Secondary processor initialization scheme.

0001This invention relates to the data processing field. More particularly, this invention is a computer system having a primary unit and a secondary unit where the primary unit controls the initialization of the secondary unit.

0002The typical microprocessor based computer system has a system board that contains a primary processor. Typi­cally, this system board contains slots into which one or more adapter cards can be plugged. These adapter cards can be for controlling a printer, a display, or communica­tions between other computers, just to name a few func­tions. Often these adapter cards contain a secondary microprocessor used to control the specific function of that particular adapter card.

0003In addition to a secondary processor, adapter cards often also contain a read only storage (ROS) or read only memory (ROM) module. This ROS (or ROM) module is used to permanently store an initialization program (otherwise known as an initial boot record) even when the power to the adapter card is off. When power is first provided to the adapter card, the secondary processor, as its first activity, executes the initialization program contained in the ROS module. This program initializes the secondary processor, the random access memory (RAM), and other hardware contained on the adapter card.

0004The requirement of having a ROS (or ROM) module on an adapter card to permanently store the initialization program is undesirable as it increases the cost of the adapter card, increases the time required to develop the adapter card, and makes field changes or updates of the adapter card initialization program quite difficult.

0005It is a principal object of the invention to provide for the initialization of a secondary processor on an adapter card or other secondary unit where an on-board ROS or ROM module is not required to store the initialization program.

0006It is another object of the invention to provide a simplified method of initializing a secondary processor on an adapter card or other secondary unit where the initial­ization program can be easily modified.

0007It is still another object of the invention to reduce the cost and time of developing adapter cards or other secondary units for computer systems.

0008These and other objects are accomplished by the secondary processor initialization scheme disclosed here­in.

0009In a microprocessor based computer system having a primary unit (such as a system board) and a secondary unit (such as an adapter card), the secondary processor (which is held in a non-running reset state) contained on the secondary unit can be initialized without requiring a read only storage (ROS) module or a read only memory (ROM) module to be a part of the secondary unit. Specifically, a window is defined from a segment of uninstalled memory address space from the primary unit and a segment of installed memory address space from the secondary unit. The segment of memory from the secondary unit includes the secondary processor instruction addresses which are ini­tially executed. The primary processor on the primary unit retrieves the secondary processor's initialization program from the primary unit storage. This initializa­tion program is then placed in the window. The secondary processor does not need to be running for primary proces­sor window accesses. The primary processor then disables the reset line of the secondary processor. Upon having its reset line disabled, the secondary processor begins executing the initialization program. In this manner, the secondary unit is initialized without the requirement of an on-board ROS or ROM module. After the initialization program has completed, the primary unit can send other preparatory programs, such as diagnostic pro-grams, to the secondary unit through the window. When all of the pre­paratory programs have been completed, the contents of the window are overwritten so the primary unit's memory and the secondary unit's memory can be fully utilized for other interprocessor communications. <ul id="ul0001" list-style="none"><li>Fig. 1 shows a block diagram of the primary unit and the secondary unit of the invention.</li><li>Fig. 2A shows how the window is created from the primary memory address space and the secondary memory address space in the preferred embodiment of the inven­tion.</li><li>Fig. 2B shows how the window is created from the primary memory address space and the secondary memory address space in an alternate embodiment of the invention.</li><li>Fig. 2C shows an alternate embodiment of the inven­tion using a shared memory scheme.</li><li>Fig. 3 shows the I/O registers of the secondary unit in more detail.</li><li>Figs. 4A-4B show a flow chart of the programmed primary processor of the invention.</li><li>Figs. 5A-5B show a flow chart of the programmed secondary processor of the invention.</li></ul>

0010Fig. 1 shows an overall block diagram of the primary unit and secondary unit of the invention. In the pre­ferred embodiment, primary unit 10 is the system board of an IBM PERSONAL SYSTEM/2 computer, although primary unit 10 could also be the primary unit of any computer. Secondary unit 20 in the preferred embodiment is an adapt­er card such as the IBM 69X6279 adapter card used on the IBM Personal System/2 computer, although secondary unit 20 does not necessarily have to be an adapter card. Secon­dary unit 20 could conceivably be a section of a main system board that performs a specialized function that requires a secondary processor. In addition, secondary unit 20 could be a remote device, such as a terminal, connected to primary unit 10 via a communications line.

0011Primary unit 10 will now be described in more detail. Primary unit 10 contains a primary central processor unit 11. In the preferred embodiment, primary processor 11 is a microprocessor such as an INTEL 80286 or 80386 micropro­cessor, although other processors could also be used. Processor 11 is connected to primary random access memory 12, non-volatile storage 17, and read only storage (ROS) 18. Non-volatile storage 17 is typically a disk, diskette, tape, optical disk or other secondary storage device that retains its stored information after power is turned off. ROS 18 is used in a conventional manner to permanently store the initialization program for the primary unit. Primary processor 11, primary memory address space 12, storage 17 and ROS 18 are all connected to system bus 19 which connects to secondary unit 20. Secondary unit 20 contains secondary processor 21, I/O registers 30 and secondary memory address space 22. Secondary processor 21, in the preferred embodiment, is a INTEL 8088 micropro­cessor although other processors could also be used. Registers 30 are shown in more detail in Fig. 3 and will be discussed later.

0012Fig. 2A shows primary memory address space 12 and secondary memory address space 22 of the preferred embodi­ment of the invention in more detail. Note that primary memory address space 12 is shown to be larger than secon­dary memory address space 22. This would typically be the case but is not necessarily required for this invention. Primary memory address space 12 contains uninstalled primary memory 14 and installed primary memory 15. Uninstalled primary memory 14 is addressable despite the fact it is not installed. A segment of uninstalled prima­ry memory 14 is designated as primary memory window loca­tion 13. Secondary memory address space 22 contains installed secondary memory 25. A segment of installed secondary memory 25 is designated as secondary memory window location 23.

0013During the initialization and preparatory process, as will be discussed in more detail later, memory locations 13 and 23 contain the same data and will be referred to collectively as window 50. It is important to keep in mind, however, that window 50 is defined by primary memory window location 13, as accessed by primary processor 11, and secondary memory window location 23, as accessed by secondary processor 21 (once it runs), during the initial­ization and preparation of secondary unit 20. When prima­ry processor 11 accesses primary memory window location 13, it is accessing data contained in the real memory of secondary memory window location 23.

0014Fig. 2B shows primary memory address space 12 and secondary memory address space 22 of a first alternate embodiment in more detail. Note that primary window loca­tion 13 is in installed primary memory 15 instead of uninstalled primary memory 14, as was the case in the preferred embodiment. When secondary processor 21 writes data to secondary window location 23, the same data must be written to primary window location 13. This is not required in the preferred embodiment.

0015Fig. 2C shows the shared memory scheme of a second alternate embodiment. In this alternate embodiment, prima­ry processor 11 and secondary processor 21 both access shared memory address space 62 via bus 19. Only one processor can access shared memory address space 62 at a time, so that data integrity is preserved.

0016Fig. 3 shows I/O registers 30 of secondary unit 10 in more detail. Registers 31 and 33 are used to define the beginning address of primary memory window location 13. In the preferred embodiment, primary memory window loca­tion 13 is 16 kilobytes in size, although this could be considerably smaller or larger or dynamically variable and still fall within the scope of the invention. Also, in the preferred embodiment, primary memory address space 12 is one megabyte in size and secondary memory address space 22 is 256 kilobytes in size, although these also could be much smaller or larger and still fall within the scope of the invention.

0017Registers 31 and 33 contain 10 address lines, AM14-­AM23. These 10 address lines are sufficient to address any 16 kilobyte segment of the one megabyte of total memory addressability in primary memory address space 12. Clearly, more or less address lines could be used for addressing if either primary window location 13 or primary memory address space 12 were of different sizes than that of the preferred embodiment.

0018Register 32 is used to address the beginning of secondary window location 23 in secondary memory address space 22. Note that register 32 contains four address lines PS1-PS4. These four address lines are sufficient to address any 16 kilobyte segment in the 256 kilobytes of secondary memory address space 22. Again, the number of address lines in register 32 can be changed if the size of either primary window location 23 or secondary memory address space 22 is different than that of the preferred embodiment.

0019Register 34 is used to tell the primary unit if the secondary unit is installed, and contains the identifica­tion of the secondary unit.

0020Register 36 contains processor reset line 37 for resetting secondary processor 21. In the preferred em­bodiment, when processor reset line 37 is disabled, secon­dary processor 21 automatically jumps to secondary memory address space 22 and begins executing the first instruc­tion contained there.

0021Figs. 4A-4B show the flow chart of programmed primary processor 11. Block 101 asks if secondary unit 29 is installed. This is determined by looking at the contents of register 34. If register 34 contains the ID of secon­dary unit 20, then primary unit 10 knows that secondary unit 20 is installed and block 101 is answered affirma­tively. If secondary unit 20 is not installed, an error message is displayed in block 102 and the program exits in block 103. If secondary unit 20 is installed, flow of control moves to block 104 where the beginning address of primary memory window location 13 is sent to registers 31 and 33 of secondary unit 20. In this manner, secondary unit 20 is informed of the primary memory window location. Block 105 sets the secondary memory window location by writing the address of the beginning of the secondary memory window location 23 to register 32. Secondary memory window location 23 must include the secondary processor power-on reset execution location.

0022Block 106 tests secondary window location 23 to assure that this memory is operating properly. Block 107 then reads the secondary processor initialization program from storage 17 of primary unit 10. This initialization program could also be located in ROS 18 or some other non-volatile storage device part of primary unit 10. Block 108 then copies the secondary processor initializa­tion program into window 50 created by primary memory window location 13 and secondary memory window location 23. Flow of control then moves to Fig. 4B.

0023Block 109 places a jump instruction to the start of the initialization program at the secondary processor's power-on reset execution address in secondary memory address space 22 contained in window 50. Block 110 re­moves the reset condition from the secondary processor. This is done by writing to register 36, thereby disabling processor reset line 37. When processor reset line 37 is disabled, secondary processor 21 is started as shown in block 200. The operation of secondary processor 21 is shown in more detail in Figs. 5A and 5B and will be dis­cussed later.

0024Primary processor 11 waits for secondary processor 21 to complete the initialization process and put an initial­ization completion code in window 50 as shown in blocks 112 and 113. In the event the time out period elapses before the initialization completion code appears in window 50, an error code is displayed in block 114 and the program exits in block 115. Otherwise, primary processor 11 loops between blocks 112 and 113 waiting for the ini­tialization completion code from secondary processor 21. Once secondary processor 21 places an initialization completion code in the window, block 116 checks to see if this completion code indicates that secondary processor 21 has been initialized properly. If not, an error code is displayed in block 114 and the program exits in block 115. If so, block 117 reads a diagnostics or other preparatory program from storage 17 and transfers this program to window 50, one block at a time. Note that primary proces­sor 11 will receive a completion code via window 50 from secondary processor 21 after each block has been success­fully transferred.

0025After primary processor 11 receives a completion code from the program that the program has successfully com­pleted (blocks 112 and 126), control loops back to block 117 and the next preparatory program (if block 127 deter­mines there is one) is transferred to window 50. In the event the time out period elapses before the completion code appears in window 50, an error code is displayed in block 124. The capability described above allows pro­grammers much flexibility in controlling the initializa­tion and preparation process of secondary unit 20. When all of the diagnostics or other preparatory programs have been transferred through window 50 to secondary processor 21 and successfully executed, primary processor 11 over­writes the last block of the last preparatory program in window 50 but continues to use window 50 to communicate with secondary processor 21 by sending requests and data and receiving requests, status and data from secondary processor 21.

0026Note that if the preparatory program is less than or equal to the size of window 50, it would not be necessary to break it into blocks and move to a different location in secondary memory. Instead, the program can be executed directly from window 50 in a manner similar to how the initialization program is executed.

0027Figs. 5A-5B show a flow chart of programmed secondary processor 21. Block 201 indicates that processor reset line 37 of register 36 has been disabled and secondary processor 21 has been reset. In block 202, processor 21 jumps to the start of the initialization program that is the destination of the jump instruction placed in secon­dary memory address space 22 by processor 11 in block 109. Block 203 tests the operation of secondary processor 21 and its associated registers. In block 204, secondary processor 21 tests the portion of secondary memory address space 22 not part of secondary memory window location 23. Block 205 performs any other initialization steps that may be required, and block 206 informs the primary processor of the completion of the initialization. This is done by placing a completion code in window 50. If secondary processor 21 has completed initialization successfully, a successful completion code is placed in window 50. Howev­er, if secondary processor 21 has not completed initial­ization successfully (due to a detected failure of one of the tests in blocks 203 or 204), an unsuccessful comple­tion code is placed in window 50. This completion code is analyzed by primary processor 11 in block 116 of Fig. 4B, as has been discussed. Flow of control then moves to block 207 in Fig. 5B.

0028Block 207 waits for a block of a diagnostic or other preparatory program to be placed in window 50 by primary unit 10. Block 209 copies this block of a diagnostic or other preparatory program to a designated area of secon­dary memory address space 22. Block 214 notifies the primary processor of the completion of the block transfer operation by writing a completion code into window 50. Block 210 then looks to see if this is the last block of the program to be transferred. If not, block 211 loops until another block has been loaded. When another block has been loaded, flow of control returns back to block 209 where the additional block is then copied into secondary memory address space 22.

0029When block 210 indicates that the last block of the program has been transferred, block 212 passes control to the entry point of the program just loaded in secondary memory address space 22.

0030In block 213, the loaded program runs and sends a completion code to primary processor 11 indicating suc­cessful execution. The loaded program then returns con­trol to secondary processor 21 if it is not the last preparatory program to be executed. If block 213 returns control, flow of control returns to block 207 where secon­dary processor 21 waits for a block of an additional diagnostic or other preparatory program to appear in the window from primary unit 10. If block 213 does not return control, the loaded program was the last preparatory program, and secondary window location 23 is overwritten and used by secondary processor 21 to communicate with primary processor 11, as has already been discussed.

0031If the second alternate embodiment shown in Fig. 2C is used, the flow charts shown in Figs. 4A-4B and 5A-5B would be modified slightly. Specifically, the window concept would be replaced by the shared memory concept. For example, block 108 would read "copy secondary proces­sor initialization program into shared memory address space." The jump instruction placed at the POR execution addres in block 109 would cause the secondary program to jump to the shared memory. Likewise, all completion codes would be sent to shared memory 62 by secondary processor 21 and read from shared memory 62 by primary processor 11.

0032While the invention has been described with respect to the preferred embodiment, it will be understood by those skilled in the art that various changes in detail may be made therein without departing from the spirit, scope and teaching of the invention. For example, the primary and second processor could operate more in a master/slave relationship, where the primary processor completely controls the transfer of blocks of diagnostic or preparatory programs to the secondary memory without assistance of the secondary processor. This could be done by moving the window around the secondary memory. In addition, the initialization program could be larger than the size of window 50. In this event, the initialization program would be broken into blocks and sent to secondary unit 20 in a manner similar to how the diagnostics or other preparatory programs are sent, as described in blocks 117-127 in Fig. 4B.