This invention relates to computer system architecture and particularly to new instructions which augment the IBM z/Architecture and can be emulated by other architectures.
Before our invention IBM has created through the work of many highly talented engineers beginning with machines known as the IBM System 360 in the 1960s to the present, a special architecture which, because of its essential nature to a computing system, became known as “the mainframe” whose principles of operation state the architecture of the machine by describing the machine instructions which may be executed upon the “mainframe” implementation of the machine instructions which had been invented by IBM inventors and adopted, because of their significant contribution to improving the state of the computing machine represented by “the mainframe”, as significant contributions by inclusion in IBM'S Principles of Operation as stated over the years. The First Edition of the z/Architecture Principles of Operation which was published December, 2000 has become the standard published reference as SA22–7832–00. We determined that further new instructions would assist the art and could be included in a z/Architecture machine and also emulated by others in simpler machines, as described herein.
It is a primary object of the present invention to disclose CPU-synchronous, problem-state, and sectioning instructions which digest data in computer storage wherein the instructions compute a condensed representation of messages or data.
It is another object of the invention to disclose instructions which specify a unit of storage to be digested, and wherein a secure hashing algorithm digests the data in the specified unit of storage.
It is another object of the invention to disclose a COMPUTE INTERMEDIATE MESSAGE DIGEST instruction and a COMPUTE LAST MESSAGE DIGEST instruction.
It is a further object of the invention to disclose a digest instruction wherein complete blocks of data are processed, and after processing all complete blocks, a padding operation is performed to include the remaining portion of the specified storage.
It is another object of the invention to emulate the digest instruction in a computer architecture which is different than the computer architecture of the instruction.
These and other objects will be apparent to one skilled in the art from the following detailed description of the invention taken in conjunction with the accompanying drawings in which:
The message digest instructions discussed herein are for computing a condensed representation of a message or data file. The compute intermediate message digest and the compute last message digest instructions will first be discussed, followed by a discussion of the preferred computer system for executing these instructions. In the alternative, a second preferred computer system which emulates another computer system for executing these instructions will be discussed.
Compute Intermediate Message Digest (KIMD)
Compute Last Message Digest (KLMD)
A function specified by the function code in general register 0 is performed.
Bits 16–23 of the instruction and the R1 field are ignored.
Bit positions 57–63 of general register 0 contain the function code.
The function codes for COMPUTE INTERMEDIATE MESSAGE DIGEST are shown in
The function codes for COMPUTE LAST MESSAGE DIGEST are shown in
All other function codes are unassigned.
The query function provides the means of indicating the availability of the other functions. The contents of general registers R2 and R2+1 are ignored for the query function.
For all other functions, the second operand is processed as specified by the function code using an initial chaining value in the parameter block, and the result replaces the chaining value. For COMPUTE LAST MESSAGE DIGEST, the operation also uses a message bit length in the parameter block. The operation proceeds until the end of the second-operand location is reached or a CPU-determined number of bytes have been processed, whichever occurs first. The result is indicated in the condition code.
The R2 field designates an even-odd pair of general registers and must designate an even numbered register; otherwise, a specification exception is recognized.
The location of the leftmost byte of the second operand is specified by the contents of the R2 general register. The number of bytes in the second-operand location is specified in general register R2+1.
As part of the operation, the address in general register R2 is incremented by the number of bytes processed from the second operand, and the length in general register R2+1 is decremented by the same number. The formation and updating of the address and length is dependent on the addressing mode.
In the 24-bit addressing mode, the contents of bit positions 40–63 of general register R2 constitute the address of second operand, and the contents of bit positions 0–39 are ignored; bits 40–63 of the updated address replace the corresponding bits in general register R2, carries out of bit position 40 of the updated address are ignored, and the contents of bit positions 32–39 of general register R2 are set to zeros. In the 31-bit addressing mode, the contents of bit positions 33–63 of general register R2 constitute the address of second operand, and the contents of bit positions 0–32 are ignored; bits 33–63 of the updated address replace the corresponding bits in general register R2, carries out of bit position 33 of the updated address are ignored, and the content of bit position 32 of general register R2 is set to zero. In the 64-bit addressing mode, the contents of bit positions 0–63 of general register R2 constitute the address of second operand; bits 0–63 of the updated address replace the contents of general register R2 and carries out of bit position 0 are ignored.
In both the 24-bit and the 31-bit addressing modes, the contents of bit positions 32–63 of general register R2+1 form a 32-bit unsigned binary integer which specifies the number of bytes in the second operand; and the updated value replaces the contents of bit positions 32–63 of general register R2+1. In the 64-bit addressing mode, the contents of bit positions 0–63 of general register R2+1 form a 64-bit unsigned binary integer which specifies the number of bytes in the second operand; and the updated value replaces the contents of general register R2+1.
In the 24-bit or 31-bit addressing mode, the contents of bit positions 0–31 of general registers R2 and R2+1, always remain unchanged.
In the access-register mode, access registers 1 and R2 specify the address spaces containing the parameter block and second operand, respectively.
The result is obtained as if processing starts at the left end of the second operand and proceeds to the right, block by block. The operation is ended when all source bytes in the second operand have been processed (called normal completion), or when a CPU-determined number of blocks that is less than the length of the second operand have been processed (called partial completion). The CPU-determined number of blocks depends on the model, and may be a different number each time the instruction is executed. The CPU-determined number of blocks is usually nonzero. In certain unusual situations, this number may be zero, and condition code 3 may be set with no progress. However, the CPU protects against endless reoccurrence of this no-progress case.
When the chaining-value field overlaps any portion of the second operand, the result in the chaining value field is unpredictable.
For COMPUTE INTERMEDIATE MESSAGE DIGEST, normal completion occurs when the number of bytes in the second operand as specified in general register R2+1 have been processed. For COMPUTE LAST MESSAGE DIGEST, after all bytes in the second operand as specified in general register R2+1 have been processed, the padding operation is performed, and then normal completion occurs.
When the operation ends due to normal completion, condition code 0 is set and the resulting value in R2+1 is zero. When the operation ends due to partial completion, condition code 3 is set and the resulting value in R2+1 is nonzero.
When the second-operand length is initially zero, the second operand is not accessed, general registers R2 and R2+1 are not changed, and condition code 0 is set. For COMPUTE INTERMEDIATE MESSAGE DIGEST, the parameter block is not accessed. However, for COMPUTE LAST MESSAGE DIGEST, the empty block (L=0) case padding operation is performed and the result is stored into the parameter block.
As observed by other CPUs and channel programs, references to the parameter block and storage operands may be multiple-access references, accesses to these storage locations are not necessarily block-concurrent, and the sequence of these accesses or references is undefined.
Access exceptions may be reported for a larger portion of the second operand than is processed in a single execution of the instruction; however, access exceptions are not recognized for locations beyond the length of the second operand nor for locations more than 4K bytes beyond the current location being processed.
Symbols Used in Function Descriptions
The symbols of
KIMD-Query (KIMD Function Code 0)
The locations of the operands and addresses used by the instruction are as shown in
The KIMD-Query parameter block has the format shown in
A 128-bit status word is stored in the parameter block. Bits 0–127 of this field correspond to function codes 0–127, respectively, of the COMPUTE INTERMEDIATE MESSAGE DIGEST instruction. When a bit is one, the corresponding function is installed; otherwise, the function is not installed.
Condition code 0 is set when execution of the KIMD-Query function completes; condition code 3 is not applicable to this function.
KIMD-SHA-1 (KIMD Function Code 1)
The locations of the operands and addresses used by the instruction are as shown in
The parameter block used for the KIMD-SHA-1 function has the format shown in
A 20-byte intermediate message digest is generated for the 64-byte message blocks in operand 2 using the SHA-1 block digest algorithm with the 20-byte chaining value in the parameter block. The generated intermediate message digest, also called the output chaining value (OCV), is stored in the chaining-value field of the parameter block. The KIMD-SHA-1 operation is shown in
KLMD-Query (KLMD Function Code 0)
The locations of the operands and addresses used by the instruction are as shown in
The parameter block used for the KLMD-Query function has the format shown in
A 128-bit status word is stored in the parameter block. Bits 0–127 of this field correspond to function codes 0–127, respectively, of the COMPUTE LAST MESSAGE DIGEST instruction. When a bit is one, the corresponding function is installed; otherwise, the function is not installed.
Condition code 0 is set when execution of the KLMD-Query function completes; condition code 3 is not applicable to this function.
KLMD-SHA-1 (KLMD Function Code 1)
The locations of the operands and addresses used by the instruction are as shown in
The parameter block used for the KLMD-SHA-1 function has the format shown in
The message digest for the message (M) in operand 2 is generated using the SHA-1 algorithm with the chaining value and message-bit-length information in the parameter block.
If the length of the message in operand 2 is equal to or greater than 64 bytes, an intermediate message digest is generated for each 64-byte message block using the SHA-1 block digest algorithm with the 20-byte chaining value in the parameter block, and the generated intermediate message digest, also called the output chaining value (OCV), is stored into the chaining-value field of the parameter block. This operation is shown in
If the length of the message or the remaining message is between one byte and 55 bytes inclusive, then the operation in
Additional Symbols Used in KLMD Functions
The following additional symbols are used in the description of the COMPUTE LAST MESSAGE DIGEST functions.
Symbol Explanation for KLMD Function Figures
A specification exception is recognized and no other action is taken if any of the following occurs:
Program Exceptions:
1. Bit 56 of general register 0 is reserved for future extension and should be set to zero.
2. When condition code 3 is set, the second operand address and length in general registers R2 and R2+1, respectively, and the chaining-value in the parameter block are usually updated such that the program can simply branch back to the instruction to continue the operation.
For unusual situations, the CPU protects against endless reoccurrence for the no-progress case. Thus, the program can safely branch back to the instruction whenever condition code 3 is set with no exposure to an endless loop.
3. If the length of the second operand is nonzero initially and condition code 0 is set, the registers are updated in the same manner as for condition code 3; the chaining value in this case is such that additional operands can be processed as if they were part of the same chain.
4. The instructions COMPUTE INTERMEDIATE MESSAGE DIGEST and COMPUTE LAST MESSAGE DIGEST are designed to be used by a security service application programming interface (API). These APIs provide the program with means to compute the digest of messages of almost unlimited size, including those too large to fit in storage all at once. This is accomplished by permitting the program to pass the message to the API in parts. The following programming notes are described in terms of these APIs.
5. Before processing the first part of a message, the program must set the initial values for the chaining-value field. For SHA-1, the initial chaining values are listed as follows:
6. When processing message parts other than the last, the program must process message parts in multiples of 512 bits (64 bytes) and use the COMPUTE INTERMEDIATE MESSAGE DIGEST instruction.
7. When processing the last message part, the program must compute the length of the original message in bits and place this 64-bit value in the message-bit-length field of the parameter block, and use the COMPUTE LAST MESSAGE DIGEST instruction.
8. The COMPUTE LAST MESSAGE DIGEST instruction does not require the second operand to be a multiple of the block size. It first processes complete blocks, and may set condition code 3 before processing all blocks. After processing all complete blocks, it then performs the padding operation including the remaining portion of the second operand. This may require one or two iterations of the SHA-1 block digest algorithm.
9. The COMPUTE LAST MESSAGE DIGEST instruction provides the SHA-1 padding for messages that are a multiple of eight bits in length. If SHA-1 is to be applied to a bit string which is not a multiple of eight bits, the program must perform the padding and use the COMPUTE INTERMEDIATE MESSAGE DIGEST instruction.
Crypto Coprocessor:
The preferred embodiment provides a crypto coprocessor which can be used with the instructions described herein and to execute cipher messages and assist in a variety of chaining message tasks which can be employed for chained and cryptographic use with the appropriate instructions.
The cryptographic control unit provides a cryptographic coprocessor directly attached to a data path common to all internal execution units of the central processing unit on a general purpose microprocessor providing the available hardware (E0 . . . En), or from a combination thereof in the preferred embodiment having multiple execution pipelines) for the central processing unit. When a cryptographic instruction is encountered in the command register (3), the control unit (2) invokes the appropriate algorithm from the available hardware. Operand data is delivered over the same internal microprocessor bus via an input FIFO register (4). When an operation is completed the a flag is set in a status register (6) and the results are available to be read out from the output FIFO register (5).
The illustrated preferred embodiment of our invention is designed to be extensible to include as many hardware engines as required by a particular implementation depending on the performance goals of the system. The data paths to the input and output registers (7) are common among all engines.
The preferred embodiment of the invention cryptographic functions are implemented in execution unit hardware on the CPU and this implementation enables a lower latency for calling and executing encryption operations and increases the efficiency.
This decreased latency greatly enhances the capability of general purpose processors in systems that frequently do many encryption operations, particularly when only small amounts of data are involved. This allows an implementation that can significantly accelerate the processes involved in doing secure online transactions. The most common methods of securing online transactions involve a set of three algorithms. The first algorithm is only used one time in a session, and may be implemented in hardware or software, while the other operations are invoked with every transaction of the session, and the cost in latency of calling external hardware as well as the cost in time to execute the algorithm in software are both eliminated with this invention.
In
As illustrated by
In
More particularly, these architected instructions can be used with a computer architecture with existing instruction formats with a 12 bit unsigned displacement used to form the operand storage address and also one having additional instruction formats that provide a additional displacement bits, preferably 20 bits, which comprise an extended signed displacement used to form the operand storage address. These computer architected instructions comprise computer software, stored in a computer storage medium, for producing the code running of the processor utilizing the computer software, and comprising the instruction code for use by a compiler or emulator/interpreter which is stored in a computer storage medium 501, and wherein the first part of the instruction code comprises an operation code which specified the operation to be performed and a second part which designates the operands for that participate. The long displacement instructions permit additional addresses to be directly addressed with the use of the long displacement facility instruction.
Furthermore, the preferred computer architecture has an instruction format such that the opcode is in bit positions 0 through 15.
While the preferred embodiment of the invention has been illustrated and described herein, it is to be understood that the invention is not limited to the precise construction herein disclosed, and the right is reserved to all changes and modifications coming within the scope of the invention as defined in the appended claims.
Number | Name | Date | Kind |
---|---|---|---|
4578530 | Zeidler | Mar 1986 | A |
5200999 | Matyas et al. | Apr 1993 | A |
5666411 | McCarty | Sep 1997 | A |
5765030 | Nachenberg et al. | Jun 1998 | A |
6134592 | Montulli | Oct 2000 | A |
6496932 | Trieger | Dec 2002 | B1 |
6542981 | Zaidi et al. | Apr 2003 | B1 |
6996725 | Ma et al. | Feb 2006 | B1 |
20020184046 | Kamada et al. | Dec 2002 | A1 |
20020191790 | Anand | Dec 2002 | A1 |
20020191792 | Anand | Dec 2002 | A1 |
20030002666 | Takahashi | Jan 2003 | A1 |
20030028765 | Cromer et al. | Feb 2003 | A1 |
Number | Date | Country |
---|---|---|
0354774 | Aug 1989 | EP |
A2 0725511 | Jan 1996 | EP |
1 494 750 | Dec 1977 | GB |
WO 0111818 | Aug 1999 | WO |
Number | Date | Country | |
---|---|---|---|
20040230814 A1 | Nov 2004 | US |