Specialized communications processor for layered protocols

• US 4,991,133 A
• Filed: 10/07/1988
• Issued: 02/05/1991
• Est. Priority Date: 10/07/1988
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Van Voorhis, David C., Mandalia, Baiju D., Davis, Gordon T., Landa, Robert E., van den Berg, Jan W.
• Primary Examiner(s): Fleming, Michael R.
• Assistant Examiner(s): CHUN, DEBRA
• Application Number:
  US07/254,986
• Time in Patent Office: 851 Days
• Field of Search: 364/200 MS File, 364/900 MS File, 370/94.1, 370/94.2, 340/825.5, 340/825.52
• US Class Current: 709/234
• CPC Class Codes: G06F 13/128 for dedicated transfers to a network for protocol converters G06F13/387
• CPC Class Codes: H04L 69/12 Protocol engines
• CPC Class Codes: H04L 9/40 Network security protocols

Abstract

A special purpose communications protocol processor (CPP) provides more efficient processing of layered communications protocols--e.g. SNA (Systems Network Architecture) and OSI (Open Systems Interconnection)--than contemporary general purpose processors, permitting hitherto unavailable operations relative to high speed communication links. The CPP contains special-purpose circuits dedicated to quick performance (e.g. single machine cycle execution) of functions needed to process header and frame information, such functions and information being of the sort repeatedly encountered in all protocol layers, and uses instructions architected to operate these circuits. The header processing functions given special treatment in this manner include priority branch determination functions, register bit reshaping (rearranging) functions, and instruction address processing functions. Frame processing functions so handled include CRC (cyclic redundancy check) computations, bit insertion/deletion operations and special character detection operations.

Claims

1. In a communication processor for efficient execution of data processes associated with layered communication protocols, said processor having discrete machine cycles of operation and cooperating with an external memory system containing programs of instructions and data related to said instructions, said data comprising header and frame information for said layered communication protocol processes and said instructions comprising at least instructions of first and second types, instructions of said first type pertaining to the processing of said header and frame information data, said instructions of said second type comprising instructions other than those of said first type, the improvement comprising:

means coupled to said external memory system for retrieving and decoding said instructions;

special purpose logic circuit means coupled to said retrieving and decoding means and said external memory system for performing processes designated by said instructions of said first type, said special purpose logic circuit means being adapted specifically for performing all operations required by each instruction of said first type in a single said machine cycle of operation of said processor; and

general purpose arithmetic logic circuit means coupled to said retrieving means and said external memory system for cooperating therewith to execute processes designated by retrieved instructions of said second type, said general purpose logic circuit means operating under control of said retrieving means for performing functions designated by individual instructions of said second type processor;

said instructions of said first type and the related operations of said special purpose logic circuit means being characterized further in that execution of each such instruction would have required multiple said machine cycles if performed via said general purpose arithmetic logic circuit means.

2. A communication processor in accordance with claim 1 wherein said instructions contained in said external memory system include instructions of a third type associated with block data movement processes in said layered communication protocols, said processor being characterized in that:

said special purpose logic circuit means for executing said instructions of said first type comprises separate header and frame processing circuit units for respectively processing said header and frame information; and

said processor further comprising;

memory access unit means coupled to said external memory system and said retrieval and decoding means for performing only data block movement processes designated by said instructions of said third type.

3. A processor in accordance with claim 2 wherein said header processing circuit unit comprises separate units for performing in single said machine cycles:

priority branching operations, register reshape operations and address processing operations;

said priority branching operations serving to test header bits denoting branch conditions in a predetermined order of priority ranking and to produce as end result an address of an instruction suited to the conditions found in the test;

said register reshape operations being useful for manipulating header data for reordering or masking bit sequences; and

said address processing operations serving to translate addresses in information frames to corresponding route and/or memory addresses for further transfer of information.

4. A processor in accordance with claim 2 wherein said frame processing circuit unit comprises separate circuit units for performing cyclic redundancy check computations, bit insertion/deletion operations and character detection operations.

5. A processor in accordance with claim 3 wherein said priority branching unit comprises:

an input register for storing a predetermined number n of data bits carried in said header segments;

said bits individually when active denoting conditions requiring special branching action;

a stack of n address registers for storing addresses associated with individual bit positions in said input register;

priority encode circuitry coupled to said register and stack for selecting an address from said stack associated with a highest priority ranked bit in said input register having a value denoting an active condition requiring branching action; and

means for presenting said selected address to said memory means for designating the next instruction to be fetched therefrom.

6. A processor in accordance with claim 3 wherein said register reshape circuit unit comprises:

an input register for storing a predetermined number n of data bits carried in said header segments;

said bits having predetermined positional ordering and carrying variable header information;

a reshape code register for defining options for swapping, masking, and shifting relative to data in said input register;

means for multiplexing and decoding the said input register and said reshape code register to extract information logically related to the contents of said input register in accordance with functions defined in said reshape code register;

an output register; and

means for transferring the information extracted by said multiplexing and decoding means into said output register.

7. A processor in accordance with claim 3 wherein said unit for performing address processing operations comprises:

an input register for storing a predetermined number n of data bits carried in said header segments;

said bits carrying address information for information data frames;

a channel identification register for defining logical channels to which values in said input register belong;

a stack of n addresses for associative lookup of data relating to the data in said input register to generate route addresses

a comparator and priority encoder to select the appropriate route address and store it in the destination address register.

8. A processor in accordance with claim 4 wherein said unit for performing cyclic redundancy check (CRC) operations comprises:

an input register containing data on which CRC computation is to be performed;

a polynomial code register which contains a polynomial used for the CRC computation;

a CRC computation unit for performing a CRC computation on data in said input register in a single said machine cycle, using the polynomial stored in said polynomial register; and

a CRC accumulator register for receiving results of said CRC computation.

9. A processor in accordance with claim 4 wherein said unit for performing bit insertion/deletion operations comprises:

an input register for data on which bit insertion or deletion operations are to be performed;

a pattern register for storing a pattern used for determining where to insert or delete data;

a bit insertion/deletion unit for operating in a single said machine cycle to insert bits into or delete bits from data received from said input register, at positions determined by said pattern register, said unit performing its operation in a single said machine cycle; and

an output register for storing result data produced by said bit insertion/deletion unit.

10. A processor in accordance with claim 4 wherein said unit for performing character detection operations comprises:

an input register for storing a byte of data carried in said header segments;

a stack of n branch addresses; and

a comparator/priority encoder to select one of said branch addresses from said stack in association with the data in said input register.

11. A communication processor in accordance with claim 2 wherein said memory access unit means comprises:

block move circuit means including means to transfer blocks of data between source and destination addresses in said memory system with logical translation of said data;

each said source address designating a first location in said memory system defining the beginning of a space in said system in which a block of source data resides;

each said destination address designating a first location in said memory system defining the begining of a space in said system for storing a block of destination data;

means including a counting register for storing buffer length information;

said buffer length information denoting a number of data bytes to be transferred from said memory system beginning at a said source address;

means including a register stack to queue block move requests designated by individual instructions of said third type in said stack;

said means to queue having queueing capacity to store a plurality of said block move requests on a first come first served basis using queue pointers to indicate stages of processing of said queued requests;

said queue pointers comprising both hardware based and software based pointer elements;

said hardware based elements denoting location of a request currently being processed by said block move circuit;

said software based elements denoting location of a last request submitted to the queue for block move processing;

said means to transfer blocks of data selectively designating bit insertion and deletion operations or cyclic redundancy check calculation operations in conjunction with individual block move operations;

said means to transfer blocks of data further comprising a block move control unit for executing said queue requests in predetermined order.

12. A communication processor in accordance with claim 3 wherein said header processing circuit unit contains a priority branch unit for performing said priority branching operations, said priority branch unit comprising:

input latches for storing n condition bits defining active and inactive branching conditions;

means responsive to a BBD (Branch on Bit Detect) instruction for loading said latches with one of k sets of n condition defining bits;

an address register stack comprising k sets of n addresses representing destinations in said external memory system determinable by bits in respective said sets of condition bits; and

means responsive to each said BBD instruction for selecting a set of addresses from said stack associated with the set of condition bits currently stored in said latches and for selecting from the selected set of addresses an address associated with a bit stored in said latches;

said address associated with said stored bit designating location of the next instruction to be fetched from said external memory system by said retrieval and decoding means for execution by said processor and said associated stored bit denoting an active condition having highest priority for attention.

13. A processor in accordance with claim 12 wherein said means responsive to said BBD instruction for selecting said set of addresses and said address comprises:

a stack of k (k>

1) sets of n (n>

1) selection control registers, each set of registers associated with a respective one of k sets of condition bits loadable from said external memory system into said input latches;

a priority encoder circuit coupled to said selection control register stack and said input latches, said encoder circuit responding to each said BBD instruction to select one of the k sets of selection control registers associated with bits currently latched in said input latches and to further select one control register in the selected set in association with a highest priority active bit in said latches; and

means responsive to contents of the selected selection control register to select an associated one of the addresses in an associated one of said k sets of addresses in said address stack.

14. A method of conducting data processes pertaining to communications in a communication network--said network comprising data processing nodes at each of which data processes pertaining to a plural-layered communication protocol may be performed, said processes requiring processing of header and frame information accompanying message data units in each layer of said layered communication protocol, said header and frame information accompanying and delineating message data units being communicated over said network, said header information comprising discrete first, second and third header information units, said first header information unit comprising a set of bits representing status of a respective set of communication conditions which when active require attention in a predetermined order of priority, said second header information unit comprising address information for determining routing of the accompanying message data unit, and said third header information unit comprising a set of bits representing message variable parameters associated with respective message data units, the bits in said third unit requiring rearrangement for accurate extraction of the associated message unit information, said frame information to be processed in each layer comprising cyclic redundancy check (CRC) characters requiring cyclic redundancy check computation processing and control characters uniquely demarking frame positions, said control characters necessitating that accompanying message data units be subjected to bit insertion and deletion processes in order to ensure unique recognizability of the control characters at receiving nodes of said network whereby message data characters do not conflict with said control characters--, said method comprising:

processing information in multiple said protocol layers at a single processing center at each node of said network in a parallel processing mode;

each said center having a predetermined machine cycle period of operation;

each said center including a general purpose processing section and a plurality of special purpose processing sections, said sections being capable of operating in parallel during any said machine cycle period;

each special purpose section being designed for performing a complex operation in a single said machine cycle which operation if performed in said general purpose section would require multiple said machine cycles for completion; and

directing performance of complex processing operations relative to each of said header units as well as said CRC computation and said insertion and deletion processes into selected ones of said special purpose sections, whereby individual header unit processing operations, CRC computations and bit insertion and deletion operations are each performed in a single said machine cycle period.

15. The method of claim 14 wherein said concentrated handling of said first set of header unit information includes:

within one of said machine cycle periods, performing a priority encoding operation on the bits of said first set to produce a priority code indication associated with a highest priority one of the bits representing an active condition requiring special action; and

within the same cycle period and as a function of said associated priority code indication locating an associated first instruction of a program routine for effecting said special action.

16. The method of claim 14 wherein said concentrated handling of said second set of header unit information includes:

within a single said machine cycle period, extracting an address associated with said second header unit by performing an associative table lookup translation.

17. The method of claim 14 wherein said concentrated handling of said third set of header unit information includes:

preloading said information of said third set into a register; and

within a single said machine cycle period performing a selected one of a plurality of register reshaping functions on said information to present said information in a required rearranged format.

18. The method of claim 14 wherein said concentrated handling of said frame unit information includes:

performing each said CRC operation in a single said machine cycle period.