CA1056458A - Common diagnostic bus for computer systems to enable testing concurrently with normal system operation - Google Patents

Abstract

A COMMON DIAGNOSTIC BUS FOR COMPUTER
SYSTEM TO ENABLE TESTING CONCURRENTLY
WITH NORMAL SYSTEM OPERATION
Abstract A common reliability and serviceability (RAS) bus is connected to each functional combination of LSI apparatus, i.e., storage control, central processing unit (CPU), input/output (I/O) attachments and devices, and system control adapter constituting a computer system. The common RAS bus includes a unique address line to facilitate discrete addressing of each functional unit from the system control adapter. Logic is included in each I/O type of functional unit which is responsive to the address signal and a test mode signal for degating or blocking data normally transferring between the functional units.
Other logic provides control signals for causing the addressed functional unit to go thru one clock cycle, one shift cycle, one machine cycle or one instruction cycle. This enables the addressed unit to go thru an operation after a test pattern has been loaded into the functional unit while the system continues to operate concurrently, if the addressed unit is other than the CPU. The status of the addressed unit is then checked by the system control adapter.

Description

26 Background of the Invention 27 1. Field of the Invention 28 This invention relates to computer 29 systems and more particularly to computer systems having system control adapters and still more particularly to :, . . . .- -: :-: : . . , - : , . -.-, : . . , . :

lOS~i458 1 computer systems having functional units constructed from LSI cir-cuitry and organized so as to be responsive to signals from a sys-tem control adapter via a common RAS bus to go thru an operation.

2. Description of the Prior Art Prior art arrangements for testing functional units within a computer system such as described in the IBM* Technical Disclosure Bulletin, Volume 12, No. 12, March 1970, page 1614 do not have direct addressing of the individual units to be tested and require simulation hardware. Some prior art arrangements such as in U.S.
patent 3,806,878** dated April 23, 1974, for Concurrent Subsystem Diagnostics and I/O Controller require use of a diagnostic instruc-tion and do not have a separate RAS bus.
Other prior art such as in U.S. patent 3,825,901** dated July 23, 1974, for Integrated Diagnostic Tool does not have direct acc-ess to all triggers, registers and other elements of the computer system. Also~ these prior art systems are unable to test the fail-ing unit by means of a test pattern capable of testing the opera-tion of individual logic elements. Rather, they use a functional test pattern and exercise the functional unit being tested, such as an ALU, to see if it operates in its intended manner. For example, a functional test pattern would be sent to the ALU to determine if it operates properly in the add mode, etc. There is no capability to test the operation of the individual logic ele-ments irrespective of the operational mode of the unit being tested.

* Registered Trade Mark ** Assigned to International Business Machines Corporation ..,~, .,. ~

~056~58 1 ~dditionally, these prior art systens are 2 not capable of concurrent operation because they do not

3 degate thc failing functional unit and do not provide

4 special system clocks to the failing functional unit.
In tlle presellt invention each I/O type of functional 6 unit can be degated from the remainder of the system.
7 If the CPU is failing, it is not degated from the rest 8 of the system. The degated functional unit can be 9 separately addressed and furnished with system clocks for effecting one clock cycle, one shift cycle, one machine 11 cycle or one instruction cycle, while the remainder of the ~2 system is operating with system clocks occurring in the 13 normal system operation mode. Thus, the present invention 14 is particularly suitable for computer systems operating with work stations. A failing work station can be 16 tested while the remainder of the system continues to 17 run in its normal mode. This is particularly advantageous 18 over prior approaches where the entire system is 19 dedicated to the test mode.
Summary of the Invention 21 The principal objects of the invention 22 are to provide improved apparatus in a computer system 23 for loading- test data serially into a functional unit and 24 causing the same to go thru an operation which:
(a) has access to all internal storage 26 devices on a LSI chip or replaceable unit, 27 (b) has a common RAS bus connected to 28 each functional unit of the computer system 29 (c) can separately address each 30 individual functional unit of the computer system, and Ro975-004 -3-' .: .

~OSG458 1 (d) can exercise the addressed functional 2 unit withc)ut utilizing the resourceS of the central 3 processing unit (CPU) so as to enable concurrent 4 operation of the CPU in the normal mode and operation of the failing unit in a test mode.
6 The foregoing objects are achieved 7 by providing in the computer system a common bus which 8 has an address line for each functional unit incorporated g into the computer system. The common bus also includes lines for carrying test and control signals for causing 11 the addressed functional unit to perform an operation.
12 The test and control signals originate from a system 13 control adapter which forms part of the computer system.
14 Each I/O type of functional unit of the computer system includes logic for isolating the functional unit from the 16 rest of the system upon being addressed and commanded to 17 perform a test operation. Access to all internal storage 18 devices of a functional unit is achieved by connecting the 19 internal storage devices serially as one or more shift registers tif more than one shift register, a 21 discrete address is required for each shift register).
22 Hence, a test or functional pattern can be serially 23 entered into the functional unit which is then caused to 24 go through an operation and then the results are . ~
serially retrieved from the functional unit. Logic is .
26 contained in the system control adapteL for providing 27 each functional unit with system clocks controlled for 28 testing purposes. In this manner, ~he addressed failing 29 functional unit is provided with system clocks controlled for test purposes and the remainder of the system 31 receives system clocks in the normal manner.
Ro975-004 -4-~, `":

~056~58 IN THE D~AWINGS
2 Fig. 1 is a block diagram illustrating 3 a computer system incorporating the present invention;
4 Figs. 2a, 2b and 2c taken together as S shown in ~ig. 2 represent a block diagram illustrating the 6 system control adapter of Fig. 1 and showing the lines 7 forming the common RAS bus;
8 Fig. 2 is a diagram illustrating the 9 arrangement of Figs. 2a, 2b and 2c;
Fig. 3a is a diagram illustrating 11 the instruction format of an instruction in RO~ for 12 operating in the hard core mode;
13 Fig. 3b is a diagram illustrating the 14 instruction format of instructions in ROM or RA~ for operating in the IOC mode;
16 Fig. 4 is a logic diagram illustrating 17 those elements of the system control adapter pertinent 18 to providing signals for the common RAS bus and a typical 19 functional unit including the logic for degating the functional unit and logic for controlling application of 21 timing pulses to the functional unit, including a 22 showing of the logic of the functional unit connected 23 in a shift register ring configuration;
24 Fig. 5 is a schematic logic diagram illustrating a typical latch of the type used in a typical 26 functional unit such as shown in Fig. 4;
27 Fig. 6 is a schematic logic diagram 28 illustrating how the latches of a typical functional 29 unit are connected for normal operation and in a shift register ring configuration for diagnostic purposes;

Fig. 7 is a schematic logic diagram illustrating the A, B
clock control logic in the system control adapter;
Fig. 8 is a timing diagram illustrating the signals of Fig. 7 and showing how a B clock can be followed by an A clock and how an A clock can be followed by a B clock;
Fig. 9 is a schematic 1G9jC diagram of the Cl, C2 clock control and distribution logic;
Fig. 10 is a timing diagram illustrating the control of the Cl, C2 clock signals for a stop sequence followed by an allow clock Cl, C2 step sequence;
Fig. 11 is a timing diagram illustrating the control of the Cl, C2 clock signals for a stop sequence followed by Cl, C2 clocks for a machine cycle sequencej and, Fig. 12 is a timing diagram illustrating the control of the Cl, C2 clock signals for a stop sequence followed by Cl, C2 clocks for an instruction cycle sequence.
Description With reference to the drawings and particularly to Fig. 1, the invention is shown by way of example as being incorporated into a 20 stored program computer system, which includes a conventional read/
write main storage 10 for storing data. Storage 10 is accessed under control of a main storage control unit 15 located in central processing unit (CPU) 20. Thé data path between CPU 20 and storage 10 is repre-sented by bus 11 while the address path is represented by bus 12.

:`

DLM/II`il O

~.

~056458 1 Although storage 10 could contain instructions and data, the instructions are stored in a conventional read/write control store 25 which is accessed by control store control 30 and CPU 20. The address path is represented by bus 26 and the data path by bus 27.
The CPU 20 interfaces to a channel 40 via bi-directional bus 35.
Channel 40 is conventional and contains registers for buffering data transferring between CPU 20 and the I/O subsystem.
The I/O subsystem includes a disk storage drive 50 which stores programs and data for the system and subsystem. Programs and data can be entered into the system or unloaded therefrom by means of load dump unit 60. This unit is quite conventional and usually takes the form of a disk storage drive having one or more removable disks or a magnetic tape unit.
Other I/O devices are connected in the system by means of I/O
controllers (IOC) 70 and 80. The IOC's 70 and 80 are structured as small control computers and include storage and a central processing unit.
The disk storage drive 50, load dump unit 60 and IOC's 70 and 80 are connected to channel 40 by bi-directional bus 45. IOC 70 controls I/O devices including a key~oard and cathode ray tube (CRT) represented by block 76, a printer 77 and system control adapter 100 which are connected to IOC 70 by busses 75 and 78. IOC 80 controls communication devices represented by block 90 which are connected to IOC 80 by bi-directional bus 85.
:, R03-75-004 _ 7 ~056458 1 Functional units of the system, i.e., main storage control 15, CPU 20, control storage control 30, channel 40, disk storage drive 50, load dump 60 and IOC's 70 and 80 are connected to a common re-liab;lity and serviceability (RAS) bus 110 which also connects to system control adapter 100.
The system control adapter 100 is shown in detail in Fig. 2 and is structured like a small control computer except that it does not include an arithmetic and logic unit (ALU). It depends upon IOC 70 or 80 for such functions. If the computer system does not have IOC's, then the system control adapter 100 would have its own ALU, otherwise it would require resources of CPU 20 and this would not permit concur-rent operation of the system and the system control adapter 100.
System control adapter 100 contains the clock for the computer system. Oscillator 101 is a conventional crystal oscillator and pro-vides pulses to clock logic 102. Clock logic 102 includes amplifiers and pulse shapers for forming clock pulses Cl and C2 which are fed to clock control and distribution logic 103. At this time it is suf-ficient to note that clock control and distribution logic 103 enables the system control adapter to selectively discontinue cl ock pul ses to any unit within the system and thereafter furnish Cl and C2 clocks for a clock step sequence, a machine cycle sequence or an instruction cycle sequence. The system control adapter 100 is also capable of providing A and B clocks for shift operations. These A and B clocks are under program control within the system 1(~56458 1 control adapter and can occur at a rate determined by the program.
The clock lines 111 and 112 are representative for the clocks Cl and C2 for each functional unit within the computer system, i.e., there is a pair of clock lines for the clocks Cl and C2 for each functional unit and the clock lines 111 and 112 are part of the RAS bus 110. RAS
bus 110 also includes lines 113 and 114 for the A and B clocks, line 116 for a test signal, line 117 which is actually a group of 16 lines for the functional unit or shift ring addresses, line 118 which is a serial data out line and line 119 which is a serial data in line. It should be noted that line 117 which consists of 16 conductors provides a discrete address for addressing each functional unit or each shift ring of each functional unit within the computer system for diagnostic purposes.
This addressing capability is independent of addressing any functional unit in the computer system when executing the program in CPU 20.
The A and B clocks on lines 113 and 114, respectively, are used for diagnostic purposes and their particular function will be described in detail later herein. It is sufficient to note at this time that the A
and B clocks occur at a rate under program control where the control is contained in block 104 which will be described in detail later herein.
The A and B clocks originate from a mode 1 register 120 which is loaded under program control. The instructions forming the program for the system control adapter 100 reside in a read only memory (ROM) 130 and in a random 1 access memory (RAM) 140. The instructions in ROM 130 provide the basic control for loading instructions into RAM 140 so as to render the system control adapter 100 operational, i.e., certain hardcore instructions in ROM 130 are very basic in nature and are for enabling the system control adapter to start functioning, for initializing the system control adapter, for establishing communication with an IOC and for loading RAM 140 with instructions from the disk storage drive 50 or the load dump device 60 via the IOC.
ROM 130 is addressed under control of address register 131 which is initially loaded or forced to a fixed address under control of an INIT CSAB signal on line 132 which comes from run control 150. The INIT CSAB signal in one instance results from a power on sequence for the computer system. The hardcore instruction in ROM 130 is shown in detail in Fig. 3a as consisting of 16 bits. Bit O is a data flow gate or branch bit. Bits 1-4 inclusive define an operation field or the branch operation to be performed if bit O indicates a br~nch condition.
Bits 5-7 inclusiYe, when the operation is a data flow gate operation, define the destination, source or control field which is contained in bits 8-15 inclusive. If the operation is a branch operation, then bits 5-15 inclusive contain the address for the branch to instruction.
The instructions in ROM 130 are decoded by logic block 135 which includes logic for determining the type of operation to be performed, i.e., 1 a data flow gate operation or a branch operation, logic for interpretingthe operation field or the branch operat;on condition, logic for detect-ing the destination, source or control field, and logic for forming the branch to address.
Address register 131 is also used for addressing RAM 140. Select lines 136 and 141 from control block 160 determine whether ROM 130 or RAM 140 will be addressed. Control block 160 receives in~uts from registers 161, 162 and 163. Register 161 is lcaded with address data from IOC 70 via bus 75. It should be noted that bus 75 actually con-sists of lines for data and lines for addresses. Whether or not bus 75 includes both types of lines depends upon the particular implementation ;~ for the invention. For example, ROM 130 and RAM 140 could be contained in the IOC 70 and with such an implementation bus 75 would not include ; address lines.
The three high order bits of the address on bus 75 are entered into , register 161, These three high order bits in this particular instance are used for developing the ROM and RAM select signals.
RAM 140 is loaded with instructions by executing an IOC instruction from ROM 130. In th1s instance, the instruction in ROM 130 is not de-coded by decode logic 135 but rather it is passed by gates 79 to the IOC
; - 70 via bus 78 and the instruction from ROM 130 is decoded in the IOC 70.
, When instructions from ROM 130 are decoded by logic 135, this is defined " as the hard-core mode of operation and this hard-core .; :.
~i"..: ~

~; .

~, ~' : ' -:

10564S~
1 ~ode of operation is to, as previously indicated, initialize the sys-tem control adapter lCO, the IOC 70 and to initialize the disk storage drive 50 and the load dump device 60.
After the hard-core mode of operation is complete, the system con-trol adapter switches into the IOC mode of operation where instructions from ROM 130 and from RAM 140 are decoded in the IOC 70. Hence the instruction from ROM 130 is decoded in IOC 70 and this causes IOC 70 to send an address via bus 75 to CSAB register 131. It will also be re-called that the high order bits of the address are placed into register 161 whereby control 160 develops select signals and in this instance a RAM select signal is developed on line 141 for selecting RAM 140. IOC
70 then provides data on bus 75 to DBO register 142 and this data is passed via funnel gates 143 into shift register 170 which can operate in either a ser;al or parallel mode and in this instance is operating in a p~rallel mode whereby the data passes from register 170 to RAM
140 via bus 171. Thus, RAM 140 is loaded with a byte of data written therein under control of a write signal WR on line 144, there being a write signal for writing a low byte and a write signal for writing a high byte, i.e., RAM 140 is two bytes wide and the low byte is written first and another cycle is taken for writing the high byte. Thus an-other cycle of operation takes place for writing the high byte and re-petitive cycles of operation for writing low and high bytes into RAM
140 are taken until ROg-75-004 - 12 -1056~58 1 RAM 140 is completely loaded. Instructions are then retrieved from RAM 140 for controlling the operation of the system control adapter. RAM 140 is in a read mode in the absence of a write signal W~ on line 144.
When operating in the IOC mode, instructions have a format as shown in Fig. 3b. It should be noted that the format for the ROM
and RAM instructions when in the IOC mode ;c the same, i.e., the ROM
instruction in the IOC mode consists of 17 bits, ROM 130 is 18 bits wide. These 18 bits when in the hard cord mode include 16 instruc-tion bits and 2 parity bits. When in the IOC mode, these 18 bitsinclude 17 instruction bits and one parity bit. RAM 140 is likewise 18 bits wide and essentially these 18 bits include two data bytes plus two parity bits; however, when in the IOC mode of operation, the 17 bits are used as instruction bits and one bit is used as a parity bit.
It should be noted that when in the IOC mode of operation, the 17 bits require three cycles of operation, i.e., one cycle is used for writing the low byte, one cycle for writing the high byte and one cycle for writing the 17th bit and the parity bit. It should be noted that the present invention is not restricted to a 17 bit instruction but in this particular implementation, the IOC 70 operates with a 17 bit instruction. The significant aspects of the invention, as previously indicated, pertain to the arrangement for concurrent operation of the system and the failing functional unit.
In order to appreciate the present invention, it not only is necessary to understand 1 the operation of the system control adapter lon~ but it is also necessary to note that the elements of each functional unit are con-nected as shift registers in a manner as set forth in U.S. patent 3,806,891 dated April 23, 1974, by Eichelberger et al for Logic Circuit t`or Scan In/Scan Out. A typical functional unit 200 is illustrated in Fig. 4 where the triggers and registers of that unit are shown as being connected in a shift register configuration 210. It should be understood that the triggers and registers of the functional unit only operate as one long shift register when in the diagnostic mode.
This can be best understood by referring to Fig. 5 which illustrates a typical latch of functional unit 200 which operates in both the shift register and non-shift register modes. Latch 215 would be some latch in functional unit 200 for performing some typical latch function. Latch 215 when operating in the non-diagnostic mode is con-trolled for being set and reset by clocks Cl and C2. Clock Cl is ap-plied to line 216 which feeds inverter 217 and AND Invert block 220 which forms part of the Ll latch of latch 215. The output of inverter 217 feeds AND Invert block 218 which also receives the data O (DO) input.
AND Invert block 219 is used for the serial shift operation and is fed by a tSERIAL DATA OUT signal and the output of inverter 221. In-verter 221 is fed by an A clock signal which is also applied to AND
invert circuit 220. The outputs of AND Invert circuits 218, 219 and 220 are dot ORed to provide a minus level R09-75-004 _ 14 ~056458 1 output on line 222. The plus level on line 224 is taken from inver-ter 223 which is fed by the dot OR output of AND Invert circuits 218, 219 and 220. The output of inverter 223 is also fed back into AND
Invert circuit 220 and to AND Invert circuit 228 of the L2 latch por-tion of latch 215. AND Invert circuit 228 is also fed by the output of inverter 227 which receives an input from OR circuit 226. OR cir-cuit 226 receives clock signals C2 and B clock. The output of OR
circuit 226 is also applied to AND Invert circuit 229. The outputs of AND Invert circuits 228 and 229 are dot ORed whereby the minus level output from the L2 portion of latch 215 is taken on line 230 and the plus level appears on line 232 which is connected to the output of inverter 231, same being fed by the dot OR output of AND Invert circuits 228 and 229. The output of inverter 231 is also fed back into AND Invert circuit 229.
Latch 215 essentially consists of two latches Ll and L2 which are connected without any control lines between them. The data in Ll and L2 are identical after clocks Cl and C2 have been applied to latch 215. However, when operating latch 215 as an element or posi-tion of a shift ring, clocks Cl and C2 are not applied, rather the A
clock is applied to shift data into the Ll portion of latch 215 and the B clock is applied to transfer the data from the Ll portion to the L2 portion.
The connection of latches 215 to form a shift register is shown in greater detail in Fig.

DLM/Wl9 1 6. More particularly Fig. 6 shows how four latches 215 are con-nected to form the shift ring. The output of inverter 231 of the latch 215 for position 1 is connected to the SERIAL DATA OUT input of the AND Invert block 219 of latch 215 for position 2. Similarly, the output of inverter 231 of latch 215 for posit;on 2 is connected to the SERIAL DATA OUT input of the AND Invert block 219 of latch 215 for position 3 and the output of inverter 231 for this latch posi-tion is applied to the SERIAL DATA OUT input of AND Invert block 219 of latch 215 for position 4. The SERIAL DATA IN line 232 is taken from inverter 231 of pos;tion 4. The SERIAL DATA OUT signal comes from the system control adapter 100 over line 118 of the RAS bus 110 and the data appearing on the SERIAL DATA IN signal on line 232 is returned to the system control adapter 100 via line 119 of RAS bus 110.
The portion of the system control adapter 100 which is particu-larly significant with respect to the present invention is shown in Fig. 4. The diagnostic address register 175 is an 8 bit register and the outputs thereof are encoded by means of high and low gate circuits 176 and 177, gated by HIGH LATCH and LOW LATCH signals, re-spectively, into 16 lines for discretely addressing any one of 16functional units. The SHIFT RING ADDRESS signal on the address line 117 in combination with other signals from the system control adapter 100 is used for degating the functional unit from the system and for gating the B clock signal to the addressed functional unit.
The A clock and B clock signals originate from mode register 120.
The signals are applied to A

1 and e clock control lQ4 which is shown in detail in Fig. 7. The A clock latch and B clock latch signals directly feed OR circuits 113a and 114b, respectively. This arrangement enables the A clock alnd B clock signals to be generated under program control by re-petitively setting and resetting register 120. The A clock signal can also be generated via AND circuits 105 and 106. These AND
circuits are conditioned by an Allow A/B CLK signal which comes from register 190, Fig. 2. The sequence of occurrence for the A
clock and B clock signals is determined by the setting of the A/B
latch position of register 190. The output of this position is applied to AND circuit 105 via inverter 191. The other input to AND circuit 105 is the Cl clock signal. AND circuit 106 has an in-put directly connected to the output of the A/B latch of register 190 and an input connected to receive the C2 clock signal. Thus, if the A/B latch position of register 190 is set, then the A clock signal is produced by a C2 clock signal whereas if this position is reset the A clock signal is produced by the Cl clock signal.
The B clock signal can also be generated via AND circuits 107 and 108 in a manner similar to which the A clock signal is generated via AND circuits 105 and 106. AND circuit 107 is conditioned by the A/B latch signal from register 190 whereas AND circuit 108 is conditioned by the output of inverter 191. The Cl clock signal gener-ates a B clock signal via AND circuit 107 when the A/B latch is set and the C2 signal generates the B clock signal via AND circuit 108 when the A/B latch is in the reset condition. Thus when the A/B
R09-75-004 _ 17 _ DLM/~22 1 latch ;c set, the Cl clock generates the B clock signal and the C2 clock generates the A clock signal. The result is that a B
clock signal is generated first and is followed by the A clock signal. If the A/B latch is reset, the Cl clock signal generates the A clock and the C2 clock signal generates the B clock whereby the A clock signal occurs first and is followed by a B clock sig-nal. The different A clock and B clock sequences are shown in the timing diagram of Fig. 8. The A and B clock signals on lines 113 and 114 are used as previously indicated for serially entering and retrieving test data from the triggers and registers connected as a shift register 210 in the functional unit 200.
The details of the clock control and distribution logic 103, Figs. 2 and 4, for the Cl and C2 clock signals are shown in Fig. 9.
Generally speaking, the clock control logic 103 provides a way for stopping the clock signals C1 and C2 with respect to the addressed functional unit and also enables the addressed functional unit after having been stopped to receive clock signals Cl and C2 selectively.
Clock control logic 103 includes a 6 position register 300. The first position of register 300 is position 301 and it provides a STOP
signal for stopping the Cl and C2 clocks to the particular addressed functional unit. Position 301 is controlled by a gate clock control signal GT CLK CONTROL from decode 195, Fig. 2, and the DBO O bit of bus 75. The Cl and C2 clock signals from clock logic 102 directly feed each functional unit clock control logic block 310 of clock control logic 103. Logic 103 includes as many clock DLM/~23 ~056458 1 control logic blocks 310 as there are addressable sh;ft rings, i.e., at least one for each functional unit.
Position 302 of register 300 provides a MACHINE CYCLE signal depending upon the bit condition of bit DB02. The MACHINE CYCLE
signal is applied to the functional unit clock control logic 310 via AND circuit 311 and OR circuit 312. AND circuit 311 is con-ditioned by the output of inverter 313 which is connected to the L2 latch of position 303. The Ll latch of position 303 is fed by the MACHINE CYCLE signal and the C2 clock signal. The Cl clock signal feeds the L2 latch of position 303 whereas for all other positions of register 300, the Cl clock signal feeds the Ll latch of those positions. Thus the MACHINE CYCLE signal is present during a Cl clock signal and a C2 clock signal and drops with the next Cl clock signal because inverter 313 will then decondition AND circuit 311.
Position 304 of register 300 is set by a DBO 3 bit and when set provides an INSTRUCTION STEP signal which is applied to AND cir-cuit 314 which feeds OR circuit 312. The INSTRUCTION STEP signal also feeds an instruction step counter 316. Counter 316 is loaded from DBO bits 0-2 inclusive under control of a LOAD COUNTER signal from decode logic 195, Fig. 2. Counter 316 is decremented by the clock signals Cl and C2 when the INSTRUCTION STEP signal is available from position 304. The outputs of counter 316 feed OR circuit 317 and its output is applied to AND circuit 314. AND circuit 314 is conditioned by a signal from OR circuit 317 whenever ~09-75-004 _ 19 ~056458 1 whenever the value in counter 316 is other than zero. When counter 316 is zero, AND circuit 314 is inhibited. Thus the INSTRUCTION STEP signal is passed by AND circuit 314 to OR cir-cuit 312 so long as counter 316 is at a value other than zero.
Position 305 of register 300 is set by DBO bit 4 and pro-vides a BLOCK C2 signal which feeds inverter 318. The output of inverter 318 feeds AND circuit 331 of the functional unit clock control logic blocks 310. Position 306 of register 300 is set by the DBO 5 bit and it provides a BLOCK Cl signal to inverter 319. The output of inverter 319 is fed directly to AND circuit 336 of the functional unit clock control logic blocks 310.
Each functional unit clock control logic block 310 is identi-cal and as previously indicated there are at least as many blocks 310 as there are functional units. Thus each functional unit clock control logic block 310 has the same inputs except for the discrete SHIFT RING ADDRESS input line 117. The functional unit clock control logic block 310 for the SHIFT RING ADDRESS used for addressing functional unit 200 is shown in detail and includes AND
circuit 325 ~hich receives the STOP signal from position 301 of register 30Q, the C2 clock signal, the SHIFT RING ADDRESS signal and the output of an inverter 327 which is fed by AND circuit 326.
AND circuit 326 receives the output of OR circuit 312 and the SHIFT
RING ADDRESS signal. AND circuit 325 controls the setting of the Ll latch 328. The L2 latch 329 of the latch pair DLM~W25 iO56~58 1 is set b~y the output of the Ll latch 328 and the Cl clock signal.
The output of the L2 latch 329 feeds inverter 330 which provides aln input to AND circuit 331. AND circuit 331 also receives the C2 clock signal and the output of inverter 318. AND circuit 331 pro ~ides a C2 clock signal on line 112 if it is not inhibited by either inverter 318 or 330.
Ll latch 333 is set under control of AND circuit 332 which re-ceives the C2 clock signal, the SHIFT RING ADDRESS signal, the output of inverter 327 and the STOP signal from position 301. The L2 latch 334 of the latch pair receives the Cl clock signal. The output of the Ll latch 333 feeds inverter 335 which in turn provides an input ~-to AND circuit 336. AND circuit 336 also receives the Cl clock signal and the output of inverter 319. AND circuit 336 provides a Cl clock signal on line 111 if it is not inhibited by inverters 319 or 335.
By the arrangement just described, control is provided to block the Cl and C2 signals and thereafter to selectively allow the occur-rence of either in a predetermined sequence as seen in Figs. 10, 11 and 12. Specifically, when position 301 provides the STOP signal to AND circuits 325 and 332, AND circuit 325 sets Ll latch 328 and AND
circuit 332 sets Ll latch 333 upon the occurrence of a clock C2 sig-nal. Latch 333 then by means of inverter 335 inhibits AND circuit 336 so as to block the passage of the Cl clock signal to the particu-lar functional unit being addressed. The Cl clock signal, however, transfers the condition of latch ~056458 1 328 into latch 329 whereby latch 329 by means of inverter 330 in-hibits AND circuit 331 from passing the C2 clock signal to the functional unit being addressed.
Clock steps, i.e., a Cl clock or a C2 clock, can be sent to the functional unit being addressed by appropriately loading posi-tions 302, 305 and 306. Position 302 provides the MACHINE CYCLE
signal whereby AND circuit 326 inhibits AND circuits 325 and 332 via inverter 327. Thus, AND circuits 331 and 336 are no longer inhibited by this path, however, positions 305 and 306 could still provide inhibiting signals to these AND circuits respectively.
The Cl clock step signal under the conditions just described would be effected by having position 306 set to a 0 condition. A C2 clock step signal is provided by having position 305 set to a 0 condition. A clock step sequence is shown in Fig. 10.
A machine cycle is a sequence of Cl clock signal followed by a C2 clock signal. Thus, for a machine cycle, positions 305 and 306 must be both set to the 0 condition. A machine cycle sequence is illustrated in Fig. 11.
An instruction step signal is variable depending upon the value set into counter 31~. In this particular example, the instruction cycle, Fig. 12, can be a sequence of a Cl clock signal followed by a C2 clock signal, a second Cl clock signal followed by a second C2 clock signal, if counter 316 is set to 4, or 3 Cl signals and three C2 signals if counter 316 is set to six, where each Cl signal DLM/~27 lOS6458 1 of the 3 Cl signals is followed by a C2 signal of the 3 C2 signals.
The Cl and C2 clock signals from the functional unit clock control logic 310 feed the functional unit 200 via lines 111 and 112, respectively. Specifically clocks Cl and C2 are shown as feeding clock generation logic 250, Fig. 4. Whether or not a functional unit would have clock generation logic 250 depends upon the require-ments of the particular functional unit. Hence by way of example, functional unit 200 is shown as having a need for clocks Tl - T4 for its normal operation. However to simplify the drawing, functional unit 200 is considered to require only clocks Cl and C2. Clock Cl is applied to AND circuit 252 which passes the Cl clock signal via OR circuit 253 to the triggers, latches and registers of functional unit 200 for their normal operation. The output of OR circuit 253 is shown as being applied to the alternate L2 latches of the func-tional unit because the Ll, L2 latch design concept does not permit the transfer of data between latches that are gated with the same clock pulse. This requirement is set forth in detail in U.S. Patent 3,783,254 dated January 1, 1974, by Eichelberger for "Level Sentitive Logic System". The Cl and C2 clock signals are also applied to alternate Ll latches so as to operate the Ll, L2 latches in the nor-mal manner. Clock signal C2 also feeds AND circuit 256 which is conditioned by the output of inverter 264. AND circuit 256 feeds OR circuit 257 and its output is applied to alternate L2 latches of the latch pairs having the Ll latches RO~-75-004 - 23 -DLM~28 ~056458 1 thereof clocked by the Cl clock signal. AND circuit 255 func-l;ions in a manner similar to AND circuit 251 in that it has the same inputs; however, it feeds OR circuit 257 rather than OR cir-~:uit 2.$3.
In order to operate the Ll and L2 latches as a shift register 210, AND circuit 251 which is conditioned by the SHIFT RING ADDRESS
signal also receives the B clock signal from A B clock control 104 over line 114 whereby the signal passed by AND circuit 251 is passed by OR circuit 253 to the L2 latches as shown. The A clock signal feeds the first Ll latch of the shift register 210 which also re-ceives the SERIAL DATA OUT line from register 170. The output of the last position of shift register 210 is fed to AND circuit 280 which is conditioned by the SHIFT RING ADDRESS signal on line 117. The out-put of AND circuit 280 feeds OR circuit 281. OR circuit 281 also has inputs from other functional units and its output is the SERIAL DATA
IN input to register 170.
When the functional unit 200 is being operated in other than its normal mode, register 175 provides a SHIFT RING ADDRESS signal on line 117 and register 185 provides a TEST signal on line 116 to AND
circuit 261. AND circuit 261 feeds OR circuit 263 which in turn feeds inverter 264. OR circuit 263 also receives an input from AND
circuit 262. AND circuit 262 receives the SHIFT RING ADDRESS signal and a power on reset (POR) POR signal. The output of inverter 264 provides. an inhibit signal to AND circuits 265 so as to degate func-tional unit 2QO from the T/O channel interface which could be the busses 75 or 85 leading to the IOC's 70 or 80 or the bus 45 leading to channel 40. The inhibit signal from in-verter 264 is also applied to AND circuits 275 for degating the functional unit from its associated I/O device.
When in the non-norm~l mode, the shift register 210 of the functional unit is loaded serially with data patterns from register 170. Register 170 can be loaded in parallel with the desired data pattern. Then A and B clcck pulses are applied to register 17Q
over lines 113 and 114, respectively, for shifting the data out over line 118 to the Ll latch of the first position of shift register 210 of the functional unit 200. Shift register 210 is shifted by A and B clocks and the output of register 210 is applied to AND circuit 280 which is conditioned by the SHIFT RING ADDRESS signal as pre-viously indicated. The output of AND circuit 280 feeds OR circuit 281 and its output is the SHIFT DATA IN line 119 which is returned to shift register 170. OR circuit 281, which is merely representa-tive of the OR logic function, receives inputs from the other func-tional units 200. OR circuit 281 would normally be a dot OR con-nection.
The SERIAL DATA OUT line 118 and the SERIAL DATA IN line 119 are also applied to exclusive OR circuit 400. This arrangement enables the data retrieved from shift register 210 to be compared serially, bit by bit, with an expected result pattern loaded into register 17Q. The output of exclusive OR circuit feeds AND circuit 401 which is conditioned by an allow error signal from regis ter 180 . The output of 2 AND circuit 401 is an error signal which is used by the 3 system control adapter 100.
4 What is claimed is:

Ro975-004 -26-

Claims (9)

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:

1. In a computer system having a central processing unit (CPU), an I/O channelunit with I/O adapters units connected thereto with at least one I/O adapter unit being a system control adapter, said units including internal storage ele-ments connected for operation in a normal mode and connected serially as a shiftring for operation in a diagnostic mode, the improvement comprising:
means in said system control adapter for discretely addressing each unit of said computer system, means in said system control adapter for providing a test signal to each unit of said computer system, and means in each unit of said computer system responsive to being addressed and to receiving said test signal for inhibiting normal mode operation and enabling diagnostic mode operation of said internal storage elements in said responsive unit.

2. The computer system of claim 1 wherein a unit enabled for diagnostic mode operation of said internal storage elements is tested by said system control adapter concurrently with operation of said computer system provided said unit enabled for diagnostic mode operation of said internal storage elements is otherthan said CPU.

3. The computer system of claim 1 wherein said system control adapter includesmeans for providing data and clocking signals to said internal storage elements of a unit enabled for diagnostic mode operation, said clocking signals causing said enabled unit to perform a predetermined operation.

4. The computer system of claim 3 wherein said system control adapter further includes means for retrieving data from said internal storage elements of a unitenabled for diagnostic mode operation.

5. The computer system of claim 3 wherein said system control adapter providesclocking signals to said internal storage elements of a unit enabled for diagnos-tic mode operation to effect one clock cycle of operation.

6. The computer system of claim 3 wherein said system control adapter provides clocking signals to said internal storage elements of a unit enabled for diagnos-tic mode operation to effect one shift cycle of operation.

7. The computer system of claim 3 wherein said system control adapter provides clocking signals to said internal storage elements of a unit enabled for diagnos-tic mode operation to effect one instruction cycle of operation.

8. In a computer system having a CPU unit and including an I/O channel unit with I/O adapter units connected thereto with at least one I/O adapter unit being a system control adapter with means for addressing all other units, means within said system control adapter for providing serial data to an addressed unit and means within said system control adapter for providing a plurality of common diagnostic control signals to an addressed unit, said units having internal serially connected storage elements which can be set to predetermined states by said system control adapter, the improvement comprising:
first logic means includes in each of said units responsive to an address signal and one of said control signals from said system control adapter to block data from passing between said addressed unit and all other units except said system control adapter, second logic means in each of said units responsive to other of said con-trol signals for initiating a diagnostic operation in said addressed unit, sensing means for detecting completion of said diagnostic operation, and means within said system control adapter responsive to said sensing means detecting completion of said diagnostic operation for retrieving serial data from said addressed unit after said diagnostic operation has been executed.

9. The computer system of claim 8 further comprising means within said system control adapter for testing said retrieved serial data.