITUTE SHEET - to the transactions generated by the automatic transaction generator 38 in a known manner. When the buffer is transferred the application 24 is notified.

API CONVENTIONS AND BUS MANAGEMENT OPERATIONS An application calls a routine in the API 20 either synchronously or asynchronously. If an application calls a routine synchronously, then at the time that the routine returns to the application, the API has completed the requested operation or the API returns a completion status indicating that the chosen request could not be completed. Alternatively, if an application calls a routine asynchronously, then the requested action is most likely not complete at the time that the routine returns control to the client. In order to call a routine asynchronously, the application provides a completion callback routine. The API may call this completion routine before returning from the original call. However, in most cases the API completes the requested operation after returning from the original call that initiated the operation, then calls the application's completion routine to indicate that the operation is done.

Before using any of the services provided by the API, an application must first initialize the API. Each application must initialize the API separately. An application initializes the API by calling an ActivateSonyAPI subroutine. This subroutine establishes a connection between the API and the application. When calling the ActivateSonyAPI, the application may specify indication routines which the API calls when a bus reset or other bus event occurs. The ActivateSonyAPI subroutine returns a value to the application which the application then uses on subsequent calls to the routines of the API.

Applications which expect a large number of indications during the course of their operation may call the AddlndBuffers routine in order to pass additional indication buffers to the API for its exclusive use. The client can first call the CountlndBuffers routine in order to check the number of buffers that the API currently owns. Prior to deactivating the API, the application may release the indication buffers previously given to the API by calling a RellndBuffers routine.

When an application is finished using the API, it calls a DeactivateSonyAPI routine. This routine breaks the connection between the application and the API and

- 16 - - SUBSTITUTE SHEET - releases any indication buffers or other resources in use by the API on behalf of the application. Note that the API may not be able to disassociate from a given application immediately if some of the application's buffers are currently in use by the API. During the time that the API is active for a given application, that application has access to all of the services that the API provides.

After initializing the API, an application may perform various IEEE 1394 bus management functions, as defined in section 8 of the IEEE 1394 standard, and described below. An application may allocate and deallocate isochronous channel numbers from the currently active isochronous resource manager using the MGMTAllocateChannel and MGMTDeAllocateChannel routines, respectively. Using these applications, the application may request to allocate a specific channel number, if it is available. Alternatively, an application may request to allocate any currently available channel number. These API routines follow the requirements of the IEEE 1394 standard with regard to allocating and deallocating isochronous channel numbers. When using isochronous channel numbers, the application is responsible for following any other requirements which may apply in the

IEEE 1394 standard or any other governing protocol document.

An application may allocate and deallocate isochronous bandwidth from the currently active isochronous resource manager using the MGMTAllocateBandwidth and MGMTDeAUocateBandwidth routines, respectively. These API routines follow the requirements of the IEEE 1394 standard with regard to allocating and deallocating isochronous bandwidth. When using these routines, the application is responsible for calculating the correct amount of isochronous bandwidth needed and allocating exactly that much. The application is also responsible for following any applicable rules as documented in the IEEE 1394 standard and any other governing protocol documents, with regard to allocating, deallocating or owning any isochronous bandwidth.

When an application deactivates the API, the API does not attempt to deallocate any bus resources that the application previously allocated. This permits the application to relinquish ownership of these resources easily, as required in the IEC AV protocols standard. However, this places complete responsibility on the application to follow the governing protocols when allocating and deallocating isochronous bus resources.

- 17 - - SUBSTITUTE SHEET - An application may retrieve the current topology map and speed map information from the active bus manager, if present, and any other available bus information using the MGMTBusInfo routine. This routine retrieves the most current information from the bus manager, whether or not the node on which the application is running is the active bus manager. Note that this routine will fail if there is no currently active bus manager.

Section 8 of the IEEE 1394 standard defines the format of the topology map and speed map, and the conditions under which a bus manager exists or does not exist.

After initializing the API, the application may call the ASYNDataRequest routine to initiate asynchronous data transfer requests over the IEEE 1394 serial bus. The application may use this routine to initiate any asynchronous transaction that is defined in the IEEE 1394 standard, including data block read or write requests, quadlet read or write requests or any lock request. When the application calls the ASYNDataRequest routine, the routine passes a descriptor for a buffer in the application's address space, a starting address in the 64 bit IEEE 1394 address space, a data transfer length and a transaction code. The ASYNDataRequest routine then generates one or more IEEE 1394 transactions to satisfy the request. When the API finishes the requested data transfer operation, or if it encounters an error, the API returns to the application or calls the application's callback routine, depending on whether the application called this routine synchronously or asynchronously. In order to perform a lock transaction over the IEEE 1394 serial bus, the application calls the ASYNDataRequest routine and passes an argument value, a data value, and a lock operation code. The API generates the requested lock operation and returns to the application or calls the application's callback routine, as determined by the type of call, e.g., synchronously or asynchronously. After initializing the API, the application may source or sink a channel of isochronous data on the IEEE 1394 serial bus. Before transferring isochronous data, the application must first open an isochronous port using the ISOCHOpen routine. When calling this routine, the application specifies the direction and other information about the stream of isochronous data that the application intends to transfer. The ISOCHOpen routine determines if the necessary system resources are available then returns to the

- 18 - - SUBSTITUTE SHEET - application. When this routine completes successfully, the application then has all necessary system resources reserved for its exclusive use to transfer a stream of isochronous data.

When an application talks or listens on an isochronous channel, the source or destination of the isochronous data in the host system is one or more data buffers owned by the application and described in a data structure. The application passes these buffers to the API by calling the ISOCHAttach routine. This routine "attaches" the application buffers to the isochronous stream in preparation for transferring application data into or out of these buffers. If the application wishes to reclaim its buffers before the API has finished with them, the application may call the ISOCHDetach routine, specifying those buffers that the application wishes to reclaim.

The API defined buffer descriptor which the application uses to describe its isochronous data buffers permits the application to specify one, more than one, or a list of data buffers to use for receiving or transmitting isochronous data. Each buffer may be contiguous or fragmented, logical or physical and the application may specify callback routines on a buffer by buffer basis. This permits extremely flexible buffer handling in the API on behalf of the application, thereby supporting a wide range of application requirements.

When the application has opened an isochronous port and has attached buffers to this port, then the application may control its stream of isochronous data. It does this by calling the ISOCHControl routine. This routine permits the application to start or stop an isochronous stream into or out of the application buffers. When calling this routine, the application may specify an event on which to start or stop the stream, e.g., immediately, on a particular isochronous cycle or other event. When the application is finished transferring a stream of isochronous data, it releases the system resources associated with the open port by calling the ISOCHClose routine.

The ActivateSonyAPI and DeactivateSonyApi routines provide the initialization mechanism which makes the IEEE 1394 specific services provided by the API available to the calling application. The ActivateSonyAPI routine establishes a connection to the services provided by the API. The DeactivateSonyAPI routine removes the specified

- 19 - - SUBSTITUTE SHEET - connection to the services provided by the API. The result of an activation is a valid activateReq structure. The calling application passes a pointer to this structure as part of all subsequent calls to the API. As part of activating the API for an application, the application may provide indication routines which the API uses to inform the caller that something has happened on the associated IEEE 1394 bus, such as a bus reset or request indication from a remote node. The result of deactivation is that the indication routines, if any, which were registered at activation time are de-registered. Following deactivation, the caller may not use any of the API services, unless the API is first reactivated. The following function activates the API for further operations:

STATUS ActivateSonyAPI(ActivateReqPtr activateReq);

This routine takes one parameter and returns a status value. The caller fills in the activateReq data structure, as defined below, and passes a pointer to the ActivateSonyAPI routine. After this routine returns with a GOOD status, the caller saves the resulting activateReq data structure. This data structure represents the connection between the application and the API. In order to identify this connection, the caller passes a pointer to the activateReq data structure on subsequent calls to the API. The possible status return values are GOOD, signalling that the API is now activated and the activateReq data structure is valid to use for further operations, PENDING, signalling that the API has accepted the request, but is not active at this time, and UNDEFINEDERROR, signalling that an unexpected error was encountered while attempting to activate the API and that the API is not activated. After a PENDING value is returned, the API calls the AsyncCompletion routine when the activation request is complete. At that time the request status field will contain the completion status for the activate request.

The single parameter contains the address of an activatReq data structure. This data structure provides information necessary to activate the API, as defined in Table I below:

- 20 - - SUBSTITUTE SHEET TABLE I typedef struct ActivateReq { void (*BusResetHandler)(BusResetPtr); /* Bus Reset Handler */

STATUS (*IndicationHandler)(IndicationPtr); /* Indication Handler */ void *RefPtr; / * for use by above routines */ void *SonyAPIPrivate; /* the cookie */ void (*AsyncCompletion)(struct ActivateReq *req); /* completion routine */ void "UserPtr; /* for use by completion routine*/

STATUS Status; /* completion status */

} ActivateReq, *ActivateReqPtr;

When the BusResetHandler filed is not equal to null, it contains the address of the routine to call upon receiving a bus reset event. When a bus reset occurs on the IEEE 1394 bus, the API calls the BusResetHandler routine, passing the address of a data structure containing bus reset information. When the IndicationHandler field is not equal to null, it contains the address of the routine to call upon the occurrence of an indication that is not handled by the API. When the API receives a request subaction from a remote node, it calls the IndicationHandler routine, passing the address of a data structure which describes the request. The API fills in the SonyAPIPrivate field as part of the activation process. The API uses the value in this field on subsequent calls. The calling application shall not modify the value in this field. When the AsyncCompletion field is not equal to null, it contains the address of the routine to call when the API is active and available for use by the invoking application. Note that the calling application may specify a completion routine whether the request is asynchronous or synchronous. The UserPtr field is available for use by the calling application's completion routine. The API does not modify this field. The Status field contains the status of the activation request.

The following function terminates the instantiation of the API represented by request:

status DeactivateSonyAPI(ActivateReqPtr request);

- 21 - - SUBSTITUTE SHEET The possible status return values for this function are GOOD, signalling that the API is now deactivated and the activateReq data structure is invalid to use for further operations, INVALIDCONNECTION, PENDING, signalling that the API has accepted the deactivate request, but is still active at this time, and UNDEFINEDERROR, signalling that an unexpected error was encountered while attempting to deactivate the API and that the API may be active. After a PENDING value is returned, the API calls the AsyncCompletion routine when the deactivation request is complete. At that time, the request status field will contain the completion status for the activate request.

The single parameter contains the address of the activateReq data structure used to activate the API previously. The section above defines this data structure and describes its fields. Note that when deactivating the caller must use the same data structure that was used to activate the API previously. The caller may modify the values in the AsyncCompletion field and the UserPtr field. The caller should not modify any other field in the activateReq data structure following the initial call to the ActivateSonyAPI routine and prior to the call to the DeactivateSonyAPI routine. In addition to deactivating the API for a specific application, this routine also releases any indication buffers that the application previously passed to the API. If there are outstanding indication buffers owned by the application and the application attempts to call this routine synchronously, this routine will return an error. If this happens, the application may call this routine again specifying a completion routine. The API will complete the deactivate request and call the application's indication routine when all of the application's indication buffers have been released.

The API calls the indication handling routines, BusResetHandler and IndicationHandler, asynchronously, and they may have limited system services available to them. Depending on the environment and possibly some other circumstances, the API may call these routines at interrupt level. In the BusResetHandler routine, the handler is passed a pointer to bus reset information. In the IndicationHandler routine, the handler is passed a pointer to indication information. The application passes the address of one or both of these indication routines to the API at the time that it activates the API. The

- 22 - - SUBSTITUTE SHEET - application may provide either one of these handlers, both handlers or no handler routines at all.

The bus reset handling routine has the following calling convention:

void BusResetHandler(BusResetBlockPtr busResetBlock);

The busResetBlock data structure contains the address of a data structure which describes the bus reset event, as defined in Table II below.

TABLE II typedef struct {

ActivatereqPtr activateReq; /* the session */ QUADLET generation; /* bus generation */ QUADLET numNodes; /* number of nodes on the bus */

TopologyMapPtr topology; /* bus topology */

... other? } BusResetBlock, *BusResetBlockPtr;

The API calls the bus reset handling routine any time that a bus reset occurs on the IEEE 1394 bus while the API is active for the application that provided the bus reset handler routine. When a cluster of resets occurs due to the physical nature of bus connection and disconnection, the handler will be called once. The handler will not be re- entered, but may be called several times in succession. As the result of the bus reset, all asynchronous transactions which were pending at the time of the bus reset will be completed with an error status. Isochronous traffic will resume per the IEEE 1394 specification, and may produce indications during the execution of the bus reset handler. The asynchronous transaction request indication routine has the following calling convention:

void IndicationHandler(IndicationBlockPtr indicationBlockPtr)

- 23 - SUBSTITUTE SHEET The IndicationBlockPtr data structure contains the address of an indication block, defined in Table III below.

TABLE III typedef struct {

ActivateReqPtr activateReq; /* the session */

LocalBufferPtr indicationBuf; /* the info */

} IndicationBlock, * IndicationBlockPtr;

The API calls the indication routine when it receives an asynchronous request subaction that is not handled by either the API itself, or by the IEEE 1394 interface hardware. For each such event, the API calls the indication routine of each application, beginning with the first application to activate the API and provide an indication handler. Each indication handler returns a value to the API to indicate whether or not it handled the indication. When the API receives a status from an indication routine indicating that it handled the indication, then the API does not call any other indication routines for this indication.

The API does handle some request subactions itself. For these transactions, the API does not call any indication handler routine. The API passes all IEEE 1394 transaction information that caused the indication and the additional information necessary for the indication handler routine to generate a response subaction through the API.

The application may contribute buffers to the Indication Handler. This facility allows the application to expand the default set of indication buffers in order to accommodate application specific requirements. A larger set of indication buffers allows more outstanding indications without causing a busy ack signal at the IEEE 1394 interface. The application is not guaranteed to receive a buffer belonging to it when it receives an indication from the API. Furthermore, the API may pass an application indication buffer to another application, if necessary, when reporting an indication. The Current Indication Buffer Count function returns the total count of indication buffers in the indication buffer pool. The returned value is the current count of indication buffers.

- 24 - - SUBSTITUTE SHEET - The Add Indication Buffers function contributes buffers to the indication buffer pool. Buffer elements are described as a LocalBuffer. The caller of this function cedes ownership of the storage represented by this request to the API and must regain ownership prior to disposing of the storage.

STATUS AddIndBuffers(ActivateReqPtr context, BufMgmtBlockPtr bufBlock);

The possible status return values for an AddlndBuffers function are GOOD, signalling that the API has accepted the request and will complete it at a later time, INVALIDCONNECTION, PENDING, signalling that the API has accepted the request, UNSUPPORTEDOP, signalling that the buffers cannot be added on this platform, and UNDEFINEDERROR, signalling that an unexpected error was encountered while attempting to honor the request, even though some data may have been transferred. When a pending value is returned the API calls the AsyncCompletion completion routine when the request is complete. At that time, the status field of the BufMgmtBlock will contain the completion status for the request.

The first parameter of an AddlndBuffers function contains the address of a valid ActivateReq data structure. The second parameter contains the address of a BufMgmtBlock data structure. This data structure describes the buffers, as defined in Table IV below.

TABLE IV typedef struct BufMgmtBlock { BMIdata APIprivate; /* API private */

LocalBufferPtr buffs; /* the buffers to contribute */ void (*AsyncCompletion) (struct BufMgmtBlock *req);

/"completion routine*/ void *UserPtr; /*for use by the completion routine*/ STATUS Status; /"completion status for operation*/

} BufMgmtBlock, * BufMgmtBlockPtr;

The APIprivate field includes private data for management of the request. The LocalBufferPtr field contains descriptors for the buffer(s) to contribute. These buffers can

- 25 - - SUBSTITUTE SHEET - be of any size. The API may use none, part or all of the contributed buffers at its discretion. When the AsyncCompletion field is not equal to null, it contains the address of the routine to call upon completing the operation. The UserPtr field is available for use by the calling completion routine. The API does not modify this field. The Status field contains the status of the requested data transfer operation. The Status field contains status

"pending" until the asynchronous operation is completed. When the completion routine is invoked, the Status field will contain completion status.

The Release Indication Buffers function returns previously added indication buffers to the invoker. Buffer elements are described as a LocalBuffer. The invoker of this function may specify a subset of the buffers added by an AddlndBuffers function request.

When all of the requested buffers are released, the completion routine is invoked.

STATUS RelIndBuffers(ActivateReqPtr context, BufMgmtBlockPtr bufBlock);

The possible status return values for a RellndBuffers function are PENDING, signalling that the API has accepted the request and will complete it at a later time, INVALIDCONNECTION, UNSUPPORTEDOP, signalling that buffers cannot be added on this platform, and UNDEFINEDERROR, signalling that an unexpected error was encountered while attempting to honor the request, even though some data may have been transferred.

The first parameter of a Release Indication Buffer contains the address of a valid activateReq data structure. The second parameter contains the address of a BufMgmtPtr data structure. This data structure describes the buffers, as defined above. When the application requests the API to release a buffer, it must describe that buffer using the same description as when the buffer was first given to the API.

The Bus Management routines perform IEEE 1394 bus management functions. These functions include allocating and deallocating isochronous bus resources and retrieving information about the topology or configuration of the IEEE 1394 bus.

- 26 - SUBSTITUTE SHEET - The MGMTAllocateChannel routine uses the protocols defined in section 8 of the IEEE 1394 standard to allocate a single isochronous channel number. The MGMTAllocateChannel routine calling convention is as follows:

status MGMTAllocatChannel (ActivateReqPtr context, MGMTAllocateChBlockPTr allocateChBlock);

The possible status return values for a MGMTAllocateChannel routine are GOOD, signalling that the channel was successfully allocated, INVALIDCONNECTION, signalling that the context parameter does not contain the address of a currently active connection to the API, PENDING, signalling that the API has accepted the request, CHUNAVAILABLE, signalling that the requested channel number is currently not available, and UNDEFINEDERROR, signalling that an unexpected error was encountered. If a pending value was returned the API calls the MGMTCompletion routine when the allocation request is complete and at that time, the status field will contain the completion status for this request.

The first calling parameter of a MGMTAllocateChannel routine is the address of an active ActivateReq data structure. The second parameter contains the address of a data structure as defined in Table V below.

TABLE V typedef struct MGMTAllocateChBIock {

QUADLET channel; /* channel number to allocate, or all ones */ QUADLET allocateCh; /* actual channel number allocated*/

OCTLET chAvailable; /* bit mask of available channel numbers */ void ("MGMTCompletion) (struct MGMTAllocateChBIock *req);

/* client completion routine */ void "UserPtr; / "for use by the completion routine */ STATUS Status; / "completion status */

} MGMTAllocateChBIock, "MGMTAIIocateChBlockPtr

The channel field contains the channel number to allocate. If the channel number is in the range of 0 to 63, inclusive, then the routine attempts to allocate the specified channel number. If the channel number is equal to all ones, then the routine chooses a

- 27 - - SUBSTITUTE SHEET - channel number to allocate. If the channel field contains any other value, then the routine fills in the chAvailable field and returns the chUnavailable status. Note that this can be used to determine the current value of the channels available bit mask from the currently active Isochronous Resource Manager. The allocatedCh field is filled with the actual allocated channel number, or all ones if a channel was not allocated as a result of calling this routine. The chAvailable field is filled with the current value of the channels_available CSR at the Isochronous Resource Manager. Note that the value in the CSR may change at any time, so the value in this field is only a snapshot and may be different on subsequent calls. If the value in the MGMTCompletion field is not equal to NULL, then this field contains the address of the routine to call upon completion. The

UserPtr field is available for use by the application's completion routine. The API does not modify this field. The Status field contains the completion status for this call. If the application calls this routine asynchronously, this field contains PENDING status until the completion routine is called. The MGMTAllocateBandwidth routine uses the protocols defined in section 8 of the IEEE 1394 standard to allocate isochronous bandwidth. The MGMTAllocateBandwidth routine's calling convention is as follows:

status MGMTAllocateBandwidth (ActivateReqPtr context, MGMTAIIocateChBlockPtr allocateChBlock);

The possible status return values of the MGMTAllocateBandwidth routine are GOOD, signalling that the bandwidth was successfully allocated, INVALIDCONNECTION, signalling that the context parameter does not contain the address of a currently active connection to the API, PENDING, signalling that the API has accepted the request,

BWUNAVAILABLE, signalling that the requested bandwidth is currently not available, and UNDEFINEDERROR, signalling that an unexpected error was encountered. If a pending value was returned the API calls the MGMTCompletion routine when the allocation request is complete and at that time, the status field will contain the completion status for this request.

- 28 - - SUBSTITUTE SHEET - The first calling parameter of a MGMTAllocateBandwidth routine is the address of an active ActivateReq data structure. The second parameter contains the address of a data structure as defined in Table VI below.

TABLE VI typedef struct MGMTAllocateBWBlock {

QUADLET bandwidth; /"bandwidth to allocate, or all ones*/ QUADLET bwAvailable /*actual value of BWAvailable register in IRM*/ void ("MGMTCompletion) (struct MGMTAllocateBWBlock *req);

/"client completion routine*/ void *UserPtr; /"for use by the completion routine*/

STATUS Status; /"completion status*/ } MGMTAllocateBWBlock, *MGMTAllocateBWBlockPtr;

The bandwidth field contains the amount of bandwidth to allocate. If this number is equal to all ones, then the routine fills in the bwAvailable field and returns the BWUNAVAILABLE status. Note that this can be used to determine the current value of the bwavailable field from the currently active Isochronous Resource Manager. The bwAvailable field is filled with the current value of the bandwidth available CSR at the Isochronous Resource Manager. Note that the value in the CSR may change at any time, so the value in this field is only a snapshot and may be different on subsequent calls. If the value in the MGMTCompletion field is not equal to NULL, then it contains the address of the routine to call upon completion. The UserPtr field is available for use by the application's completion routine. The API does not modify this field. The Status field contains the completion status for this call. If the application calls this routine asynchronously, this field contains PENDING status until the completion routine is called. The MGMTDeAllocateChannel routine uses the protocols defined in section 8 of the IEEE 1394 standard to deallocate a single isochronous channel number. The

MGMTDeAllocateChannel routine's calling convention is as follows:

status MGMTDeAllocateChannel (ActivateReqPtr context, MGMTAIIocateChBlockPtr allocateChBlock);

- 29 - - SUBSTITUTE SHEET The routine takes two parameters and returns a status value. The possible status return values for the MGMTDeAllocateChannel are GOOD, signalling that the channel was successfully deallocated, INVALIDCONNECTION, signalling that the context parameter does not contain the address of a currently active connection to the API, PENDING, signalling that the API has accepted the request, CHUNAVAILABLE, signalling that the requested channel number was not allocated, and UNDEFINEDERROR, signalling that an unexpected error was encountered. If a pending value was returned the API calls the MGMTCompletion routine when the allocation request is complete and at that time, the status field will contain the completion status for this request. The first calling parameter of a MGMTDeAllocateChannel routine is the address of an active ActivateReq data structure. The second parameter contains the address of a MGMTAllocateChBIock data structure. This routine deallocates the channel specified in the channel field of that data structure and fills in the chAvailable field with the current value of the channels available bit mask from the currently active isochronous resource manager.

The MGMTDeAllocateBandwidth routine uses the protocols defined in section 8 of the IEEE 1394 standard to deallocate isochronous bandwidth. The MGMTDeAllocateBandwidth routine's calling convention is as follows:

status MGMTDeAllocateBandwidth (ActivateReqPtr context, MGMTAIIocateChBlockPtr allocateChBlock);

The routine takes two parameters and returns a status value. The possible status return values for a MGMTDeAllocateBandwidth routine are GOOD, signalling that the bandwidth was successfully deallocated, INVALIDCONNECTION, signalling that the context parameter does not contain the address of a currently active connection to the API, PENDING, signalling that the API has accepted the request, BWUNAVAILABLE, signalling that completing this request would cause the bandwidth_available register in the isochronous resource manager to become invalid and no action was taken, and UNDEFINEDERROR, signalling that an unexpected error was encountered. If a pending

- 30 - - SUBSTITUTE SHEET - value was returned the API calls the MGMTCompletion routine when the allocation request is complete and at that time, the status field will contain the completion status for this request.

The first calling parameter of a MGMTDeAllocateBandwidth routine is the address of an active ActivateReq data structure. The second parameter contains the address of a

MGMTAllocateBWBlock data structure. This routine deallocates the bandwidth contained in the bandwidth field and fills in the bwAvailable field with the current value of the bandwidth_available register in the currently active isochronous resource manager.

The MGMTBusInfo routine returns information about the node on which the application is running and the connected IEEE 1394 bus. Such information includes the current setting of the PHY gap count, the number of nodes on the connected IEEE 1394 bus, a pointer to the bus topology map and a pointer to the bus speed map, as defined in the IEEE 1394 standard.

The ASYNDataRequest routine generates one or more IEEE 1394 asynchronous read or write transactions in order to transfer data between the application's data buffer and a linear range of addresses in the 64 bit IEEE 1394 address space. The ASYNDataRequest routine has the following calling convention:

STATUS ASYNDataRequest (ActivateReqPtr request, asyncTransportPtr xfrBlock);

The possible status return values for an ASYNDataRequest routine are GOOD, signalling that the API has successfully completed the data transfer request, PENDING, signalling that the API has accepted the request and will complete it at a later time, INVALIDOPCODE, signalling that there is an unknown code in the flags.opCode field, INVALIDCONNECTION, signalling that the activateReqPtr field does not represent an active connection, and UNDEFINEDERROR, signalling that an unexpected error was encountered while attempting to honor the request even though some data may have been transferred.

The first parameter of an ASYNDataRequest routine contains the address of a valid activateReq data structure. The second parameter contains the address of an

- 31 - - SUBSTITUTE SHEET - asyncTransport data structure. This data structure describes the requested data transfer operation, as defined in Table VII below.

TABLE VII typedef struct AsyncTransport {

ASYdata APIprivate; /* API private */ OPTION OPCode 4; /"defines the operation to perform*/

OPTION BusSpeed 4; /"bus speed to use for xfr*/

OPTION Nonlncr 1 ; /"do not increment remote addr*/

BUFSIZE BlockSize /"for block read or write requests - size to use for all block requests 0 means use max for bus speed*/

LocalBuffer AppIBufPtr; /"buf descr for application data*/

RemoteAddr RemoteBufPtr; /*64 bit address on IEEE 1394 bus*/

BUFSIZE Length; /"number of bytes to transfer*/ void ("AsyncCompletion) (struct AsyncTransport* req); /"crnpl routine*/ void *UserPtr; /"for use by the completion routine*/

STATUS Status; /"completion status for operation*/

} AsyncTransport, "AsyncTransportPtr; enum OpCodes {

/"asynch data transfer operations */

BLOCKWRITE , /"transfer data using block write requests*/

BLOCKREAD, /*transfer data using block read requests*/

QUADLETWRITE , /"transfer data using QUADLET write transactions*/ QUADLETREAD , /"transfer data using QUADLET read transactions*/ lock transactions*/

MASKSWAP , /*mask swap lock operation*/ COMPARESWAP , /"compare swap lock operation*/ FETCHADD , /"fetch and add lock operation*/ LITTLEADD , /* little endian fetch/add lock operation*/ BOUNDEDADD /* bounded add lock operation*/ WRAP ADD /*wrap add lock operation*/

};

The ASYdata field includes private data for management of the request. The OpCode field contains a code describing the requested operation which must be one of the values defined in the asyncOpCodes enum. The Nonlncr field when set to one, instructs the routine to transfer all data to the same IEEE 1394 address contained in the

- 32 - SUBSTITUTE SHEET - remoteBufPtr field and when set to zero, instructs the routine to transfer data to an incrementing range of IEEE 1394 addresses, beginning with the address contained in the remoteBufPtr field. The BlockSize field contains the maximum size, in bytes, to use for all block read or write request subactions. A value of zero means to use the maximum request size for the chosen bus speed. The APPLBufPtr field contains the descriptor for the application data buffer. The RemoteBufPtr field contains the 64 bit address of the data buffer in a remote IEEE 1394 node. The Length field contains the number of bytes to transfer which may be less than or equal to the length of the application data buffer. When the AsyncCompletion field is not equal to null, it contains the address of the routine to call upon completing the data transfer. The UserPtr field is available for use by the calling application's completion routine and is not modified by the API. The Status field contains the status of the requested data transfer operation. This field contains status "pending" until the asynchronous operation is complete. When the completion routine is invoked, this field will contain completion status. The ASYNLockRequest routine generates one lock transaction on the IEEE 1394 bus. The ASYNLockRequest routine has the following calling convention:

STATUS ASYNLockRequest (ActivateReqPtr request, AsyncLockBlockPtr lockBIock);

The possible status return values for an ASYNLockRequest routine are GOOD, signalling that the API has successfully performed the lock transaction and the results are contained in the AsyncLockBlock, PENDING, signalling that the API has accepted the request and will complete it at a later time, INVALIDOPCODE, signalling that there is an unknown code in the OpCode field of the AsyncLockBlock data structure,

INVALIDCONNECTION, signalling that the activateReqPtr field does not represent an active connection, and UNDEFINEDERROR, signalling that an unexpected error was encountered while attempting to honor the request even though some data may have been transferred.

- 33 - - SUBSTITUTE SHEET - The first parameter of an ASYNLockRequest routine contains the address of a valid activateReq data structure. The second parameter contains the address of an AsyncLockBlock data structure. This data structure describes the requested data transfer operation, as defined in Table VIII below.

TABLE VIII

typedef struct AsyncLockBlock {

ASYdata APIprivate; /* API private */

OPTION OPCode :4; /"defines the operation to perform*/

OPTION BusSpeed :4; /"bus speed to use for xfr*/ struct { union {

QUADLET Arg32;

OCTLET Arg64;

} ar ; /* 32 or 64 bit lock argument */ union {

QUADLET Data32; OCTLET Data64;

} data; /* 32 or 64 bit lock data */

} ArgData;

RemoteAddr remoteBufPtr; /*64 bit address on IEEE 1394 bus*/ void (* AsyncCompletion) (struct AsyncLockBlock *req); /"completion routine*/ void *UserPtr; /*for use by the completion routine*/

STATUS Status; /"completion status for operation*/

} AsyncLockBlock, "AsyncLockBlockPtr;

The APIPrivate field contains private data for management of the request. The OpCode field contains a code describing the requested operation and must contain one of the values defined in the asyncOpCodes enum. The ArgData struct field contains the argument and data for this lock transaction. The remoteBufPtr field contains the 64 bit destination address on the IEEE 1394 bus. When the AsyncCompletion field is not equal to null, it contains the address of the routine to call upon completing the lock transaction. The UserPtr field is available for use by the calling application's completion routine and is not modified by the API. The Status field contains the status of the requested data transfer operation. This field contains status "pending" until the asynchronous operation is

- 34 - - SUBSTITUTE SHEET - complete. When the completion routine is invoked, this field will contain completion status.

The Isochronous Resource Management routines allocate and deallocate system resources needed to transfer isochronous data over the IEEE 1394 interface into or out of application data buffers. Allocation and deallocation is necessary to avoid conflicts among multiple potential applications of isochronous data in the system. Whenever isochronous data flows over the IEEE 1394 bus, there is an entity on the IEEE 1394 bus which owns the necessary bus resources, namely channel numbers and bandwidth. Each application which uses isochronous data has its own set of rules for who must allocate and deallocate these resources and when. The bus management routines in the API permit an application to allocate these resources according to the requirements of the IEEE 1394 standard. Note that the routines in this section do not allocate IEEE 1394 bus resources; these routines only allocate system level resources necessary to transfer isochronous data into or out of application data buffers. These resources include a DMA channel and the system resources to sustain it, such as the low level interrupt handler and dispatcher.

The ISOCHOpen routine opens and initializes an isochronous port. An isochronous port is a collection of hardware and software resources in the local node on which the application of the API and the API are running. This collection of resources constitutes everything in the local node which is needed to transfer a single stream of isochronous data into or out of the node. This collection of resources does not include the IEEE 1394 bus resources which the application must allocate separately, according to the bus management rules defined in section 8 of the IEEE 1394 standard, and the governing application rules and requirements. The routines which enable an application of the API to allocate and deallocate IEEE 1394 bus resources have been described above. The port open routine has the following calling convention:

status ISOCHOpen (ActivateReqPtr connectionPtr, ISOCHOpenBlockPtr openBlock);

The first parameter of the port open routine contains the address of a valid activateReq data structure. The second parameter contains the address of an

- 35 - - SUBSTITUTE SHEET - ISOCHOpenBlock data structure. Upon successful completion of this routine, the application uses this ISOCHOpenBlock data structure to reference this opened isochronous port on future calls to the API which affect this port.

The possible status return values for a port open routine are GOOD, signalling that an open request was completed successfully, PENDING, signalling that the API has accepted the request and will complete it at a later time, NORESOURCES, signalling that an isochronous port or other necessary resource is not currently available and the request is denied, INVALIDREQUEST, signalling that the requested bus speed is not supported, INVALIDCONNECTION, signalling that the ActivateReqPtr field does not represent an active API connection, and UNDEFINEDERROR, signalling that the request could not be honored, but the error could not be identified.

The calling parameter of a port open routine contains the address of an ISOCHOpenBlock data structure. This data structure describes the request, as defined in Table IX below.

TABLE IX

typedef struct {

ISOIink APIprivate; /* API private */ OPTION Direction :2; /*source/sink*/

OPTION BusSpeed :4; /"requested bus speed*/ void (*AsyncCompletion))struct ISOCHOpenBlock* req); /"compl routine*/ void *UserPtr; /*for use by the completion routine*/

STATUS Status; /"completion status for operation*/ } ISOCHOpenBlock, *ISOCHOpenBlockPtr; enum Direction {

INPUT , /*specifies input to application to data buffer*/

OUTPUT ^specifies output from application data buffer*/

};

The direction field indicates the direction of the isochronous data transfer. When the AsyncCompletion field is not equal to null, it contains the address of the routine to call upon completion. The UserPtr field is available for use by the calling application's

- 36 - SUBSTITUTE SHEET - completion routine and is not modified by the API. The Status field contains the status of the requested data transfer operation. This field contains status "pending" until the asynchronous data transfer operation is complete. When the completion routine is invoked, this field will contain completion status. When the application is finished with the isochronous port, it passes the isochPortPtr to the ISOCHClose routine.

The ISOCHClose routine closes an isochronous port that was previously opened using the ISOCHOpen routine. This routine has the following calling convention:

STATUS ISOCHClose (ActivateReqPtr connectionPtr, ISOCHOpenBlockPtr openBlock);

This routine takes a single parameter and returns a status value. This routine executes asynchronously, and invokes the completion routine defined in the ISOCHOpenBlock upon completion. The possible status return values of an ISOCHClose routine are GOOD, signalling that the operation was completed successfully, PENDING, signalling that the

API has accepted the request and will complete it at a later time, INVALIDCONNECTION, signalling that the connectionPtr does not represent an active API connection or openBlock does not represent a currently open isochronous port, and UNDEFINEDERROR, signalling that the request could not be honored, but the error could not be identified.

The first calling parameter of ISOCHClose routine is the address of a valid activateReq data structure. The second calling parameter is the address of the ISOCHOpenBlock used to open the port with the ISOCHOpen routine.

The isochronous data control routine controls a stream of isochronous data into or out of application data buffers. For applications which listen to isochronous data, these control routines only affect the flow of isochronous data into the system; they do not affect the isochronous data on the IEEE 1394 bus itself. For applications that transmit isochronous data from application data buffers, these control routines also affect the flow of isochronous data on the IEEE 1394 bus. The ISOCHControl routine has the following calling convention:

- 37 - - SUBSTITUTE SHEET - STATUS ISOCHControl (ISOCHOpenBlockPtr opensBlockPtr, ISOCHCtlBlockPtr ctlReqPtr);

The possible status return values of an ISOCHControl routine are GOOD, signalling that the operation was completed successfully, INVALIDCONNECTION, signalling that the openBlock field does not represent an active isochronous port, PENDING, signalling that the operation is currently pending, INVALIDOPCODE, signalling that the opCode field of the ISOCHControlBlock contained an illegal value, UNSUPPORTEDOP, signalling that the operation is not supported due to a limitation of the IEEE 1394 interface hardware or a limitation of the software or operating environment, and UNDEFINEDERROR, signalling that the operation could not be performed, but the error could not be identified. If a pending value is returned, at the time the callback routine is called, the status field in the ISOCHControlBlock will contain the completion status of the control request.

The first parameter of an ISOCHControl routine contains the address of a valid ISOCHOpenBlock data structure. The second parameter contains the address of an

ISOCHCtlBlock data structure. This data structure describes the requested control operation, as defined in Table X below.

TABLE X

typedef struct ISOCHCtlBlock { ISOIink APIPrivate; /* API private */ OPTION IsoOpCode 4; /"operation to perform*/ OPTION IsoEvent 4; /"trigger event for start stop*/ OPTION Sy 4; /*sy field value, if needed*/ OPTION Tag 2; /"tag value to use when starting*/ OPTION Channel 6; /"channel value to use when starting*/ BusTime Time; /"specifies when an event should occur*/ void (*AsyncCompletion)(struct ISOCHCtlBlock *req);

/"completion routine*/ void "UserPtr; /"for use by the completion routine*/ STATUS Status /"completion status for operation*/ } ISOCHCtlBlock, "ISOCHCtlBlockPtr;

- 38 - SUBSTITUTE SHEET The APIPrivate field contains private storage used by the API to manage this request. The IsoOpCode field contains a value from the IsoOpCode enum which describes the requested control operation. The IsoEvent field specifies the trigger event on which to perform the requested operation. If the IsoEvent field contains the value "SYFIELD," the Sy field contains the value in the sy field that will cause the isochronous channel to start or stop. If the IsoOpCode field contains the value "START," the value from the Tag field is used for the tag value in the isochronous data block packet header. If the IsoOpCode field contains the value "START," the value from the Channel field is used for the channel value in the isochronous data block packet header. If the IsoEvent field contains the value "TIME," the Time field contains the bus time on which the requested action is to take place. When the AysncCompletion field is not equal to null, it contains the address of the routine to call upon completion of the data transfer. The UserPtr field is available for use by the calling application's completion routine. The API does not modify this field. The Status field contains the status of the requested data transfer operation. This field contains status "pending" until the asynchronous data transfer operation is complete. When the completion routine is invoked, this field will contain completion status.

The isochronous attach routine passes application data buffer descriptors to the API software. The application may call this routine at any time to make buffers available to the IEEE 1394 interface hardware and the associated low layers of software. The calling convention for this routine is as follows:

status ISOCHAttach(ISOCHOpenBlockPtr openBlock, ISOCHAppendBlockPtr append);

The possible status return values for an isochronous attach routine are GOOD, signalling that the operation was completed and the application data buffers are accepted,

INVALIDCONNECTION, signalling that the openBlock field does not represent an active isochronous port, UNSUPPORTEDOP, signalling that the resynchronization event specified in a buffer descriptor is not supported which may be due to hardware implementation, software implementation or other system environment limitation, INVALIDREQUEST, signalling that an attempt to append a circular list of application

- 39 - - SUBSTITUTE SHEET - data buffers while isochronous data is flowing, or to append new buffers to an existing circular list while isochronous data is flowing, and UNDEFINEDERROR, signalling that the operation could not be completed, but an actual error could not be identified.

The first parameter of an ISOCHAttach routine contains the address of a valid ISOCHOpenBlock data structure. The second parameter contains the address of an ISOCHAppendBlock data structure. This data structure describes the application data buffer list as defined in Table XI below.

TABLE XI

typedef struct ISOCHAppendBlock {

ISOIink APIPrivate; /* API private */ isochBufferPtr IsochBuffList; /"start of list of isoch buffers*/ void ("AsyncCompletion) (struct ISOCHAppendBlock *req); /"completion routine*/ void *UserPtr; /"for use by the completion routine*/

STATUS status; /"completion status for operation*/

} ISOCHAppendBlock, *ISOCHAppendBlockPtr;

The APIPrivate field contains private storage used by the API to manage the request. The IsochBuffList field contains the address of the first isochronous buffer of a list to append to the specified port. If the current buffer list or the buffer list to append is circular then the operation can only be performed when the port is stopped and the append operation will replace any previously appended buffers. A non-circular list of buffers may be appended to an existing non-circular list of buffers at any time. When the AsyncCompletion field is not equal to null, it contains the address of the routine to call upon completion. The UserPtr field is available for use by the calling application's completion routine and is not modified by the API. The Status field contains the status of the requested operation. This field contains status "PENDING" until the asynchronous data operation is complete. When the completion routine is invoked, this field will contain completion status.

- 40 - - SUBSTITUTE SHEET The IsochBuffList field contains the address of an isochBuffer. The isochBuffer data structure describes a single application data buffer. Typically, isochBuffer data structures exist in a doubly linked list. This data structure is defined in Table XII below.

TABLE XII

typedef struct isochBuffer { struct isochBuffer "Next; /"ptr to next block*/ struct isochBuffer "Previous; /"ptr to pre v. Block*/

OPTION Circular : 1 /*list is circular*/

OPTION ResynchEvent :4 /"optional resynch event*/

OPTION Sy :4 /*sy field value*/ busTime Time; /"used with resynch event*/ localBufPtr ApplBufPtr; /*ptr to application data*/ void (*IsochCompletion)(struct isochBuffer *buf);

/"completion routine*/ void "UserPtr /"for use by the completion routine*/

} isochBuffer, "isochBufferPtr;

The Next field contains the address of the next buffer in the list. The Previous field contains the address of the previous buffer in the list. The Circular field indicates that the complete set of buffers is circular. The ResynchEvent field contains an optional resynchronization event. A value of IMMEDIATE in this field indicates no resynchronization. When the ResynchEvent field contains a value of "SYFIELD," the Sy field contains the sy value to pause for before transferring data into or out of this application data buffer. When the ResynchEvent field contains a value of "TIME," the Time field contains the bus time to wait for before transferring data into or out of this application data buffer. The ApplBufPtr field contains the address of the application data buffer. This address may be logical, physical, scattered or contiguous, depending on the capabilities of the operating environment. The IsochCompletion field contains the address of the completion routine. If the value in this field is not null, then this routine is called when data transfer for this buffer is complete. The UserPtr field is available for use by the calling application's completion routine and is not modified by the API.

- 41 - SUBSTITUTE SHEET The ISOCHDetach routine retrieves application data buffers from the API and returns ownership to the application. The application may call this routine at any time to detach buffers from the IEEE 1394 interface hardware and the associated low layers of software. The requested buffers are detached when the completion routine is invoked. The calling convention for this routine is as follows:

status ISOCHDetach(ISOCHOpenBlockPtr openBlock, ISOCHAppendBlockPtr unhook);

The possible status return values of an ISOCHDetach routine are GOOD, signalling that the operation was completed and the application data buffers are now detached,

PENDING, signalling that the operation is currently pending, INVALIDCONNECTION, signalling that the openBlock field does not represent an active isochronous port, INVALIDREQUEST, signalling that the buffers described in the ISOCHAppendBlock were not found, or they are not owned by the isochronous port described in the ISOCHOpenBlock structure, and UNDEFINEDERROR, signalling that the operation could not be completed, but an actual error could not be identified. If a pending value is returned, at the time the callback routine is called, the status field in the ISOCHAppendBlock will contain the completion status of the detach operation.

The first parameter of an ISOCHDetach routine contains the address of a valid ISOCHOpenBlock data structure. The second parameter contains the address of an

ISOCHAppendBlock data structure. This data structure describes the application data buffer list as defined above.

The applications programming interface of the present invention provides an interface to an application and allows the application to transfer data over a bus structure in both isochronous and asynchronous data formats. The applications programming interface supports transferring asynchronous data over the bus structure 28 during an isochronous data transfer. While the isochronous data is being transferred over the bus structure 28, the asynchronous data can be used to fill in the gaps. The IEEE 1394 standard specifies a worst case jitter for the isochronous data, thereby specifying a bounded latency for a packet of isochronous data. The API 20 therefore ensures that the

- 42 - - SUBSTITUTE SHEET - packets of isochronous data are transferred during their appropriate time period. However, in the gaps between the packets of isochronous data, asynchronous packets of data are transferred.

In contrast to systems of the prior art, the API of the present invention is capable of automating the transfer of asynchronous data by controlling an automatic transaction generator 38 which automatically generates the transactions necessary to complete an asynchronous data transfer over the memory-mapped bus structure 28. During an isochronous data transfer, a linked list of buffer descriptors, each representing a corresponding buffer, is maintained for transferring the data to or from the application over the bus structure 28. Each buffer can include a callback routine and a resynchronization event.

The present invention has been described in terms of specific embodiments incorporating details to facilitate the understanding of the principles of construction and operation of the invention. Such reference herein to specific embodiments and details thereof is not intended to limit the scope of the claims appended hereto. It will be apparent to those skilled in the art that modifications may be made in the embodiment chosen for illustration without departing from the spirit and scope of the invention.

- 43 - SUBSTITUTE SHEET

Claims

C L A I M SWe Claim:

1. An interface between an application and a bus structure for controlling data transfers to and from the application over the bus structure comprising: a. means for controlling asynchronous data transfers between the application and the bus structure for providing a memory-mapped interface to the application; and b. means for controlling isochronous data transfers between the application and the bus structure including a linked list of buffers.

2. The interface as claimed in claim 1 wherein the means for controlling asynchronous data transfers transmits or receives the data in packets each having a header including a destination identifier, an address and a length value.

3. The interface as claimed in claim 2 further comprising an automated hardware interface coupled to the means for controlling asynchronous data transfers and to the bus structure for automatically generating requests necessary to complete an asynchronous transfer of data across the bus structure.

4. The interface as claimed in claim 3 wherein the bus structure is an IEEE 1394 standard bus structure.

5. The interface as claimed in claim 1 wherein each of the linked buffers includes a forward pointer and a backward pointer.

6. The interface as claimed in claim 5 wherein each of the linked buffers include a callback routine for calling the application at a predetermined point during a data transfer operation.

- 44 - - SUBSTITUTE SHEET -

7. The interface as claimed in claim 6 wherein each of the linked buffers further include a resynchronization event for resynchronizing the application to the data transfer.

8. The interface as claimed in claim 7 further comprising means to add buffers to the linked list.

9. The interface as claimed in claim 8 further comprising means to remove buffers from the linked list.

10. An interface between an application and a bus structure for controlling data transfer operations between the application and the bus structure comprising: a. an applications interface for communicating with the application and receiving requests from and sending instructions to the application; b. an automatic transaction generator for automatically generating transactions necessary to complete an asynchronous data transfer operation between a buffer assigned by the application and the bus structure; c. means for controlling an isochronous data transfer including a linked list of buffers provided by the application with which the data transfer is to be completed; and d. a bus interface for providing a physical interface to the bus structure for sending and receiving data packets over the bus structure.

11. The interface as claimed in claim 10 wherein the linked list of buffers includes a forward pointer and a backward pointer for each buffer.

12. The interface as claimed in claim 1 1 wherein each of the buffers include a callback routine which is activated to call the application at a predetermined point during a data transfer operation.

- 45 - - SUBSTITUTE SHEET -

13. The interface as claimed in claim 12 wherein each of the buffers include a resynchronization event which is activated to resynchronize the application to a predetermined point during a data transfer operation.

14. The interface as claimed in claim 13 further comprising means to add buffers to the linked list.

15. The interface as claimed in claim 14 further comprising means to remove buffers from the linked list.

16. The interface as claimed in claim 10 wherein the automatic transaction generator is implemented in a hardware logic circuit.

17. The interface as claimed in claim 10 wherein the automatic transaction generator is implemented in software.

18. A method of providing a memory-mapped interface to an application and managing asynchronous data transfer operations between the application and a bus structure comprising the steps of: a. receiving a request for a data transfer operation from the application; b. providing an instruction to an automatic transaction generator for automatically generating transactions necessary to complete the data transfer operation across the bus structure; and c. notifying the application when the data transfer operation is complete.

19. The method as claimed in claim 18 wherein the request includes an address for an application buffer, a starting address in an address space of the bus structure, a length of data to be transferred and a direction of the transfer.

- 46 - SUBSTITUTE SHEET

20. The method as claimed in claim 19 wherein the starting address is a 64 bit address.

21. The method as claimed in claim 20 wherein the bus structure is an IEEE 1394 standard bus structure.

22. The method as claimed in claim 18 wherein the automatic transaction generator is implemented in a hardware logic circuit.

23. The method as claimed in claim 18 wherein the automatic transaction generator is implemented in software.

24. An interface between an application and a bus structure for managing data transfer operations between the application and bus structure comprising: a. means for communicating with the application for receiving requests from and sending instructions to the application; b. an automatic transaction generator for automatically generating transactions necessary to complete an asynchronous data transfer operation between a buffer of the application and the bus structure; and c. means for monitoring and communicating over the bus structure for sending and receiving data packets from the bus structure.

25. The interface as claimed in claim 24 wherein a data transfer operation is initiated when the application sends a request to the means for communicating.

26. The interface as claimed in claim 25 wherein the request includes an address for the buffer of the application, a starting address across the bus structure, a length of data to be transferred and a direction of the transfer.

- 47 - - SUBSTITUTE SHEET

27. The interface as claimed in claim 26 wherein the bus structure is an IEEE 1394 standard bus structure.

28. The interface as claimed in claim 24 further comprising means for controlling an isochronous data transfer operation between the application and the bus structure.

29. The interface as claimed in claim 28 wherein the means for controlling an isochronous data transfer operation maintains a linked list of buffers provided by the application with which the data transfer is to be completed.

30. The interface as claimed in claim 29 wherein the linked list includes a forward pointer and a backward pointer for each buffer in the list.

31. The interface as claimed in claim 29 wherein each of the buffers included in the list include a callback routine which is activated to call the application at a predetermined point during a data transfer operation.

32. The interface as claimed in claim 31 wherein each of the buffers included in the list include a resynchronization event which is activated to resynchronize the application to a predetermined point during a data transfer operation.

33. An interface between an application and an IEEE 1394 standard bus structure for managing data transfer operations between the application and the bus structure comprising: a. an applications interface for communicating with the application and receiving requests from and sending instructions to the application; b. an automatic transaction generator for automatically generating transactions necessary to complete an asynchronous data transfer operation between a buffer assigned by the application and the bus structure; and

- 48 - - SUBSTITUTE SHEET - c. a bus interface circuit for providing a physical interface to the bus structure for sending and receiving data packets over the bus structure.

34. The interface as claimed in claim 33 wherein a data transfer operation is initiated when the application sends a request to the applications interface.

35. The interface as claimed in claim 34 wherein the request includes an address for the buffer of the application, a starting address across the bus structure, a length of data to be transferred and a direction of the transfer.

36. The interface as claimed in claim 35 wherein the bus structure is an IEEE 1394 standard bus structure.

37. The interface as claimed in claim 33 wherein the automatic transaction generator is implemented in a hardware logic circuit.

38. The interface as claimed in claim 33 wherein the automatic transaction generator is implemented in software.

39. A method of providing an interface to an application and managing isochronous data transfer operations between the application and a bus structure comprising the steps of: a. receiving a request for a data transfer operation from the application; and b. managing a linked list of buffers provided by the application for completing the data transfer operation, wherein data is transferred between one buffer and the bus structure while the application processes another buffer.

40. The method as claimed in claim 39 wherein during an isochronous send operation the application processes a buffer by loading data into the buffer.

- 49 - - SUBSTITUTE SHEET -

41. The method as claimed in claim 39 wherein during an isochronous receive operation the application processes a buffer by obtaining data from the buffer.

42. The method as claimed in claim 39 wherein the linked list includes a forward pointer and a backward pointer for each buffer within the linked list.

43. The method as claimed in claim 39 wherein each of the buffers in the linked list include capability for a callback routine which is activated to call the application at a predetermined point during a data transfer operation.

44. The method as claimed in claim 39 wherein each of the buffers included in the list include capability for a resynchronization event which is activated to resynchronize the application to a predetermined point during a data transfer operation.

45. The method as claimed in claim 42 further comprising means to add buffers to the linked list.

46. The method as claimed in claim 45 further comprising means to remove buffers from the linked list.

47. An interface between an application and a bus structure for managing data transfer operations between the application and the bus structure comprising: a. means for communicating with the application for receiving requests from and sending instructions to the application; b. means for controlling isochronous data transfer operations between the application and the bus structure including managing a linked list of buffers; and c. means for monitoring and communicating over the bus structure for sending and receiving data packets from the bus structure.

- 50 - - SUBSTITUTE SHEET

48. The interface as claimed in claim 47 wherein the linked list of buffers are provided by the application.

49. The interface as claimed in claim 48 wherein the linked list includes a forward pointer for each buffer within the linked list.

50. The interface as claimed in claim 49 wherein the linked list further includes a backward pointer for each buffer within the linked list thereby making the list circular.

51. The interface as claimed in claim 47 wherein each of the buffers in the linked list include capability for a callback routine which is activated to call the application at a predetermined point during a data transfer operation.

52. The interface as claimed in claim 47 wherein each of the buffers included in the list include capability for a resynchronization event which is activated to resynchronize the application to a predetermined point during a data transfer operation.

53. The interface as claimed in claim 47 further comprising an automatic transaction generator for automatically generating transactions necessary to complete an asynchronous data transfer operation between a buffer assigned by the application and the bus structure.

54. The interface as claimed in claim 53 wherein the bus structure is an IEEE 1394 standard bus structure.

55. The interface as claimed in claim 48 wherein the linked list includes a forward pointer and a backward pointer for each buffer within the linked list thereby making the list linear.

- 51 - SUBSTITUTE SHEET -

56. The interface as claimed in claim 55 further comprising means to add buffers to the linked list.

57. The interface as claimed in claim 56 further comprising means to remove buffers from the linked list.

58. An interface between an application and an IEEE 1394 standard bus structure for managing data transfer operations between the application and the bus structure comprising: a. an applications interface for communicating with the application and receiving requests from and sending instructions to the application; b. means for controlling isochronous data transfer operations between the application and the bus structure including a linked list of buffers provided by the application; and c. a bus interface circuit for providing a physical interface to the bus structure for sending and receiving data packets over the bus structure.

59. The interface as claimed in claim 58 wherein the linked list includes a forward pointer and a backward pointer for each buffer within the linked list thereby making the list circular.

60. The interface as claimed in claim 59 wherein each of the buffers in the linked list include capability for a callback routine which is activated to call the application at a predetermined point during a data transfer operation.

61. The interface as claimed in claim 60 wherein each of the buffers included in the list include capability for a resynchronization event which is activated to resynchronize the application to a predetermined point during a data transfer operation.

- 52 - - SUBSTITUTE SHEET -

62. The interface as claimed in claim 58 wherein the linked list includes a forward pointer and a backward pointer for each buffer within the linked list thereby making the list linear.

63. The interface as claimed in claim 62 further comprising means to add buffers to the linked list.

64. The interface as claimed in claim 63 further comprising means to remove buffers from the linked list.

- 53 - - SUBSTITUTE SHEET -