WO1999009467A2

WO1999009467A2 - A transient datastream-processing buffer memory organization with software management adapted for multilevel housekeeping

Info

Publication number: WO1999009467A2
Application number: PCT/IB1998/001143
Authority: WO
Inventors: Paulus Thomas Arnoldus Thijssen
Original assignee: Koninklijke Philips Electronics N.V.; Philips Ab
Priority date: 1997-08-20
Filing date: 1998-07-27
Publication date: 1999-02-25
Also published as: TW432321B; KR100594953B1; WO1999009467A3; JP2001505752A; KR20000068786A; US6230163B1; EP0932862A2

Abstract

A datastream-processing buffer memory organization has a datastream input, storage for transiently storing the datastream, accesses the buffer memory, and means for memory housekeeping. In particular, the management means is executed in software.

Description

A transient datastream-processing buffer memory organization with software management adapted for multilevel housekeeping.

BACKGROUND OF THE INVENTION

The invention relates to a datastream-processing buffer memory organization as recited in the preamble of Claim 1. US patents 4,314,361, 4,236,225, and 4,222,102 respectively describe FIFO organizations with fixed input and variable output, with variable input and fixed output, and with both variable input and variable output. For effecting such elementary FIFO functionality the references use exclusively hardware-based housekeeping. The present inventor has experienced a need for an enhanced and more flexible functionality, in particular to allow more complicated processing modes. Moreover, moving of information from one physical location to another should be kept as restricted as possible, in view of limited transfer capability viz a viz ever larger datastream volumes, so that no undue delays will be incurred. This applies even more when the same data items are used several times for respective functions. A final argument against physical movements is that power consumption for physical data transport should be kept low.

SUMMARY TO THE INVENTION

In consequence, amongst other things, it is an object of the present invention to allow the mapping of various logical buffers on shared physical memory space, whilst dimishing physical data transfer to an extremely low amount. Now therefore, according to one of its aspects the invention is characterized as recited in the characterizing part of Claim 1.

Advantageously, the memory management is arranged for multilevel housekeeping. The definition of the data format often has specified various successive levels. Through assigning the housekeeping with respect to a particular format level to an associated housekeeping level, flexibility is attained in an easy manner. Further advantageous aspects of the invention are recited in dependent Claims.

BRIEF DESCRIPTION OF THE DRAWING

These and further aspects and advantages of the invention will be discussed more in detail hereinafter with reference to the disclosure of preferred embodiments, and in particular with reference to the appended Figures that show:

Figure 1 , exemplary hardware for thereon mapping the organization; Figure 2, an example of a data flow organization; Figure 3, an exemplary sector header format; Figure 4, an exemplary user data format of an audio sector.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

Figure 1 shows an exemplary hardware embodiment for thereon mapping the organization of the invention. The processor is centred around bus facility 80 that has sufficient capacity for the application and has an appropriate width for transferring data, addresses, and control signals between the various subsystems connected thereto. I/O subsystem 88 interfaces to the external world through interconnection 90. Processing facility 82 executes necessary processing operations. RAM memory 92 stores the data proper according to an organization that will be disclosed more in detail with reference to Figure 2 hereinafter. Memory Management Unit 84 executes various housekeeping tasks with respect to memory 92. ROM memory 86 may allow storage of the appropriate software, inasfar as this is not stored in processing facility 82.

Figure 2 shows an example of a data flow organization. Herein, serial data are received on datastream input 20 from an appropriate source, such as an optical high volume disc or a remote broadcast or line source. Multiple sources may be connectable in parallel or in alternation. The width of the data stream may be defined in appropriate manner. In the embodiment hereinafter, the data source is a DVD or VCD disc that stores digital video complete with associated digital audio and also various extra data to enable further functionalities to the system and to the user. Module 22 reads the data and stores it in buffer 24. In particular, this module has a DMA feature. Here, the storage is physical, inasmuch as the data is received from the external world, such as on connection 90 in Figure 1. Subsequent writing of the data to other buffers is generally executed through logical assigning of the data as will be discussed hereinafter. If applicable, read-modify-write memory cycles can be executed, in which case there is physical data transport between processing facility and memory. Furthermore, module 22 may perform error checking on the data, and if necessary and feasible, error correction. Upon finding or reaching an appropriate level of correctness, user data are presented to decryption control module 51. These user data exclude sector headers and error checking information ECC. Next to decryption control module 51, the corrected data are assigned to buffer 30, that is used for a pre-navigation feature. This allows a user to navigate through the data in a fast manner, such as fast forward or fast reverse, through using non-encrypted data, and without necessitating any further operations on the data. In decryption control module 51, the data are decrypted if necessary, through an appropriate decryption key. Next, the decrypted data are assigned to buffer 26. This buffer receives data at a uniform high rate, and outputs data at a variable rate, in accordance with various source compactions that may be present. Module 28 accesses the data in buffer 26, and dispatches the readout data to further buffers 34, 38, 42, 48, 49, the first three of which interface to modules 36, 40, 44, respectively.

In the embodiment, module 40 is an MPEG video decoder that requires a packetized elementary stream PES as input. Further, module 36 is an MPEG audio decoder that requires an elementary stream as input. Module 44 is a subpicture decoder. Buffer 48 allows similar user navigation as buffer 30, but in a more precise manner such as through so- called seamless jumping, whilst also taking into account various presentation control modes, such as information that has been copy-protected. Buffer 49 allows storage of subcode items that can be used for various overlaying functionality types, such a timecode display and further user navigation.

The housekeeping with respect to buffers 34, 38, is such that only the necessary parts of the information are read by modules 36, 40, 44. A final buffer 46 is fed by video decoder 40, for buffering user data such as closed-caption information. The various buffer sizes will be specified hereinafter. These sizes are based on the organization of the related packetizing of the datastream. For example, the amount of video information is relatively large with respect to the amount of audio information, whereas the amount of subcode information is still much less.

Various one of the modules shown, not necessary exclusively on the rightmost or lowest level, access the stored data for outputting the result to a user or user device, which has been shown by respective arrows. Such device may be a television monitor for video, an audio presentation device for audio, and similarly for still further streams of user information data. Another example would be a further control device not shown such as one that may serve for effecting further types of user navigation.

Figure 3 shows an exemplary sector format; this pertains to the format as relevant on interface 20. As shown, a sector contains a sector header, that delimits the sector, indicates a time value, an identifier, and possibly other parameters that may be used for subsequently organizing the storage proper. Part 54 contains error protection code, such as the redundancy symbols of a Reed-Solomon code that can be used for correcting a percentage of the symbols that have been received in an incorrect manner. After executing the correction, part 54 generally is of no use for further processing. Generally, the header and ECC information are sequestered and not used in subsequent accessing.

Figure 4 shows an exemplary user data format of an audio sector. Here, the sector contains two packets of audio data, that each comprise a relevant packet header information, and two parts of the relevant audio data. The number of audio parts may be different, and their sizes may be non-uniform. The function of the headers on this level may be variable. Sometimes, the header information is sequestered and not used in subsequent accessing. Other decoder types may use the header for easier delimitation of the information.

Further legends of Figure 2 are as follows:

SBM: Stream Buffer Manager

SP : Sector Processor

DCRYP: DeCRYPtion Control

DMX : DeMultipleXer

AUD : AUDio decoder

VID : VIDeo decoder

SUBP: SUBPicture decoder

Capacity of a DVD disc: 14 GByte Input data rate: 11 Mb it/ sec Buffer size indications: SBM ERRCHECK 10 KByte SBM_VBR 500 KByte SBM_PRENAV 20 KByte SBM_AUDIO 20 KByte SBM VIDEO 250 KByte SBM_SUBPIC 64 KByte SBM_POSTNAV 20 KByte SBM_SUBCODE 1 KByte SBM_VID_USER_DATA 5 KByte

The SBM controls all buffer accesses in a manner that is invisible for producer as well as for consumer processes, and in particular controls:

• a pool of buffer units

• for each unit an administration of the number of references to it from any buffer

• a linked list indicating the free buffer units

• for each buffer a linked list of buffer units plus additional administration for the applicable range(s) inside a buffer unit.

All buffers are assumed to be FIFO's, with an additional non-destructive 'peek' facility; in other applications, a different organization such as random access may apply. In particular, the scenario is given as follows in the form of a list, as an alternative of a standard flow-chart: • data comes in on the left-hand side of the pictures in chunks of N bytes (sector size);

• the size of the units is also N bytes

• SP requests SBM to write N bytes to SBM_ERRCHECK

• SBM takes the first unit from the free list and sets up DMA to copy the data in

• SBM sets the number of references to the unit to " 1 " • SBM adds the unit to the list of buffer SBM ERRCHECK with range indication [0,N); generally, the lower buffer boundary is included, the upper one is not

• if after M new sectors no error has been indicated by the hardware, SP assumes the sector to be correct

• SP reads X bytes (sector header size) from SBMJΞRRCHECK • SBM updates the range indication to [X,N)

• via a 'peek' read, SBM determines if it is a NAV pack

• SP requests SBM to copy N-X-Y bytes from SBM_ERRCHEGK to SBM_VBR (Y equals the ECC size)

• if it is a NAV pack, SP also requests SBM to copy the (N-X-Y) bytes to SBM_PRENAV • SBM adds the unit to the list of SBM VBR with range indication [X, N-Y) and increments the number of references to the unit

• if it is a NAV pack, SBM also adds the unit to the list of SBM_PRENAV with range indication [X,N-Y) and increments the number of references to the unit again

• SBM updates the range indication in SBMJERRCHECK to [N-Y,N) • SP requests SBM to read Y bytes from SBM ERRCHECK

• SBM updates the range indication in SBM_ERRCHECK to [N,N)

• this means that there is no data in the unit for this buffer, so SBM removes the unit from the list of SBM_ERRCHECK and decrements the number of references to the unit • DMX determines the packet type and size S by 'peeking' SBM VBR

• in case of a NAV pack, DMX requests SBM to copy S bytes from SBMJVBR to SBM POSTNAV

• SBM adds the unit to the list of SBM_POSTNAV with range indication [X,X + S) and increases the number of references to the unit • SBM updates the range indication of the unit in SBMJVBR to [X+S,N-Y)

• the application wants to read B bytes from SBM_POSTNAV

• SBM updates the range indication of the unit in SBM_POSTNAV to [X+B,X+S).

Suppose an audio decoder requires a PES stream: • when DMX discovers audio packets in SBM_VBR, it can request SBM to copy them completely to SBM_AUDIO

• consequently, SBM will add the corresponding unit to the list of SBM_AUDIO and set the range indication to [X,N-Y), where X and Y correspond to the sector header and ECC sizes as indicated in Figure 3 • note that the unit is also removed from the list of SBM_VBR

Suppose an audio decoder requires an elementary stream:

• suppose that the contents of the first unit in the list of SBM_VBR are as presented in Figure 4 • the range indication of this unit is [X,N-Y)

• by 'peeking', DMX discovers an audio packet in SBM_VBR and determines packet header size Ql and packet size SI

• DMX requests SBM to read Ql bytes from SBM VBR

• SBM updates the range indication of the unit in SBMJVBR to [X+Q1,N-Y) • DMX requests SBM to copy Sl-Ql bytes to SBM_AUDIO

• SBM adds the unit to the list of SBM_AUDIO and increments the number of references to the unit

• SBM sets the range indication in the SBM_AUDIO buffer to [X+Ql .X+Sl)

• SBMi sets the range indication in the SBMJVBR buffer to [X+S1.N-Y) • by 'peeking', DMX discovers an audio packet in SBMJVBR and determines packet header size Q2 and packet size S2

• DMX requests SBM to read Q2 bytes from SBMJVBR

• SBM updates the range indication of the unit in SBMJVBR to [X+S1 +Q2,N-Y) • DMX requests SBM to copy S2-Q2 bytes to SBM_AUDIO

• SBM adds the unit to this list of SBM_AUDIO and increments the number of references to the unit (this means that the same unit is referenced multiple times from the same buffer; also see note below)

• SBM sets the range indication in the SBM_AUDIO buffer to [X+S1 +Q2,X+S1 +S2) • note that in this example, X+SH-S2 equals N-Y

• SBM sets the range indication in SBMJVBR to [N-Y, N-Y)

• this means that the unit can be removed from SBMJVBR and the number of references to the unit can be decremented

In the above, an optimization may be to indicate multiple ranges. In the case of an audio decoder requiring an elementary stream, this could apply to the situation that the range indication of an element in the list of SBM_AUDIO would have a range indication {[X+Q1,X+S1),[X+S1 +Q2,N-Y)}, meaning that a single unit would point at two respective ranges.

Claims

CLAIMS:

1. A datastream-processing buffer memory organization comprising a datastream input, storage means for transiently storing said datastream, accessing means for accessing said buffer memory, and memory management means for memory housekeeping, characterized in that said memory management means are executed in software.

2. An organization as claimed in Claim 1, wherein said memory management means are arranged for multilevel housekeeping.

3. An organization as claimed in Claim 2, wherein an initial level housekeeping executes error protection measures and sequesters datastream formatizing items.

4. An organization as claimed in Claim 2, wherein a non-initial level housekeeping sequesters user data formatizing items.

5. An organization as claimed in Claim 2, wherein said multilevel housekeeping numbers at least three levels.

6. An organization as claimed in Claim 2, wherein at least one level separates various monomedia information categories.