Home

asmintro - Jay Moseley

1. IOREAD2 EQU LOOP HEAD FOR READING GET IODSKDCB GET LOCATE Rl NEXT RECORD LR RO R1 MOVE TO PLACE BEST FOR PUT PUT IOPRTDCB 0 ILLUSTRATE REGISTER FOR FOR PUT MOVE B IOREAD2 LOOP UNTIL DONE SPACE 1 IOEOF2 EQU BRANCH HERE SEE EODAD IOEOF2 CLOSE IODSKDCB DONE WITH IT CLOSE IT LA R2 IOTITLE3 PUT IN amp TITLE AREA PUT IOPRTDCB 2 ILLUSTRATE REGISTER FORM CLOSE IOPRTDCB AKE SURE LAST PRINTED DONE SPACE 1 PRINT NOGEN XRETURN SA TR NO SPACE 1 IOCARD DS OD CL80 80 BYTES D ALIGNED DC CLSD3Z a PAD TO 133 BYTES FOR PRINTING IOTITLE1 DC CL133 1 ECHO CHECK OF INPUT CARDS BELOW IOTITLE2 DC CL133 0 FIRST PASS DONE RECORDS FROM DISK FOLLOW BELOW IOTITLE3 DC CL133 0 END OF EXAMPLE LAST LINE PRINTED SPACE 1
2. C BLDLCNT F 3 IS BLDL SECTION COMPLETE BL UPDATE YES GO TO UPDATE FIND MACRO POINTER T TERMFLAG X FF OTHERWISE IS TERMINATION FLAG SE BNO DRECTSCH NOT SET GO FOR DIRECTORY SEARCH EXAMPLE SPACE 2 OTHERWISE IT IS TIME TO END EXECUTION SO CLOSE THE OPEN DATA SET AND RETURN TO THE SYSTEM SPACE 2 CLOSE DCBADD SPACE 2 TERMINATE NORMALLY SPACE 2 PRINT NOGEN XRETURN SA PRINT GEN SPACE 10 LA FNDADD 76 FNDADD UPDATE POINTER TO NEXT LIST ENTRY AR HEADING NEXTHD UPDATE HEADING POINTER REG TO NEX HEADING B CFIND DO NEXT FIND ETC SPACE 5 FOLLOWING SECTION DOES A DIFFEREN TYPE OF FIND MACRO INSTRUCTION WHICH FORCES A SEARCH OF THE DATA SETS DIRECTORY IMMEDIATELY IT TENDS TO BE SOMEWHAT LESS LESS EFFICIENT SLOWER THAN A BLDL FIND COMBINATION BUT MAY BE EASIER TO ORGANIZE AND CODE SPACE 5 OI TERMFLAG X FF SET T
3. SAMPLE PROGRAM INPUT OUTPUT QSA LE 1 READ FROM CARD READER ECHO TO PRINTER WRITE ON DISK Lil 2 READ FROM DISK WRITE TO PRINTER VIK ILLUSTRATES GET MOVE PUT MOVE GET LOCATE PUT LOCATE STEP1 EXEC ASGCG PARM SOURCE NOESD NOXREF PARM DATA MAP SOURCE INPUT DD TITLE QSAM SAMPLE PROGRAM IOTESTQS CSECT PRINT NOGEN EQUREGS XSAVE TR NO NO TRACING TO BE DONE PRINT GEN GEN SO CAN SEE OPENS ETC SPACE USE E OPEN MACRO TO INITIALIZE FOR INPUT OUTPUT SPACE 1 OPEN IOCRDDCB INPUT IOPRTDCB OUTPUT HE ABOVE GIVES 2 DCB NAMES AND DIRECTIONS FOR I O OPEN IODSKDCB OUTPUT OPEN WHERE WE WILL PUT DATA SPACE 1 PRINT A TITLE BEFORE DOING ANYTHING ELSE PUT IOPRTDCB IOTITLE1 PUT THE MESSAGE PUT MOVE FORM SPACE 2 THE FOLLOWING LOOP READS A CARD FROM CARD READER PRINTS IT AS AN ECHO CHECK THEN WRITES IT ON DISK CONTROL IS TRAN
4. HE PURPOSE OF THIS PROGRAM IS TO DEMONSTRATE FOLLOWING MACROS TIMI S HE BASIC FLOW OF THIS PROGRAM IS AS FOLLOWS G E CURRE TIME AND DATA USING E MACRO HEN CONVER TO OU FORM AND PRINT 2 EN DO 10 000 CALLS TO A RANDOM NU ERATOR WHICH USES HE TIME RO IN BINARY TO SET THE HEN OUTPU THE RES OF THE CALLS TO THI UMBER GENERATOR 3 ISS A STIMER MACRO FOR 5 SECON DO 1000 BC O TO D INE THE NUMBER OF DS THAT A BCT INSTRUCTION TAKES THEN OUTPUT E 4 THE USE TIMER WITH AN EXIT CHECK A LOOP THAT IS ISSUE AN STIMER FOR 1 04 H AN EXIT ADDRESS WHEN T EXIT ADDRESS GETS CONT CAUSES THE LOOP TO COME TO A LT AND OUTPUTS THAT TAKES ONGER THAN 1 04 SECONDS HEN CANC HE H A TTIMER MACR EJECT PRINT NOGEN EQUREGS CSEC XSAVE WHEN TIME IN DEC IS ISSUED IT RETURNS THE EG 1 IN THIS FORM OOYYDDDF WHERE YY IS LAST TWO DIGITS OF THE YEAR AND DDD IS THE ULIAN DATA EN LEFT THIS DATE AND UNPACK THE T AND PLACE IN OU PRINTING PRINT GEN TIME EC PRINT N
5. 1019NINETEFNTH RECORD EXEC ASGCG SOURCE SYSGO DD DISP OLD PASS DELETE SOURCE INPUT DD THIS CSECT ADDS RECORD TO THE DATA SET THAT ALREADY EXISTS PRINT NOGEN EQUREGS SECOND CSEC XSAVE FIRST OPERN THE INPUT AND OUTPUT DATA SETS THEN SET REGSTER 11 TO 1000 WHICH IS USED TO RELATIVE BLOACK ADDRESS FOR THE RECORDS TO BE ADDED TO THE DATA SET OPEN INPT INPUT DIRECT OUTPUT LA R11 1000 KI READ THE INPUT AND CONVERT THE DEY TO BINARY FROM IN REGISTER 2 THEN COMPUTE THE RELATIVE BLOCK ADDRESS AND STORE THIS AT REF THEN WRITE THE NEW RECORD IN THE PROPER PLACE ON DISK NEXTREC GET INPT KEY XDECI R2 KEY SR R2 R11 ST R2 REF PRINT GEN SPACE 5 HE POSITIONAL PARAMETER ARE 1 NAME OF THE DATA EVENT CONTROL BLOCK TO BE CREATED BY THE WRITE MACRO 2 TYPE DA ADD A NEW BLOCK WHEREEVER THERE IS SPACE HE SEARCH FOR AVAILBBLE SPACE STARTS ATTHE DEVICE ADDRESS ACA CACA ACA ACA CACA F A F F SNE D
6. y E X C P PROGRAM ope CMPSC 411 12 7 71 BRK KR k k KK k K K k K K RA KOK KO K KOK K R A kk EXEC ASGCG PARM DATA MAP SOURCE INPUT DD TITLE EXCP EXECUTE CHANNEL PROGRAM TEST PROGRAM CHANPROG CSECT PRINT NOGEN XSAVE PRIN GEN OPEN INDCB INPUT OUTDCB OUTPUT OPEN DCBS I E CONNECT BLOCKS TO ALLOW IO TO TAKE PLACE SPACE 3 EXCP OUTIOB EXEC THE CHANNEL PROG BEGINNING AT LOC OUTIOB VI OUTCCW X 11 CHANGE FROM SPACING TO TOP SPACE 3 OF NEW PAGE TO SINGLE SPACING READ EXCP INIOB EXEC CHAN PGM TO READ A PIECE OF DATA AS SPECIFIED IN INIOB SPACE 3 WAIT ECB INECB GO INTO WAIT STATE AND DO NOT PROCESS ANY MORE UNTIL THIS IO HAS POSTED COMPLETION IN INECB CLI INECB X 41 CHECK FOR GOOD TERMINATION BE EOF IF TERM CODE X 41 EOF READ SPACE 3 EXCP OUTIOB EXEC CH PGM AT OUTIOB WAIT ECB OUTECB WAIT FOR TERMINATION OF IO B READ LOOP SPACE 5 EOF CLOSE INDCB OUTDCB CLOSE OPEN DATA SETS PRINT NOGEN XRETURN SA SPACE 5 INAREA DC CL80 AREA FOR INPUT INDCB DCB MACRF E USING EXCP MACROS
7. JE THIS JOB WILL RUN WITH TIME 25 SECONDS PGE RECORD 600 EXEC ASGCG SYSIN DD PRINT NOGEN EQUREGS THE PURPOSE OF THIS PROGRAM IS TO DEMONSTRATE THE GETMAIN AND FREEMAIN MACRO USING THE REGISTER CONVENTION FIRST READ IN THE NO OF BYTES TO BE OBTAINED FROM THE OS THEN READ IN NUMBER TO FILL THESE BYTES AND PLACE THEM IN THE AREA OBTAINED NEXT SNAP THESE NUMBERS AND FREE THE AREA WITH A WITH A FREEMAIN MACRO MAIN CSECT XSAVE THE FIRST SECTION PREPARES FOR THE GETMAIN MACRO FIRST READ IN THE NUMBER OF BYTES TO BE OBTAINED FROM THE OPERATING SYSTEM THEN ECHO PRINT THE NUMBER MOVE THE NUMBER OF BYTES TO BE OBTAINED INTO REG 0 AND MAKE A COPY OF THIS NUMBER IN R9 DIVIDE R4 BY 4 TO GET THE NUMBER OF NUMBERS TO BE READ IN THEN PLACE THE SUBPOOL NUMBER IN RO ALONG WITH THE NUMBER OF BYTES REQUESTED XREAD WORD READ IN THE NUMBER OF BYTES TO BE OBTAINED FROM THE OPERATING SYSTE XPRNT WORD 1 81 ECHO PRINT THE NO OF NUMBERS TO BE R XDECI R4 WORD CONVERT TO INTERNAL FOR LR RO R4 AKE A COPY OF R4 FOR GETMAIN LR R9 R4 AKE A COPY OF THE NUMBER OF BYTES SRL R4 2 DI
8. OUZ DC C OTHE AVERAGE TIME IN MICRO SECONDS FOR A BCT INSTRUC DC CATION TS OUZZ DC CL12 C MICRO SECONDS TUNUM DC F 20000 MONE DC F 1 DONE DC F O TWO DC F O THREE DC F O FOUR DC EROS FIVE DC Bot SIX DC F O SEVEN DC Fig EIGHT DC pro NINE DC F O TENT DC F O TEN DC F 10 OUT DC D 0 WORD DC 2E 0 OUTPUT5 DC C OTIME OUTPUT2 DC E wi OUTPUT3 DC Ge pa OUTPUT4 DC Gi DC C DAY OUTPUT1 DC OF 19 OUTPU DC eo y OUTT DC x 00 DC chai A DC C THE NUMBER OF TIMES THAT EACH NUMBER OCCURED IS GIVEN DC C BELOW CL62 OU DC GPO EE Er O CO ACNE Ch a DC CAT GE NESEN ETL CLIN N T CRI 3 DC Cer ETT NS G T QUETTT DC EKS 4 LTORG EJEC IAND CSEC XSAVE SA NO TR NO THIS IS A RANDOM NO GENERATOR IT HAS TWO ARGUMENTS THE FIRST THE MAX VLLUE TO BE RETURNED THE SECOND THE MIN VALUE TO BE RETURNED TI FIRST DETERMINES THE NO OF BITS IN THE MAX VALUE THEN IT GENERAES THAT MANY RANDOM BITS THESE BITS ARE PLACED IN ONE OF THE REGISTERS THEN IT DETERMINES IF THE NO GENERATED IS WITHIN THE BOUNDS OF THE THE TWO AGUMENTS IT WAS GIVEN IF IT IS NOT THEN IT DOES THE PROCESS ALL OVER AGIAN x FIRST FIND THE MAX AND MIN VLUES TO BE CONSIDE
9. END ITLE BPAM I O EXAMPLE HE DCB DSECT IHADCB FOLLOWS PACE 2 CBD DSORG PO DEVD DA PACE 2 SECT PACE 2 RINT NOGEN SAVE TR BPAM I O EXAMPLE RINT GEN PACE 5 NEMONIC REGISTER EQUATES FOLLOW PACE 2 QU 1 WORK REGISTER QU 2 DCB ADDRESS POINTER QU 3 TRACK ADDRESS POINTER QU 4 ALWAYS POINTS TO THE DECB QU 5 ALWAYS HAS INTERNAL BUFFER ADDRESS QU 6 USEFUL CONSTANT 1 QU 7 EACH PRINT HEADING ADDRESS QU 8 OFFSET TO NEXT HEADING QU 9 COUNTING REGISTER FOR PROGRAM CNTRL QU 10 RECORD POINTER REGISTER QU 11 BUFFER END POINTER QU 12 BASE REGISTER QU 13 SAVE AREA POINTER QU 14 RETURN ADDRESS POINTER U 15 ENTRY POINT ADDRESS POINTER PACE 5 NITIALIZE REGISTERS WHERE POSSIBLE PACE 2 A DCBADD BPAMDCB GET DCB ADDRESS A FNDADD BLDLLIST 12 GET BLDL RELATIVE ADDRESS A DECB MAC GET DECB ADDRESS A AREA BUFFER GET INTERNAL BUFFER AREA ADDRESS A ONE 1 USEFULL CONSTANT A HEADING CALLHD FIRST PAGE HEADING ADDRESS A NEXTHD 81 OFFSET TO NEXT HEADING A BLDLCNT 0 INITIALIZE COUNT OPDCB SKIP BLDL LIST SPACE PACE 5 EXT USE OUR MACROS TO GENERATE BLDL LIST SPACE PACE 2 ENERATE LIST HEAD DISCRIPTOR WORD ISTHD BLDLLIST 3 76 ENERATE EACH LIST ENTRY IST CALL PACE 5 IS RETURN PACE 5 ES SAVE PACE 5 P
10. BZ SEC IF ZERO THEN SECOND CALL LA 5 SHEAD1 GET ADD OF OUTPUT FOR FIRST CALL B OUTPUT GO TO DO OUTPUT SEC LA 5 SHEAD2 GET ADD FOR OUTPUT OF SECOND CALL OUTPUT PUT 0 2 0 5 XRETURN SA TR NO SHEAD2 DC CL132 OTHIS IS SECOND CALL TO SECOND SHEAD1 DC CL132 OTHIS IS FIRST CALL TO SECOND PRINT GEN END LOG OBJECT SYSLMOD DD DSNAME 66LOADMOD SECOND EXEC ASGCL SOURCE SYSGO DD DISP OLD PASS SOURCE INPUT DD PRINT NOGEN ECBDSECT DSECT ECBADD DS F HE PURPOSE OF THIRD IS GAIN CONTROL AND OUTPUT A MESSAGE HEN TO POST THE ECB THEN RETURN THIRD CSECT XSAVE TR NO L 2 0 1 GET ADD OF DCB L 3 4 1 GET ADDRESS OF ECB PUT 0 2 THEAD PRINT GEN POST 0 3 240 PRINT NOGEN XRETURN SA TR NO THEAD DC CL132 OTHIRD NOW EXECUTING END OBJECT SYSLMOD DD DSNAME amp amp LOADMOD THIRD DISP OLD PASS EXEC ASGCLG SOURCE SYSGO DD DISP OLD PASS SOURCE INPUT DD PRINT NOGEN THIS IS THE MAIN JOB STEP MAIN ATTAHCES SECOND TWICE AND THIRD ONCE IT USES ATTACH DETACH EXTRACT CHAP WAIT AND POST DETACH WAIT AND EXTRACT IT DOES ALL THREE ONE STEP AT A TIME MAIN CSECT XSAVE TR NO OPEN OTPT OUTPUT PUT OTPT MHEAD
11. DSECT 2 THE FOLLOWING DSECT FORMATS THE SAVE AREA SAVEAREA DSECT UNUSED DS F BACKLINK DS F PTER TO HIGHER S A FORELINK DS F PTER TO LOWER S A REG14SAV DS F SAVE AREA FOR REG 14 REG15SAV DS F START OF S A FOR REG 15 12 x lt THE FOLLOWING DSECT FORMATS THE BEGINNING OF A CSECT IF THE NAME CONVENTION IS FOLLOWED THE FIRST INSTR MUST BE A BR ON R15 AS A BASE REG FOLLOWED BY A LENGTH AND A NAME NAMECONV DSECT BRANCH DS XL2 XL2 SPACE FOR BSC INSTR 4 bytes LENGTH DS C NAME DS C SPACE FOR NAME MARK BEGINNING ADDR ONLY END MAIN Following is the output from this example KKK AIN ENTERED KKK NEXT ENTERED KKK AST ENTERED KKK KKK TRACE ENTERED KKK BACK TRACE OF SAVE AREAS TRACE LAST NEXT MAIN IEWLCTRL BACK TRACE COMPLETED TRACE EXITED LAST EXITED KKK NEX EXITED KKK MAIN EXITED xxx Fol DS ECT 3 lowing is the actual assembler listing of the TRACE Notice those instructions which referenc E esects labels from the SAVEAREA and NAMECONV dsects Look at the object code and
12. BALR R12 RO SET R12 FOR BASE FEG DROP R15 USING R12 ib R4 0 R3 GET ADDRESS OF PAR L R5 0 R4 GET PARM XDECO R5 N CONVERT TO OUTPUT FORM XPRNT OUT 52 LA R6 1 SET R6 TO 1 FOR SUBTACTION SR R5 R6 DECREASE RECURSIVE COUNTER BY ONE BZ OUTPUT IF RECURSIVE COUNTER 0 THEN OUTPUT AT THIS POINT WE HAVE NOT DONE ENOUGH RECURSIVE CALLS SO CALL RECRSIVE AGAIN WITH THE NEW PARM IN NUMBER WHICH WAS OBTAINED FROM THE OPERATING SYSTEM ST R5 NUMBER SET VALUE FOR NEXT CALL TO RECRSIVE LA R9 72 R2 PUT ADD OF NUMBER IN R9 ST R9 76 R2 PUT ADD OF PARM IN PARM LIST LA R1 76 R2 PUT ADDRESS OF PARM LIST IN R1 CALL RECRSIVE B DONE GO TO RETURN FROM THIS CALL AT THIS POINT WE HAVE DONE 5 RECURSIVE CALLS TO RE RSIVE SO OUTPUT THE SAVEAREAS AND AND RETURN OUTPUT LA R6 5 SET R6 TO 5 FOR COUNTER ON LOOP FOR a OUTPUTTING THE SAVEAREAS LR R11 R13 GET R13 TO R11 FOR LOOP LA R10 80 R11 GE HE END OF THE AREA OBTAINED XPRNT HEADING 47 LOOP XSNAP STORAGE 0 11 0 10 T NOREGS L R11 4 R11 GET ADD OF HSA LA R10 80 R11 GET ADD OF END OF AREA BCT R6 LOOP RETURN FOR NEXT SAVEAREA PRINT GEN AT THIS POINT WE HAVE FINISHED WITH THIS VERSION OF THE CSECT
13. PRINT NOGEN MAIN CSEC XSAVE TR NO PRINT GEN THE LOAD MACRO INSTRUCTION IS CODED SO THAT THE LOAD MODULE WITH ENTRY POINT SECOND IS LOADED INTO CORE SINCE DCB IS OMMITED IT SEARCHES THE STEPLIB WHICH IS INCLUDED BY DATA STEPLIB DD DSNAME amp amp LOADMOD DISP OLD PASS THE ENTRY ADDRESS OF SECOND IS RETURNED IN REGISTER 0 SO THAT SECOND CAN BE CALLED OAD EP SECOND PRINT NOGEN LR 15 0 GET ADD OF SECOND CSECT IN 15 CALL 15 BRANCH TO SECOND PRINT GEN NOW THAT WE HAVE FINISHED WITH SECOND WE DELETE LOAD MODULE AGAIN THE ENTRY POINT IS SPECIFIED THE DELETE MACRO MUST BE ISSUED IN THE SAME TASK AS THE LOAD MACRO DELETE EP SECOND IN THE LINK MACRO WE CODE THE ENTRY POINT OF THE LOAD MODULE TO WHICH WE SISH TO LINK THE PARAM SPECIFIES A PARAMETER LIST WHICH IS PASSED IN REGISTER 1 THE VL 1 INDICATES A VARIABLE NUMBER OF ARGUMENTS THESE ARGUMENTS WILL NEVER BE USED IN THIRD THEY ARE FOR EXAPMLE LINK EP THIRD PARAM PARMI PARM2 VL 1 BEFORE WE CAN PASS CONTROL TO FOURTH WE MUST RELOAD REGISTERS 13 AND 14 TO POINT THE WAY THEY DID BEFORE ENTRY TO MAIN HEN SINCE WE USE THE ENTRY POINT CONVENTION TO FIND THE LOAD MODULE WE LET THE MACRO RESOTRE REGISTER 2 THRU 12 HE XRETURN
14. READBEYONDEND 100 RD 0 0 READ A CARD WD 0 0 WRITE IT BACK OUT BCS 15 LOOP LOOP FOREVER OR UNITL CARDS RUN OUT END THIS IS THE ONLY DATA CARD PROGRAMLOOPS 10 BCS 15 8 SAME AS B END BRANCHTOBADOPCODE 5 BCS 15 41 BRANCH TO NEXT STMT RES W O NOT A LEGAL OPCODE END ADDRESSOUTSIDE 5 tl 1 100000 BIG NUMBER SW 0 5 1 TOO BIG ADDRESSING ERROR END ADDRESSINGTEST 20 LI 1 1 FOR INDEXING Br 2723 FOR INDEXING LIW 8 X 1 LOAD AN 8 INTO R8 LW 7 X 2 LOAD A 7 INTO R7 LW 6 X LOAD A 6 INTO R6 LW 5 X 1 LOAD A 5 INTO R5 LI 3 512 ANOTHER INDEX VALUE AWM 1 Y 2 BOMB OUT OF RANGE WAIT 0 0 NEVER REACH HERE RES W 7 WAIT 0 X6 FOR INDIRECT ADDRESSING RES W 8 RES W 6 RES W 5 RES W 10000000 BIG END COMPUTEXCUBED 100 RD 0 0 GET NUBMER IN STW O X SAVE THE VALUE W 0 X MULITPLY 1 TIME TW 15 CNT DECREMENT CNT TO CNT 1 BCS 7 LOOP BRANCH IF NOT ZERO WD 0 0 WRITE RESULT OUT RES 1 RES W 2 END CS102M1 01 COMPUTER SCIENCE 102 RUN ASSIGNMENT T 1 Punch up the following program and run YOUR JOB CARD EXEC ASACG SYSIN DD MAIN CSECT THIS PROGRAM ILLUSTRATES XDUMP AND PROGRAM INTERUPTION BALR 12 0 THESE TWO ST S ARE FOR USING 12 COMMON LI
15. SOURCE INPUT DD x lt K HE THIRD CSECT IS ENTERED VIA THE LINK MACRO THE LINKAGE lt CONVENTION APPEARS STANDARD BUT ACTUALLY REGISTER 14 POINTS TO AN ADDRESS IN THE CONTROL PROGRAM OF THE LINK MACRO E HEREFORE THE LINK CONTROL PROGRAM GETS CONTROL WHEN CONTROL E IS PASSED FROM THIRD TO MAIN AND FROM MAIN TO THIRD x lt PRINT NOGEN THIRD CSECT XSAVE TR NO OPEN ATPT OUTPUT PUT ATPT THEAD CLOSE ATPT XRETURN SA TR NO THEAD DC CL80 OTHIRD ENTERED VIA LINK MACRO RETURN TO MAIN ATPT DCB DSORG PS MACRF PM LRECL 80 BLKSIZE 80 RECFM FA DDNAME FT06F001 EROPT ACC PRINT GEN END OBJECT SYSLMOD DD DSNAM EXEC ASGCL El amp amp LOADMOD THIRD DISP OLD PASS DELETE SOURCE SYSGO DD DISP OLD PASS SOURCE INPUT DD THE FOURTH CSECT IS GIVEN CONTRLO THROUGH THE XCTL MACRO WHEN CONTROL IS PASSED VIA THE XCTL MACRO CONTROL IS NOT RETURNED TO THE STEP ISSUING THE XCTL MACRO AND THE STEP ISSUING THE XCTL MACRO IS DELETED BY THE XCTL MACRO THE
16. XDECI R10 AREA COMPARE CR R9 R10 CHECK FOR KEY BNE DUMMY IF NOT GO TO OUTPTT DUMMY RECORD x lt ba AT THIS POINT WE KNOW THAT THIS IS THE NEXT KEY TO BE WRITTEN SO WRITE THE INPUT TO KISK AND THE CHECK THE DECB WITH A CHECK MACRO THEN INCREASE THE MIN KEY VALUE AND RETURN FOR NEXT INPU x lt x SPACE 5 FOR THE WRITE STATEMENT THE DCEB1 IS THE NAME OF THE DATA EVENT CONTROL BLOCK THE SECOND PARAMTTE SF INDICATES NORMAL WRITE CONDITION OTPT IS THE DCB NAME AREA IS THE ADDRESS WHERE THE OUTPUT DATA IS STORED THE PARAMETER TO THE CHECK MACRO IS THE DECB NAME FOR THE WRITE STATEMEN ka x WRITE DECB1 SF OTPT AREA SPACE 5 CHECK DECB1 PRINT NOGEN LA R9 1 R9 ADD ONE TO KEY COUNT B LOOP RETURN FOR NEXT INPUT A AT THIS POINT THE KEY JUST READ IS GREATER THAN THE MIN VALUE IN REGISTER 9 NOW CHECK THE EXPECTED KEY VALUE TO SEE IF IT IS GREATER THAN THE MAX KEY VALUE IN REGISTER 8 x DUMMY CR R9 R8 CHECK TO SEE IF LAST INPUT BH EOJ IF HIGH THEN DONE PRINT GEN a AT THIS POINT THE LAST KEY READ IN WASA GREATER THAN MIN KEY VALUE BUT LWSS THAN MAX KEY VALUE THEREFORE WRITE A DUMMY RECORD TO THE DATA SE x lt SPACE 5 x THE FIRST POSITIONAL PARAMETER IS THE DECB NAME TO BE CREATED THE SD INDICATES THAT THIS IS A DUMMY RECORD O
17. DATA TYPE MEANING COMMENTS e Character usuallv a contiguous string of bvtes in memorv treated as printable characters or a string of bits FORTRAN LOGICAL 1 EX MVC CLC OC IC STC USUALLV IMPLIES SS INSTRUCTION FORMAT all except IC STC D Double precision floating point Doubleword 64 bit FORTRAN REAL 8 EX AD SD LTDR LD IMPLIES RR OR RX INSTRUCTION FORMAT E Exponent single precision floating point fullword 32 bit FORTRAN REAL 4 EX AE LER ME IMPLIES RR OR RX INSTRUCTION FORMAT H Halfword 16 bit binary number FORTRAN INTEGER 2 EX AH H STH CH IMPLIES RX FORMAT P Packed decimal format 2 decimal digits per byte EX AP SP CP IMPLIES SS INSTRUCTION FORMAT OF TWO LENGTH TYPE 4 MACHINE FORMATS Several characters are used to denote the specific type of operand format being used note that the data types can also imply specific formats If they imply one of several the last character distinguishes among them FORMAT MEANING EXAMPLES I Immediate IMPLIES SI FORMAT EX MVI CLI Ol R Register IMPLIES RR FORMAT EX AR BCR DDR LIE EXAMPLE This section lists all OF FAMILY OF RELATED OPCOD the members of the menemonics nam The letters V machine Format CS102M1 05 S Compare famil
18. ASPRGTC1 08 III CIRCUMSTANCES FAVORING USE OF THE VARIOUS FORMS A MACRO INSTRUCTIONS In general a pure macro instruction is used as follows VARYING CODE the required cod varies radically from call to call For example XSAVE and XRETURN SHORT CODE if a macro can generate less in line cod to perform the required function than is needed to generate a call to the routine then it should be written as a macro In some cases it takes as much work to set up the arguments as it does just to perform the operations For example the code to obtain the minimum or maximum of several arguments is probably most efficiently written as a in line macro LINKAGE CODE code for linking to routines is almost necessarily written as macros since it makes little sense to call a routine in order to perform linkage unless the linkag cod required is very complex in which case the program is probably going to be SLOW B INTERNAL SUBROUTINES Internal subroutines are usually used as opposed to macros which generate code in line under the following circumstances CODE WITH LITTLE VARIANCE if the code is not going to be much different from macro call to macro call it may be better to let the macro call generate a BAL to one copy of the code as an internal subr Internal subroutines are usually used as opposed to EXTERNAL subroutines under these circumstances SHORT CODE HEAVILY USED
19. Add PRINT NOGEN S to delete X MACROS and run program with data GBLB amp QDUMPO CSECT CSECT XSAVE LA 11 AREA3 LOOP XREAD CARD 1 XPRNT CARD 81 B DONE QHEXI 2 CARD 1 BALR 3 0 QHEXO 2 AREA2 QHEXO 3 0 11 QHEXI 4 0 1 QHEXO 4 AREA3 10 XPRNT AREA AREASL CALL SUBI B LOOP DONE XRETURN SA CARD DC CL81 0 20CL100 AREA DC C O VALUES OF REGISTERS 2 3 4 AREA2 DC CELO 4 AREA3 DC 2CL10 AREASL EQU AREA LENGTH SUBI CSECT XSAVE SA NO BR 15 WATCH OUT FOR THIS BASE REGISTER ODUMPO 2 HEX VALUE OF 2 H QDUMPO 2 DEC VALUE OF 2 QDUMPO 4 DEC VALUE OF 4 C H SUB1H QDUMPO 4 QDUMPO 1 H L 0 F amp QDUMPO QDUMPO 0 VALUE OF amp amp QDUMPO amp QDUMPO SETB 1 QDUMPO 4 ERROR ERROR ERROR ERROR XPRNT CL30 VALUE OF amp amp QDUMPO amp QDUMPO 30 XRETURN SA NO LTORG END DATA CARDS AS FOLLOWS BEGINNING IN COLUMN 1 00001234 FFFFFFFF OTABCDBE Y 6 ee kaa ARA pi ra i 12345 12345678 lt x amp gt ABCD ONLY 1 67 FEF CS411MC2 01 COMPUTER SCIENCE 411 MACRO ASSIGNMENT DUE III LINKED LIST MACROS QDEFL QHED ONEXT QPOP QPUSH QSRCH QSNAP This assignment covers the followi
20. CS411MC1 01 COMPUTER SCIENCE 411 MACRO ASSIGNMENT DUE This writeup pages 01 03 I LINKAGE MACROS QSAVE QRETURN QCALL This set of macros covers the following items OS 360 linkage conventions and some basic items of macro programming referencing macro arguments obtaining items in sublists use of local and some global set variables concatenation of set variables and other items amp SYSNDX and amp SYSECT A QSAVE macro for entering subroutine Your macro will be similiar to XSAVE or SAVE and will accept the following operands REGS BASE SA as follows REGS regl reg2 will store registers regl reg2 at appropriate locations will be specified as numbers and default to REGS 14 12 BASE number will set up register number as a program base register Defaults to BASE 12 VAlues 13 14 15 are illegal should be flagged by an MNOTE and then use 12 instead SA value controls save area linkage and save area name SA NO means the subroutine has no save area so that R13 should not be modified and no inter save area linkage created SA name save area linkage will be done and the address of name will be placed in register 13 as the save area SA the macro will make up a unique name composed of the first 3 characters of the current CSEC name followed by a unique number followed by S and will refer to this when settin
21. UNIT RECORD DEVICES 1403 LINE PRINTERS of various models printing with maximum rated speed of 1100 lpm lines minute for 1403 N1 600 lpm for others Use removable TRAINs so that different character sets can be obtained upper case only ON upper lower TN Attached to 2821 control units on multiplexor 2540 CARD READ PUNCH one unit contains a card reader and card punch treated logically as separate addresses for example 00C for reader 00D for attached punch READS cards optically at 1000 cpm cards minute maximum PUNCHES cards at 300 cpm maximum Attached to 2821 control unit 2671 PAPER TAPE READER reads punched paper tape at up to 1000 cps characters per second attached also to 2821 control unit SUMMARY OF DEVICE CHARACTERISTICS DEVICE CAPACITY PER UNIT TRANSFER RATE AVERAGE DELAY TYPE megabytes KB second seek latency ms 2301 DRUM 4 09 1200 0 8 6 2319 29 17 per pack 312 60 12 5 2314 29 17 per pack 312 15 12 5 2400 DRIVE varies 20 per 90 2400 ft tape OK 1403 RINTER 132 bytes line 2 4 u u 2540 ER 80 bytes card Pes _ 2540 H 80 bytes card 0 4 _ 2671 ER TAPE 1 0 REFER GA22 6810 IBM S 360 SYSTEM SUMMARY GA27 2719 IBM S 360 MOD LT 67 FUNCTIONAL CHARACTERISTICS 01 08 73 date of last update INDEX102 01 CMPSC 102 INDEX OF BAT FILES J R MASHEY The following provides a brief i
22. ES WHICH AR INSTRUCTIONAL FILES OF MATERIAL TO BE INCLUDED ON THE ASSIST DIS TION TAPE THE FORMAT OF THIS FILE ALLOWS IT TO BE ED TO PRODUCE PSU JCL INCLUDI RDS TO BE USED TO COPY THESE FIL ES TO TAPE FILENAME IS PRECEDED BY sr A PSU RJE ID IF THI E NOT SAVED UNDER RJE ID JR EFAULT ROG JRMO5 BATCOPY READS T RODUCES JOBS TO CO E ES TO TAPE ON TAPE THE FORMAT OF ES IS gt FI E BEGINNING IN COLUMN 1 ON D PRECEDING EACH FILE gt IS NOWHERE ELSE USED IN HAT IT IS EASY TO SEARCH THE FILE PECIFIC SECTION AND PUNCH OR PRINT IT HE FOLLOWING COMMENTS MAY BEGIN THE RIPTIONS JCL E FILE CONTAINS JOB CONTROL LANGUAGI R USE ON A 0S 360 370 SYSTEM PLUS TYPICAL SA TEXT HE E CONTAINS TEXT MATERIAL EGUN BY A HAVING IN COLUMN 1 THESE R CASE LETTERS D SO TYPICALLY REQUIRE A TN PRINT TRAIN OR 1 ENT FOR EST AE RANCE gt AAAINDI OV
23. The control u HARDWAR1 R SUBCHANNEL 04 DRIVES read write tape 9 trac ks per tape Each group of the tape gaps of nits for thes 90KB drives ar using tape 6 inch between UNIT RECORD DEVICES 1403 LINE PRINTERS of various models printing with maximum rated speed of 1100 lpm lines minute for 1403 N1 600 lpm for others Use removable TRAINs so that different character sets can be obtained upper case only ON upper lower TN Attached to 2821 control units on multiplexor 2540 CARD READ PUNCH one unit contains a card reader and card punch treated logically as separate addresses for example 00C for reader 00D for attached punch READS cards optically at 1000 cpm cards minute maximum PUNCHES cards at 300 cpm maximum Attached to 2821 control unit 2671 PAPER TAPE READER reads punched paper tape at up to 1000 cps characters per second attached also to 2821 control unit SUMMARY OF DEVICE CHARACTERISTICS DEVICE CAPACITY PER UNIT TRANSFER RATE AVERAGE DELAY TYPE megabytes KB second seek latency ms 2301 DRUM 4 09 1200 0 8 6 2319 DISK 29 17 per pack 312 60 12 45 2314 DISK 29 17 per pack 312 75 1 2 5 2400 TAPE DRIVE varies 20 per 90 2400 ft tape OK 1403 PRINTER 132 bytes line 2 4 E 2540 READER 80
24. PLACE THE SUBPOOL NUMBER AND THE NUMBER OF BYTES TO BE FREED IN RO AND THEN USING THE REGISTER CONVENTION FREE THE AREA OBTAINED WITH THE REGISTER CONVENTION PUT THE ADDRESS OF THE AREA TO BE FREED IN A REGISTER XSNAP STORAGE 0 7 4 6 T NOREGS X ABEL THIS IS A SNAP OF THE AREA OBTAINED FROM GETMAIN OR RO R3 PLACE SUBPOOL NUMBER IN FIRST BYTE LR RO R9 PU HE PS NUMBER AND LENGTH IN RO FOR THE MACRO CALL PRINT GEN USING A FREEMIAN MACRO FREE THE AREA OBTAINED WITH THE REGISTER CONVENTION THE SUBPOOL NUMBER IS PLACED IN THE FIRST BYTE OF RO THE ADDRESS OF THE AREA TO BE FREED IS PLACED IN A REGISTER DESIGNATED BY THE A PARAMETER FREEMAIN R LV 0 A 7 PRINT NOGEN XRETURN SA DC F O RD DC 20F 0 END LOG DATA XSNAPOUT DD UNIT AFF FTOG6FOOL DATA SYSUDUMP DD SYSOUT A DATA INPUT DD HARDWAR1 01 FEB 1972 T PENN STATE UNIVERSITY COMPUTATION CENTER 360 67 CONFIGURATION this writeup pages 01 04 plus Diagram A separate INTRODUCTION This writeup briefly describes the devices included in the PSU 360 67 system and shows how they are connected together Each device is described below and diagram A shows the connections References are made to DEVICE ADDRESSES Each individually addressable devic such as a single disk drive card reader etc has a 3 digit hexadecimal number which uniquely identifies it to the system and is use
25. T introduction to information representation in computer memorv addressing similiaritv to FORTRAN vector witl index beginning at O rather than 1 elements of memory in S 360 byte halfword fullword doubleword positional notation number systems binary octal decimal hexadecimal conversion between them uses representations of binary numbers Two s complement One s complement Sign magnitude advantages and disadvantages TC 1 zero but harder for people OC 2 zeroes but easier to handle SM asiest to handle but slower circuitry NG STRUBLE CHAPTER 1 Look at ASSIST PART III more on information representation introduction to machine structure meanings of bit patterns 1 2 4 byte binary numbers charcters packed decimal good for people but wastes space floating point sign characterisitc and fraction structure of a very simple machine memory of 16 bit words 1 register 1 program coiunter a few instructions each with opcode and address explanation of basic instruction cycle 1 Fetch instruction from where program counter points 2 Increment program counter Decode instruction into its parts Execute instruction Loop back to 1 Owe Ww S 360 machine structure memory note abbrev K GP and floating point registers PSW refer to GREEN CARD x Begin instruction types 1 RR names with R examples
26. tA Ti UO ETACH HE PRIORITY OF MAIN IS RESTORED ECOND IS ATTACHI AGAIN USING THE ATTACH MACRO U ONLY THIS TI ROUTINE SECOND AGAIN THERE IS IN MAIN 7 HEN THIRD 15 ATTACHED BUT IT IS NOT GIVEN AN ECB AN ADDRESS IS PASSED IN THE ATTACH MACRO FOR THE ECB AND THIRD USES A POST MACRO TO SET HE ECB OND IS GIVEN AN EXIT ROUTINE IN THE EXIT S DETACHED AND CONTROL IS RETURNED TO MAIN ECB GIVEN TO SECOND AND A WAIT MACRO ISSUED CJ 2 FF FF 32 FF F HF X F F ACA F F FF FF ARCA H H H n EJEC PRINT NOGEN x Es WHEN SECOND OBTAINS CONTROL IT DETERMINES WHICH CALL IS BEING as ADE THEN IT OUTPUTS THE APPROPRIATE HEADING AND RETURNS S ECOND CSECT XSAVE TR NO L 2 0 1 GET ADD OF PARM LIST T SECOND PARM 0 4 GET VALUE OF SECOND PARM 4 CHECK FOR 0
27. OSHASP 04 When an allowable job becomes available the initiator obtains a REGION for the job from the FREE area also called the DYNAMIC area uses the information from the RDR to allocate DASD storage tape drives and other I O devices It then ATTACHes the first module of the program to be executed thus creating the JOB STEP TASK and WAITs until the job step completes T When a job step is finished the TERMINATOR part of the INITIATOR really so that the whole unit is called an INITIATOR TERMINATOR ffectively cleans up performing disposition of I O devices DISP parameter in JCL and releasing the REGION which had been acquired for the job step During this process job steps are essentially independent i e they could require different sizes of regions and might execute in different locations Note that the INITIATOR TERMINATOR must also control the skipping of steps as controlled by the JCL COND option During execution SYSOUT datasets are written to DASD to be printed punched later When the last job step of a job completes the INIT creates a work queue entry calling for the job s output to be printed punched C WRITING SYSTEM OUTPUT A program called a SYSTEM OUTPUT WRITER WTR can be STARTed by the operator to transcribe output from DASD to printers or punches or even tapes to be printed punched later Output can originally be gr
28. VARIABLES USED BY THIS SUBROUTINE VARIABLES CHANGED BY THIS SUBROUTINE etc DO block off large sections of comment cards Large blocks of comments can begin in whatever column is appropriate but in general should use most of the card since they will otherwise add a great deal of length to a program For the sake of appearance comments should be blocked off by blank lines SPACE n or lines of continuous characters The most common characters used for this purpose are asterisks in columns 1 71 or in just the odd columns An esthetic appearance can be obtained by placing an asterisk in column 71 of each comment card in a major block with lines of asterisks before and after the entire block of documentation DO flag instructions which will be modified during execution in order to make programming logic obvious This may be accomplished by using Ix x or S the latter EQU ed to zero for any modified field For example EQU 0 S gt INST MODIFIED IN EX STARS other statements STE 2 MVC 1 SET BUFFER LEN FOR LATER E USE AA other statementS MVC MVC OUTPUT S 0 5 MOVE VARIABLE BYTES INTO A OUTPUT BUFFER The above methods have been derived both from the examination of many professionallv written programs and from the authors own experiences Thus they are not arbitrary rules but techniques which have been widely use
29. using actual values For example SACTIVE EQU X 01 AJLSTATS gt job using CPU VI AJLSTATS SACTIVE show this job now active CLI AJLSTATS SACTIVE is this job active The above is much clearer than MVI AJLSTATS X 01 etc Even MORE IMPORTANT is the fact that it is infinitely safer and less error prone A useful method is to have 1 different DSECT for each distinct type of control block list element table etc plus perhaps 1 DSECI for a global control table A TYPICAL METHOD IS THEN TO LOAD AN address constant of a table CSECT into a specified register very early in the main program connect it to the DSECT with a USING then NEVER modify that register during the program Registers 10 11 12 seem to be the most favored for this type of use CS411FP3 07 IX SPECIFIC IMPLEMENTATION TECHNIQUES This section discusses the specific project covering specific modules data structures and techniques which may be of use in writing this project RA Kakaku RA NOTE SAKSISKE THIS INFORMATION IS ONLY TO ELP GETTING STARTED IT IS NOT REQUIRED THAT THE PROJECT BE STRUCTUR N THE WAY DESCRIBED HERE ALTHOUGH THE PROJECT MUST BE CAPABLE OF HE REQUIRED ACTIONS REGARDLESS OF HOW IT IS WRITTEN E ie HH EJ d Please note that the data structures and modules described ar desig
30. READINGS STRUBLE CHAPTER 6 pp 110 121 problems 7 9 10 ASM LANG 3 7 9 10 18 except variable symbols sequence symbols 19 21 29 33 section 5 EQU DC all except Bit Length Modifier Scale Modifier Exponent Modifier all types except E D L P Z Y S Q complex relocatability DS ORG LTORG END SPACE EJECT TITLE POP pp 24 34 except CVB CVD Logical instructions except TR TRT ED EDMK Branching except EX 02 04 give out final project discuss assembler interpreters concepts of assemblers 2 pass assemblers how to set up opcode and symbol tables indexed jump methods output desired go over structure of SIGMA 4 5 computer and its interpreter noting indirect addressing in particular HANDOUTS CS102FP1 01 08 general assembler interpreter descr CS102FP2 01 06 specific material for final project ASSIGN Final project due 13 March described in CS102FPx 02 07 decimal numbers and conversions zoned packed decimal to and from binary PACK UNPK CVB CVD equivalent codes using M D loops for decimal binary decimal examples of various formats conversions T READINGS STRUBLE Chapter 5 106 110 Chapter 218 228 228 233 02 09 misc review misc instructions program mask SPM instruction use of program mask review BXLE BXH etc 02 11 MIDTERM covered data representations most standard instru
31. I O REQUEST REPORT gives time jobname whenever an executing job stops to do I O CHANNEL ALLOCATION REPORT done every time a job actually receives control of a channel and gives time jobname and channel number It might optionally show the length of time the job will use the channel BATCH REPORT this report occurs at the end of a complete BATCH of jobs when all jobs in the BATCH have been INITIATED EXECUTED and terminated and the system is completely empty O At this time statistics can be printed about the BATCH Needless to say the appropriate counters must have been kept during the simulation T following items are to be printed time for entire BATCH TOTAL time last job terminated number of jobs run called NJOBS For each CPU or CHANN Total in the system HANNEL was being used to EXECUTI CJ CJ DO G bi ka in time during which CP T used 100 TIMEUSED TOTAL Degree of Multiprogramming gives the average number of jobs in memory over entire batch Needs an accumulator MULTI and a job counter INCORE both initially 0 Each time a job is initiated or terminated set MULTI MULTI INCORE interval since last INIT or TERM and increment decrement INCORE as is appropriate Finally at the end of a batch MULTI TOTAL gives the ave
32. XPRNT 0 1 SYNADRLS BR 14 DC F O DC 10F O DC CL100 4 DC F O DCB DSORG PS MACRF GM LRECL 80 BLKSIZE 80 RECFM F DDNAME INPUT EROPT ACC EODAD EOJ PRINT GEN SPACE 5 THE KEYWORD PARAMETERS FOR DCB AR E DSORG DA DIRECT ACCESS DDNAME S BUFL BUFF OPTCD BUFNO SYNAD DCB DS OP LTORG END DA UNIT SYSDA DAT 1002C 1010C EX A A EC SOU SOU F x FOUR D I HANGE T 1020CGANGE RECOR a HANGE R IRECTDD D NPU TANDARD DIRECTDD R READ I SEARCH TO BE MADE BY BLOCK IDENTIFICATION S DYNAMIC BUFFERING C CHECK ABSENCE DENOTES WALI W WRITE 1 SEARCH TO BE MADE BY BLOCK IDENTIFICATION C HCECK ABSECCE DENTOES WAIT ER LENGTH 40 R SEARCH TO BE MADE BY BLOCK IDENTIFICATION NUMBER OF BUFFERS ERROR ROUTINE ADDRESS ORG DA DDNAME DIRECTDD MACRF RISC WIC BUFL 40 TCD R BUFNO 1 SYNAD CHECKER D SYSOUT A D SYSOUT A D DSNAME amp amp TEMP DISP OLD PASS SPACI DJ 44 21 Tj Q O ps OOO ASGCG SYSGO INPUT U U U THIS JOB OUT PRINT NO EQUREGS CSEC XSAV CJ WENTX E a D DISP OLD PASS
33. 2 DISADVANTAGES SLOW ASSEMBLY macro processing can be very slow LARGE CODE if used improperly macros can generate larg amounts of code very easily If there are many copies of large blocks of code much space can be wasted OBJECT DECKS a macro cannot be assembled and an object deck of it gotten like a subroutine can i e if a call is made to a macro the macro definition must be included in the program or in a library while a CSECT may be saved as an object deck which is usually much smaller than the source deck B INTERNAL SUBROUTINES 1 ADVANTAGES SPEED although not as fast as in line code from a macro the code for an internal subroutine is usually faster than the linkage to an external one In particular values can be passed in registers and usually registers will not have to be saved SPACE internal subroutines require less space than generating the same code several times via macro expansions Zi DISADVANTAGES E if the same function is performed by internal subroutines in several CSECTS code is thus duplicated and space wasted COMPLEXITY in some cases in order to make efficient use of a number of internal subroutines it is necessary to set up fairly xtensive rules on usage of registers in a CSECT so that the linkage among them may be fast and small ASPRGTCI 07 C EXTERNAL SUBROUTINES 1
34. Briefly a job is to be selected from the list of waiting jobs by scanning down the list until the first one meeting desired criteria is found If no such job is found within the range of the list scanned then no job is initiated at this time The following options are a few Of the ways in which this range can be specified JSRI Initiate the first job if it fits in memory according to the memory allocation rules given below If it does not fit do not initiate any other job even though it may fit JSRT Initiate the first job on the entire list which fits CJ JSRv Initiate the first job which fits scanning up to amp JSRANG Note the effects of the above rules JSR1 ensures that the jobs are initiated in the order given by the JO option JSRT does the best in keeping memory full but it may also keep a large memory job waiting for a relatively long time JSRv is then a compromise between the others D MEMORY ALLOCATION FOR JOB INITIATION any different ways exist for allocating storage to jobs the ways used depend strongly on the purposes of the system and also on the computer hardware being used For example the S 411 computer series includes computers with a wide range of addressing relocation circuits The smaller ones require that a program be loaded as one contiguous block in memory and never moved from that area The medium systems use bounds registers so that although a program must be loa
35. CJ CS102AS1 01 T T COMPUTER SCIENCE 102 ASSIGNMENT 1 DUE This assignment covers simple input output binary arithmetic for fullword and halfword numbers and basic data movement and testing codes for handling such numbers AI BASIC PROGRAM The basic program should do the following A Read a card XREAD and print it out immediately called an ECHO CHECK standard practice The card contains 5 numbers punched on it which are to be scanned and converted XDECI to binary form and placed in 5 consecutive fullwords in memory Print the hexadecimal values of these 5 words 20 bytes using XDUMP B Perform the following computations in a straightforward way storing each result in name given using RX instructions where you can 1 F A B C 2 G A B C LCR useful 3 H A BE 4 I A B be careful watch for negativ s 5 J MOD A B i e remainder from 4 6 K A E B C D C Print all of the above values F K in hexadecimal XDUMP then also print them in decimal using XDECO and XPRNT print their values an headings all on one line D According to the sign of result H print one of the 3 messages H IS LESS THAN ZERO H IS GREATER THAN ZERO H IS ZERO 11 EXTENDED VERSION OF PREVIOUS PROGRAM Modify the previous program which
36. Execute cycle The it easy to share code i e it nstruction separately As an consider the following related instructions MNEMONIC HEX CODE BINARY COD E SA CR 19 0001 1001 CR CH 49 1000 1001 CH C 59 1001 1001 E Examine the bit patterns above Machine Format 00 RR 10 RX 01 Fullword 00 Halfword in this case Verb 1001 algebraic Compare In E INSTRUCTION ASSEMBLED ri 1901 2 3 4 49034002 4 5 6 59056004 The first teo bits give the the third and fourth give a Data Type The fifth eighth bits give the ther ssence is only 1 Compare which is branched to after the operands are obtained CS102TPA 01 LT COMPUTER SCIENCE 102 TOPICS COVERED HANDOUTS WINTER TERM 1972 MASHEY The handouts given are described in file CS102HN DATI CJ TOPICS HANDOUTS READING ASSIGNMENTS 1 2 01 07 READI 01 10 READI HANDO 72 introduction to course prerequistites 101 401 equiv listed text materials for course 1 STRUBLE ASSEMBLER LANGUAGE PROGRAMMING IBM SYSTEM 360 2 IBM SYSTEM 360 PRINCIPLES OF OPERATION POP 3 IBM s 360 OS ASSEMBLER LANGUAGE 4 PSU ASSIST INTRODUCTORY ASSEMBLER USER S MANUAL 25 cents at 426 McAllister 5 IBM S 360 REFERENCE CARD GREEN CARD BRING TO CLASS E tA
37. I MAXIM FUNCTION SUBPROGRA Write a function in assembler language consisting of 1 CSECT named MAXIM which accomplishes the following A Follows standard 0S 360 calling conventions receiving the address of an argument list in register 1 There may be a variable number of arguments in the list Each address in the argument list points to a fullword somewhere in storage B The program should algebraicly compare the fullwords addressed by the argument list and place the value of the maximum one in register 0 as a result then return control to the caller C In writing this program DO NOT USE XSAVE OR XRETURN macros II WRITE FORTRAN TEST PROGRAM TO TEST MAXIM his program should test MAXIM by using it as a function i e it will have statements of the form I MAXIM 1 2 5 10 4 5 in it It should have at least two tests of this form Print out the results to show they are correct Use the above set of data plus another of you own choosing III WRITE ASSEMBL T R MAIN PROGRAM TO TEST MAXIM Write an assembler main program which has the same logic and test values as does the previous FORTRAN program It should use standard 0S 360 linkage its own save area of course and utilize the IBM macros SAVE and RETURN It should do the following A As given by the LINKAGE writeup obtain the PARM field see if any exists by testing for z
38. A A A F FX F NP F Ax HE ANSWER PARM IS A FULL WORD IN CORE STORAGE FOR THE RESULT OF THE EXTRACT MACRO S PARM INDICATES THIS JOB HE FIELDS PARM SPECIFIES WHICH ONE a Ti lw ESIRE EXTRACT ANSWER S FIELDS PRI SR 10 10 ZERO REGISTER 10 LA 9 ANSWER GET ADDRES OF ANSWER I 10 3 9 GET PRIORITV OF MAIN SRL 10 1 DIVIDE DISP PRI BV 2 LNR 10 10 MAKE DISPATCH PRI NEG THE EP IS THE ENTRY POINT FOR THE LOAD MODULE TO BE ATTACHED PARAM SPECIFIES A LIST OF PARM TO BE PASSED TO THE LOAD MODULE AND VL INDICATES AN INDEFINITE NUMBER OF PARMS ECB SPECIFIES THE ADDRESS OF ANDEVENT CONGROL BLOCK TO BE ECB SPECIFIES THE ADDRESS OF AND VENT CONGROL BLOCK TO BE ECB SPECIFIES THE ADDRESS OF AN EVENT CONGROL BLOCK TO BE POSTED WHEN SECOND COMPLETES EXECUTION HE LPMOD SPECIFIES AN INTEGER VALUE TO BE SUBTRACTED FROM HE LIMIT PRIORITY OF SECOND HE DPMOD GIVES ANVALUE TO BE ADDED TO DISPATCH PRIORITY OF MAIN IN THIS EXAMPLE A REGISTER VALUE HE ADDRESS OF THE TCB OF S
39. ECT XSAVE RETURN SA RG MSUB4 X 1000 D THE FOLLOWING STMTS ARE JCL KLUDGES REQUIRED TO ET EVERYTHING IN THE RIGHT PLACE GIVEN THE WAY THE ROCEDURES ARE YSLIN DD NCLUDE OBJ GET FOR MODO NO OVERLAY AME ODO OVERLAY OPT WILL BI ULLED OUT NCLUDE OBJ GET ANOTHER COPY OF THE OBJ VERLAY ALPHA DEFINE BEGINNING NSERT SUB1 SUB1A SUB1B SUBIC SUBID SUB1 amp ITS SUBRS VERLAY ALPHA BACK TO SAME PLACE AS ABOVE NSERT SUB2 SUB2A SUB2B SUB2C SUB2 amp ITS SUBRS AME MOD1 3 SEGMENT OVERLAY E NCLUDE OBJ COPY FOR 10 SEGMENT ERLAY VERLAY ALPH ORIGIN AS BEFORE NSERT SUBI POSITION SUBI AFTER ROOT VERLAY BETA ORIGIN FOR SUBIA B NSER SUBI PUT SUBIA AT END OF SUBI VERLAY BETA BACK TO END OF SUBI NSERT SUBI PUT SUBIB AT END OF SUB1 ALSO VERLAY CHI1 END OF SUBIB NSERT SUBI PUT SUBIC AT END OF SUBIB VERLAY CHI1 BACK TO END OF SUBIB NSERT SUBI PUT SUB1D BEGIN LIKE SUBIC VERLAY ALPHA BACK TO WHERE SUB1 BEGAN NSERT SUB2 BEGIN SUB2 WHERE SUB1 DID VERLAY BETA2 ORIGIN FOR SUB2 S SUBROUTINI NSER SUB2A PUT SUB2A AT END OF SUB2 VERLAY BETA2 BACK EVEN WITH SUB2A NSERT SUB2B PUT SUB2B IN SAME VERLAY BETA2 ONCE MORE NSERT SUB2C PUT C IN SAME AS A AND B AME MOD2 10 SEGMENT SINGLE REGION MODUL NCLUDE OBJ COPY FOR 3 REGION OVERLAY VERLAY ALPHA ORIGIN AS BEFORE NSERT SUBI POSITION SUBI AFTER ROOT VERLAY BETAL ORIGIN FOR SUB1A B NSER SUBI PUT SUBIA AT END OF SUBI VERLAY BETAL BACK TO END OF SUB1 NSERT SUBI PUT SUBIB AT END OF SUB1 ALSO VERLA
40. Lt LIE THIS PROGRAM DEMONSTRATES THE BSAM I O TECHNIQUES JES 1 USES MACROS READ WRITE CHECK POINT DCB OPEN CLOSE AND A READ POINT READ FORM 2 WE WILL READ A CARD FROM THE READER THEN ECHO PRINT IT TO THE PRINTER THE CARD IMAGE IS THEN WRITTEN TO DISK USING THE READ POINT FORM Pik AT EOD A POINT MACRO IS ISSUED TO REPOSITION THE DISK SO THAT THE FIRST RECORD CAN BE RECOVERED fi AT EOD FROM THE DISK THE DCBS ARE CLOSED AND ff CONTROL IS RETURNED TO THE CALLING PROGRAM EJE 3 TIMEING CONSIDERATIONS 50 SECONDS 1000 RECORDS EXEC ASGCG PARM DATA MAP SOURCE INPUT DD TITLE BSAM I O EXAMPLE BSAMIO CSECT SET STANDARD OS LINKAGE XSAVE TR NO SPACE 5 OPEN ALL DCBS CARDCB TO READ DATA PRNTDCB TO ALLOW WRITE TO THE PRINTER AND THE INTERMEDIATE DISK STORAGE OUTIN OPTION IN OPEN MACRO ALLOWS A WRITE OPERATION FOLLOWED BY A READ OPERATION WITHOUT AN ADD
41. subroutines a save area just ab since it will alway conditions across a The problems described above are typically nt cod segm ove these cod s have that same nts set up the same USING setting a specific register to point to the EXecuted instructions and data sections value DDRI handled either by making conditions or by beginning of the internal If register 13 points to it can be used this way Getting the same USING n entire multi entry CS ENTRYX XSAVE L BASEREG A CSECTNAM USING CSECTNAM BASEREG Note that the above can be accomplished with the XSAVE ECT can be done AD operand ASPRGTCI 04 D COMBINED FORMS In some cases it is convenient to combine the ease of use of the macro with the small size of internal or external subroutines In this case the macro expansion sets up any needed arguments saves registers etc then generates code to invoke the subroutine The subroutine then provides the major portion of the processing code any needed large data areas etc macros XDECI XXXXPRNT and Examples of the combined form are the XDECO XPRNT XSNAP which call XXXXDECI XXXXSNAP respectively following XXXXDECO Two different extremes exist in writing combined forms 1 COMBINED FORM STANDARD LINKAG CJ In some case the calling sequence to invoke an external subrou
42. 5 NUMBI DOUBLE 5 DOUBLI 15 V BEGIN 14 15 15 V OUT 14 15 dy E 1 REPEAT 0 4 5 EJ co wW Ej BEGIN NOGEN 7 0 ST BEGIN OF 4c 16F F 14 ASTERICK 17 14 F3 6 6 READ 6 F 16 F2 6 0 INPUT 64 273 BLNK 2 65 11 INPUT 6 LoL Er 10 8 6 1 6 10 07 ASTERICK CHECK ASPRGTC1 S 360 Assembler Language Programming Techniques Topic Using Macros It is generally important in any computer program to John R Mashey Winter 1972 Program Modularity and Parametrization Methods Internal Subroutine External Subroutines This topic pages 01 08 avoid any procedure mor times than necessary It is generally best t something one tim of a program this macros In assembler language internal subroutines then have it available for later use in man there are thr main ways and external subroutines r writeup describes disadvantages of each DESCRIPTION D A MACRO INSI A macro instruction is defined language program EM macro language statements at the point of i ly from call to call of an assembly library a SYS may vary great DEFINIT TION A macro definitio which gives STATEMENT includes 0 or more machine instructions to be ge to direct the expansio is terminated by the M ach of these techniques gives the advanta and not
43. E RALL OUTLIN J A OV This section gives a very brief outline of lecture topics by day Several empty days are left for exams problems and possible expansion of any of the topic areas DAY TOPICS AG Introduction to Course Prerequisistes Review X Macros 2 0S 360 Linkage Conventions 3 Dump reading Debugging amp good programming technqiues Common 4 interrupts and interpretation of dumps 5 Overall macro concepts amp structure FORTRAN analogies 6 Details on macros examples most statement Lypes 7 Finish macros Aspecial argument handling character scanning 8 Miscellaneous cleanup on S 360 Assembler CNOP V cons etc 9 Addressibility DSECTS Multiple USINGS Useful techniques at this point S 360 Assembler Language is finished 10 Assembler Comparison 1 2 4 pass SPASM ASSIST Assembler G 11 Introduction to operating systems history basic types 12 Architecture summary CPU Memory Devices Channels Memory 13 structure and communication memory protection 14 I O devices sequential DASD capacities characteristics 15 Finish I O devices I O Channels types and programming S 360 16 Channel level programming Interrupt handling show examples 17 Final Project explain simulation concepts system overview 4 8 odule management types of modules REENTRANT etc loader 19 Link editor Program Fetch Overlays 20 User overview of 0S 36
44. due 02 02 pwan HANDOUTS CS102AS1 03 labeled CS 102 AS2 also indexing READINGS STRUBLE CHAPTER 5 start on STRUBLE CHAPTER 7 01 21 condition code branching instructions loops condition code values and encoding BCR BC Extended Mnemonics subroutine example of flowcharti READINGS STRUB 01 24 finis Explain BX forward BX show need recommended for use over BC 4 BALR BAL and s BCT BCTR usage including decrementing regs basic loop to sum array of numbers ng and good design versus kludge programming E Chapters 7 8 5 h loop control begin on USING DROP linkage H BXLE instructions give typical setups a E loop backwards BXH loop BXH scan loop for USING command give rules for computation of base displacements minimum base displacement for those which are available higher numberred register if several have same begin conventions exaplin registers 15 14 usage on entry HANDOUTS LINKAGE 0S 360 linkage conventions READINGS STRUBLE Chapter 5 LINKAGE HANDOUT 01 26 savearea linkage ans ome review Describe 1 8 fullword save area go through the standard code used at beginning and end of a routine calling methods Do not work on argument passing just normal code isc instructions IC STC start on Shifts Various review for problems Note general usage of registers get students into g
45. T LT 1 MEMORY REGISTERS AND PROGRAM STATUS BITS The MEMORY of the SIGMA 4 5 model W72 is composed of 1024 WORDS addressed at locations 0 l 2 1023 Each is 32 BITS long The SIGMA 4 5 contains 16 REGISTERS each 32 bits long numbered from 0 to 15 Of These all may be used for holding operands and doing arithmetic while registers 1 7 only may be used as INDEX REGISTERS in address calculations A PROGRAM STATUS WORD is used to keep the current status of the SIGMA 4 5 computer Of the 32 bits in this word the following uses are made of BITS NAME USAGE 0 1 CONDITION CODE exactly like S 360 condition code 2 14 UNUSED 15231 PROGRAM COUNTER 17 bit address of the next instruction in memory to be executed NOTE all numbers ar ncoded as 32 bit signed numbers using two s complement notation 2 INSTRUCTION FORMATS ENCODING ACTIONS The SIGMA 4 5 has two basic instruction formats RX Register to Storage Indexed almost exactly like S 360 RX RI Register Immediat uses Immediate operand in instruction The layout of the RX format instruction is as follows 0 000000 000 00 EE 1 ET q T 202420 DA 2 2 2 2 282343 BIT 0 1234 5678901234 5678901234 56 7 89 0 1 NAME I OP R X M SIZE 1 7 4 3 17 bits The various fields above are used as follows I INDIRECT
46. lists the mn RX RI whet exactly CS102FP2 06 emonic opcode for each machine hexadecimal 7 bits her or not the instruction sets the what the instruction does If it sets the CC in the same way unless otherwise specified NAME CODE TYPE CC SET ACTION AI 20 RI YES ADD IMMEDIATE add Immediate field to reg R LI 22 LI NO LOAD IMMEDIATE extend immedaite field to 32 bits load that value into register R MI 23 RI YES MULTIPLY IMMEDIATE extend immedaite field multiply with register R truncate to botain low order 32 bits in R setting CC for result WAIT 2E RX NO WAIT execution terminates EA placed into specified register as a completion code AW 30 RX YES ADD WORD like S 360 A CW 31 RX YES COMPARE WORD like S 360 C LW 32 RX NO LOAD WORD like S 360 L MIW 33 RX YES MODIFY AND TEST WORD the R field does not refer to a register instead it is taken as a signed 4 bit number 8 to 7 extended to 32 bits and added to word at EA setting CC STW 33 RX NO STORE WORD like S 360 ST MW 37 RX YES MULTIPLY WORD like MI xcept operand taken from EA location instead of immediate field SW 38 RX YES SUBTRACT
47. 08AA ONEXT RWK2 END QQ amp SYSNDX BB not expanded XSNAP LABEL amp MSG T NO STORAGE 0 RWK2 amp LEN RWK2 BCT RWK1 000008AA 08BB EQU He OT HER USEFUL MACROS OPTIONAL CS411MC2 06 LNMBR The following macros may be required amp LABEL LREMV amp RND amp RLK found to be useful but are not LREMV removes the NODE addressed by amp RND from a LIST assuming that register amp RLK addresses the LIN NODE add LINK field addressed by amp R ressed by amp RND i K fiel the val LK are the same ES present ld of the NODE which points to the in amp RND and the value in the HDR amp MSG amp RGW RWK1 RWK2 in the list headed by amp HDR placing result in first register given by amp RGW and using second amp LABEL LNMBR amp LNMBR counts the number of NOD one as a work register I f MSG printed with amp MSG as a heading I TE STING THE MACROS WHAT TO HAND IN specified also the number is Write a program to test your macros as follows 1 Define a list of 1 8 byte K and LISTB i e these ar ZE a single NODE in columns 1 5 empty nodes EY and 12 byte DATA areas names of NODE from the FREE list f you must TE tn Eb 3e 4 LISTA fo first remove the node from LISTA Read in 10 data cards each called FREE list each with Also define two
48. 1K blocks of memory available for entire system up to IK blocks allocated to operating system and thus not available to be allocated for user programs page sized round value i e all requests are to be rounded up to this size EX if 1 every 1K block of memory can be allocated separately if 128 then each job is allocated at least 128K i e jobs occupy 128K 256K 348K This unit will be referred to as a PAGE an arbitrary limit on the maximum number of jobs in memory at any one time Note that the special case amp JOBLIM1 1 implies UNIPROGRAMMING while amp JOBLIMI gt 1 implies that MULTIPROGRAMMING is at least possible a calculated limit on the number of jobs possible in memory assuming that each job must be allocated at least amp MEMPGR K bytes of storage This can be calculated as follows amp JOBLIM2 amp MEMSIZE amp MEMOS amp MEMPGR final limit on number of jobs MIN amp JOBLIM1 SJOBLIM2 This variable would be the one used to actually control code generation MEMORY ALLOCATION ALGORITHMS The following are common techniques for determining WHICH block of MAA1 MAA2 MAA3 memory will be used to satisfy a request In each case a block is given to a job for the duration of its execution and then returned to unused status when the job terminates In every case this block just returned must be MERGED with an
49. S S 0 REF 1 SPACE 5 CHECK DECB LOAD REGISTER 3 WITH BUFFER ADDRESS THEN MOVE NEW DATA TO BUFFER AND STORE BUFFER ADDRESS IN DCEB OF WRITE STATEMETN L R3 DECB 12 MVC 0 30 3 DATA ST R3 DECBW 12 SPACE 5 POSITIONAL PARAMETERS OF WRITE MACRO ARE 1 DECB NAME TO BE CREATED BY WRITRE MACRO 2 YPE DI WRITE THE BLOCK AT THE DEVICE ADDRESS PROVIDED AT THE BLOCK ADDRESS OPERAND DATA AND KEYS ARE WRITTEN 3 DCB ADDRESS DIRECT 4 AREA ADDRESS S INDICATES DYNAMIC BUFFERING ADDRESS PROVIDED IN DECB 12 PREVIOUSLY BY READ AND STORE 5 LENGTH S INDICATES LENGTH CMMES FROM BLKSIZE IN DCB 6 KEY ADDRESS 0 INDICATES KEY NOT WRITTEM 7 BLOCK ADDRESS ADDRESS OF THREE BYTES OF MAIN CORE CONTIANING RELATIVE BLOCK ADDRESS WRITE DECBW DI DIRECT S S 0 REF 1 SPACE 5 CHECK DECBW B LOOP PRINT NOGEN CLOSE DATA SETS AND RETURN TO 0S CLOSE INPT DIRECT XRETURN SA ERROR ROUTINE PROVIDED BY SYNAD PARM IN DCB CHECKI DATA pu EF INPT C x F X HA DATA XSNAPOUT D DATA SYSUDUMP D ER SYNADAF ACSMETH BDAM
50. XSNAP XSET process EXECUTION TIME suppose that a program runs a reasonable length of time appears to be running correctly then bombs in a given situation If debug output is produced for the earlier part of the run record limits are exceeded Therefore a convenient way is needed to determine whether debug output is needed during execution One such way is to 1 Read counters flags etc into a global area of the program from the SSDEBUG card 2 Use XSNAP IF option to test these counters flags etc and either print debug output or skip it In particular the above approach is of considerable merit when OBJECT DECKS are being used since this way it is possible to get an object deck which is loaded with debug output code but never actually prints such output unless requested from input cards Note that SSDEBUG cards can then be inserted in the input stream in position just BEFORE trouble occurs and removed when debugged CS411FP3 05 VIII GENERAL IMPLEMENTATION METHODS This section presents general methods helpful not only for this project but for any other assembler language program of more than trivial size Some of these methods have analogues in high level languages These hints may save much time and effort A MODULAR PROGRAMMING The program should be constructed of a number of different CSECTs For OS 411 the appr
51. accessing of suboperands usage of amp SYSLIST many examples Program design 5 minutes worth importance of good design flowcharting debugging techniques 9 01 25 macro problems multiple entry CNOP external symbols answer questions on macros note problems with k amp SYSECT give examples of ENTRY usage multiple entry points and the usual ways to use them go over CNOP and how to use it note differences between external symbols which exist as symbols after assembly and internal ones which do not note USING across CSECTS and fact that CSECTS may not be in same order as in assembly HANDOUTS ASPRGTC1 01 08 macros internal subroutines and extenals and when to use them ASBROPS2 01 03 ASSIST base register assignment ASREPLGD 01 11 ASSIS REPLACEMENT USER S GUIDE ASSIGNMENT write base register routine ASBROPS2 using either method A worth 5 points or method B worth 10 points due 02 18 10 01 25 addressbility USING DROP DSECTS Methods for addressing large data areas using A type adcons Typical setup of 1 2 local bases plus one for global table DSECTS write a simple DSECT then reference in code showing that each reference meerely requests a base displacement to be computed Show equivalent code written explicitly and show why DSECTS are better for readability and ease of modification also show method of computing leng
52. each of t control unit AVERAGE 25 millisec HI o move 60 or 330 337 430 433 IS AS FOLLOWS 2314 8 drv 233 megabytes 2319 8 drv 233 megabytes 2314 4 drv 116 megabytes 2301 2 drums 8 megabytes TOTAL 590 megabytes approx doa DATA TRANSFER RATE AVERAGE AT MAX o EADS n n be removed ENCY EADS to correct cylinder 130 or 135 DRIVE n 20 OVABLE hich t total of 22 holds one disk surfaces HEADS here are 200 Each track can and another 12 5 millisec millisec 312 000 bytes sec on channels 2 3 he three storage facilities contains its own and each drive is numbered accordingly Tae 4 HARDWARI 04 SECONDARY STORAGE SEQUENTIAL DEVICES 240x 2402 III 2403 III MAGNETIC TAPE DRIVES read write tape at maximum density of 800 BPI bits inch 9 tracks per tape 2 of the drives also read write 7 track tapes Each group of 4 drives is connected to one SELECTOR SUBCHANNEL of the MULTIPLEXOR CHANNEL The control units for these drives ar contained in the 2403 units MAXIMUM TRANSFER RATE 90 000 bytes sec 90KB using tape speed of 112 5 inches per second tape gaps of 6 inch between blocks of data
53. some pointers for storage management gives a pointer to each routine associated with a defined SVC initializing the SVC table which number It also sets up to be able to obtain modules from the SYS1 LINKLIB which co establis ntains the heaviest used load modules for the system hes communications with the operator and also 5 For any system having one NIP loads reentrant modules into the LINK PACK These modules can be used during following execution and are located at the high end of memory In a system with fast core LCS the LINK and the PACK can be split high end of LCS residing at both 6 With the addition of various other miscella NIP prepares a REGION which will contain the MASTER SCHED ULER the high end of fast core neous operations which is the program doing overall job scheduling and operator communication to the MASTER SCHEDUL It then can pass control LINK or XCTL ER the system is finally ready to run jobs and At this point memory layout fast core only is as follows HIGH ADDRESS LINK PACK reentrant modules MS MASTER SCHEDULER FREE AREA dynamic for problem programs SOS SYSTEM QUEU SPACE contains space for system control blocks TCB s etc LOW ADDRESS NUCLEUS NOTE in systems with HIARCHY SUPPORT FREE AREA MS and LINK PACK would
54. the following general philosophy If all other things are equal between two jobs then the one placed earlier on the list should be the one which arrived in the system earlier and has thus waited longer having the higher priority lower PRIO value smallest storage requirement smaller value of SP smaller running time requirement T value smaller category number CAT JO1 JO2 JO3 JO4 JO5 JO6 JOT JO8 CS411FP1 08 The following order options can be considered FIFO First In First Out or FCFS First Come First Served A new JOB is always added to the end of the list so that the earliest arriving JOB is listed earliest STRAIGHT PRIORITY JOBS are kept in order from highest to lowest low PRIO values to HIGH PRIO values If two JOBS have equal priority then th arlier arriving one is first CATEGORY PROGRESSION JOBS are kept in order by CATEGORY from lowest CAT to highest Within each category JOBS are listed either FIFO or by PRIORITY or both LENGTH OF TIME REQUIRED BY JOB either JO4S SJF SHORTEST JOB FIRST in order by T small to big JO4L LJF LONGEST JOB FIRST in opposite order from JO4S SPACE REQUIRED BY JOB either JO5S smallest job first JO5L largest job first INPUT OUTPUT REQUIREMENT SELECTION Distinguish between I O BOUND JOBS IORL large relative to IO
55. then t If PRE job CPU s control of CPU s the process of allocating CPU s is as follows until it must perform input output IOIN paramet When a the use of I immediately er JOB requires 1 0 O CHANNEL s it must WAIT until When a for the leng its 1 0 oper JOBs may use use of the C PREEMPT a JO by whatever E When a JOB obtains control th of time calculat ation is complete it of ed TE If CPU s at any or when the RJQ changes for are IDLE i e not being on the RJQ are given control described below a job hich currently has job which s n control which can be calculated from the as described in the IOIN options it must then compete with other jobs for If it cannot obtain the desired CHANNEL becomes available a CHANNEL it keeps control of it from the IOIN parameter As soon as releases the channel so that other it and then reent rs the RJQ so that it may compete for E PU s again If PR EMP B CURRENTLY HAVING CONTROL OF A CPU ng scheme is being used orderi job first enters t the T param number of mi ter W lliseconds used a he RJO henever a JOB uses the CPU nd depending on the option ION is allowed it may immediately if it ranks earlier it is allowed a total time from it is charged the it may be
56. to do what it is supposed i e In front nvenient to omit th later ones than the nts for values which may not be needed always ha ar arguments make it positional UBLISTS or amp SYSLIST for variable numbers of E TUA BODY OF MACRO BUILDING MACRO TYP L CJ ATEM ED ENTS TO BI GENI ERAT ROTOTYPE e macro nds and h serve finition typical enerally enerated needed be a goo punch it macro to do if an group since earlier or Lor re most guments a COMMANDS ASPRGTCI 02 2 INVOCATION A macro can be called merely by writing its name and supplying it with any needed arguments Note that a label on a macro call is never generated and is thus UNDEFINED unless the macro definition is made to generate it on some model statement B INTERNAL SUBROUTINES Internal subroutines are sections of code written as parts of a given control section CSECT and are only used inside that CSECT Like external subroutines internal subroutines can of course call others They are typically used for small to medium sections of code which are needed at several places in a CSECT but are not needed by any other or are not big enough to warrant the overhead in making them external subroutines 1 DEFINITION It is often typical to place a group of internal subroutines near the end of the code section
57. 0 5 MOVE VARIABLE BYTES INTO x OUTPUT BUFFER The above methods have been derived both from the examination of many professionally written programs and from the authors own experiences Thus they are not arbitrary rules but techniques which have been widely used and proven to be effective aids in programming assembler language LINKAGI ti l ps STANDARD LINKAGE CONVENTIONS Charles Pfleeger Under OS 360 certain conventions have been established regarding the use of registers These conventions will have been followed when you the problem programmer receive control from the system they should be followed for any routines which you call or for communi cating with the system e g system macro calls SVC s returning control etc Following these conventions will make your code easier for someone else to follow Certain debugging aids are also available for those who adhere to standard conventions In general unless there is a strong reason to deviate thes conventions should be employed REGISTER 14 is called the return register and contains the address to which this routine is to return upon exit REGISTER 15 is called the entry point register and contains the address through which this routine was entered Note that tempor ary addressability may be established by USING entrypointname 15 If this routine calls no other routines register 15 may be used
58. 2 Q O x x OV O x YOM Xx OW Pl O d O x t t r t t gt Q ti ti U D Q ti R R p ti a SUB2B URN SA SUB2B X 2000 2 a 4300 U DA SUB2C a WU aa E E a ti lt CJ TURN SA SUB2C X 2000 LE MSUB1 2 3 4 CSECTS CALLED FROM ALL OVER MSUB1 CSECT d O x x O O x lt O O x lt H W W 1 MSUB3 ALL MSUB4 ACE 1 ETURN SA G MSUB1 X 3000 A 2 ot Q El MSUB2 Q H ti ti 1 MSUB3 1 MSUB4 ACE 1 ETURN SA G MSUB2 X 4000 A gt gt 2 Q Li MSUB3 a A gt 1 lt EE TURN SA RG MSUB3 X 1000 2 Q O x lt O QQ O xa O O x O G O x lt Q O Wa D Q ti MSUB4 OBJI FO x Q FU CJ HOHOHOHOHOHOHOHOHOHOHTZHOHOHOHOHOHOHOHOHOHZHOHOHZHb O
59. 3 2 MTW 1 3 2 STW 1 2 ORG LABEL MW 1 2 SW 122 BDR 1 3 2 AWM 1 3 2 BCS 1 3 2 RD 1 3 2 WD 1 3 2 WAIT 2 2 WORD RES 2 RES W 2 RES 2W 2 END S JOB LEGALOPS2 0 THIS PROGRAM TEST THE RECOGNITION OF L I 172 E 1 2 I 1 2 AW 2 AW 2 AW 2 3 AW 1 2 3 AW 1 5 AW Lato AW 1 3 AW 1 5 3 AW 1 2 AW 1 2 AW 1 S 2 3 AW 1 2 3 AW 1 SYMBOL AW 1 SYMBOL AW 1 SYMBOL 7 AW 1 SYMBOL 7 AW 1 SYMBOL 1 2 AW 1 SYMBOL 1 2 AW 1 SYMBOL 1 2 SYMBOL RES 5 END S JOB PASS1ERR 0 THIS PROGRAM CONTAINS ERRORS IN LAB LI 5 0 LI 3 0 SWR 3 0 2 ILLEGAL OPCODE AI 2 4 LOOP MI 2 CONST 7LAB SW 2 CONST 1 ILLEGAL LABE STW 2 SAVE UNDEF IND SYMBOL ORGLINE ORG LOOP NO LABEL ALLOWED RD 3 AREAL UNDEFINED LABE STW 3 SUM OUT WD 3 AREA AWM 3 SUM BRANCH BDR 2 LOOP ELS AND ASSE T BL ERANDS ER OPS 2 BADLABI EA 5 SUM LW OUT NOT TW SU ARE Dnw Ww tis ti jann S R THISLABE T WD FOLLOWING R ES LISTOO 5 SUM 0 OUT 100 1 W 9999 1100 0w 1 10W w 10 W W 12345A 10W ONG ANDS MTW ENDLABI END IL S JOB AI LI LI LI LI LI LI LI LI LI AW AW AW AW AW AW AW AW AW AW AW AW AW AW AW AW AW AW AW AW AW PASS2 THIS PROGRA RI FORMA OMMI MULTI E BLANK CARD FOLLOWS ORG TO
60. ADVANTAGES SPACE if written as an external subroutine code can be usefully called from almost anywhere in a program Thus there is only one copy of it and it generally will occupy the least space SEPARATE COMPILE ASSEMBLY a routine written as a CSECT can be assembled separately from the rest of the program an object deck can be obtained and translation time generally saved The routine may of course be written in a different language than the rest of the program 2 DISADVANTAGES LINKAGE TIME if standard OS 360 linkage is followed a fair amount of execution time and object code space can be consumed by this linkage More efficient nonstandard linkage can be used instead but this brings with it the disadvantage of nonuniformity and lack of generality D COMBINED FORMS 1 ADVANTAGES In general the combined forms can possess all the advantages of the separate forms especially since the macro portions can generate different code depending on circumstances thus the code for the same macro might expand in line in one case and generate an out of line call to a routine in another 2 DISADVANTAGES COMPLEXITY it of course requires somewhat more planning and code to set up a good combined form system since both a macro and module must be created and meshed together properly
61. ASSIGNMENT T 1 Punch up the following program and run YOUR JOB CARD EXEC ASACG SYSIN DD MAIN CSECT THIS PROGRAM ILLUSTRATES XDUMP AND PROGRAM INTERUPTION BALR 12 0 THESE TWO ST S ARE FOR USING 12 COMMON LINKAGE CONVENTIONS LA 3 CARD PTR TO CARD IMAGE READ IN XREAD CARD 80 READ DATA CARD XPRNT CARD 80 ECHO PRINT XDECI 4 0 3 CONVERT DECIMAL TO INTERNAL HEX XDECI 5 0 1 CONVERT NEXT ON CARD THE NEXT STMT PRINTS CONTENTS OF USERS REGISTERS NOTE REG 4 5 XDUMP B 4000 ABEND BRANCH OUT OF PROGRAM CARD DS 80C END DATA INPUT DD 100 1024 2 This next program is a batch run of 5 jobs each terminating abnormally The program is stored on RJE file Punch up the follwoing cards EXACTLY to run the program YOUR JOB CARD EXEC ASACG PARM BATCH SYSIN DD INCLUDE RABO1 BATCH 3 To merely get a listing of the prog in YOUR JOB CARD INCLUDE RABO1 PRINT INCLUDE RABO1 BATCH 2 use the following cards CS102M1 02 A GUIDE TO S 360 MNEMONIC OPERATION COD T E a l INTRODUCTION The beginning programmer facing the varietv of operations available on a modern large computer is often overwhelmed bv the large number of operations and complexitv thereof In some case
62. DON T bury code with too many interspersed comments If so many comments are necessary place them in blocks ahead of the program segments and not in the middle DO put a comment on nearly every machine instruction Comments are also helpful for explanations of variables and flags Each comment should describe the function of its statement and generally it alone If a comment is needed to describe the function of a block of half a dozen cards it probably should be placed on a comment card preceding the block of code These comments should be punched when the program is originally punched A good technique is to add these comments while keypunching the program Often this results in catching many mistakes at that point It is noted that few novice programmers do this while most experts do It is also noted that many programmers who do this wish they had started doing so earlier since they realize how much time they had wasted by not commenting the original deck DO use TITLE SPACE and EJECT commands The command ITLE A HEADING MESSAGE skips the listing to a new page and prints the heading message at the top of every page until another TITLE command is issued This not only clearly labels your listing but it saves time in looking through a listing which is more than a few pages long The command EJECT skips the listing to a new page and is useful in b
63. For all jobs 1 an I O REQUEST or FACH job the I O operation is completed the and it returns to the RJQ to once time until a job gives up the CPU is always found is given to t amp IOINTVL gives the time until EST is taken from the IOIN parameter on For example UNIFORM the actual interval is computed from a UNIFORM distribution using the IOIN parameter as the M any value from O to 2 IOIN is equally likely EAN La Seg CS411FP3 01 B I O REQUEST QUEUE SCHEDULING When I O request occur for channels currently in use the jobs making the requests must enter queue s to obtain channel use These queue s can of course be ordered in various ways corresponding to the JO and EO options given previously However for this type usually only several of the possibilities are actually done 1001 FIFO the first job requesting a channel gets it first 1002 PRIORITY the highest priority lowest PRIO job waiting for a channel obtains it even if others arrived earlier 10061 I O BOUND JOBS are favored over CPU bound jobs with this calculation made either statically or dynamically NOTE FIFO is the most commonly used but sometimes PRIORITY is used especially if there are one or more jobs in the system which absolutely require fast I O service usually in Real Time or Process Control systems also containing batched jobs Es I O REQUE
64. PRINT GEN DATA CONTROL BLOCKS FOLLOW SPACE 1 DATA CONTROL BLOCK FOR THE CARD READER IOCRDDCB DCB DDNAME CARDS JCL DDNAME DEFINING THE DATA DSORG PS PHYSICAL SEQUENTIAL DEVICE ACRF GM GET MOVE MACRO FORM USED EODAD IOEOF1 END OF DATA EXIT ADDRESS TO GOTO i RECFM F RECORD FORMAT FIXED NOT BLOCKED RECL 80 OGICAL RECORD LENGTH CARD SIZE 1 BLKSIZE 80 BLOCK SIZE CARD SIZE UNBLOCKED NOTE THE LAST 3 PARAMETERS COULD BE LEFT OFF HERE SPACE 1 DCB FOR THE PRINTER PSECIFYING ALL NEEDED IOPRTDCB DCB DDNAME PRINT JCL DDNAME FOR THE PRINTER DSORG PS DATA SET ORGANIZATION PHYS SEO ACRF PM ACRO FORMAT IS PUT MOVE RECFM FA RECORD FORMA FIXED HAS CARR CONTRL RECL 133 OGICAL RECORD LENGTH PRINT LINE BLKSIZE 133 BLOCK SIZE SAME AS LRECL UNBLOCKED SPACE 1 NOW HAVE DCB FOR DISK WILL SPECIFY SOME DCB VALUES IN THE JCL CARDS DD CARD FOR WORKDISK IODSKDCB DCB DDNAME WORKDISK JCL DDNAME USED DSORG PS ALSO PHYSICAL SEQUENTIAL DATASET ACRF GL PL BOTH GET AND PUT LOCATE MACRO FORME EODAD IOEOF2 FOR WHEN END OF DATA END PER NOW EXECUTE THE PROGRA WIE DATA WORKDISK DD UNIT SYSDA GIVE ME SPACE ON ANY DASD SPACE 160 20 SPACE FOR 20 RECORDS OF 2 CARDS i ET DCB RECFM FB LRECL 80 BLKSIZE 160 SUPPLY DCB DSNAME amp amp TEMP GIVE DATA SET A NAME UNNEC HERE DISP NEW DELETE CREATE IT NOW GET RID WHEN DONE DATA PRINT DD SYSOUT A PUT THIS ON A PRINTER SOMEWHERE DATA CARDS DD WE HAVE A
65. Reading a JOB card begins a ls during the run until the BATCH of long sequence of actions which must be performed to simulate the actions needed to run a JOB First and EXECUTED the JOB card is scanned and all information is recorded from it The JOB is entered on a queue of JOBS which are not yet able to obtain use of a Central Processing Unit CPU Basically this step represents the process of reading in a complete JOB and storing it on magnetic disk before it can be executed Whenever it might be possible that some JOB be loaded into memory the queue of waiting JOBS is scanned and a JOB is perhaps selected according to one of many possible rules This JOB is then INITIATED i e allocated MEMORY and made ready for EXECUTION It is thus free to compete with ot or one of several her such ACTIVE JOBS for the use of the CPU if such exist The list of ACTIVE JOBS ones in memory is periodically scanned according to one of several algorithms anda JOB is selected to receive control of a CPU for some DISPATCHED period of time i e the JOB is After some length of time during which the JOB has use of the CPU it may make an Input Output Request in which case it relinquishes control of the CPU The IO REQU so that another ACTIVE JOB if any may get it EST needs the use of an IO CHANNEL of which there are one or more Depend
66. SO FREE THE STORAGE OBTAINED WITH A FREEMAIN AND RETURN TO CALLING PROGRAM FREEMAIN E LV 80 A ADDRESS SP 1 DONE ST R2 ADDRESS SET ADD FOR FREEMAIN PRINT NOGEN LA R5 1 R5 INCREASE R5 FOR OUTPUT XDECO R5 NN PLACE IN OUTPUT STREA XPRNT IN 52 L R13 4 R13 GET ADD OF HIGHER SAVE AREA E R14 R12 12 R13 RESTORE THE REGISTERS BR R14 RETURN TO CALLING PROGRA HEADING DC C OTHE SAVEAREAS AND NUMBERS ARE OUTPUTTED BELOW ADDRESS DC F O OUT DC CL40 RECURSIVE CALLED WITH VALUE N DC 3F 0 IN DC C RETURNING FORM RECRSIVE WITH NUMBER NN DC 3F O LTORG END LOG DATA SYSUDUMP DD SYSOUT A DAT TA XSNAPOUT DD UNIT AFF FTOG6FOOL Lt THIS JOB WILL RUN WITH TIME 35 SECONDS RECORDS 1000 JJ EXEC ASGCG SOURCE INPUT DD OF THIS PROGRAM IS TO DEMONSTRATE THE FOLLOWING lia lt THE PURPOSI T MACROS SPIE STAE HE BASIC FLOW OF THIS PROGRAM IS AS FOLLONS FIRST ISSUE A SPIE MACRO FOR ABEND CODES 1 5 THEN CAUSE AN ABEND SOC1 AND INTERCEPT IT WITH A SPIE 2 RETURN CONTROL TO THE NEXT INSTRUCTION FROM THE SPIE EXI ROUTINE 3 CANCEL THE SPIE MACRO 4 ISSUE A STAE MACRO THEN CAUSE A SOC6 AND INTERCEP THE ABEND WITH THE STAE EXIT ROUTINE THEN ALLOW THE ABEND TO CONTINU
67. The operator makes sure the disk pack called SYSRES SYStems RESidence is mounted on a disk drive The LOAD UNIT switches are set to show the device address of the SYSRES disk pack and the LOAD button pressed This causes the CONTROL RECORD to be read from the first record on the disk pack consisting of a PSW and two CCW s placed at location 0 in memory Execution of this record causes the IPL BOOTSTRAP record to be read into memory The BOOTSTRAP record consists of a set of CCW s which are used to read the IPL program into memory beginning at location 0 It ends with a LPSW to give control to the IPL program 2 IPL selects which NUCLEUS will be loaded there may be a choice which can be given by switches on the operator console 3 IPL clears all memory above itself to zeroes also obtaining the size of memory i e it stores until addressing interrupt occurs 4 IPL clears the floating point registers thus finding out if the floating point feature is installed 5 IPL brings the NUCLEUS into memory First it relocates the part of itself not yet executed into high memory near 252K so that the NUCLEUS can be placed beginning at 0 It then simulates Program Fetch loading the csects of the NUCLEUS load module into memory The first csect loaded is the NIP loaded just below IPL followed by the I O Interrupt Handler at 0 which thus defines all of the special PSW s in low core IPL
68. also charged for use of a CHANNEL A EXECUTION CHARGING METHOD The following option determines whether a JOB is charged for the use of CPU only or for both CPU and CHANNEL amp ECCH 0 The user is charged only for CPU usage Thus whenever the JOB is given control of the CPU the time until the next I O request is determined and an interrupt scheduled to occur at that time However if the time remaining to be used for the job lt time until I O it is used instead and the job terminated rather than performing I O at that time amp ECCH 1 In addition to the above charging there is a charge for the use of a CHANNEL and the I O duration is then time until 1 0 done MIN time left I O duration A job never gets to use anything if it has 0 milliseconds left CS411FP2 05 EX ECUTION ORD ERING M ETHODS These methods of course are similiar to the JO options for deciding which JOB s are to be initiated For further explanation on similiar ones refer back to CS411FP1 07 Note that one difference is that the JOB s are generally serviced in the order they are on the list without having to worrying about whether they fit in memory or not naturally since they already ARE in memory EOL FIFO the JOBs are ordered on the list according to their arr
69. contains detailed descriptons of day by day lectures contains index to CMPSC 102 files which in some cases may overlap with 11 or 411 above files are held under RJE ID JRMO2 LINKAGI ti l ps STANDARD LINKAGE CONVENTIONS Charles Pfleeger Under OS 360 certain conventions have been established regarding the use of registers These conventions will have been followed when you the problem programmer receive control from the system they should be followed for any routines which you call or for communi cating with the system e g system macro calls SVC s returning control etc Following these conventions will make your code easier for someone else to follow Certain debugging aids are also available for those who adhere to standard conventions In general unless there is a strong reason to deviate thes conventions should be employed REGISTER 14 is called the return register and contains the address to which this routine is to return upon exit REGISTER 15 is called the entry point register and contains the address through which this routine was entered Note that tempor ary addressability may be established by USING entrypointname 15 If this routine calls no other routines register 15 may be used as a permanent base register If this routine calls any other routines however register 15 will be changed and should not be used as a permanent base register In
70. cover STRUBLE cahpter 3 pp 50 56 symbols self defining terms literals location counter reference absolute and relocatable terms xpressions READINGS STRUBLE Chapter 4 to page 78 ASSIGNMENT STRUBLE Chapter 1 problems 5 6 7 8 9 Chapter 2 problems 2 3 Chapter 3 problems 1 2 3 4 6 INFORMAL ASSN modify dump program to use XDECO and DUMP storage use program with START to check relocatable vs absolutes modify one of batch programs to get 0C6 rather than 0C4 5 O1 17 introduction to arithmetic and data movement instructions introduce idea of instruction families and regularity of mnemonics Go thru following instructions LR LPR LCR LNR LTR L LH LA AR ALR A AL AH SR SLR 8 SL SH mention M and D also briefly note existence of Condition Code and show how to test it without worrying about encoding 20 minute question answer and review questions occurred on differences between literals and self defining terms and on use of symbolic register equates READINGS STRUBLE Chapter 5 HANDOUTS CS102AS1 pages 01 02 first assignment input output of numbers calculations in binary CS102M1 pages 02 05 S 360 mnemonic construction 6 10 CS102TPA 03 01 19 quiz and finish up data movement and binary arithemetic Twenty minute quiz diagnostic mainly base conversions 2 8 10 16 negative numbers base index displacement addr
71. external one In particular values can be passed in registers and usually registers will not have to be saved SPACE internal subroutines require less space than generating the same code several times via macro expansions Zi DISADVANTAGES E if the same function is performed by internal subroutines in several CSECTS code is thus duplicated and space wasted COMPLEXITY in some cases in order to make efficient use of a number of internal subroutines it is necessary to set up fairly xtensive rules on usage of registers in a CSECT so that the linkage among them may be fast and small ASPRGTCI 07 C EXTERNAL SUBROUTINES 1 ADVANTAGES SPACE if written as an external subroutine code can be usefully called from almost anywhere in a program Thus there is only one copy of it and it generally will occupy the least space SEPARATE COMPILE ASSEMBLY a routine written as a CSECT can be assembled separately from the rest of the program an object deck can be obtained and translation time generally saved The routine may of course be written in a different language than the rest of the program 2 DISADVANTAGES LINKAGE TIME if standard 0S 360 linkage is followed a fair amount of execution time and object code space can be consumed by this linkage More efficient nonstandard linkage can be used instead but this brings with it the disadvantage of no
72. illustrates most floating point instructions by computing a function in FORTRAN then using some equivalent assembler code T EMAIN program to illustrate use of GETMAIN FR program to illustrate load module management LINK LOAD XCTL DELETE macros etc NAME CARDS SOURCI E CS411GI1 09 DESCRIPTION OVLY1 PTPCHMAC QSAM RECURAS SPIESTAE TIME WTOWTL 276 22 126 22252 222 2 252 PE BAT BAT BAT BAT OVERLAY example complete run uses link editor to edit a single object module several different ways showing the amount of storage which can be saved by using overlay methods includes multiple REGIONS uses the IBM utility program to print 3 macros from SYS1 MACLIB illustrates use of IEBPTPCH allows stude entire Queued writes JCL DD nts to look at macros if they want to run setup shows SAVE RETURN CALL Sequential Access Method reads from cards to disk blocked illustrating use of DCB from card reads from disk prints shows GET information for DCB n PUT OPEN CLOSE DCB shows all sources of illustrates recursive assembler program using ETMAIN FREEMAIN macros E STAE hows use of error interception macros SPII test program to show use of timing macros TIME ir ST
73. priority of this job with 0 highest and 255 lowest i e if all other things are equal prioirty 0 gets preference over 255 CS411FP1 06 CHAN CHANnel _ specifies a number from 1 to the maxium number of channels in the system This specifies a particular channel which the JOB needs to use to perform I 0 0 specifes that the JOB may use ANY available channel in the system when it needs to perform 1 0 Of course if option CHNI is used this parameter can be ignored If CHAN specifies a channel number higher than the maximum one amp NUMCHNS CHAN 0 should be assumed instead i e this is NOT an error but allows use of any available channel CAT CATegory is a number from 1 to a maximum of 15 although some OS 411 s may allow a smaller maximum Specifies in some way the type of processing a JOB is to receive Some versions of OS 411 may not utilize this option at all The above options include many things often found on JOB cards for various systems Others MIGHT be required such as a date or time by which a JOB must be finished or limits on output lines printed cards punched or limit on the sum of these B SSCLEAR CARD INDICATES END OF A BATCH When this card is found it requests the following actions 1 The simulation is continued until all JOBS currently executing or waiting for execution are completely processed However no more S JOB cards are read during this period of tim
74. without the knowledge of 0S 360 which believes the job disappeared whenever it finished execution Only when a job is finished punching is its disk space released This allows for jobs to be saved across system crashes and for such useful services as repeating output by operator control C DASD STORAGE MANAGEMENT IN HASP HASP manages its DASD storage quit fficiently not only needing NO accesses to DASD to allocate or deallocate space but also doing a good job of minimizing arm movement on moveable head devices HASP requires the use of entire volumes normally 2311 or 2314 disks For example the PSU CC s 360 67 has 3 2314 disk packs for HASP The management of this storage works as follows A MASTER CYCLINDER BIT MAP is maintained in HASP This is a string of bytes in which each bit represents 1 CYLINDER on the SPOOL disks for example 600 bits for the cylinders on 3 packs A one bit represents a FREE CYLINDER while a zero bit shows that the given cylinder is allocated to some job HASP also remembers for each disk which cylinder was the last referenced thus always noting the current position of the read write heads OSHASP 08 Two JOB BIT MAPS exist for each job one for SYSIN data and the other for SYSOUT data Whenever a cylinder is required for a job HASP searches for a free one in the following fashion 1 It first searches the master bit map for a free cylinde
75. 0 OS DATA MANAGEMENT MACRO INSTRUCTIONS d This is recommended form and describes in detail the various ways of coding the data management macros The manual just previous is read for understanding this one is needed for actually writing such programs R lt SS gt GC28 6646 5 S 360 OS SUPERVISOR SERVICES a This is explanatory part of lt SS amp M gt older version R lt DMS gt GC26 3746 1 S 360 OS DATA MANAGEMENT SERVICES c This is explanatory part of lt DMSG gt older version and not as well written as lt DMSG gt R lt S amp DMM gt GC28 6647 4 IBM S 360 OS SUPERVISOR AND DATA MANAGEMENT MACRO INSTRUCTIONS b d Older version of material in lt SS amp M gt and lt DMM gt R lt S amp DMS gt GC28 6646 3 IBM S 360 OS SUPERVISOR AND DATA MANAGEMENT SERVICES a c Older version same as SS and DMS put together D lt LE amp L gt 6 GC28 6538 9 IBM OS LINKAG This describes how to use the named programs CJ CS411GI1 04 EDITOR AND LOADER and is particularly useful and necessary for anyone writing overlay programs or concerned with management of program descriptions of object and D lt JCLR gt Les libraries GC28 6704 2 IBM S 360 OS JOB CONTROL LANGUAG The beginning contains fair load modules and their processing Gl gives various coding forms and somewhat di
76. 06 TION calculated for each job at the time it is number used to determine its relative priority for using a CPU Note that the relative priority of two jobs does NOT change while t E06 are examples of t 08 DYNAMIC ORD he two Jobs are in memory Note that all options his type of scheme EOL ERING COMPUTATION In this th and can thus desired The cas computation of y EO8F at in at to EO8V The dynamic ordering methods thus include the static ones overhead but can also give improved performance involv mor relativ hange during execution fixed intervals memory have order numbers computed for them variable intervals give a job control of a CPU ordering of jobs is obtained dynamically according to whatever criteria are order numbers is typically done either such as every 2 milliseconds All jobs or whenever a decision must be made They since they can respond better to the changing state of the system A number of typical dynamic ordering methods follow including some counterparts of the previous static ordering ones EO81 entered RJQ but a job only enters doing 1 0 possibly the midd more fair to all arlier ones mor a EO84 ENGTH OF TIM L T JOB cards it rejoins the queue at the end FIFO jobs are ordered not according to FIRST time jobs last i e the it at the end RJQ is inde
77. 2 RX give first explanation of base displacement 3 RS NGS STRUBLE CHAPTER 2 POP pp 7 15 UTS CS102M1 page 01 run some ASSIST programs for dumps 3 01 12 READI CS102TPA 02 finish operands formats and introduce assembly language 4 SI instructions examples MVI CLI 5 SS instructions examples MVC CLC machine language easy for machine to execute hard to write assembly language converted by assembler to machine code format of assembly language label opcode operand comments machine instructions actual operations to be executed assembler instructions pseudo ops give information to the assembler ex CSECT DS DC some basic functions of the assembler 1 location counter 2 convert mnemonic opcodes a machine ops translate to codes increm location cntr b assembler ops take actions specified increm loc cnt 3 operands convert to internal binary base displacement 4 print out a listing 5 make program ready for execution and pass control to it stepped through complete test program XREAD XPRNT XDECI XDUMP and explained listing and contents of dump NGS STRUBLE Chapter 3 ASSIST MANUAL PARTS II and IV ASSEMBLER LANGUAGE pp 1 18 HANDOUTS DOCUMENT documentation techniques for assembler 4 01 14 go over some dumps and errors discuss operand fields go through various dumps showing 0C1 0C4 and 0C6 errors
78. BDAM XPRNT 0 1 SYNADRLS BR 14 REF DC F O DC x 00 INPTT DC 10F 0 OTPT DCB DSORG PS MACRF PM LRECL 40 BLKSIZE 40 RECFM FA DDNAME FTO6F001 EROPT ACC PRINT GEN SPACE 5 INPT DCB DSORG DA MACRF RIC RECFM F BLKSIZE 40 OPTCD ER LIMCT 3 DDNAME DIRECTDD SYNAD CHECKER LTORG END DATA FTO6FOO1 DD SYSOUT A DATA DIRECTDD DD DSNAME 66TEMP DISP OLD DELETE SPACE 44 21 UNIT SYSDA DATA SYSUDUMP DD SYSOUT A Lt iL ENE THIS IT US P E En ROGRA S THE BLD JJ EXEC ASGCG SYSIN DD THIS p p TIMING AND PAR PROG RECO RD NOXRI RAM IS IS AN MACRO AN E ah EXAMP OF IHE BPAM ACC E E SS METHOD D RAT H E CONSIDI a F PAR SEC IT Es ECTIO THIS P EACH O REGIST ION IONED ACCE HOD I 1 USES THE B E TIONS D TO ILLUS TWO KINDS OF 40 SECOND FIND MAC S 2500 R ROS ECORDS E THE RATE T USE E OF THI HAS TWO SECT CONTAINS A RO INS IVE RYG THE UST B E SER SUPP E FULL WO LF WORD 1 ORD 2 ER E ONTAIN T
79. CHECK PRNTDECB IS 1 0 OPERATION COMPLETE SPACE 5 B READ LOOP TO GET ALL DATA SPACE 5 PRINT A LITERAL INDICATING EOD ON CARD READER BEGIN READING FROM THE DISK SPACE 5 ENDATA WRITE PRNTDEC1 SF PRNTDCB LITERAL SPACE 5 ISSUE CHECK AS ABOVE SPACE 5 CHECK PRNTDEC1 SPACE 5 SINCE DISK DCB WAS NOT CLOSED AT EOD A POINT MACRO MUST BE ISSUED TO REPOSITION THE THE DISK AT THE FIRST RECORD SPACE 5 POINT DISKDCB POINT REPOSITION DISK TO GET 1ST REC SPACE 5 NOW READ FROM DISK AND WRITE TO PRINTER USING SAME DATA SPACE 5 DISKRD READ DISKDEC1 SF DISKDCB BLOCK SPACE 5 CHECK FOR END OF I O OPERATION SPACE 5 CHECK DISKDECI CHECK I O COMPLETION SPACE 5 WRITE TO PRINTER AS CARDS COME OFF DISK SPACE 5 WRITE PRNTDEC2 SF PRNTDCB BLOCK 1 SPACE 5 ai CHECK I O DONE SPACE 5 CHECK PRNTDEC2 I O COMPLETION CHECK SPACE 5 BCT 5 DISKRD WRITE ONLY THE CARDS THERE SPACE 5 END CLOSE CARDCB PRNTDCB DISKDCB CLOSE ALL ACTIVE DCBS SET UP RETURN TO CALLER OS XRETURN SA TR NO SPACE 5 ABTERM ABEND 500 DUMP GET ABEND DUMP IN CASE OF TROUBL DC OF O POINT DC X 00000100 CONTROL WORD FOR POINT MACRO DC G A ASA CARRIAGE CONTROL CHAR BLOCK DC soc BUFFER AREA LITERAL DC C REPEAT READ DATA FROM DISK 80C SPACE 5 CARDCB DCB RECFM F DATA FORMAT IS FIXED BLKSIZE 80 PHYSICAL BLOCK SIZE IS 80 BYTE RECL 80 OGICAL RECORD SIZE IS 80 BYTES DDNAME INPUT OGICALLY CONNECT TO INPUT DATA EODAD ENDATA ON EOD GO TO ENDATA SYNAD ABTERM I O ERROR
80. CL CLC SIR B AL And Link form of branch for doing linkage to subroutine so it can return EX BAL BALR E Condition branch or not depending on a previously set condtion IF GOTO EX BC BCR GT Count branch form used to decrement a register and branch if value not zero DO LOOP EX BCT BCTR X indeX branch form for incrementing and testing index quantities DO LOOP EX BXH BXLE L Complement used to set a register to complement itself or other Y ABS X EX LNR LNDR P Positiv set register to positive value from self or other Y ABS X EX LPR LPER E Test set register to value from self or other L ST M Multipl several registers are loaded or stored in one operation EX LM STM CS102M1 04 3 DATA TYPES As noted previously a data type character is usually the same as that used in a DC or DS statement to obtain a given type of data If a type character is omitted it usually implies that the instruction operates on 32 bit fullword binary quantities such as A C S etc DATA TYPE MEANING COMMENTS e Character usuallv a contiguous string of bvtes in memorv treated as printable characters or a string of bits FORTRAN LOGICAL 1 EX MVC CLG 0C IE STC USUALLV IMPLIES SS INSTRUCTION FORMAT all except IC STC D Double precision floating point Doubleword 64 bi
81. ES D handle nd formats are given linked toget of the NEXT NODE QPOP POPS t and modi it EFA the ne n of EA altering the LINK of a UMPS a list ULT e D h type r il e s n fies Q ER to S CS411MC2 of li below QHED a first cr LIST NO 02 st just Briefly eates a given from r DI H E the A POP it do SO F an ORD RE then already E ARE Gl INSE there or each VEN FOR THE ry AR ry SUGG ESTED AS EPEAT NO WILL B RE UIRED YO U ry O CONVENI ENT DEFAULT For t YO ONIC AND coded OF MNI CJ On Nr AH bi tA tA he U D FJ bi E examples HAV e NOT ti EJ EJ EJ FJ El IO O O O O GGG GG ao o e naaa vn U statem e executable code in a macro expansion POSSIBLE data ge nless nt label ot nerated EQU 2 ah given assume that B OT EING CONV AY SUBSTITUTE F H iR ry NI B ENT VALU ES S the TO USE TH E ES THEY ARE F ON followi Y INCLUD ng EQ UT THEY IF YOU U S are ED AS EXAMPLES Nu fs O A
82. ES O R Gi B ER ENTRI F ES MUST BE IN N 2 ISSUES A FIN ACRO INST IS DIRECTED RELATIVE ADD TO DO TH E PDS DI HEM INTO ETS THE DCB THIS PUTS TO DEA T RIC ORD T ALPHAMI ry R RUCTION ONLY RECTORY SI RI ESS IS CONVERTE nr m F F HIS TI THE ARCH ITSELF FIN TH T Ho D AS IN SI F E INSERTED INTI ROGRAM XSAV RET N D USES bb B DO D IND HECK LISTHD LIST DCB DCBD CCURRANCE X X R B E C PROG PROGRA T IS in O THI DCB ACROS RAM LOCAL M LOCAL EXPLAINI F ER USAGE IS yy ROGRA FF FF CACA F CACA FF F FF FF F F FF F F FF F FF F FF F FF F FF AC F FF F F A F F FF F FF F FF F va F F F KF OF ED IN THI ja in T RAM ECTION 1 AND TEXT E EXPLAIN D IN LOGIC THIS PROGRAM PROG EQUATE F THE S ECTION R ECOV ERS FROM TH ah SYS1 MACLIB AND THI E ACA A F CACA HF F CACA F S F F s F x Qh Z D lt E AR O x CMACLIB PROGRA JIBRARIES THE DEFINITIONS FOR FOUR SYSTEM MACROS CALL RETURN SAVE AND DCB MACROS CALL RETURN AND SAVE ARE RECOVERE USING THE BLDL FIND MACRO COMB
83. ES STMTS A TOO LARGI EGAL LAB ILLEGAL OPCO ABEL TOO DI e ONG PLY DI EFINE D LABEL N EGATIVE ADDR ESS IN ER Hd OITTHISOPCOD Li EL FIEL ROR ERR 0 CONTAINS B tr PS T ED OP T RAND E RRORS IN THE OPERANDS OLLOWS 16 0 EGIST ER TOO LARGI p OP p 15 RAND 157 OPERAND RI E E m E OMI OMMITTED D REGISTER 0 OMITTED 15 524288 15 524289 15 524287 15 524288 15 0 IN RX FORMAT TOO BIG EGAL EGAL LID DIS VAI PLACE 16 0 R EGIS R TOO ES 1 5 OM ITTE 15 OMI 15 1 15 5 X LS 5 1 15 2 ILL TED O W RAN AW 15 X 3 15 X 3 15 X 3 y X X 15 0 8 15 X 5 8 15 8 15 2 15 0 7 15 20000 1 0 OM AW Byt 0 1 1 0 131071 0 131072 END IND ITTE EGIS INDEX EX R ERG EG TOO O BAD A DDRI O LA ARGI ESS ILI EGAL REG SPECIFICATION LEGAL TOO BIG S SSJOB LOOP S JOB S JOB S JOB S JOB X6 Y S JOB LOOP
84. F CL8 H NODI herwis specified to be generated ig offset RROR STOPPING WAYS TO DO CERTAIN of the THINGS INK field in common offset to key field most common KI temporary wor 2nd temporary work reg link to next node K first data item second data item E1 3 4 4 node le AB EY ordering field EL ngth EY length k reg Cannot be canno to 4 the it BUT NOT NE r C abel amp her on or before or usuallv on ESSARILV THE B for at least one sample of each macro is an the first the first EST CODE i NO ti a reg 0 t be 0 multiple optional byte of byte of s given A QDEFL amp LABEL QDEFL amp NUMBER amp LENGTH amp MESSAGI CS411MC2 03 DEFINE A LIST CJ This macro generates amp NUMBER nodes each of total length LENGTH bytes including 4 byte link linked together with kev data areas initialized to blanks and preceded by an optional message amp NUMBER is a self defining term of value gt 0 giving the number of nodes to be generated amp LENGTH is any absolute expression giving a length for each node Code generated must allow for rounding this to 4 multiple MESSAGE is a st
85. FOR RX OR RS BM RX GO TO SET PSW LA R7 6 GET INSTRUCTION LENGTH B ILCSE GO TO SET IL RR LA R7 2 SET ILC FOR RR B ILCSET GO TO SET I RX LA R7 4 STE IL OF 4 ILCSET AR R10 R7 GET NEW ADDRESS FOR PSW SLL R7 29 SET R7 TO IL CODE LA R8 63 SET R8 TO HEX 0000003F SLL R8 24 SET R8 TO 3F00000000 NR R8 R2 GET CC AND PROGRAM MASK OR R10 R7 GET IL CODE IN NEW PSW OR R10 R8 NOW WE HAVE NEW PSW ST R10 8 R1 SET NEW PSW PRINT NOGEN XSNAP T NOREGS STORAGE 0 1 36 1 X ABEL PROGRAM INTERUPT ELEMENT IN SPIE L R2 0 1 GET PICA ADDRESS XSNAP T NOREGS STORAGE 0 2 6 2 X ABEL PICA FOR SPIE MACRO LM RO R15 SAVE GET ADDRESSES TO RETURN BR R14 RETURN TO CONTROL DROP 15 FIXSTAE XSAVE SA NO XSNAP STORAGE 0 1 104 1 T NOREGS X LABEL THIS IS THE 104 BYTE WORKAREA PROVIDED BY STAE LA R15 0 MMDONE R14 12 13 RESOTRE REGISTER 14 FROM SAVEAREA LM R0 R12 20 R13 RESTORE REG 0 THRU 12 BR R14 RETURN TO PPERATING SYSTEM MHEAD1 DC CL80 OSTAE MACRO HAS ISSUED AND RETURNED CONTROL SAVE DC 18F O HOLD DC F O DS OF DC H O HALFWORD DC C NONO MHEAD DC CL80 OTHE INTERUPT HAS OCCURED AND FIX CALLED ON SPIE END LOG DATA SYSUDUMP DD SYSOUT A DATA XSNAPOUT DD SYSOUT A jif tA SOURC LOG HANNA MAIN A A F x THIS JOB WILL RUN WITH XEC ASGCG E INPUT DD HE PUT TIME 35 RECORDS ECONDS
86. GO TO ABTERM ACRF R BSAM I O USES READ WRITE MACROS DSORG PS DATA SET ORGANIZATION PHYSICAL SEQUENTIAL SPACE 5 PRNTDCB DCB RECFM FA RECORD FORMAT FOR OUTPUT X BLKSIZE 80 PHYSICAL BLOCK SIZE X RECL 80 OGICAL RECORD SIZE X DDNAME FT06F001 CONNECT TO OUTPUT DEVICE x SYNAD ABTERM I O ERROR GO TO ABTERM x DSORG BS DATA SET ORGAN IS PHYS SEQU X ACRF W END PRINTER DCB SPACE 5 DISKDCB DCB RECFM F RECORD FORMAT FOR DISK I O X BLKSIZE 80 PHYSICAL BLOCK 80 BYTES X RECL 80 OGICAL RECORD LENGTH X DDNAME FTO9F001 OGICALLY CONNECT TO DEVICE X EODAD END ON EOF GO TO END X SYNAD ABTERM PERMANENT I O ERROR GO TO ABTERMX DSORG PS DATA ORGAN IS PHYS SEQU x ACRF RP W READ WRITE COMBINATION SPACE 5 END OF DCBS FOR I O OPERATION END BSAMIO DATA FTO9F001 DD UNIT SYSDA SPACE CYL 1 DSN amp amp TEMP x DISP NEW DELETE LOG DATA XSNAPOUT DD SYSOUT A DATA SYSUDUMP DD SYSOUT A DATA INPUT DD I AM THE FIRST CARD TO BE READ I AM THE SECOND CARD TO BE READ I AM THE THIRD CARD TO BE READ CARDS ARE ECHO PRINTED AND WRITEN TO DISK THEN ARE READ FROM THE DISK AND PRINTED AGAIN THIS IS THE LAST CARD
87. IS INCLUDED TO PROVIDE A SAVEAREA FOR THE XSAVE L 13 4 13 GET ADD OF SYSTEM SAVEAREA L 14 12 13 RESTORE REGISTER 14 XCTL 2 12 EP FOURTH PRINT NOGEN XRETURN SA TR NO PARM1 DC F 0 PARM2 DC F 1 END OBJECT SYSLMOD DD DSNAME amp amp LOADMOD MAIN DISP OLD PASS DELETE DATA STEPLIB DD DSNAME 66LOADMOD DISP OLD PASS DATA SYSUDUMP DD SYSOUT A OSHASP 01 OVERVIEW OF OS 360 WITH HASP This writeup gives a quick overview of the process by which any 0S 360 sytstem is initialized how storage is used particularly in 0S 360 MVT and describes how OS 360 is modified by the use of HASP Houston Automatic Spooling Priority system The storage layout is described for the PSU CC 360 67 system I INITIALIZATION getting a system up and running Consider a computer with no operating system currently in it The first necessity is to get a workable operating system in it so that jobs can be run This is NOT a trivial process note that there is no Program Fetch resident in the machine no I O Access Method routines and not even a correct set of PSW s in low core for directing interrupt actions For OS 360 the initialization process is composed of two parts IPL and NIP IPL Initial Program Loader initializes memory and some other things and brings the NUCLEUS the core of the OS into memory NIP Nucleus Initialization Program performs the remaining actions required to set up a specific NU
88. JRMO2 TESTDECK 4 DUMP CARDS If your program runs out of time or records you do not normally receive a completion dump Inserting a card with DUMP in columns 1 6 anywhere in your deck like LOG will allow you to get a dump assuming you have a SYSUDUMP card also in the correct place 4 PARM FIELDS PARM fields can be used to pass information to a processing program with up to 1 PARM for each STEP in a cataloged procedure The following options may be useful to you SOURCE STEP ASSEMBLER G NOESD Deletes lst page External Symbol Dictionary which is often not too useful NOLIST Deletes entire source listing especially good for a debugged program when you want to save records NOXREF Deletes the Cross Reference from end of listing saves time and records DECK Requests that an OBJECT DECK be punched You can then run the program or include it with another one without having to assemble it again saving time OBJECT STEP LINK EDITOR AP Requests a MAP of the modules in your program DATA STEP LOADER AP Requests a MAP of where modules are loaded into memory EXAMPLES EXEC ASGCG PARM SOURCE NOESD NOXREF DECK PARM DATA MAP EXEC ASGCLG PARM SOURCE NOLIST PARM OBJECT MAP missing s ok LL used to run an assembler program using Assemble
89. LCLx only in core when needed ordinary symbols ordinary symbols pass 2 One Pass FORWARD R EE ERENCE PROBLEM ASMTUTI CS411MC1 tutor macro write macros for CS411MC1 ial on assemblers assignment sequenc symbolic pass 3 4 hex conversions and dumps due in 3 weeks day 02 01 TRT usage types of operating systems history TRT table manipulation reference to handout CS411MC1 HEX conversions using ORG instructions in tables how TRT works including code equivalent OPERATING SYSTEM TYPES BY HISTORY 1 HANDS ON 2 BATCH PROCESSING uniprogramming 3 MULTIPROGRAMMING SYSTEMS with SPOOLing 4 TIMESHARING SYSTEMS Other classifications A UNIPROCESSING and B MULTIPROCESSING HANDOUTS CS411MC2 pages 01 06 Linked List assignment ASSIGN assignment given by CS411MC2 due 02 22 02 01 operating system types programs begin hardware give comparison table for OS types program attributes non reusable serially resusable reent REAL TIME SYSTEMS OMPUTER ARCHITECTURE CPU S PRIMARY MEMORY CHANNELS IO access writeback cycle times EADINGS C memory fetch store physical word versus POP pages 5 7 logical word 15 22 68 83 IO et E x interleaving CS411TPA 04 14 02 03 memory protection input output devices DASDS HANDOUTS HARDWAR1 01 04 A PSU 36
90. MAXIM EXEC ASGCG PARM DATA MAP 01 AM A PARM FIELD SOURCE INPUT DD jus h main program sysudump and xsnapout cards if needed B Using the object deck produced by the previous program run this test of MAXIM with FORTRAN EXEC FGCG PARM DATA MAP SOURCE INPUT DD elijes FORTRAN main program DATA DECK DD sleeve r object deck from MAXIM sysudump and xsnapout cards as needed VI THOUGHT QUESTIONS A Suppose that you also wrote MAXIM as a FORTRAN function What would you do to obtain an object deck of it and use it as a subprogram of your assembler test program B Does MAXIM need to have its own save area If so why If not why not C How does the object deck of MAXIM compare with its source deck Does the answer tell you why people use object decks D What reasons can you think of for using a mixture of FORTRAN programs an assembler programs What does FORTRAN do well that assembler does not and vice versa CS411GI1 01 01 09 73 date of last revision COMPUTER SCIENCE 411 GENERAL INFORMATION This writeup provides general information regarding CMPSC 411 SYSTEMS ORGANIZATION AND PROGRAMMING as currently taught at PSU It notes the prerequisites text materials handouts assignments and generally describes what is taught in this course and what is expected of the students taki
91. NOW READ THE FIRST BLOCK OF THE RETRIEVED DATA INTO SOME INTERNAL BUFFER AREA SO THAT IT CAN BE HANDLED IN SOME USEFULL FASHION SPACE 2 BREAD READ MAC SF DCBADD AREA S SPACE 5 NOW CHECK THE COMPLETION OF THE 1 0 INITIATED BY THE READ AS NO DE BLOCKING OF THE RECOVERED DATA CAN BE DONE UNTIL THE TRANSFER IS COMPLETE SPACE 2 CHECK DECB SPACE 2 THE READ RETURNS THE BLOCK SIZE INTO THE DCB FOR USE IN RECORD DEBLOCKING AND OTHER DATA MANAGEMENT SPACE 5 USING IHADCB DCBADD NOTE USING ON DCB DSECT SPACE 5 NEXT FOUR INSTRUCTIONS ALLOW THE PROGRAM TO CORRECTLY ACCESS NEXT BUFFER FULL OF INFORMATION ALSO TO DETECT SHORT BLOCKS AT END OF THE CURRENT MEMBE SPACE 2 H END DCBBLKSI GET BLOCK SIZE FROM DCB I ENTRY 16 DECB GET IOB ADDRESS SH END 14 ENTRY GET RELATIVE END OF NEW BUFFER AR END AREA GET ABSOLUTE ENDING ADDRESS SPACE 2 LR RCRDPT AREA COPY AREA POINTER INTO RECORD POINTER REGISTER DPRINT VC PRNTAREA 1 80 0 RCRDPT DE BLOCK RECORDS INTO 81 BYTE PRINTING BUFFER PRINT NOGEN XPRNT PRNTAREA 81 PRINT THE LOGICAL RECORD PRINT GEN SPACE 5 LA RCRDPT 80 RCRDPT UPDATE POINTER BY 80 BYTE LRECL CR RCRDPT END FIND OUT IF BUFFER IS EMPTY BNL BREAD IF AT END READ NEW BLOCK B DPRINT OTHERWISE CONTINUE PRINTING OUT OF BUFFER AREA SPACE 10 AEODAD AR BLDLCNT ONE INCREMENT MEMBER COUNTER EXIT E UPDATI AR A xXx DRECTSCH XXX A xXx NAME BPAMDCB CALLHD
92. RO OVERFLOW BCTR RO 0 SUBTRACT 1 FROM RO BEGIN LR RI RO LOAD Rl FROM RO ALR RO RO CREATE A RANDOM BIT BO ONETWO BRANCH IF A BIT IS CREATED ALR RO R1 CREATE A RONDOM BIT BO ONE IF BIT IS CREATED BRANCH LA R1 0 PUT A 0 IN R1 FOR THE RNADOM BIT WHICH WAS NOT CREATED B FINISH GO TO THE END OF THE SECTION ONETWO ALR RO R1 CREATE A ROANDOM BIT BO BEGIN RETURN AND START AGAIN ONE LR R1 R5 LOAD THE BIT FROM THE RANDOM CHOICE FINISH ST RO SAVEODD SAVE THE ODD NO SHIFT THE BIT INTO R6 WHICH WILL CONTAIN THE RANDOM NO LR R7 R1 MOVE THE RANDOM BIT TO R7 SLDL R6 1 MOVE THE RANDOM BIT INTO R6 IF THIS IS NOT THE LAST BIT GENERATE ANOTHER BCT R4 AGAIN FIND THE NEXT RANDOM BIT DETERMINE IF THE NO EXCEEDES THE MAX VALUE CR R6 R2 DETEMMINE IF THE NO GENERATED EXCEEDES THE MAX VALUE BP OVERFL IF MAX EXCEEDED RETURN AND DO AGIAN DETERMINE IF THE NO EXCEEDES THE MIN VALUE CR R6 R8 DETERMINE IF THE RANDOM NO IS L5SS THIN THE MIN VALUE B OVERFL IF OUT OF RANGE DO OVER LR RO R6 PLACE RESULT IN RO FOR RETURN XRETURN SA NO RGS 14 15 1 12 TR NO L DS 0D SAVEODD DC F Q END DATA SYSUDUMP DD SYSOUT A bt LL Lhe LT tA SOURC LOG A A F A F F F F MAIN AX A X x FF ACA ACA F HF F FF F ACA CACA CA F ACA F ACA ax a KF OF THIS JOB WILL RUN WITH XEC ASGCG E INPUT DD EI WTL WTO WTOR ti HE PURPOSE OF THIS PROGRAM IS TO DI MACROS
93. SET OF CARDS FOLLOWING O FIRST INPUT TEST CARD SECOND INPU EST CARD GOES IN BLOCK WITH FIRST THIRD INPU EST CARD FOURTH INPU EST CARD GOES IN BLOCK WITH THIRD FIFTH INPU EST CARD WILL BE IN TRUNCATED BLOCK BY ITSELF FRE JE THIS JOB WILL RUN WITH TIME 25 SECONDS IIF RECORDS 600 EXEC ASGCG SYSIN DD THE PURPOSE OF THIS PROGRAM IS TO DEMONSTRATE THE USE OF GETMAIN AND FREEMAIN WITH THE E CONVENTION RCALL DSEC SAVEAREA DS 18F NUMBER DS F PRINT NOGEN EQUREGS THE MAIN PROGRAM ONLY CALLS THE RECURSIVE SUBROUTINE RECRSIVE MAIN CSECT XSAVE CALL RECRSIVE NMBR XRETURN SA NMBR DC F 5 THE CSECT RECRSIVE IS A RECURSIVE CSEC HE SAVEAREA IS OBTAIN FROM THE OPERATING SYSTEM BY A GETMAIN HEN T
94. T WTO ROUTC E PRIN CSECT XSAV PRINT Li d THI L MACRO THE TIME 2 RECORDS 600 ESSAGE APPEARS AT THE ti lav D O Q BUT ON WTL WTO IF ACTUA WTO 1 jam EE 11 WE ARE WRITING TO THE HIS IS ALSO THE SYST 5 SECONDS EMONSTRATE T THE B HIS IS AN HIS IS AN tA OF WTL WTO WTO OF WTO ROUTCD Ej ULD Bl FIRST tA R IS THE MESSAGE TO BE ERATOR jan ka E ROUTCDE E M EXAMPLES OF W AND WTO RTCDI 11 ROUTCDE 11 O AND WTOR EN TO TH CONSOLE NUMBER TO BE EANING OF n G U PERVISOR AND IF JOB DOI ROUTCDE 1 2 CJ tA R IS THE MESSAGE TO BI RMINATI CJ ESC 1 CJ T E T PTOR IS GIVEN IN APP ANAGEMENT MACROS IN 10 SI ED REPLY INDICAT HE NAME OF AN E HE REPLYADDRESS T ANSWER GOES E MAXIMUM LI IF STANDAR ERATING CON ECB ROUTCDE 2 DESC 2 PROGRAM IN THE AS FOR THE WTO MAC E IMMEDIETE ACTI
95. UJL element from the fills been placed in the UJL by RDR completed and removed by the Pp a job which was UJL obtains an empty it in with required Lement R or takes car amp INITDI This module implements the following groups EADY then either of requesting an he initiator then loops through the above until it can no longer for any reason whatsoever JSR MA and lled whenever ca ither becaus of an I O request CPU If PR switch a REA LT DY job for s finished can occur either when one completes I O and is thus REA It may call IOI ENG a new job DY again to determine how given control of the an I O request he Leme DISPATCH nts AJL el CPU for some time wil his module determines which EMPTION is allowed if any job should this module should an EX ECUTING one which being long a ll exe e loaded or when an old job being DISPATCH cute until it makes Li ER probably manipulates various clock values in the decrements time remaing counters tc and generally performs bookkeeping operations needed to generate statistics E If PREEMPTION is modify clock elements fairly does not lose appropriate amount of allowed the DISPATCH accordingly time and generates a time ER must take care of it so that a preempted job is trea
96. XSNAP WITHOUT XSNA assembler at all The DATA a POUT WILL R SYSUDUMP CARD TO G ET ESULT IN ABI and just executes the program ANY DUMP AT ALL END UO300 DUMPSJCL 02 D ADDITIONAL USEFUL JCL Ty SGLEVEL 1 If is often useful to see the JCL of cataloged procedures used Punch a comma after the programmer name on your JOB card punch anything in column 72 of you JOB card then follow the JOB card with the following Jl SGLEVEL 1 or for even more information use SGLEVEL 1 1 2 LOG CARDS A card having LOG in columns 1 5 can be put anywhere in your deck except in the middle of continued cards or in data or source decks It causes additional information to be printed in the system log which is the very beginning of your output This is strongly recommended for multi step jobs since it shows how far a a program progressed and how much time each step needed 3 INCLUDE CARDS Your instructor may sometimes place cards on magnetic disk such as test programs and make them accessible to you Each FILE on disk has a FILENAME and you can essentially copy each such file by using a INCLUDE card referencing that file Files may contain both data and JCL Your instructor has an identification which must be used to refererence the files The form is INCLUDE ident filename for example INCLUDE
97. also have areas in LCS OSHASP 03 II RUNNING JOBS IN AN OS 360 SYSTEM This section describes how jobs are run in a standard 0S 360 system using either OS MFT or OS MVT Note that OS PCP runs jobs one at a time sequential scheduling uniprogramming with no SPOOLing of jobs to and from disk before and after execution OS MFT and OS MVT are generally similiar in that they both can SPOOL input onto DASD execute jobs in priority order and write the output out later The main difference is in the handling of storage in which OS MVT is much more dynamic Note that all scheduling of jobs and communication with the operator is effectively under the control of the MASTER SCHEDULER A READING INPUT STREAMS For each existing input stream card reader or input on tape the operator can issue a START RDR command This causes a copy of the READER INTERPRETER program referred to herafter as a RDR to read cardimages from the requested input device During its operation a RDR reads an input stream scans JCL cards nd converts them to a standard internal text form and also obtains ataloged procedure definitions from the procedure library PROCLIB rom the internal text it builds INPUT QUEUE entries representing he information on the user JCL cards It also writes any input data ards onto disk while placing pointers to the data into the INPUT UEUE entries so that it can be found later Th
98. as a permanent base register If this routine calls any other routines however register 15 will be changed and should not be used as a permanent base register In this latter case the sequenc LR BASEREG 15 USING entrypointname BASEREG where BASEREG is any of registers 2 12 may be used to establish permanent addressability On return register 15 may be used to return a code to indicate normal or error return One frequently used technique is to set R15 zero on a normal return and set it non zero if some error condition occurred prior to return REGISTER 0 is used to return the single result from some process as in a Fortran function subprogram Note although you will probably not use this convention much it is heavily used by the operating system Register 0 cannot be guaranteed to be intact after executing some call to the system as a system macro or an SVC REGISTER 1 is the pointer to an argument list It contains the address of the first of one or more full word entries on con secutive f w boundaries Thes ntries ar th addresses of arg uments to be used by the calling routine If there may be an indefinite number of arguments as with a routine which would accept one two or any number of arguments c f Fortran MAX0 the first bit of the last address is set to a l This bit will not interfere with ordinary S 360 addresses since an address can be fully
99. be able to return the registers intact to the calling program This save area has a set format Word 1 Used by PL I and FORTRAN Word 2 address of the save area used by the calling program Word 3 address of the save area set up by the called program Word 4 address to which to return reg 14 Word 5 address of entry point reg 15 Word 6 contents of register 0 Word 18 contents of register 12 Save areas are chained in a doubly linked list At any low level routine by tracing back through a chain of save area links one can eventually return to the system at the original point of call LINKAGE 3 When your routine is entered first you should save registers and then establish and link your own save area STM 14 12 12 13 save regs 14 15 and 0 12 in calling program s save area LA 5 MYSAVE get addr of my save area ST 5 8 13 link calling pgm s a to mine ST 13 4 5 link my s a to calling pgm s LR 135 5 transfer pointer to s a On return L 13 4 13 retrieve addr of calling pgm s save area LM 14 12 12 13 restore registers as they were BR 14 MYSAVE DC 18F 0 A calling program is known as a higher routine and the routine called is the lower routine Register 13 is always to point to an area whose contents may be destroyed An exception to the requirement that a routine must always establish a save area is that the lowest level rou
100. by the calling program Word 3 address of the save area set up by the called program Word 4 address to which to return reg 14 Word 5 address of entry point reg 15 Word 6 contents of register 0 Word 18 contents of register 12 Save areas are chained in a doubly linked list At any low level routine by tracing back through a chain of save area links one can eventually return to the system at the original point of call LINKAGE 3 When your routine is entered first you should save registers and then establish and link your own save area STM 14 12 12 13 save regs 14 15 and 0 12 in calling program s save area LA 5 MYSAVE get addr of my save area ST 5 8 13 link calling pgm s a to mine ST 13 4 5 link my s a to calling pgm s LR 135 5 transfer pointer to s a On return L 13 4 13 retrieve addr of calling pgm s save area LM 14 12 12 13 restore registers as they were BR 14 MYSAVE DC 18F 0 A calling program is known as a higher routine and the routine called is the lower routine Register 13 is always to point to an area whose contents may be destroyed An exception to the requirement that a routine must always establish a save area is that the lowest level routine the one which calls no others need not set up a save area The reason for this is the save area is for the use of any called routines but that the lowest lev
101. compare single precision floating numbers CH Cc H RX compare a register algebraicly with halfword from storage with sign extension CE E P ss compare two packed decimal numbers CR Cc R RR compare two fullword values algebriacly gotten from C by adding R CLC C L C ss compare logically character strings CLI C L I SI compare logical immediate a byte in memory with the one inside the instruction CDR CD R RR compare double precision in registers CER CE R RR compare single precision in registers The Sytem 370 computers have some additional opocdes CLM CL M RS compare logical masked from register to mem CLCL CLCL RR compare logical character strings long up to 16 million bvtes in one compare Consider the problem of writing a FORTRAN program which would simulate the operation of the instructions above Memorv GP Registers variables representing PSW the Fetch Instruction arrangement Decod r Fetch Operands of the opcodes would make would not be necessarv to code each i example MNEMONIC HEX CODE BINARY CODE SAMPL CR 19 0001 1001 CR CH 49 1000 1001 CH C 59 1001 1001 E Examine the bit patterns above Machine Format 00 RR 10 RX 01 Fullword 00 Halfword in this case Verb 1001 algebraic Compare In maintain and go through 4 83 etc Execute cycle The it easy to share code i e it nstruction separately As an consider the follo
102. for ASSEMBLER G COMPILE AND GO and it has two JOB STEPS SOURCE and DATA The typical deck setup is EXEC ASGCG PARM DATA MAP SOURCE INPUT DD Griy gia 360 assembly language source deck DATA SYSUDUMP DD SYSOUT A required for a dump DATA XSNAPOUT DD SYSOUT A required if XSNAPs are used The above procedure is the most efficient way to assemble and run an assembler program which cannot be run by ASSIST B LINK This procedure is somewhat slower than the above EDITOR CATALOG ED PROC E EDURE ASGCLG but can be used It and contains three JOB required if XSNAPs are used rather than a to run somewhat large programs and offers additional features stands for ASSEMBLER G COMPILE LINK and GO STEPS SOURCE OBJECT and DATA EXEC ASGCLG PARM OBJECT MAP SOURCE INPUT DD sawas 360 assembly language source deck DATA SYSUDUMP DD SYSOUT A required for a dump DATA XSNAPOUT DD SYSOUT A C EXECUTION ONLY CATALOGED PROCEDURE ASGG In some cases the user may have an OBJECT DECK SOURCE DECK in which case he does not need th procedure ASGG has only 1 step EXEC ASGG PARM DATA MAP DATA DECK DD input object deck DATA SYSUDUMP DD SYSOUT A DATA XSNAPOUT DD SYSOUT A NOTE YOU MUST INCLUD ALSO USING
103. may be useful for the programmer to create a DSECT which describes the control block generated by the macro expansion This would permit the module to refer to arguments and return points using symbols rather than absolute displacements A typical DS dsectnam DSECT ECT might be DS V routine space for adcon DS 3F space for regs 14 15 0 argument DS F argument value placed here if any A further arugment DS statements follow return LM 14 0 4 14 return label YES THIS IS LEGAL it does NOT generate code but it makes the point clear as to description of block If such a DSECT were used the routine code would include USING dsectnam 14 to set up DSECT addressibility B return return instead of B number 14 ASPRGTCI 06 II ADVANTAGES AND DISADVANTAGES The following lists the good and bad points of each type A MACRO INSTRUCTIONS 1 ADVANTAGES Code can be tailored to each individual request i e the code generated bv each macro call can varv from a great deal to nothing such as debug code eliminated by testing a global set variable SPEED macro generated code can be the fastest in execution since it can perform its actions without having to set up linkage to another section of code VARIABILITY generated code can vary depending on the nature of arguments passed to a macro such as testing the TYPE of arguments to generate different instructions
104. memory allocation algorithms may be affected by the requirement of reports R2MJ R2MM R2MT and RAM i e it may be the Case that it is NOT actually necessary to implement the algorithms in MAA MAC and MAM IF it is NOT required that memory maps be printed Thus it may only be necessary to know it there is enough space to load a job NOT where it actually is R5 R6 R7 R8 R9 CS411FP3 03 TERMINATION REPORT whenever a job has consumed all of its 0 allotted time it is terminated removed from memory This report should show Job name Current time JOB turnaround time current time time job was read into the system Memory residence time current tim time when the JOB first entered the AJL Depending on the version of 0S 411 used these might be useful R5R he Total Total 1 0 called TIMEUSED Percentage of time each CPU CHANNE especially for Random type systems total time spent using a CPU total I O time total time WAITING for CPU and total time WAITING for a channel ratio of CPU time used to total time showing CPU I O BOUNDEDNESS TYPE DISPATCH REPORT each time a job gets control of a CPU it is said to be DISPATCHED This is done whenever a job gets control of a CPU and displays time jobname and CPU number for multiple CPU systems Might show length of time the job is to be given control of the CPU
105. move jobs or not is heavily dependent on the size of the amp MAMOVE factor which is essentially a move cost Ii INITIATION DELAY FACTOR T SINITDEL INITiation DELay give the number of milliseconds required to load a job into memory once it has been chosen i e it cannot begin execution until time NOW SINITDEL JOB SCH When a This list ma jobs which a serviced by gain control the methods EDULING EX DURING JOB is INITIATE y be called a re in memory a CPU D Ready Job Queue ready to execute he order of this list determines which jobs may CS411FP2 04 ECUTION it is added to a list of jobs in memory RIO since it is a list of but not currently being of a CPU when on described in IV B becomes available itself can be ordered in any way desired the JO options Naturally the list and generally includes all of plus some additional others commonly used only for CPU allocation Briefly time when a JOB enters or leaves memory the RJQ is examined he first job s EMPTION is allowed which would appear earlier on the RJQ than a job w a CPU is allowed to seize control of the CPU from the other is then returned to the RJQ A table is also kept of the JOB Whenever a JOB is first give it keeps control of it for AT MOST the length of time any reason used for ANY of the idle currently in of a CPU r
106. nanosec 2361 II 1 unit Large Core Storage LCS 2048K bytes organized physically of 2 way interleaved doublewords CYCLE TIME 8000 nanosec 8 microsec ACCESS TIME 3 2 mic Of the two types of storage the first contains user programs and heavily used parts of system programs while the LCS contains less used system programs tables and buffer areas HARDWAR1 02 CHANNI E H a 2870 ULTIPLEXOR CHANNEL includes 2 SELECTOR SUBCHANNELS used for magnetic tape drives generally handles LOW SPEED devices card readers printers etc AXIMUM TOTAL TRANSFER RATE 426 KB kilobytes per second 2860 SELECTOR CHANNELS 5 total 2 in 2860 II 3 in 2860 III used for HIGH SPEED devices disk drum etc AXIMUM DATA TRANSFER UNIT EACH SELECTOR 1250KB LT All CHANNELS and the CPU contend for use of memory modules The BCU arbitrates among them using a simple priority scheme in following order SERVED EARLIER gt SERVED LATER CHANNEL il 2 0 3 4 5 CPU drums disk mx disk disk ADAGE The above order is used since the drums cannot wait very long and have the highest transfer rate the multiplexor channel 0 is fairly early because it may have a large number of things to do and the CPU is always last because it never hurts it to wait CONTROL UNITS Each control unit can at
107. necessarily written as macros since it makes little sense to call a routine in order to perform linkage unless the linkag cod required is very complex in which case the program is probably going to be SLOW B INTERNAL SUBROUTINES Internal subroutines are usually used as opposed to macros which generate code in line under the following circumstances CODE WITH LITTLE VARIANCE if the code is not going to be much different from macro call to macro call it may be better to let the macro call generate a BAL to one copy of the code as an internal subr Internal subroutines are usually used as opposed to EXTERNAL subroutines under these circumstances SHORT CODE HEAVILY USED if code must be used many times by a CSECT then the faster linkage of internal subroutines usually makes it worth writing it that way CODE NEEDED ONLY BY ONE CSECT if not too long it is fairly logical to incorporate it as part of that CSECT It will probably be much more efficient since it will have access to the internal variables of the CSECT and be able to communicate via register values easily rather than requiring long operand lists Ex EXTERNAL SUBROUTINES LONG CODE if something is long and complex enough it may be a good idea to make a separate module of it test it get an object deck then leave it along thereafter CODE OF GENERAL USE NEEDED MANY PLACES in this case it is
108. next node time this event to occur data for event type might be addr of control block address of routine or code giving index to it anything else which might be useful UN INITIAT ED JOB LIST UJL L element is needed for each job waiting on disk for execution ht be up to amp MAXJOBS of these needed EAch node must contain all useful information from a JOB card plus anything else desired A typical dsect might be UJLNODE DSECT UJLLINK DS UJLORDR DS UJLNAM DS UJL DS UJLENTR DS addr of next node order field compute according to JO option name of job from JOB card number of job showing order in which read time at which job was read appropriate entries for T SP IOIN IORL PRIO CHAN CAT 4 One AJLNODE ACTIVE CS411FP4 05 JOB LIST AJL is used for each job currently in memory and would typically include all items present in the UJLNODE or a pointer to a having them set of areas like those of UJLNODE There are a maximum of amp JOBLIM3 of these and time when job first loaded into memory initiated time when status of job last changed useful for statistics handling time remaining until job is completed this is decremented and job terminated when it 0 time until a job either requests I O or is scheduled to be terminated useful in I O handling address of CPU element or channe
109. now whatever time th vent is to occur c Uses a code value from the node to determine which of several routines RDR JBINIT etc to call to do required action d Loops through a until the EQ is empty i e nothing more is to occur THE ABOVE LOOP IS AN EXCELLENT CANDIDATE FOR EMBEDDED DUMPING CODE WHICH CAN BE TURNED ON OR OFF ACCORDING TO VARIABLE SWITCHES B DATA STRUCTURES AND CS411FP4 03 TABLES his program is essentially a huge list processing program which of course makes a set of list processing macros very useful It is also useful to write simple macros then combine calls to them in more powerful and asier to us for this program static is initialized all of th of node can be pushed onto the free list for that kind of node and thus they will be ready for the next BATCH although they may not be in exactly the same ORDER as limits on the number of any type of node this method should work and be easy to do The dynamic method implies maintaining a single FR macros here are two basically different ways of allocating the storage and dynamic he static method is the easiest to program Whenever the system e nodes in an active list for a specific kind they were before Since there are upper T ELIST of all unused memory then alloc needs one then returning the n
110. numbers Two s complement One s complement Sign magnitude advantages and disadvantages TC 1 zero but harder for people OC 2 zeroes but easier to handle SM asiest to handle but slower circuitry NG STRUBLE CHAPTER 1 Look at ASSIST PART III more on information representation introduction to machine structure meanings of bit patterns 1 2 4 byte binary numbers charcters packed decimal good for people but wastes space floating point sign characterisitc and fraction structure of a very simple machine memory of 16 bit words 1 register 1 program coiunter a few instructions each with opcode and address explanation of basic instruction cycle 1 Fetch instruction from where program counter points 2 Increment program counter Decode instruction into its parts Execute instruction Loop back to 1 Owe Ww S 360 machine structure memory note abbrev K GP and floating point registers PSW refer to GREEN CARD T Begin instruction types 1 RR names with R examples 2 RX give first explanation of base displacement 3 RS NGS STRUBLE CHAPTER 2 POP pp 7 15 UTS CS102M1 page 01 run some ASSIST programs for dumps 3 01 12 READI CS102TPA 02 finish operands formats and introduce assembly language 4 SI instructions examples MVI CLI 5 SS instructions examples MVC CLC machine language easy for machine
111. of a program just before the data areas It is a good idea to set up conventions for the use of internal subroutines before writing any The following ar often needed return register either one standard one or several different ones argument registers and work registers which can be used without saving In general internal subroutines should not need to do much saving and restoring of registers They should be able to return via BR REG 2 INVOCATION Calling an internal subroutine is usually done by first filling any argument registers with needed values then coding BAL REG INSUB This type of linkage can be fast and small C EXTERNAL SUBROUTINES External subroutines are used for major program segments and can usually be assembled separately from the rest of the program In fact they can be written ina different language i e FORTRAN and ASSEMBLER combinations 1 DEFINITION An external subroutine may be written in either of two ways in assembly language as a CSECT or as an ENTRY within a CSECT In the first case the subroutine is entered at the CSECT statements and return at one or more places depending on the desired code In the second case each entrypoint may be given control and may share code or be totally separate from the other entries This form is often used for a group of related routines like SIN and COS which are both entries in a CSECT or for a routine r
112. of the fourth word of the save area the place register 14 was stored Although this technique does not seriously affect the contents of the save area for reading in a dump this technique quickly shows what save areas are active and which are not active when reading a dump Register 13 must be kept as the save area pointer however by careful programming it can also double as a base register Consult the appropriate section from XSAVE and XRETURN documentation for the coding sequence using these macros You may set up your own save area for this purpose by setting it high in a program and following it by a USING on register 13 referencing the name of the save area For reserving the 18 fullwords of storage for a save area use DC instead of DS A constant of F O or F 1 will quickly show in a dump if the save area was ever used SAVE and RETURN are two system macros which will eliminate much of the coding for saving and returning conventions SAVE generates th cod necessary to save a specified series of registers The registers are specified as they would be for a STM instruction In addition the operand T will cause registers 14 and 15 to be stored regardless of what other registers may also be saved from the pair specified The following example will cause registers 5 6 10 and 14 and 15 to be saved SAVE 5 10 T The RETURN macro will generate code to re
113. output as requested The run will use control cards like EXEC ASACG PARM BATCH SYSIN DD SJOB ASSIST PROGRAM VERSION II AR program II pk teste test data repeat above starting at JOB for programs III and IV The following test data should be used for each program A B Cc D E 5 2 4 2 2 2 si 10 1 l 4096 1 1 iso DI Note that the columns they are punched in should not matter CS102AS1 03 COMPUTER SCIENCE 102 ASSIGNMENT 2 This assignment uses the concept of indexing into an array of elements Es BASIC PROGRAM A Read a card and echo print containing a maximum of 20 numbers Convert the numbers to hex XDECI and store them in successive fullwords in memory Use a loop to eliminate redundant coding Then for each card find the maximum value and the minimum value printing out these numbers with appropriate labels B Form of data 1 Each card contains a maximum of 21 numbers where the first number the number of numbers on the card You will need the first number for a counter in the loop in part A 2 There are an unspecified of data cards i e make your program general to accept any of data cards L DATA FOR YOUR PROGRAM 96 LO 76 11 123 432 123 748 9087 0 33 33 45 10 6 90 145 1024 6698 1024 345 Ur O Ju PRIME INDEX J R MASHEY E PROVIDES THE PRIME INDEX
114. overall purpose of the program and gives some indication of the general structure Each major section has a block of comments describing it as does each of the section s subsets 2 At least 95 of machine instruction statements have comments 3 The program is easy to read and blocked off into logical sections so that anyone may look at it and understand it easily 4 Good programs typically have 15 25 of the total statements as comment cards in addition to the comments on the individual statements S 360 ASSEMBLER DOCUMENTATION HINTS DO S and DON TS DON T punch statements in random columns This makes a program very unreadable Use a drum card and if you do not know how ask your assistant The following is a defacto standard for S 360 Assembler statements Col 1 LABELS Col 103 OPERATION CODES COLL L6 OPERAND FIELD Col 36 COMMENTS col 40 is preferred by some people Col 724 CONTINUATION COLUMN Col 73 80 SEQUENCE NUMBERS very useful ask your assistant how to sequence a deck if you are unsure This layout can be obtained by the use of the following drum card Cols 1 10 16 36 73 punch l gives tab stops at these cols Col 722 punch skips col 72 automatically unless AUTO DUP SKIP is off All other columns punch A If for some reason these columns are not wanted a standard set should be decided upon and then held to comple
115. practically necessary to make code an external subroutine so that it can be accessed where needed D COMBINED FORMS These are useful anywhere the others ar The nondestructive form is specially useful if it is to De used by beginning programmers FRE JE THIS JOB WILL RUN WITH TIME 75 SECONDS jif RECORDS 1000 EXEC ASGCL SOURCE INPUT DD tA E PUTPOSE OF THIS JOB IS TO DEMONSTRATE THE MACRO S LISTEC B PUTPOSE OF THIS JOB IS TO DEMONSTRATE THE MACRO S LISTECD HT ATTACH DETACH WAIT POST EXTRACT THE OVERALL FLOW OF THIS PROGRAM IS 1 HE CSECTS SECOND AND THIRD ARE COMPILED AND LINKED 2 MAIN IS COMPILED AND LINKED EDITED AND EXECUTES 3 DURING THE EXECUTION OF MAIN 1T ATTACHES SECOND WICE USING THE ATTACH MACRO BEFORE SECOND 15 ATTACHED THE DISPATCH PRIORITY OF MAIN 15 OBTAINED USING THE EXTRACT MACRO AFTER 1T HAS BEEN OBTAINED DIVIED BY 2 AND WHEN SECOND IS ATTACHED THE PRIORITY OF MAIN IS HALVED USING THE DPMOD PARM WHILE BOTH MAIN AND ECOND ARE COMPETING FOR CPU USE THE TCB S FOR MAIN AND ECOND ARE SNAPPED USING THE EXTRACT MACRO HEN SECOND AND HEN SECOND 15 EXECUTING 15 ATTACHED 1T 15 GIVEN AN ECB AND N MAIN A WAIT MACRO 15 ISSUED FOR THIS ECB THEN SECOND 15 DETACHED IN MAIN USING THE Bi p
116. read and blocked off into logical sections so that anyone may look at it and understand it easily 4 Good programs typically have 15 25 of the total statements as comment cards in addition to the comments on the individual statements S 360 ASSEMBLER DOCUMENTATION HINTS DO S and DON TS DON T punch statements in random columns This makes a program very unreadable Use a drum card and if you do not know how ask your assistant The following is a defacto standard for S 360 Assembler statements Col 1 LABELS Cot 10 OPERATION CODES Col 164 OPERAND FIELD Col 36 COMMENTS col 40 is preferred by some people Col 124 CONTINUATION COLUMN Col 73 80 SEQUENCE NUMBERS very useful ask your assistant how to sequence a deck if you are unsure This layout can be obtained by the use of the following drum card Cols 1 10 16 36 73 punch l gives tab stops at these cols Col 72 punch skips col 72 automatically unless AUTO DUP SKIP is off All other columns punch A If for some reason these columns are not wanted a standard set should be decided upon and then held to completely DOCUMENT 3 DON T Place a comment card before every statement This bad habit makes programs absolutely unreadable Embedded comments should be used to block programs into logical sections not to explain the function of individual statements
117. s and DROP s because they us DSECTs more than do beginners SPACE optimize to produce the smallest complete program PROGRAMMING SIMPLICITY optimize to produce a running program which is simple and understandable and can be programmed quickly i e simulating the conditions which require a program to be finished in a short period of time Ei AA4 OUTR AA5 DSECTI DS1 DS2 DS 4 DS5 DS6 DS8 DSECT2 CARD DS2A BASE AA6 MAIN RROR COMMENTS ED ON 8 ERROR IT PROVIDDI EJ n ITLE ASSIS BASE REGISTER USING DROP TEST PROGRAM THIS PROGRAM PROVIDES VARIOUES ERROR TESTS FOR BROPS2 ALL STATEMENTS LEGAL EXCEPT THOSE WITH E CSECT BALR 12 0 USING 12 L 0 AA5 AS100 L 5 AA4 USING AA5 9 ST 3 AA5 USING AA5 10 A 2 AA5 USING AA4 10 SL 3 AA5 ST 3 AA4 DROP 10 USING DSECTI 7 STH 4 DSI LH 5 DS6 DROP 8 10 TWO AS003 MESSAGES CVB 6 DS8 USING DSECT1 9 CVD 6 DS8 USING DSECT2 10 8 DS4 LA 5 CARD D 8 DS5 DROP 10 B TI EC 10 DS2A AS100 ADDRESSIBILITY LA 11 OUTR AS100 ADDRESSIBILITY USING 13 ST 5 AA4 DS F DS 1500F DS E DS F EJEC DSEC DS H DS HL8 DS A DS F DS CL6 DS D DSEC DS CL80 DS P CSEC USING 13 14 15 DROP 8 1
118. see what the base register and displacement by which they were assembled is LOC OBJECT CODE ADDR1 ADDR2 STMT SOURCE STATEMENT 000300 192 TRACE CSECT 193 XSAVE SA TRACESA 000000 220 USING SAVEAREA R13 000000 221 USING NAMECONV R15 222 XPRNT CL15 BACK TRACE 000386 58DD 0004 00004 232 L R13 4 R13 00038A 12DD 233 LOOP LTR R13 R13 00038C 4780 COTE 003E0 234 BZ DONE 000390 58F0 D010 00010 235 L R15 REG15SAV 000394 D503 F000 C1C6 00000 00528 236 CLC BRANCH 2 X 47F0 00039A 4770 C13A 0049C 237 BNE ERROR 00039E 4370 F004 00004 238 IC R7 LENGTH 0003A2 0670 239 BCTR R7 0 0003A4 4470 C170 004D2 240 EX R7 MOVE 241 XPRNT OUT 40 0003CE D226 C177 C176 004D9 004D8 251 VC OUT 1 39 OUT 0003D4 98EB DOOC 0000C 252 L R14 R11 REG14SAV 0003D8 58DO D004 00004 253 L R13 BACKLINK 0003DC 47F0 C028 0038A 254 B LOOP 255 DONE XPRNT CL20 TRACE COMPL 000406 41D0 COF2 00454 265 LA R13 TRACESA 266 XRETURN SA TRACESA 285 ERROR XPRNT CL20 ERROR IN TR 295 ABEND 999 DUMP 0004D2 D200 C177 F005 004D9 00005 303 MOVE MVC OUT 1 NAME 0004D8 4040404040404040 304 OUT DC CL40 000500 305 LTORG 310 000000 311 SAVEAREA DSECT 000000 312 UNUSED DS F 000004 313 BACKLINK DS E 000008 314 FORELINK DS B 00000C 315 REG14SAV DS F 000010 316 REGISSAV DS F 317 000000 318 NAMECONV DSECT 000000 47F0 F000 00000 319 BRANCH B 0 15 000004 320 LENGTH DS C 000005 321 NAME DS ZZK 000000 323 END MAIN COMPUTER SCIENCE 102 RUN CS102M1 01
119. since the drums cannot wait very long and have the highest transfer rate the multiplexor channel 0 is fairly early because it may have a large number of things to do and the CPU is always last because it never hurts it to wait CONTROL UNITS Each control unit can attach to a number of devices and it is used to control greatly different devices in a such a way as to make them appear more alike as far as the channels are concerned Each device must be attached to a particular type of control unit and each control unit normally can control a group of related devices 2820 STORAGE CONTROL UNIT controls the 2301 drum units attached to channel 2821 CONTROL UNIT controls UNIT RECORD devices card readers printers punches attached to multiplexor channel 2848 DISPLAY CONTROL controls the 8 2260 scopes which display system status to the operators 2701 DATA ADAPTOR controls a small number of high speed transmission lines i e high speed terminals 4800 bits sec transmit rate such as 360 20 s at various locations 2103 TRANSMISSION CONTROL controls a larger number of lower speed terminals including typewriter teletype terminals and read print punch terminals at Commonwealth Campuses such as IBM 2780 DCS CP 4 etc DISPLAY DI EVICES 1052 2260 SECONDARY 2301 231x TOTAL DASD STORAGE CONSOLE TYPEWRITER action by co
120. specified in 3 bytes byte 1 is ig nored on an address constant LINKAGE 2 The following example illustrates how to use the address list passed through register 1 LA 1 ARGLIST get argument list address L 15 V CALLRIN get entry address BALR 14 15 Call routine ARGLIST DC A ARG1 DC A ARG2 DC X 80 AL3 ARGn Note the length factor does not provide auto matic alignment CALLRTN CSECT L 2 0 T get addr of next arg LTR 1 1 Last arge ia list BM RETURN if yes return LA 1 4 1 else get addr of next arg When a programmer receives control from the system information from the PARM field of his EXEC card is passed via register 1 Register 1 points to a fullword of storage Bit 0 of this fullword is set to 1 to indicate the last only argument of the list This fullword contains the address of a halfword The halfword is a count of the number of characters in the parm field message and these characters follow immediately after the halfword count field The contents of the halfword may be picked up to use as a length count in an execute instruction and the address of the halfword may be used as a base to move the information characters of the PARM field REGISTER 13 is called the save area register It contains the add ress of an 18 fullword area on a f w boundary within the calling routine The routine called will use this area to save the contents of registers to
121. that job were in the RJQ the new Job may seize control of the CPU i e PEEEMPT it In this Case the old job is removed from the CPU and placed back in the RJQ with the following information computed elapsed time in CPU must be subtracted from both the job time remaining and the time interval until the next 1 0 for it and the fact that this job has been preempted must be noted so that a new interval until 1 0 is NOT computed when job regains control of the CPU later Two options are thus possible amp EOP 0 NO PR LT EMPTION amp EOP 1 PREEMPTION ALLOWED more complicated to program but usually better for utilizing CPU and CHANNELS since it avoids having CPU BOUND programs tying up CPU for long periods of time VI Normally system must WAIT INPUT OUTPUT SCH EDULING when a job receives control it keeps control of that CPU until CS411FP2 of a CPU on a R 08 FAL computer it either requests I O and or a higher priority job s I O completes and is allowed to preempt it Since the jobs in a simul whenever a job is scheduled to recei the time until an I O request occurs must be COMPUT 1 0 request scheduled to occur until the job relinquishes the CPU is always the MINIMUM of an I O INT ERVAL JOB time remaining QUANTUM When is removed from memory I O REQUEST if using at t
122. the location referenced should be 90 100 10 NOTES AND GENERAL INFORMATION The SIGMA 4 5 stops at doing only 1 indirect address step and thus uses ONE LEVEL INDIRECT ADDRESSING Some other machines would extend this by testing the I bit in a word and if on perform another indirect addressing step until a word was found witlout I 1 in which case that word would have the address of the word finally used as an operand This type is called MULTILEVEL or CASCADED INDIRECT ADDRESSING Also since SIGMA 4 5 adds the index register X AFTER any indirect addressing is done it is called POST INDEXING Some machines perform indexing before indirect addressing and are thus called PRE INDEXING systems The RI Register Immediate format instructions are basically similiar to the RX except The I bit is ignored The X and M fields together form a two s complement signed 20 bit number which is extended to 32 bits then used as the operand in combination with the register R This allows numbers from 1048576 to 1048575 to be specified in the instruction 3 The following table instruction OPCODE E TAB L the machine code for each operation the instruction format condition cod an instructio e and finally n corresponds to an S 360 instruction
123. the address of the entry name V entryname into R15 and BALR s there assuming RI is already set correctly Version 3 is like version 2 but also does a LA to get the address of the address list into register 1 before calling the routine Version 4 combines versions 1 and 3 with the address list created inside the macro expansion like the CALL macro does see CNOP instr D SAMPLE EXPANSIONS OF THE MACROS These expansions are examples IT IS NOT NECESSARY TO GENERATE THIS EXACT CODE AS LONG AS THE CODE MEETS THE REQUIREMENTS GIVEN MAIN CSECT OSAVE USING 15 TEMPORARY USING B 10 15 BRANCH AROUND IDENT DC AL1 5 CL5 MAIN STM 14 12 12 13 SAVE REGISTERS LA 12 MAI0002S GET SAVE AREA ADDRESS ST 12 8 13 POINTER TO NEW SAVE AREA ST 13 4 12 POINTER TO OLD SAVE LR re GET IN RIGHT SAVE AREA BALR 12 0 SET UP NEW BASE DROP 15 DELETE TEMPORARY USING 12 NEW USING CALL QCALL SUBX ADDRX CALL DS OH DEFINE LABEL LA 1 ADDRX ADDRESS OF ADDRESS LIST L 15 V SUBX SUBROUTINE ADDRESS BALR 14 15 CALL ROUTINE GOBACK QRETURN SA GOBACK DS OH DEFINE LABEL L 13 4 13 RESTORE PREVIOUS SA PTR LM TA FONT Sh 4 RESTORE REGS BR
124. the following reasons 1 WTO WTOR WTL HASP adds own processing as desired 2 I O to disk drum tape terminals etc HASP does not interfere but passes these on to the real I O Supervisor 3 I O to real unit record devices these have probably been issued bv HASP in the first place so it passes control to the real I O routines to let them perform the I O 4 I O to a pseudo devic these must be caused bv user program For input HASP fetches the cardimages from disk into memorv if thev are not alreadv present and feeds requested cardimage s to the user program bv MVCing them there using user protect kev for safetv For output it blocks up output and eventuallv writes it to disk In all cases HASP simulates the effect of having real card readers printers punches which are odd onlv in possessing great speed i e th ffect on OS 360 is of having issued an I O request and having had it complete immediatelv During execution HASP can also provide extra services such as monitoring time used output records etc It also reorders priorities of executing user tasks so that I O bound jobs have higher priorities than do CPU bound ones This action which is unknown to 0S 360 helps minimize time spent waiting 3 Output Stages Print and Punch after a job has been executed it enters the Print queue is printed enters the Punch queue and has punched output if any actually punched This activity occurs
125. then passes control to NIP OSHASP B NIP Nucleus Initialization Program The IPL process The NIP is generated of system and choice of options desired link edited with the nucleus nucleus via address constants initialization services It includes the following steps in different ways Note is 1 The CVT Communications Vector TAble location placed at location 16 routine whether part of the nucleus or not initialized 2 NIP determines whether th attached to it or not systems which include HIARCHY SUPPORT distinguish between main core and LCS heavily used and lesser used sections i e Ss NE checks the instructio and contai P checks the workability of operator console s workability of ready direct access devices ns It particularly checks that the SYSRES volume ns certain datasets needed by the system 4 NIP performs various housekeeping actions setting the timer to make sure it is working correctly computer has Large Core Storage This is particularly necessary for those the ability to usefully perhaps splitting programs into 02 described above applies to all versions of 0S 360 depending on the specific type NIP is a csect which is so that it can refer to sections of the and provide efficient and specific and its so that it can be accessed from any LCS and also using TIO is mounted such as checking and initial izing
126. these UCB s which is not being used and effectively changes the tables for this type of card so that it appears as XXXXXXXX DD UNIT xxx As a result the OS RDR thinks that the data set will be read from unit xxx so that it does not try to SPOOL the input In any case the input no longer follows that JCL card because HASP feeds the RDR only the JCL cards of a user job During this process HASP connects up the device address xxx to the specific input data set which had been previously SPOOLed XXXXXXXX DD SYSOUT x HASP also has a large number of UCB s for nonexistent pseudo printers punches It does the same thing to this kind of card as it does to DD cards except that it only allocates the pseudo devices and will later save the output which is written to them As soon as the RDR finishes reading a job an initiator can immediately initiate it since HASP chooses jobs appropriately When the initiator chooses i o devices it finds that it can always allocate devices for unit record i o since HASP had already checked to make sure a pseudo reader printer punch was available for each SYSIN or SYSOUT data set OSHASP 07 Finallv a job step of the user job executes When it wishes to read cards or print lines it acts as though it were using a real device attached to the system and OS 360 accepts this Whenever an SVC 0 is issued to request such I O HASP intercepts it HASP mav be entered for anv of
127. this latter case the sequenc LR BASEREG 15 USING entrypointname BASEREG where BASEREG is any of registers 2 12 may be used to establish permanent addressability On return register 15 may be used to return a code to indicate normal or error return One frequently used technique is to set R15 zero on a normal return and set it non zero if some error condition occurred prior to return REGISTER 0 is used to return the single result from some process as in a Fortran function subprogram Note although you will probably not use this convention much it is heavily used by the operating system Register 0 cannot be guaranteed to be intact after executing some call to the system as a system macro or an SVC REGISTER 1 is the pointer to an argument list It contains the address of the first of one or more full word entries on con secutive f w boundaries Thes ntries ar th addresses of arg uments to be used by the calling routine If there may be an indefinite number of arguments as with a routine which would accept one two or any number of arguments c f Fortran MAX0 the first bit of the last address is set to a 1 This bit will not interfere with ordinary S 360 addresses since an address can be fully specified in 3 bytes byte 1 is ig nored on an address constant LINKAGE 2 The following exampl
128. to execute hard to write assembly language converted by assembler to machine code format of assembly language label opcode operand comments machine instructions actual operations to be executed assembler instructions pseudo ops give information to the assembler ex CSECT DS DC some basic functions of the assembler 1 location counter 2 convert mnemonic opcodes a machine ops translate to codes increm location cntr b assembler ops take actions specified increm loc cnt 3 operands convert to internal binary base displacement 4 print out a listing 5 make program ready for execution and pass control to it stepped through complete test program XREAD XPRNT XDECI XDUMP and explained listing and contents of dump NGS STRUBLE Chapter 3 ASSIST MANUAL PARTS II and IV ASSEMBLER LANGUAGE pp 1 18 HANDOUTS DOCUMENT documentation techniques for assembler 4 01 14 go over some dumps and errors discuss operand fields go through various dumps showing 0C1 0C4 and 0C6 errors cover STRUBLE cahpter 3 pp 50 56 symbols self defining terms literals location counter reference absolute and relocatable terms xpressions READINGS STRUBLE Chapter 4 to page 78 ASSIGNMENT STRUBLE Chapter 1 problems 5 6 7 8 9 Chapter 2 problems 2 3 Chapter 3 problems 1 2 3 4 6 INFORMAL ASSN modify dump program to use XDECO and DUMP st
129. used to save regis XRETURN SA will suf nt however XRETURN one and at that one his is that SA will be created per module For further details on the parameters involved in these two macros see the appropriate PSU documentation The following example causes register 12 to base register causes all registers to be saved trace messages to be printed on entry or on exit loaded with the return code value 8 MAIN CSECT be established as a on entry cuases no and causes R15 to be XSAVE BR 12 TR NO Note default is for all XRETURN SA TR NO RC 8 registers to be saved THIS JOB WILL RUN WITH TIME 110 RECORDS 800 EXEC ASGCL SOURCE INPUT DD THE PURPOSE OF THIS PROGRAM IS TO DEMONSTRATE THE MACROS GIVEN BELOW LOAD DELETE LINK XCTL THE O
130. using LOOK UP illegal operand gt ves gt set flag also flag no for execute finish object code for stmt Deo gt print card image amp error message s update loc counter ti LOOK UP find svmbol in table gt yes gt return displacement amp flag no RETURN return 7 fa no lt xecute fals set up psedo reg amp storage execute instructions wait stmt gt no ves Print dump Return to monitor tt y s CS102FP1 08 gt print terminatio message flush cards unti S TRIVIAL CS102FP2 01 V SPECIFIC LANGUAG CJ DETAILS FOR THE TRIVIAL SYSTEM This section describes th xact details for the various parts of the TRIVIAL system The monitor main program processes various control cards which direct further processing The TRIVIAL assembly language is converted by the assembler into machine code for a computer system called the SIGMA 4 5 The interpreter portion then simulates the operation of the SIGMA 4 5 using S 360 instructions A CONTROL CARDS 1 JOB CARD SHOWS BEGINNING OF TRIVIAL PROGRAM This card has the following format starting in column 1 JOB jobname number S JOB identifies this as a JOB card jobname is a sequence of up to 20 nonblank characters which identify this JOB number gives a maximum limit on the number of instructions which can be executed by th
131. 0 67 configuration this lecture continues preparation for general concepts of operating systems and I O system operation CPU note Mode bit for supervisor problem states types of memory protection 1 NONE use in HANDSON systems only 2 PROTECT BIT IN WORD cheap but inconvenient 3 BASE LIMIT REGISTERS l pair movable programs but no reentrant code easily 4 2 PAIRS OF BOUNDS REGISTERS rrentrant code OK 5 LOCK AND KEY S 360 in detail to be continued under virtual memories I O DEVICES beginning with DASD s Drums fixed head disks movable head disks data cells information on devices CS411FP1 01 08 lst part of final project ASSIGN final project due 03 09 15 02 08 READING read on privileged operations in POP I O devices completed I O processing beginning miscellaneous questions on current addignments agnetic TApes physical records vs logical records blocking factors tape gaps parity bits EVEN ODD comparison of tape usage blocked 80 80 or 80 8000 Printers bar drum train chain I O channels control between CPU and CHANNELS CYCLE STEAL Types of channels selector burst mode only multiplexor burst and multiplex modes both S 360 actions interrrupts old and new PSW s usage Masks masking off interrupts using lst 4K of memory HANDOUTS CS411FP2 01 08 2nd part of final project 16 02 08 N
132. 0 services 21 I O concepts buffering record formats blocking etc 22 Survey of common OS 360 macros except data management 23 JCL 24 JCL 25 OS 360 and HASP internal structure IPL NIP etc ASP 26 Assorted topics microprogramming cache and virtual memories 27 Assorted topics pipeline and array computers non IBM systems 28 29 30 left for exams problems expansion etc COMMENTS the schedule above is fairly tight It has been useful especially in the first several weeks to give help sessions once a week in the evenings to go over problems review etc This has been useful especially to even out differences among 102 410 courses taken at different times and or different campuses Note that the order of topics above seems to work out fairly well since it gives the students the material needed for the assignments fairly early Note that some explanation of JCL should be given thoughout the term when relevant to specific assignments etc so that the two days allotted to JCL contain a unified explanation but assuming that the students should have some familiarity with it by then CS411GI2 02 ETAILED OUTLINE REFERENCES TOPICS Outline of course administrative details grading exams course prerequsisites review X Macros XREAD XPRNT XPNCH XDUMP XSNAP XDECI XDECO XHEXI XHEXO XSET Note that a few of these may NOT be review but new material X MA
133. 000 150800 RASP 1236 1336 100 135000 14E000 FREE 1024 1236 212 100000 135000 S 964 1024 60 0F1000 100000 HASP 876 964 88 ODBOOO OF1000 RDR 866 876 10 0D8800 ODBOOO FMGR 852 866 14 0D5000 0D8800 RJE 832 852 20 0D0000 0D5000 WATFOR 704 832 128 0B0000 0D0000 FREE 122 704 582 015800 0B0000 NUC 0 1 22 122 000000 O1E800 NOTES MS Master Scheduler includes the Link Pack areas The fast core section contains mainly modules for the various I O Access Methods while the LCS part contains reentrant parts of INITIATORS RDRS plus HASP other routines overlay supervisor special tables etc Fast core section is most heavily used sections LCS part has less r used sections plus such items as in core SYSJOBQUE HASP intercepts all RDR and INIT reads writes to SYSJOBQUE and keeps such information in about 600K of LCS Also has HASP buffers for all devices plus tables of tape names locations for user tapes FMGR RJE WATFOR RASP FREE File Manager manages synchronizes RJE BAT files Remote Job Entry handles typewriter terminals WATFOR REgion RPSS manages CAtegory W fast processors swapped in and out of memory WATFOR ASSIST PL C etc Interface between 360 67 and ADAGE AGT 30 graphics computer fast core 560K for user programs 4x140 2x280 1x280 2x 140 occasionally 1x560 rest for Sytem Queue Spac LCS currently unused except for systems pro
134. 06 II ADVANTAGES AND DISADVANTAGES The following lists the good and bad points of each type A MACRO INSTRUCTIONS 1 ADVANTAGES Code can be tailored to each individual request i e the code generated bv each macro call can varv from a great deal to nothing such as debug code eliminated by testing a global set variable SPEED macro generated code can be the fastest in execution since it can perform its actions without having to set up linkage to another section of code VARIABILITY generated code can vary depending on the nature of arguments passed to a macro such as testing the TYPE of arguments to generate different instructions 2 DISADVANTAGES SLOW ASSEMBLY macro processing can be very slow LARGE CODE if used improperly macros can generate larg amounts of code very easily If there are many copies of large blocks of code much space can be wasted OBJECT DECKS a macro cannot be assembled and an object deck of it gotten like a subroutine can i e if a call is made to a macro the macro definition must be included in the program or in a library while a CSECT may be saved as an object deck which is usually much smaller than the source deck B INTERNAL SUBROUTINES 1 ADVANTAGES SPEED although not as fast as in line code from a macro the code for an internal subroutine is usually faster than the linkage to an
135. 14 RETURN MAI0002S DC 18F 0 SAVE AREA ADDRX QCALL MAIN GOBACK ADDRX DS OF DEFINE LABEL DC A MAIN DC X 80 AL3 GOBACK CS411MC1 03 ti ADDITIONAL FEATURES If desired additional features may be added to these macros which may receive extra credit Do not however spend too much time on this assignment and especilly do not spend a great deal of time adding exotic features unless they really seem useful F WHAT TO HAND IN Hand in one run with all the macro listings well commented and showing complete macro expansions and execution of A LEAST the following program you may add more if you want and should do so to make sure all reasonably different cases are tested MAINPRG CSECT OSAVE QCALL SUBX ADDRX QCALL SUBY GOBACK QRETURN SA ADDRX QCALL MAINPRG GOBACK LTORG SUBXCS CSEC H ENTRY SUBX SUBY SUBX OSAVE SA SUBXSA BASE 11 REGS 14 11 cnop 2 4 cnop for nastiness QCALL SUBI SUBX QCALL SUB2 SUBRET QRETURN SA SUBXSA REGS 14 11 SUBY QSAVE SA NO XPRNT CL50 0 AT SUBY 50 QRETURN SA NO LTORG SUB1 CSECT QSAVE BASE 13 QRETURN SA SUB2 CSECT OSAVE BASE 15 REGS 2 12 QRETURN SA REGS 2 12 END CS411MC1 04 COMPUTER SCIENCE 411 MACRO ASSIGNMENT DUE This assignment pages 04 08 II HEXADECIMAL CONVERSION DUMPING MACROS QHEXI QHE
136. 2 AS003 REGISTER NOT USI AH 6 DS1 AH 4 AA6 L 5 AA5 AS100 ADDRESSIBILITY DROP 7 9 10 AS003 ON 10 DROP 13 14 15 DS H TITLE OVERALL TES STUDENT WRITTEN PROGRAM THIS SECTION CAN BE USED TO TEST BROPS2 STUDENT WRITTEN SAMPLE PROGRA START SUPERVR ENTRY STM MAINSAV BOUT1 BIN1 COROUT2 BLANK OUTPUT OUTI HI STORI Li STORE1 H9 SUPER CSEC ENTRY PRINT BALR USING IN OUT SUPERVR 14 12 12 13 12 0 12 13 MAINSAV 9 0 6 0 11 0 15 V COROUT2 15 14 BIN1 15 BOUT1 15 14 BOUT1 15 BIN1 15 13 MAINSAV 14 12 12 13 14 V COROUTI STORE STORE1 4 0 1 OUTPUT 5 BLANK 0 64 1 3 5 3 3 9 9 STORE 15 V IN 15 OUTPUT 4 C r 14 10 4 1 r 4 t to ONS H 15 V IN 14 15 3 STORI 343 4 1 4 4 F 63 STORE 3553 BLANK 3 65 OUTPUT 62 C SUPER 9 1 OUTI 15 V SUPERVR 15 E COROUTI H2 INI PACK CVB BALR NUMB DOUBLI NEXT QWP gt HU a h F1 F2 F3 ti A Z HAR ERICK BALR BNM DS DS DS CSECI PRINI BALR J WE dt El al HA ENTRY LTORG USING 8 0 8 IN1 15 V OUT 14 15 V NEXTCHAR 14 15 10 C 0 H2 10 NUMB
137. 28 282 150800 197000 WATFOR 1336 1346 10 14E000 150800 RASP 1236 1336 100 135000 14E000 FREE 1024 1236 212 100000 135000 S 964 1024 60 0F1000 100000 HASP 876 964 88 ODBOOO 0F1000 RDR 866 876 10 0D8800 ODBOOO FMGR 852 866 14 0D5000 0D8800 RJE 832 852 20 0D0000 0D5000 WATFOR 704 832 128 0B0000 0D0000 FREE 122 704 582 01E800 0B0000 NUC 0 122 122 000000 01E800 NOTES MS Master Scheduler includes the Link Pack areas The fast core section contains mainly modules for the various I O Access Methods while the LCS part contains reentrant parts of INITIATORS RDRS plus other routines overlay supervisor special tables etc HASP Fast core section is most heavily used sections LCS part has lesser used sections plus such items as in core SYSJOBQUE HASP intercepts all RDR and INIT reads writes to SYSJOBQUE and keeps such information in about 600K of LCS Also has HASP buffers for all devices plus tables of tape names locations for user tapes FMGR File Manager manages synchronizes RJE BAT files RJE Remote Job Entry handles typewriter terminals WATFOR WATFOR REgion RPSS manages CAtegory W fast processors swapped in and out of memory WATFOR ASSIST PL C etc RASP Interface between 360 67 and ADAGE AGT 30 graphics computer FREE fast core 560K for user programs 4x140 2x280 1x280 2x 140 occasionally 1x560 rest for Sytem Queue Spac
138. 28 6821 8 IBM S 360 PRINCIPLES OF OPERATION This also will be used heavily The students should be familiar with much of the material but it may be needed for review of some operations such as TR TRT ED EDMK It will also be needed for the following topics system structure protection features I O status switching program states protection PSW instructions interruptions all in this section input output operations most of this material CAW CSW CCW basic operation of channels R lt INTRO gt 4 GC28 6534 3 IBM S 360 OS INTRODUCTION OR R lt C amp F gt GC28 6535 7 IBM S 360 OS CONCEPTS AND FACILITIES These manuals INTRO is effectively a reworking of the older C amp F give an overall view of operating system services using 0S 360 terms in particular They do not explain things in detail but give general concepts and vocabulary The student must be familiar with most of the concepts and terms in these manuals by the end of the course CS411GI1 03 The following group of manuals is continually changed around by IBM with various parts shuffled among manuals of same or changing names or numbers There ar ffectively 4 distinct modules of information a Supervisor Services description of general concepts b Supervisor Macro Instructions coding details for these macros c Data Management Services description of general concepts d Data Management Macro In
139. 4 A dump should be provided at the end of each job Information included will be the value of the program counter the contents of th registers and the storage tt CS 102FP 05 IV FLOWCHARTS A General Flow of Monitor TRIVIAL MONITOR read a card flush is it gt S or anything else or job card 2 jif gt print end of processing message job card STOP PASS 1 assemble opcode res create symbol table call TRIVIAL assembler PASS 2 assemble operands EXECUTE xecute assembled call TRIVIAL interpreter program Ht CS102FP1 06 B TRIVIAL assembler Initialization and Pass 1 TRIVIAL ASSEMBLER initialize symbol table to blanks set flag that prog is executable read next card lt B END card gt yes gt Pass 2 no is there a a label gt yes gt LOOK UP gt in gt no exec symbol set flag not in enter symbol amp set flag displacement in print err table Note whether message i an RES PASS 2 determine typ RES of stmt create convert operand to binary partial obj code update loc update location counter fo counter illegal opcode proper boundary if necessar B store number set flag for update loc counter false for execution B B tt tt ti TRIVIAL Assembler Pass 2 PASS 2 gt get next card image CS102FP1 0 end stmt gt yes gt return to Monitor no process operands
140. ACT get data from TCB Storage Management EAch job is given REGION but space is left in it to be used GETMAIN obtain main storage dynamically supply length and get back address of area s FREEMAIN return storage to the system Synchronization needed for both TASKs and I 0 WAIT for an event to occur POST occurrence of event ECB s ENQ DEQ less important to user Timing TIMI STI E get time and date ER set timer get interrupt REAL or TASK IMER tests timer gets amount left hardware types of timers SPII STA Debug El El Error Handling get program interrupts get all ABEND s of any kind SNAP dump storage and registers DOCUMENT 1 Ll S 360 Assembler Language Documentation and Listing Techniques by John R Mashey and Andrea Rhodes Goals of Good Documentation 1 Aid in designing good programs 2 Aid in debugging programs 3 Make programs clear and understandable once written 4 Make structure of program well organized Good documentation is a great aid to producing clear well written and understandable programs and can save much programming and computing time Good documentation is especially necessary for programming projects requiring either a long period of time by one programmer any period of time by more than one programmer or modifications to any code by anyone other than the original auth
141. ADDRESS 0 gt direct addressing 1 gt indirect see below OP OPCODE number from 0 to 127 noting which operation to be done R REGISTER number from 0 15 specifying a register usually or sometimes a Branch Mask or other value x INDEX REGISTER number from 0 7 specifying an index register with 0 specifying NO indexing M MEMORY ADDRESS number from 0 128K specifying the actual location in memory of an operand CS102FP2 05 EFFECTIVE ADDRESS CALCULATION RX INSTRUCTIONS In general the SIGMA 4 5 calculates an EFFECTIVE ADDRESS EA using the I X and M fields in the procedure below then operates in some way on the word at the EA and the word in register R The EA is determined by the following steps First the M field is taken either as the address of a word in memory or the address of an address of a word in memory depending on the I bit as follows If I 0 then EAl M Direct addressing If I 1 then EAl low order 17 bits of word at location M This is called INDIRECT ADDRESSING Now EAl is either taken as the EA or is modified by adding to it the contents of an index register X as follows If X 0 EA EA1 no indexing If X 0 EA EA1 contents of register X NOTE EAl is a positive 17 bit number while value in register X may be either positive or negative Thus if M 100 X 1 and register 1 contains 10
142. AME INPUT EN PRIN SPACI U Q ti ti E HE KEYWORD PARAMETERS FOR DCB AR SORG IS DA SPECIFYING DIRECT ACCESS p ECFM IS FIXED EYLEN KEY LENGTH IS 4 BYTES KSIZE IS 40 BYTES ACRF WRITE BLOCKS ARE TO BE ADDED TO THE DATA SET PTCD E EXTENDED SEARCH R RELATIVE BLOCK ADDRESSING OEWADOH LIMCT NUMBER OF BLOCKS TO BE SERRCHED FOR EKY DDNAME STANDARD SYNAD ADDRESS OF ROUTINE FOR ERROR DURING WRITE DIRECT DCB DSORG DA RECFM F KEYLEN 4 BLKSIZE 40 MACRF WA OPTCD ER LIMCT 3 DDNAME DIRADD SYNAD CHECKER LTORG END DATA DIRADD DD DSNAME amp amp TEMP UNIT SYSDA DISP OLD PASS SPACE 44 21 DATA SYSUDUMP DD SYSOUT A DATA XSNAPOUT DD SYSOUT A DATA INPUT DD 1002SEDOND RECORD 1004FOURTH RECORD 1006SIXTH RECORD 1008EIGHT RECORD 1010TENTH RECORD 1012TWELTH RECORD 1014FOURTEENTH RECORD 1016SIXTEENTH RECORD 1018EIGHTEENTH RECORD 1020TWENTIETH RECORD EXEC ASGCG SOURCE SYSGO DD DISP OLD PASS SOURCE INPUT DD THIS CSECT UPDATES THE DATA SET OPEN THE DATA SET AND SET R
143. AR1 TEXT DESCRIBES TYPICAL DEVICES USED ON LARGE S 360 SYSTEM WITH DATA RATES CAPACITIES ETC gt INDEX102 TEXT INDEX TO MATERIALS FOR 1ST ASM COURSE gt INDEX411 TEXT INDEX TO MATERIALS FOR SYSTEMS COURSE gt LINKAGE TEXT TUTORIAL ON OS 360 LINKAGE CONVENTIONS gt LINKLOAD JCL 08 360 ILLUSTRATES USE OF LOAD MODULE MANAGEMENT COMMANDS LINK LOAD XCTL ETC gt OSHASP gt OVLY1 gt PTPCHMAC gt QSA gt RECURAS gt SPIESTAE gt TIM CJ gt WTOWTL AAAIND EX 03 TEXT EXPLAINS JOB SCHEDULING AND FUNCTIONING OF OS 360 WITH HASP JCL ILLUSTRATES USE OF LINK EDITOR OVERLAY FACILITIES SHOWING DIFFERENT TREE STRUCTURES JCL ILLUSTRATES USE OF UTILITY IEBPTPCH TO PRINT MACROS FROM LIBRARY JCL SHOWS USE OF 0S 360 QSAM MACROS JCL SHOWS USE OF GETMAIN FREEMAIN IN MAKING RECURSIVE ASSEMBLER PROGRAMS JCL SHOWS USE OF SPIE STAE MACROS JCL ILLUSTRATES USE OF TIME STIMER TTIMER MACROS FOR TIMING JCL ILLUSTRATES WTO WIL MACROS ASBROPS2 01 T ASSIST BASE Ti pu EGISTER ASSIGNMENT DUI CJ This assignment is essentiallv to write the base register handling routine for ASSI
144. ATA bytes contain such things as location counter value length attribute section identifier number and any other useful items like ENTRY EXTRN CSECT flags etc Given the setup described above note that an EMPTY LIST one containing NO NODES is just a HEADER CELL containing O The following example illustrates an alphabetical LIST Each NODE contains a fullword LINK and KEY plus 8 bytes of DATA resulting in a NODE length of 16 LOCATION LINK KEY DATA ITEMS HEX HEX CHAR CHAR 001000 00004000 this is the header cell 002000 00000000 CCCC 3RD ITEM 004000 00006000 AAAA 1ST ITEM 006000 00002000 BBBB 2ND ITEM The macros are to be written he exact functions a creates a LIST of empty NODES obtains the address described ODEFL HEAD ER QN EX a H a LIST i remove it form the list to the begin performs a LIST SI for the correct position for a NOD in that position QSNAP is a debugging macro which D E NOD the NODE The following sections give example prototype statements VALUE Og of the macros K A T RE NOT HINK TH E EYWORD OP R E Y ERANDS FADER or address of a current obtains the address of QPUSH is the opposite modifying the H searc according to KEY value NODE ning of a LIST EARCH AND INS ERT to NODE it Lur E NO Be ry IN SO CAS
145. B2 SUB2 1 MSUB1 0 SUB2A SUB2B SUB2C jhe fp SI MSUB1 0 MSUB3 MSUB4 MSUB2 Z 0 MSUB3 MSUB4 jie 0 MSUB3 MSUB4 SUB1A SUBID SUB2A SUB2B SUB2C J X Xx k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k LER STEP ASMLINK SOURCE OBJECT ASSEMBLE THE PROGRAM THEN USE LINKEDITOR TO PRODUCE 4 LOAD MODULES WITH THE MODULES LINKED IN INCREASING ORDER OF OVERLAY COMPLEXITY LIE ASMLINK EXEC ASGCL PARM SOURCE NOXREF PARM OBJECT MAP LIST OVLY SOURCE SYSGO DD DSN amp amp STUPID SOURCE INPUT DD OVLY TITLE TEST PROGRAM FOR OVERLAY OPTIONS MACRO amp LABEL OCALL amp ENTRY gt MACRO OCALL SPECIAL VERSION OF CALL TO SHOW V A ADCONS amp LABEL L 15 A amp ENTRY A TYPE ADCONS HAN MAIN MAIN1 MAIN2 SUB1 SUB2 SUBIA SUB1B BALR U Hi gt z au ti g O x uns n un rg Ei ti ta QH RINT x a D lt ti XSNAP SPACI OCALL OCALL OCALL OCALL XSNAP SPACE XRETU DS LTORG EQU ORG TITLE CJ 15 V amp ENTRY 14 15 4 XSAVE OFF XRETURN OFF NOGEN 15 V SUBI T NO LABEL THIS IS 2 SUB1 SUB2 SUB1CI SUB2C T NO 2 RN SA OD CJ MAIN X 1000 CSECTS SUBI
146. BLER COURSE 1ST ASSIGNMENT LINKAGE BETWEEN FORTRAN ASSEMBLER PARM FIELD ACCESS 08 360 gt CS411GI1 TEXT GENERAL INFORMATION COURSE OUTLINE INDEX ETC FOR 2ND ASSEMBLER SYSTEMS COURSE 1 OF 2 gt CS411GI2 TEXT GENERAL INFORMATION ETC PART 2 OF 2 gt CS411FP1 TEXT A FINAL PROJECT ASSIGNMENT WRITEUP FOR A SIMULATOR FOR MULTIPROGRAMMING OPERATING SYSTEMS FOR USE IN 2ND ASSEMBLER SYSTEMS COURSE PART 1 OF 4 gt CS411FP2 TEXT FINAL PROJECT PART 2 OF 4 gt CS411FP3 TEXT FINAL PROJECT PART 3 OF 4 gt CS411FP4 TEXT FINAL PROJECT PART 4 OF 4 gt CS411MC1 TEXT MACRO INSTRUCTION ASSIGNMENTS WRITE OWN VERSIONS OF CALL SAVE RETURN 2ND WRITE MACRO MODULE COMBINATIONS FOR HEXADECIMAL CONVERSIONS gt CS411MC2 TEXT MACRO INSTRUCTION ASSIGNMENT WRITE MACRO PACKAGE FOR MANIPULATION OF LINKED LISTS ALSO TO BE USED IN CS411FP1 4 gt CS411TPA TEXT COURSE OUTLINE AND DAY BY DAY NOTES FOR 2ND COURSE IN ASSEMBLER SYSTEMS gt DOCUMENT TEXT HINTS AND GOOD PRACTICES ON DOCUMENTATION OF ASSEMBLER PROGRAMS gt DSECT TEXT SAMPLE USE OF DSECTS AND EXPLANATIONS gt DUMPSJCL TEXT OS 360 BRIEF NOTES ON JCL TO BE USED FOR ASSEMBLER RUNS SAMPL DUMPS FOR VARIOUS ERRORS gt EXCP JCL SAMPLE RUN SHOWING EXCP COMMANDS gt FLOTLINK JCL ILLUSTRATES LINKING FORTRAN amp ASSEMBLER FLOATING POINT OPERATIONS IN ASSEMBLER gt GETMAIN JCL ILLUSTRATES GETMAIN FREEMAIN MACROS gt HARDW
147. CAL COPY OF C MVC I 0 R5 GET LOCAL COPY OF I AT THIS POINT LOCAL A B C I HAVE SAME VALUES AS THOSI E SUPPLIED IN ORIGINAL PROGRAM NOW MAKE CALLS TO ROUTINES AND SEE HOW THEY RETURN VALS LA RI ADCONS GET ADDRESS OF ADCON LIST L R15 V F2A ADDRE E SS OF FORTRAN ROUTIN BALR R14 R15 CALL THE ROUTINE XSNAP T FL LABEL FO CONTAINS RESULT OF FUNCTION F2A NOW CALL EQUIVALENT ASM ROUTINE USING CALL MACRO PRINT GEN SPACE 2 CALL A2 A B C I VL SPACE 2 PRINT NOGEN XSNAP T FL LABEL FO CONTAINS RESULT OF FUNCTION A2 XRETURN SA DS DS DS DS DC LTORG END D j m non A B C X 80 AL3 I ADCON LIST DATA XSNAPOUT DD UNIT AFF FT06F001 XSNAPS ON FT06F001 DATA SYSUDUMP DD SYSOUT A DATA FTOSFOO1 DD 10 Dis 1 07 2 1 10 xi 4
148. CLEUS to be ready to execute A IPL Initial Program Loader The process of getting an 0S 360 system running is called IPLing and includes the following main steps 1 The operator makes sure the disk pack called SYSRES SYStems RESidence is mounted on a disk drive The LOAD UNIT switches are set to show the device address of the SYSRES disk pack and the LOAD button pressed This causes the CONTROL RECORD to be read from the first record on the disk pack consisting of a PSW and two CCW s placed at location 0 in memory Execution of this record causes the IPL BOOTSTRAP record to be read into memory The BOOTSTRAP record consists of a set of CCW s which are used to read the IPL program into memory beginning at location 0 It ends with a LPSW to give control to the IPL program 2 IPL selects which NUCLEUS will be loaded there may be a choice which can be given by switches on the operator console 3 IPL clears all memory above itself to zeroes also obtaining the size of memory i e it stores until addressing interrupt occurs 4 IPL clears the floating point registers thus finding out if the floating point feature is installed 5 IPL brings the NUCLEUS into memory First it relocates the part of itself not yet executed into high memory near 252K so that the NUCLEUS can be placed beginning at 0 It then simulates Program Fetch loading the csects of the NUCLEUS l
149. CRO Writeups 0S 360 Linkage Conventions go over in detail CS411FP1 01 T COMPUTER SCIENCE 411 FINAL PROJECT DUE This writeup CS411FP 1 01 08 2 01 08 3 01 08 4 01 07 5 01 06 1 INTRODUCTION A PURPOSE OF ASSIGNMEN This assignment requires the student to design code debug and test a program which simulates the execution of a fairly general operating system OS 411 under a variety of circumstances It requires the student to become familiar with a number of different strategies for implementing operating system components and provides experience in working with programs of nontrivial size and complexity This project provides experience with the following operating system techngiues job scheduling storage allocation processor dispatching I O request handling and general job processing The following programming techniques can be included in this program linked list manipulation queueing methods and random number applications In particular the assignment generally requires the use of the S 360 Assembler Language items multiple CSECTS DSECTS heavy use of macro definitions and use of SET variables among macros and in open code B PROCEDURES FOR WRITING AND RUNNING THIS ASSIGNMENT 1 This writeup describes many alternate ways of performing the actions needed in an operating system Although stude
150. CSEC XSAV SPAC CALL CALL CALL CALL CALL CALL tJ FJ d Q D PACE ETU tinta A A ae U gt E E ti ACE ETU a LH H Q CG EJ Fj Hi EJ a FJ Q AAA EJ U D E QE E 1 MSUB1 MSUB2 MSUB3 MSUB4 SUBIA SUB1B SUBIC 1 RN SA SUBI X 2000 3 1 MSUB1 1 SUB2A SUB2B SUB2C 1 RN SA SUB2 X 3000 2 RN SA SUB1A X 4000 SUBIC SUB1D V TYPE ADCON CALL THE ZAP TH ROUTINI Ed WHERE ENTAB IS ENTRY POINT OF CS ABEL A V ADCONS FOR SAME E TONS OF MES SAGES NTAB STORAGE 0 15 40 15 ECT SUBIC STORAGE MAIN1 MAIN2 BEGINNING ADDRESS FOR IT ERAL DUMP ti NDING ADDRESS FOR LIT ERAL DUMP AKE SIZE NIZE SUB2 SUBIA SUBIB CALL LOW LEV 1 EL ROUTIN E ETURN SA SUB1B X 5000 CSECTS SUBIC D SUB2A B C 1 zi HD a ep ta Qe H SUB1C SUB1CI CJ A y Y Ki U D Q ti ti ALL SUB2 rJ rg FJ p Ha Gt RN SUB1CI Li ALL SUB2 U D Q ti ETURN SA G SUBIC X 1000 A 2 Q EJ SUB1D a AP lt ti TURN SA RG SUB1D X 1000 3 SUB2A B C SUBROUTINES CALLED BY SUB2 OR MAIN SUB2A CSECT XSAVE ETURN SA SUB2A X 3000
151. DDNAME IN UNDER THE DDNAME FOR JCL IN INECB DC F O BLOCK TO POST STATUS OF IO SPACE 3 INIOB DC B 01000010 X 000000 INFORMATION ON FORM OF IO ACTIVITY DC A INECB LOCATION TO POST STATUS DC 2F 0 MAGIC USED FOR STATUS DC A INCCW ADDR OF CHANNEL COMMANDS DC A INDCB PTR TO DCB ASSOCIATED WITH THIS IO TASK DC 2F 0 MAGIC USED FOR STATUS SPACE 3 COMMAND WORD FOR READ FOLLOWS INCCW DC OD O MUST BE DOUBLE WORD ALIGNED DC x 02 COMMAND READ DC AL3 INAREA WHERE TO PLACE DATA DC B 00100000 FLAGS SUPPRESS INCORRECT LEN DC x 00 UNUSED BUT MUST BE INCLUDED DC H 80 LENGTH OF DATA FIFLD TO READ EVERYTHING DOWN TO HERE HAS BEEN ONE COMMAND WORD FOR ONE IO ACTIVITY A READ SPACE 5 OUTDCB DCB MACRF E DDNAME OUT OUTECB DC F OQ EVENT CONTROL BLOCK FOR OUTPUT SPACE 3 COMMAND BLOCK FOR OUTPUT FOLLOWS OUTIOB DC X 42000000 INFORMATION FOR CHANNEL DC A OUTECB ADDR OF ECB FOR OUTPUT DC 2F 0 MAGIC STATUS DC A OUTCCW ADDR OF CHAN COMM WORD S DC A OUTDCB ADDR OF ASSOCIATED DCB DC 2F 0 MAGIC STATUS AGAIN SPACE 5 OUTPUT CHANNEL COMMAND WORD FOLLOWS DS OD OUTCCW DC X 8B SKIP TO NEW PAGE DC AL3 INAREA AREA FROM WHICH TO TRANS
152. E Gl O taf Ax x ADDRESS IN THE BLOCK ADDRESS OPERAND TYPE SEARC IS IN THE OPTCD PARAMETER IN DCB MACRO 3 DCB NAME HERE IT DIRECT 4 AREA ADDRESS ADDRESS OF MAIN CORE CONTAINING THE BLOCK TO BE WRITTEN 5 ENGTH NUMBER OF BYTE TO BE WRITTEN S INDI CATES THAT THIS IS OBTAINED FROM BLKSIZE IN DCB 6 EY ADDRESS IS ADDRESS OF MAIN CORE AREA CONTAIN ING THE KEY 7 BLOCK ADDRESS ADDRESS OF THR HAT CONTAIN THE RELATIVE BLOCK ADDRESS LT E BYTES IN AMIN COR CJ WAIT MACRO HALTS TASK TILL I O DONE E HEN RETURN FOR NEXT INPUT WRITE DECB DA DIRECT DATA S KEY REF 1 PRINT NOGEN PRINT GEN SPACE 5 WAIT ECB DECB PRINT NOGEN B NEXTREC RETURN FOR NEXT INPUT INPUT FILE CLOSE OUTPUT FILE AND RETURN TO SYSTEM Q n OSE INPT DIRECT RETURN SA RINT GEN PACE 5 tn tg x Q THIS IS THE FRROR REUTINE FOR DEBUGGING T ER SYNADAF ACSMETH BDAM A A A A A A xXx PRINT NOGEN XPRNT 0 1 PRINT GEN SYNADRLS BR 14 DS F DS CL40 DC CL100 DS F PRINT NOGEN DCB DSORG PS MACRF GM LRECL 80 RECFM F BLKSIZE 80 ROPT ACC EODAD EOJ DDN
153. E PROGRAM INTERRUPTION CONTROL AREA DISPLACEMENT F F CACA F CACA CACA F F F F F FF F FF ACA F F ACA F F F KF OF BYTES 0 ER 2 3 4 5 6 x x x x x x x x 0000 xPROGRAM x EXIT ROUTINE xINTERRUPTION x x x MASK x ADDRESS x TYPE x x x x x x x x EJECT x PROGRAM INTERUPTION ELEMENT x DISPLACEMENT E BYTES 0 1 2 3 0 xRESERVED x PICA ADDRESS x 4 xOLD PROGRAM xINTERUPTION CODES x xSTATUS WORD x x x 12 x REGISTER 14 x 16 x REGISTER 15 x 20 x REGISTER 0 x Es 24 x REGISTER 1 x x 28 x REGISTER 2 x MAIN A F AR F x A A A x FIX EJECT PRINT NOGEN EQUREGS CSEC XSAVE SA NO THE SPIE MACRO BELOW INDICATES THAT IF AN ABEND SOC1 THRU ABEND SOC5 OCCURS THAT FIX IS TO BE GIVEN CONTROL PRINT GEN SPIE FIX 1 5 PRINT NOGEN ST RI HO
154. ECOND IS RETURNED IN REG 1 ATTACH EP SECOND PARAM OTPT ONE VL 1 ECB ECB1 LPMOD 1 X DPMOD 10 ST 1 TCBADD SAVE TCB ADDRESS NEXT SNAP THI CB FOR MAIN AND SECOND USING EXTRACT MACRO SEVENI IS THE AREA FOR THE RESULT THE S INDICATES TCB FOR MAIN HE FIELDS SPECIFY THE FIELDS TO BE SNAPP HE ti ie EXTRACT SEVENI S FIELDS ALL TSO PSB COMM TJID PRINT NOGEN XSNAP T NOREGS STORAGE SEVEN1 SEVEN1 44 X LABEL TCB FOR MAIN WITH PRIORITY LOWERED PRINT GEN NEXT SNAP THE TCB FOR SECOND TCBADD IS A FULL WORD CONTAINING THE ADDRESS OF SECOND TCB EXTRACT SEVEN1 TCBADD FIELDS ALL TSO PSB COMM TJID PRINT NOGEN XSNAP T NOREGS STORAGE SEVEN1 SEVEN1 44 X LABEL THIS IS TCB FOR SECOND ON FIRST CALL PRINT GEN T ISSUE A WAIT MACRO FOR THE EVENT CONTROL BLOCK PASSED TO SECOND IN THE ATTACH MACRO CONTROL PROGRAM WILL POST ECB AR F F A A F HF x A FF ACA F F F A F AD xXx F xXx WAIT ECB ECB1 DETACHED ETACH TCBADD PRINT NOGEN U OTPT MHEADI CHAP 10 S EJECT A
155. EGISTER 11 TO 1001 WHICH IS USED TO COMPUTE THE RELATIVE BLOCK ADDRESS PRINT NOGEN EQUREGS THIRD CSECT XSAVE OPEN INPT INPUT DIRECT OUTPUT LA R11 1001 READ IN TEH INPUT AND CONVDR HE KEY TO A RELATIVE BLOCK ADDRESS THEN STORE RELATIVE BLOCK ADDRESS IN R READ THE RECORD WITH THIS RELATIVE ALOCK ADDRESS THE WRITE THE NEW INPUT IN ITS PLACE THEN RETURN FOR N INPUT LOOP GET INPT KEY XDECI R2 KEY SR R2 R11 ST R2 REF PRINT GEN SPACE 5 ACA F FF F F CACA F HF F FF CACA ACA F HF F FF F FF F F F F AR O F F x ti THE POSITIONAL PARAMETER FOR READ MACRO ARE 1 DECB NAME TO BE CREATED BV READ STATEMENT 2 YPE DI SEARCH FOR RECORD USES BLOCK IDENTIFICATION 3 DCB ADDRESS DIRECT 4 AREA ADDRESS S INDICATES THAT DYNAMIC BUFFERING IS TO BE USED 5 ENGHT S NUMBER OF BVTES COMES FROM DCB 6 KEY ADDRESS 0 INDICATES KEY NOT TO BE READ 7 BLOCK ADDRESS ADDRESS OF THREE BVTES CONTAINING RELATIVE BLOCK ADDRESS READ DECB DI DIRECT
156. EN SYS1 MACLIB AND CMACLIB DATA SETS PACE 2 PEN DCBADD INPUT BPAMDCB 48 X 10 DID DCB GET OPEN OK NO EXIT IF NOT OPEN QUIT PACE 5 ONSTRUCT BLDL LIST WITH THE BLDL MACRO INSTRUCTION PACE 2 BLDL DCBADD BLDLLIST SPACE 5 ALL INITIALIZATION AND PROGRAM SETUP IS DONE BEGIN BY PRINTING THE FIRST HEADING TO LABEL OUR OUTPUT SPACE 2 PRINT NOGEN CFIND XPRNT O HEADING 81 PRINT THE HEADING XPRNT THRESKP 3 SKIP SOME LINES SPACE 5 PRINT GEN DO FIND ON EACH LIST ENTRY TO RECOVER THE ACTUAL TRACK ADDRESS OF THE MEMBER INTO THE DCB SO THAT IT CAN BE ACCESSED BY OSAM GET PUT OR BSAM READ WRITE SPACE 2 FIND DCBADD FNDADD C SPACE 2 THE C PARAMETER INDICATES TO THE FIND MACRO THAT IT IS DEALING WITH THE LIST PRODUCT OF THE BLDL MACRO INSTRUCTION EXECUTION SPACE 5
157. ERALL INDEX gt ASBROE TEXT ASSIGNMENT USING ASSIST RI MONITOR TO REPLACE THE BASE ER PART OF ASSIST SEE ASREPLGD AND SASBROPS2 gt SBRTI TEST DATA FOR US H ASSIGNMENT ASBROPS2 gt ASPRGTC1 TEXT ASSEMBLER PROGRAMMING TI LINKAGE MACROS MODULAR PROG gt ATTACH JCL OS 360 SAMPLE PROGRAM ATTACH DETACH ETC gt BDA JCL OS 360 BDAM EXAMPL gt BDAM2 JCL OS 360 BDAM EXAMPL gt BPA JCL OS 360 BPAM EXAMPLE gt BSA JCL OS 360 BSAM EXAMPLE gt CS10 TEXT 1ST ASSEMBLER COURSE ASSIGNM gt CS10 TEXT FINAL PROJECT IN FIRST COURSI gt CS102FP2 TEXT 2ND PART OF FINAL PROJ ASSEMBLER INTERPRETER FOR SMALL gt CS102M1 TEXT MISC WRITEUPS FOR 1ST ASM COURSE gt CS102TPA TEXT DAY BY DAY OUTLINE OF 1S ER COURSE AAAINDEX 02 gt CS411AS1 TEXT 2ND ASSEM
158. ERM FLAG TO STOP SPACE 5 FIND DCBADD NAME D SPACE 2 NOTE HAT THE C PARAMETER IN THE FIND MACRO HAS CHANGED TO A D THIS INDICATES TO THE SYSTEM THAT THE EMBER NAME EXISTS ON A DOUBLE WORD BOUNDRY AND IS A DOUBLE WORD LEFT JUSTIFIED AND RIGH PADDED WITH BLANKS IF NECESSARY SPACE 2 PRINT NOGEN XPRNT DISRCH 81 PRINT NEW HEADING XPRN THRESKP 3 SKIP SOME LINES PRINT GEN B BREAD GO BACK TO PROCESS AS BEFORE SPACE 2 LTORG START LITERAL POOL HERE SPACE 2 DS OD GET DOUBLE WORD ALIGNMENT DC CL8 GET DEFINE NAME FOR FIND PACE 5 DCB DSORG PO INDICATE PARTITIONED DATA SET DDNAME MACLIB LOGICALLY CONNECT TO DD CARD EODAD AEODAD ON END OF DATA GO HERE ACRF R INDICATE MACRO TYPE ALL OTHER REQUIRED INFORMATION WILL BE TAKEN FROM EITHER THE DD CARD OR THE DATA SET LABEL SPACE 5 DC CL81 3 x x x xx CALL MACRO LISTING DC CL81 1 RETURN MACRO LISTING x DC CL81 1 SAVE MACRO LISTING SPACE 2 DISRCH DC CL81 1 GET MACRO LISTING USING DIRECT FIND PRNTAREA DC CLI PRINTING BUFFER DS 80C RECORD SPACE TERMFLAG DC x 00 INITIALIZE TERM FLAG TO OFF THRESKP DC CL33 y INE SKIPPER CNOP 0 8 BUFFER ALIGNMENT AND SPACE BUFFER DS 500D END BPAMIO DATA MACLIB DD DSN SYS1 MACLIB DISP SHR UNIT 2314 VOL SER PSUO1 DD DSN CMACLIB DISP SHR UNIT 2314 VOL SER PSU02 LOG DUMP DATA SYSUDUMP DD SYSOUT A
159. F x AR A xXx F S x X ER F F EXTR EJECT XC MECB 4 MECB CLEAN OUT MECB FOR WAIT AND POST ATTACH THIRD SNAP PRIORITY OF MAIN WAIT ON THIRD ST 1 TCBADD SAV ERE ECB IS PASSED AS PARM PMOD IS NEGATIVE INTEGER TACH EP THIRD PARAM OTPT MECB DPMOD 30 E TCB ADDRESS FOR DETACH XTRACT PRIORITY FOR MAIN AND SNAP IT PRINT NOGEN E WAIT FOR THIRD TO COMPLETE XTRACT ANSWER S FIE XSNAP T NOREGS STORAGE ANSWER ANSWER 4 LDS PRI LABEL PRIORITY SNAPPED FOR MAIN ON ATTACH TO THIRD PRINT GEN WAI ECB MECB DETACH THIRD DETACH TCBADD OSE OUTPUT FILE AND RETURN T RINT NOGEN OSE OTPT R ROP 12 J ETURN SA TR NO ECT P D EXIT ROUTINE FOR SECOND CALL TO SECOND T XSAVE TR NO SA NO RINT GEN ETACH SECOND IN EXIT ROUTINE WITH ADD OF TCB OF SECOND IN TCBADD ETACH TCBADD PRINT NOGEN XRETURN SA NO TR NO MECB DC F OQ ZERO DC F OQ ONE DC EST ECB1 DC F O TCBADD DC Erg ANSWER DC F O SEVEN DC TE O SEVENL DC TE O MHEAD2 DC CL132 OSECOND WILL ATTACHED AGAIN WITH POST ON COMPLE MHEAD DC CL132 OMAIN IS NOW EXECUTING NEXT OUTPUT SECOND MHEAD1 DC CL132 OMAIN EXECUTING W
160. FER DC B 00100000 STATUS FLAGS DC x 00 UNUSED DC H 80 LENGTH OF DATA TRANSFER SPACE 5 JCL USED EXEC ASGCG PARM DATA MAP SYSIN DD SOURCE DECK DATA OUT DD SYSOUT A FOR THE DATA SET OUT CREATED AT THE DATA LEVEL TO BE ROUTED TO A PRINTER DATA SYSUDUMP DD SYSOUT A JUST IN CASE DATA IN DD HE DATA SET IN CREATED AT THE DATA STEP WILL FOLLOW DIRECTLY DATA CARDS SPACE 5 END CHANPROG DATA OUT DD SYSOUT A DATA SYSUDUMP DD SYSOUT A DATA IN DD THIS IS DATA CARD 1 THIS IS THE SECOND DATA CARD THIS IS THE LAST DATA CARD AFTER IT THE DCB S WILL BE CLOSED LOG PROGRAM FLOTLINK THIS PROGRAM ILLUSTRATES THE FOLLOWING jix 1 JOB CONTROL LANGUAGE FOR LINKING FORTRAN ASSEMBLER fife 2 CALLING LINKAGE AMONG FORTRAN ASSEMBLER MODULES 3 FLOATING POINT INSTRUCTI
161. FTEN a new JOB arrives i e how often new JOB cards should be read IN WHAT ORDER the list of WAITING JOBS should be kept and WHAT STRATEGIES could be followed to decide which if any JOB should selected from this list and INITIATED loaded into memory and executed A JOB READING OPTIONS These options describe the intervals between successive arrivals of JOBS into the OS 411 system and are RDR1 A SSJOB card is read at an interval fixed at SYSGEN arriving every amp RDRINT milliseconds RDRR A SSJOB card is read at intervals which are random according to some probability distribution Among the possibilities RDRRU amp RDRINT is the mean of a UNIFORM distribution i e intervals vary between 0 and 2 amp RDRRINT RDRRE the intervals between successive JOBS are obtained from an EXPONENTIAL distribution with amp RDRRINT as a mean this is called a POISSON ARRIVAL PROCESS and is actually the most realistic of the arrivals given here B JOB ORDERING OPTIONS After the parameters of a JOB have been scanned it joins a list of JOBS which may of course be empty which are waiting for execution In general the position of a JOB on the list usually determines how soon it can b xecuted the first one on the list should be the next one executed if possible according to which set of scheduling rules is being used Combinations of the various options may be used with
162. GENERAL INFORMATION AND HARDWAR1 OSHASP ASPRGTC1 DOCUMENT LINKAGI DSECT CS102M1 CJ PROGRAMMING H EVICES PSU 360 67 OF OS 360 AND HASP PROG TECHNIQU ENTATION HINTS ENTIONS F IES AND NAMING SPECIFIC COURSE CS102TPA CS102HN CS102AS1 cs102FP1 2 J CS411TPA CS411HN CS411GI1 RTIAL OUTLIN OF INTRO ASSE UT LIST FOR INTRO ASS CJ EMBL R INTRO ASSEMBLI E OF SYSTI FOR ABOVE TION FOR DUMPSJCL DUMPTI CS411AS1 CS411MC1 CS411MC2 CS411FP1 2 3 SAMPLE PROGRAMS FLOTLINK EXCP BSAM BPAM QSAM OVLYI PTPCHMAC PRODUCTION SYSTEMS COURSI EST DECKS ABOV ECT AND TE S FOR SYSTEMS EMS COURSE FINA ECKS RTRAN ASSI RLAY STRUCTU i EXAMP LI S COURSI EMBLE EL PROGRAM CJ EXAMPLE HARDWAR1 01 FEB 1972 T PENN STATE UNIVERSITY COMPUTATION CENTER 360 67 CONFIGURATION this writeup pages 01 04 plus Diagram A separate INTRODUCTION This writeup briefly describes the devices included in the PSU 360 67 system and shows how they are connected together Each device is descri
163. HE PARM HAS 1 SUBTRACTED FROM IT AND IF IT IS 0 THEN WE OUTPUT THE SAVEAREAS IF NOT 0 THEN WE CALL RECRSIVE THE LAST WORD THAT IS OBTAINED HAS THE VALUE FOR THE NEXT CALL TO RECRSIVE RECRSIVE CSEC USING 15 LR R11 R15 MAKE A COPY OF ENTRY ADDRESS FIRST WE GET A COPY OF THE ADDRESS OF THE PARM LIST THEN OBATIN THE DESIRED STORAGE FROM THE OPERATING SYSTEM NEXT SET R2 FOR A DSECT THEN DO AN XSAVE WITH SA SAVEAREA NEXT TEST THE PARM TO SEE IF IT ISO LR R3 R1 MAKE A COPY OF PARM LIST ADDRESS PRINT GEN GETMAIN EU LV 80 A ADDRESS SP 1 PRINT NOGEN LR R15 R11 SET R15 FOR USING IN XSAVE E R2 ADDRESS SET R2 TO BEGINNING OF AREA OBTAINED XC 0 80 R2 0 R2 CLEAN OUT THE AREA THIS WILL BE USED WITH DSECT USING RCALL 2 STM R14 R12 12 R13 SAVE REGS OF CALLING PGM LA R5 SAVEAREA GET ADD OF CALLED PGM SAVEAREA ST R5 8 R13 SET LINK FOR LSA OF CALLING PGM ST R13 4 R5 SET HSA OF CALLED PGM LR R13 R5 SET R13 TO CALLED PGM SAVEAREA
164. HECK IF END OF CHAIN BZ DONE IF YES EXIT L R15 REGI5SAV GET P a O BEGIN OF CSECT CLC BRANCH X 47FO CHECK TO SEE IF VALID BRANCH BNE ERROR IF NOT ABORT IC R7 LENGTH PICK UP LENGTH OF NAME BCTR R7 RO SET UP FOR EXECUTE EX R7 MOVE OVE CHARS OF NAME TO OUTPUT XPRNT OUT 40 PRINT NAME OF ROUTINE VC OUT 1 39 OUT BLANK OUT OUTPUT AREA L R14 R11 REG14SAV RELOAD REGS FOR RETURN L R13 BACKLINK FOLLOW LAST LINK B LOOP DONE XPRNT CL25 OBACK TRACE COMPLETED 25 LA R13 TRACESA XRETURN SA TRACESA ERROR XPRNT CL25 0ERROR IN TRACE BACK 25 ABEND 999 ABORT OVE MVC OUT 1 NAMI INSTR FOR EXECUTE OUT DC CL40 LTORG DSECT 2 THE FOLLOWING DSECT FORMATS THE SAVE AREA x lt SAVEAREA DSECT UNUSED DS F BACKLINK DS F PTER TO HIGHER S A FORELINK DS F PTER TO LOWER S A REG14SAV DS F SAVE AREA FOR REG 14 REG15SAV DS F START OF S A FOR REG 15 12 THE FOLLOWING DSECT FORMATS THE BEGINNING OF A CSECT IF THE NAME CONVENTION IS FOLLOWED THE FIRST INSTR MUST BE A BR ON R15 AS A BASE REG FOLLOWED BY A LENGTH AND A NAME NAMECONV DSECT BRANCH DS XL2 XL2 SPACE FOR BSC INSTR 4 bytes LENGTH DS C NAME DS C SPACE FOR NAME MARK BEGI
165. HEN 1000 BCT ON R10 TO ARE DONE LA RIO 1000 STIMER TASK TUINTVL TUNUM BCT R10 ET R10 TO 1000 FOR BCT a D H H Q x lt kd WHEN A TTIMER MACRO IS ISSUED IT RETURNS THE TIMER RE IN THE TIMER INTERVAL ISSUED BY THE STIMER MACRO THEN COMPUTE THE TIME FOR 1000 BCT INSTRUCTIONS AND OUTPUT THE RESULTS NEXT CANCEL THE TIMER INTERVAL USSING THE TTIMER WITH CANCEL SPECIFIED x lt kud TTIMER L R9 TUNUM GET NUMBER OF TIMER UNITS AT STAR SR R9 RO GET NUMBER OF REMAINING TIMER UNITS SR R8 R8 ZERO REG 8 LA R6 26 SET R6 TO NUMBER OF MICRO SECONDS PER TIMER UNIT MR R8 R6 GET THE NUMBER OF MICRO SECONDS SR R8 R8 ZERO R8 FOR DIVIDE LA R6 1000 SET R6 TO 1000 FOR DIVIDE DR R8 R6 GET AVERAGE TIME PER CALL IN
166. I O to disk drum tape terminals etc HASP does not interfere but passes these on to the real I O Supervisor 3 I O to real unit record devices these have probably been issued by HASP in the first place so it passes control to the real I O routines to let them perform the I O 4 1 0 to a pseudo devic these must be caused by user program For input HASP fetches the cardimages from disk into memory 1f they are not already present and feeds requested cardimage s to the user program by MVCing them there using user protect key for safety For output it blocks up output and eventually writes it to disk In all cases HASP simulates the effect of having real card readers printers punches which are odd only in possessing great speed i e th ffect on OS 360 is of having issued an I O request and having had it complete immediately During execution HASP can also provide extra services such as monitoring time used output records etc It also reorders priorities of executing user tasks so that I O bound jobs have higher priorities than do CPU bound ones This action which is unknown to OS 360 helps minimize time spent waiting 3 Output Stages Print and Punch after a job has been executed it enters the Print queue is printed enters the Punch queue and has punched output if any actually punched This activity occurs without the knowledge of 0S 360 which believes the job disappeared whenever it finished e
167. IMER TTIMER test run to show use of WTO WTL macros for communicating with computer operator CS411GI1 10 V ASSIGNMENTS DUE DATES The foll of each DUI Li 12 15 18 27 n eac lowing lists the assignments given approximate due date in terms of day number within the 30 class days per term plus comments o li h NAME DESCRIPTION COMMENTS DUMPS run dump programs as described by DUMPSJCL bring to class familiarizes students with dump reading LINK assignment CS411AS1 gives students practice with nkage including FORTRAN aseembler linkage dumps gets them programming again quickly SAVE RETURN CS411MC1 pages 01 03 write own versions of extended SAVE RETURN CALL macros starts students on macros HEX CS411MC1 pages 04 08 write hexadecimal conversion macros amp modules like XHEXI XHEXO covers macro module linkage TR TRT PACK UNPK instrs BASEREG ASSIST base register replacement ASBROPS2 SB RTEST ASREPLGD covers base registers well in short program helps with understanding of DSECTS LINKED LIST CS411MC2 linked list macro package more practice on macros needed for final project FINAL PROJECT CS411FP1 2 3 4 5 CS411FPJ CS411FPK operating system simulation CS411GI2 01 VI COURSE OUTLINE READINGS
168. IN and CPU BOUND JOBS IORL small relative to IOIN JO6I I O BOUND JOBS earlier small values of IOIN IORL JO6C CPU BOUND JOBS earlier small values of IORL IOIN STATIC ORDERING COMPUTATION When the JOB enters the system a single number is calculated from some subset of the JOB S parameters thus weighting the various factors according to whatever the designer of OS 411 desires Note that all of the JO options above are really special cases of this The JOB is then ordered according to this number which is thus a generalized priority For example ORDER PRIO T 256 SP 4 IOIN IORL CAT favors high priority jobs shorter jobs smaller jobs 1 0 BOUND jobs and especially jobs in lower numbered categories assuming jobs are ordered from low ORDER to high ORDER DYNAMIC ORDER CALCULATION This case includes all JO options as special cases and allows the ordering of the JOBS to be varied dynamically according to any system conditions For example it might be basically a priority system but check the list of JOBS at intervals and occasionally raise the priority of a JOB if it has been waiting a long time It could also take into account the other jobs in a system such as trying to give equal service to each category of jobs or deciding to select an I O BOUND JOB if CPU BOUND ones are in memory or vice versa CS411FP2 01 Cx JOB SELECTION RANGE AND METHODS T
169. INATION TH IS RECOVERED USING THE D TYPE FIND MACRO 1 PROGRAM BUILDS THE BLDL LIST FOR CALL R E AND DOES THE BLDL ON THAT LIST A LOOP F H T Ho DCB MACRO NSTRUCTION IS ENTERED THAT DOES A FIND C TYPE ON T IST ENTRY AN INNER LOOP STARTS UP THAT R LOCK OF DATA AND DOES DE BLOCKING AND PRINTS THE TA ON E O F THE LOOP FOOT AEODAD IS ACTIVATED ICH TESTS TO SEE IF THE BLDL LIST HAS BEEN PROCESSED YES THE PROGRAM DOES THE D TYPE FIND FOR THE H B MACRO LISTING AND READS DE BLOCKS AND PRINTS NTIL IT IS FINISHED E PROGRAM THEN TERMINATES PROGRAM BY RICHARD FORD JULY 1972 EJECT ACRO LISTHD amp NAME amp NUM amp LNGTH GBLA GLEN SPACE 2 KO OK x x x x KIKKI THIS MACRO EXPANSION WILL CREATE THE LIST DISCRIPTOR FIELD AS REQUIRED BY THE BLDL MACRO INSTRUCTION KK x x k k x x x x x KKK x x x x KKK x x k
170. INTS N EV ER CALL of the routi multiple entry ENTRY1 ETURN XRI needed by more ASPRGTC1 03 et up as follows YN often CSECT routines at the points not etc ETURN etc code than one entry point the entry point routines multipl EACH OTHER In essence all level in cal m my F F ONE SAVE ONLY nes represented by the various entry points are at the ling structure of an entire program AREA IS ACTUALLY NEED the end can refer to it for the later ones constants mu the SA operand is CSECT never cal ah of the LAST section of code note that if placed o st be used instead ED same Since the routines inside l each other the so t to access it using With n the coded only o LAS E ST BE CARE MU dE AKEN WITH AD DR ESSIBILIT can of course addr ss the data er must th he programm be very careful with BASE REGISTERS USED E ro user can cod hat all of n the first XSAVE XRETURN ys areas at the e any th the save area at the previous sections it would be difficult instruction address this means that a LA XRETURN All of the nd of the CS internal ASSEMBLE IN code sections ECT However subroutines he ERNAL SUBROUTINES rites becaus UST HA
171. IPLYING TO GET C ABS I SPACE LTR R6 R6 WAS I POSITIVE BP A2NOMUL YES SO CAN SKIP F2 IS OK LD FA D 1 GET VALUE OF 1 FOR DIVIDE DER F4 F2 DIVIDE TO GET 1 0 C ABS I ER F2 F4 GET CORRECT VALVE OF C I DE F2 E 10 EZ SO O ES 1 7 104 AER FO F2 FO A 2 B C 1 10 SE FO E 2 FO A 2 B C 1 10 2 RESULT HAS BEEN COMPUTED AT THIS POINT SAVE ACROSS CALL FLOAT REGS NOT PROTECTED AND PRINT STE FO A2RESULT SAVE THE RESULT PRINT GEN SPACE 2 CALL F3A AZRESULT A2MSG A2LEN SPACE 2 PRINT NOGEN E FO A2RESULT RELOAD RESULT XRETURN SA TR NO RETURN NO TRACE DS E DC C COMPUTED BY A2 AT EXIT DC A L A2MSG LTORG TITLE MODULE Al OF FLOTLINK CSECT THIS MODULE ILLUSTRATES THE CALLING SETUPS FOR CALLING BOTH FORTRAN AND ASSEMBLER ROUTINES IT CALLS A2 F2A SPACE ON ENTRY R1 CONTAINS THE ADDRESS OF AN ADDRESS LIST FOR ARGUMENTS A B C 1 AS FOLLOWS O R1 ADDRESS OF A gt VALUE OF A 4 R1 ADDRESS OF B gt VALUE OF B 8 R1 ADDRESS OF C gt VALUE OF C 12 R1 ADDRESS OF 1 gt VALUE OF I XSAVE DUMP THE ADDRESS LIST FOUND ON ENTRY XSNAP STORAGE 0 R1 16 R1 LABEL R1 gt 4 FULLWORDS OF ADDRESS LIST MOVE THE ARGUMENTS OVER HIS IS NOT REALLY NECESSARY IT IS JUST DONE TO SHOW ACCESSING OF ARGUEMNTS LM R2 R5 0 R1 LOAD PTRS TO VALUES OF A B C I MVC A 0 R2 GET LOCAL COPY OF A MVC B 0 R3 GET LOCAL COPY OF B HOW p ADCONS MVC C 0 R4 GET LO
172. ISTER TABLE LINKED LIST FORM BY ACTIVE REGISTERS A linked list can be kept of the active registers and their values and section ID s This can make for faster searches but can require more space and more complex code C REGISTER TABLE WITH LINK Li D LIST ORDERED BY SECTION ID s A set of linked lists can be kept with one for each of the section ID s currently active Each list links together th register or registers which are active base registers and are flagged with the given section ID This is potentially the fastest method but also requires the most complex programming D REGISTER TABLE WITH SEPARATE ID TABLE AND TRT USAGI E In this method a separate search is made of a 16 byte table which contains the active section ID s possibly using the TRT instruction then computing the register number from the position in the index table and going to antoher table to computer the value This method can be fast but may require more space unless 256 bytes of TRT table are available elsewhere POSSIBLE OPTIMIZATION GOALS SPEED optimize for the fastest program possible Note that this involves determining the relative frequency of USING DROP and base displacement computations which can differ depending on the type of programmer producing the test program i e xperienced programmers usually have many more USING
173. ITH DISPATCH PRIORITY HALVED OTPT DCB DSORG PS MACRF PM LRECL 132 BLKSIZE 132 RECFM FA DDNAME FTO6F001 EROPT ACC END OBJECT SYSLMOD DD DSNAME amp amp LOADMOD MAIN DISP OLD PASS DATA STEPLIB DD DSNAME 66LOADMOD DISP OLD PASS DATA SYSUDUMP DD SYSOUT A DATA XSNAPOUT DD UNIT AFF FTO6FOOL1 Lt THIS JOB WILL RUN WITH TIME 170 SECONDS ENE RECORDS 1500 EXEC ASGCG SOURCE INPUT DD LOG EH PURPOSE OF THIS JOB IS TO DEMONSTRATE BASIC DIRECT ACCE ETHOD BDAM E BASIC FLOW OF THIS JOB IS AS FOLLOWS 1 JOB STEP 1 CREATES THE DIRECT ACCESS DATA SET 2 JOB STEP 2 ADDS RECORD TO THE DATA SET 3 JOB STEP 3 UPDATES THE DATA SET 4 JOB STEP 4 PRINTS OUT THE DATA SET IN LOGICAL ORDE EJECT IS JOB STEP CREATES THE DIRECT ACCESS DATA SET E INPUT IS CARDS WITH A KEY OF 4 DIGITS BETWEEN 1001 AND 1020 HE KEY IS CONVERTED TO A BINARY NUMBER WHICH IS THE BLOCK NUMBER FOR THE INPUT RECORD FOR EACH INPUT E NEXT 40 CHARACTERS ON THE CARDS ARE WRITTEN TO DISK R EACH KEY THAT IS NOT PRESENT A DUMMY RECORD IS WRITTI DISK THUS THERE ARE 20 BLOCKS OF DATA ON THE DISK SPACE 5 PRINT NOGEN EQUREGS MAIN CSECT XSAVE RINT GEN OPEN THE INPUT DATA SET AND THE OUTPUT DATA SET INITIALIZE REGISTER 8 9 AND 7 REGISTE
174. ITIONAL OPEN CLOSE SET SPACE 5 OPENDCBS OPEN CARDCB INPUT PRNTDCB OUTPUT DISKDCB OUTIN SPACE 5 EST THE OPEN IF OPENS DID NOT GO TERMINATE WITH ABEND GIVING USER 500 COMPLETION CODE SPACE 5 T CARDCB 48 X 10 DID THE CARD READER OPEN GO NO ABTERM ABNORMALLY TERMINATE T PRNTDCB 48 X 10 DID PRINTER OPEN GO NO ABTERM ABNORMALLY TERMINATE T DISKDCB 48 X 10 DID DISK OPEN GO BNO ABTERM ABNORMALLY TERMINATE SPACE 5 READ A CARD ON THE READER SPACE 5 LA 5 0 SET UP CARD COUNTER READ READ CARDECB SF CARDCB BLOCK READ A CARD IMAGE INTO BLOC SPACE 5 ISSUE CHECK TO TEST COMPLETION OF 1 0 OPERATION SPACE 5 CHECK CARDECB CHECK FOR I O COMPLETION SPACE 5 LA 5 1 5 COUNT CARDS COMMING IN DO A WRITE TO PRINTER FOR ECHO PRINT SPACE 5 WRITE DISKDECB SF DISKDCB BLOCK WRITE TO DISK FROM BLOCK SPACE 5 ISSUE CHECK AS ABOVE SPACE 5 CHECK DISKDECB IS I O OPERATOON COMPLETE SPACE 5 SAVE CARD IMAGE ON DISK SPACE 5 WRITE PRNTDECB SF PRNTDCB BLOCK 1 WRITE TO THE PRINTER SPACE 5 CHECK AS BEFORE SPACE 5
175. LCS currently unused except for systems programs JOBLIB DD UNIT SYSDA DSN amp amp LOADMOD SPACE CYL 5 1 3 DISP PASS KK k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k Z OVERLAY TEST PROGRA fife THIS PROGRAM ILLUSTRATES THE USE OF THE OVLY OPTION FOR THE LINKEDITOR IT FIRST PRODUCES AN OBJECT MODULE OF A NUMBER OF CSECTS THEN USES THE LINKE EDITOR TO PU THEM TOGETHER IN VARIOUS WAYS USING Pik ASMLINK ASSEMBLE THEN LINK MODULES 4 WAYS PLACING THEM IN amp amp LOADMOD AS MODULE ODO MOD3 fi THE REMAINING STEPS EXECUTE THE MODULES MODO MOD3 JAK STEPO NO OVERLAY WHATSOEVER EJE STEP1 SIMPLE OVERLAY WITH THREE SEGMENTS STEP2 COMPLEX 1 REGION OVERLAY WITH 10 SEGMENTS STEP3 THREE REGION OVERLAY MOVES SUBROUTINES OUT OF THE ROOT SEGMENT CALLING HIERARCHY CHART FOR THIS PROGRAM LEVEL MODULE CALLS ROUTINES AT LEVEL SHOWN CSECTS ONLY 5 MAIN 4 SUB1 LAKE 2 SUBIC SUB2 4 SUB1 3 SUB1B 2 SUBIC 1 MSUB1 MSUB2 LJ 0 MSUB3 MSUB4 SUBIA Pap ALE 3 SUB1B APR 2 SUB1C JAK 0 SUB1D EJE jj 2 SUB1C 1 MSU
176. LD SAVE ADDRESS OF PREVIOUS PICA DC F 0 XPRNT MHEAD 80 L R5 HOLD THE SPIE MECRO IS THE EXECUTE FORM IT CANCELS THE PREVIOUS SPIE MACRO PRINT GEN SPIE MF E 5 HE STAE MACRO BELOW INDICATES THA THE FIXSTAE ROUTINE IS O BE GIVEN CONTROL ON AN ABEND AND TO ALLOW THE 1 0 TO CONTINUE EVEN THOUGH THE ABEND HAS CCCURED ALSO CREATE THIS STEA AREA NOT OVERLAY ANY PREVIOUS STAE FIXSTAE CT PURGE NONE LA R3 HALFWORD GET ADDRESS OFHALFWORD L R2 0 R3 IMPROPER ALIGNEMNT CAUSE ABEND PRINT NOGEN XPRNT HEAD1 80 PRINT GEN STAF 0 PRINT NOGEN XRETURN SA NO DROP 12 USING 15 ST RO R15 SAVE L R2 8 R1 GET IL CC AND NEXT INSTRUCTION ADDRESS LA R11 255 SET R11 TO HEX 000000FF LR R10 R11 PUT HEX FF IN R10 SLL R11 8 MOVE OVER 1 BYTE OR R10 R11 MAKE R10 HEX 0000FFFF SLL R11 8 MOVE OBER 1 MORE BYTE OR R10 R11 MAKE R10 HEX OOFFFFFF NR R10 R2 GET INSTRUCTION ADDRESS SR R9 R9 ZERO R9 FOR IC IC R9 0 RIO GET OPCODE FOR NEXT INSTRUCTION LA R8 64 SET TO HEX 40 CR R9 R8 CHECK FOR RR INSTRUCTION B RR GO TO SET NEW PSW LA R8 192 SET R8 TO CHECK
177. LE Chapter 5 LINKAGE HANDOUT 01 26 savearea linkage ans ome review Describe 1 8 fullword save area go through the standard code used at beginning and end of a routine calling methods Do not work on argument passing just normal code isc instructions IC STC start on Shifts Various review for problems Note general usage of registers get students into good habits READINGS STRUBLE Chapter 11 01 28 logical algebraic arithmetic shifts 20 minute quiz on previous instructions differences between condition code setting aroverflow in algebraic arithmetic and logical arithmetic examples shift instructions and how they are used READINGS STRUBLE Chapter 11 begin on chapter 10 11 12 1 3 14 15 16 CS102TPA 04 01 31 bit manipulation and uses review on branching bit manipulation instructions NR XR OR N X O NI XI OI NC XC OC plus TM what they do and how to use them EQU trick for SI instructions and how to use it review prototypes on loop control advantages disadvantages READINGS STRUBLE Chapter 10 first 3 sections 02 02 assembler housekeeping misc areas go over all of DC DS operand formats in detail showing what can exist as duplication factor type length constant including multiple operands and constants expressions as duplication factors and length modifiers also cover TITLE EJECT SPACE
178. MICRO S SECONDS XDECO R9 0UZZ PUT R9IN OUTPUT XPRNT OUZ 84 TTIMER CANCEL x NEXT ISSUE AN STIMER WITH AN EXIT ROUTINE WHEN AN EXIT ROUTINE IS GIVEN AT THE END OF THE INTERNAL THE ROUTINE IS GIVEN CONTROL THEN SET CHECK TO 0 TO STOP THE LOOP HE TASK INDICATES THE TIME TO BE DECREMENTED ONLY WHEN THE TASK IS ACTIVE AGAIN THE INTERVALIS SPECIFIED IN TIMER UNITS IT IS 1 04 SECONDS APPROXIMATELY AFTER THE LOOP IS STOPED THEN CANCEL THE TIME INTERVAL WITH A TTIMER CANCEL x lt LA R9 0 SET R9 TO 0 FOR COMPARE LA R10 0 SET R10 TO 0 BCTR R10 0 SUBTRACT ONE FROM RO STIMER TASK LPINTVLP TUINTVL LPINTVL LP R9 CHECK CHECK TO SEE IF INTERVAL IS OVER BE DONE IF TIME INTERVAL OVER GO TO DONE BCT R10 LP RETURN DO LP OVER MDONE TTIMER CANCEL XRETURN SA DROP 12 x ik THIS IS THE STIMER EXIT ROUTINE IT ZEROS CHECK AND PRINTS A MESSAGE AND RETURNS CONTROL TO A PROGRAM THAT THEN RETURNS CONTROL TO THE PLACE WHERE IT LEFT OFF WHEN THE INTERVAL EXPIRED NOTE THE USE OF SAVE AREAS LPINTVLP XSAVE SA NO H PINTVL1 XRETURN SA NO LPINTVLM DC CL80 OLP TAKES LONGER THAN 1 04 SECONDS SO STOP LOOP LPINTVL DC F 4000
179. N CORE FOR UNPK UNPK OUT 6 WORD 4 CHANGE TO OUTPUT FORM MVC OUTPUT2 2 OUT PLACE HR IN OUTPUT MVC OUTPUT3 2 OUT 2 PLACE MINUTES IN OUTPUT MVC OUTPUT4 2 OUT 4 PLACE SEC IN OUTPUT XPRNT OUTPUTS 34 LA R11 4095 SET R10 TO 4095 LA R11 4095 R11 SET R11 TO 8190 LA R11 1810 R11 SET R11 TO 10 000 LA R9 DONE HE TIME MACRO IN THE RANDOM NUMBER GENERATOR RETURNS THE IME IN REG 0 IN BINARY FORM THEN IT IS STORED IN SAVEODD O BE USED FOR THE BASIS FOR RANDOM NUMBER GENERATOR THEN OUTPUT THE RESULTS OF 10 000 THROWS OF A TEN SIDED DIE CALL IAND TEN MONE LR R10 RO SET R10 TO RO BCTR R10 RO SUBTRACT 1 FROM R10 SLL R10 2 MULTIPLY R10 BY FOUR AR R10 R9 GET ADD OF NUMBER TO INCREASE L R8 0 R10 GE AST VALUE LA R8 1 R8 INCREMENT COUNT BY ONE ST R8 0 R10 PLACE IN COUNTER BCT R11 LOOP RETURN FOR NEXT CALL XPRNT OUTT XPRNT OUTTT LA R10 10 SET R10 TO NUMBR OF SIDES ON DIE LA R11 OUTTTT 1 L R8 0 R9 GET NUMBER OF TIMES A NUMBER OCCURED XDECO R8 0 R11 PLACE VALUE OF COUNTER IN OUTPUT LA R9 4 R9 GET ADDRESS OF NEXT COUNTER LA R11 12 R11 INCREASE OUTPUT POINTER BCT R10 LOOP1 RETURN FOR NEXT COUNTER XPRNT OUTTTT WHEN AN STIMER MACRO IS ISSUED WITH TASK IT ONLY DECREMENTS HE TIME INTERVAL WHEN THE TASK IS ACTIVE THE TUINTVL IS A FULLWORD ON A FULLWORD BOUNDRY THAT GIVES THE TIME IN IMER UNITS ONE TIMER UNIT APPROXIMATELY 26 MICRO SECONDS HIS STIMER SETS THE INTERVAL TO APPROXIMATELY 5 SECONDS
180. NE ERROR 4370 F004 00004 238 IC R7 LENGTH 0670 239 BCTR R7 0 4470 C170 004D2 240 EX R7 MOVE 241 XPRNT OUT 40 D226 C177 C176 004D9 004D8 251 VC OUT 1 39 OUT 98EB DOOC 0000C 252 L R14 R11 REG14SAV 58D0 D004 00004 253 L R13 BACKLINK 47F0 C028 0038A 254 B LOOP 255 DONE XPRNT CL20 TRACE COMPL 41D0 COF2 00454 265 LA R13 TRACESA 266 XRETURN SA TRACESA 285 ERROR XPRNT CL20 ERROR IN TR 295 ABEND 999 DUMP D200 C177 F005 004D9 00005 303 MOVE MVC OUT 1 NAME 4040404040404040 304 OUT DC CLag 305 LTORG 310 311 SAVEAREA DSECT 312 UNUSED DS F 313 BACKLINK DS E 314 FORELINK DS E 315 REG14SAV DS F 316 REGISSAV DS F 317 318 NAMECONV DSECT 47F0 F000 00000 319 BRANCH B 0 15 320 LENGTH DS C 321 NAME DS 322 323 END MAIN 000000 The first assignment ASS IGNM not to be turned in is e DUMPSJCL 01 ENT A DUMPS AND JCL INTRODUCTION ssentiallv to use the Job Control Language and deck setups to be used most often during term and also to become familiar with the completion dumps issued bv 0S 360 I JCL AND DECK SETUPS 0S 360 A LOADER CATALOGED PROCEDURE ASGCG The following is the RECOMMENDED PROCEDURE for any assembler run which cannot be run under ASSIST and which does not require some of the special facilities available using the LINK EDITOR The procedure ASGCG stands
181. NEXT CSECT XSAVE ESTAB H STANDARD LINKAGE CALL LAST CALL LOWEST ROUTINE XRETURN SA ESTAB H SAVE AREA LTORG LAST CSECT XSAVE ESTAB H STANDARD LINKAGE CA TRACE CALL TRACE RTN TO PRNT S A XRETURN SA GENERATE SAVE AREA LTORG THE ABOVE ROUTINES DO NOTHING BUT ESTABLISH LINKS TO TRACE THROUGH THE SAVE AREAS kud x ROUTINE TRACE PROVIDES A PRINTED TRACE OF THE NAMES OF THE CSECTS OF ACTIVE S A S IT USES DSECTS SAVEAREA AND NAMECONV TO FORMAT THE SAVEAREA AND FIRST FEW BYTES OF THE PROGRAM TRACE CSEC XSAVE SA TRACESA ESTABLISH LINKS USING SAVEAREA R13 USING NAMECONV R15 XPRNT CL25 OBACK TRACE OF SAVE AREAS 25 L R13 4 R13 CONNE O FIRST ACTIVE S A LOOP LTR R13 R13 CHECK IF END OF CHAIN BZ DONE IF YES EXIT L R15 REG15SAV GET P a O BEGIN OF CSECT CLC BRANCH X 47FO CHECK TO SEE IF VALID BRANCH BNE ERROR IF NOT ABORT IC R7 LENGTH PICK UP LENGTH OF NAME BCTR R7 RO SET UP FOR EXECUTE EX R7 MOVE OVE CHARS OF NAME TO OUTPUT XPRNT OUT 40 PRINT NAME OF ROUTINE VC OUT 1 39 OUT BLANK OUT OUTPUT AREA L R14 R11 REG14SAV RELOAD REGS FOR RETURN L R13 BACKLINK FOLLOW LAST LINK B LOOP DONE XPRNT CL25 OBACK TRACE COMPLETED 25 LA R13 TRACESA XRETURN SA TRACESA ERROR XPRNT CL25 0ERROR IN TRACE BACK 25 ABEND 999 ABORT OVE MVC OUT 1 NAME INSTR FOR EXECUTE OUT DC CL40 LTORG
182. NKAGE CONVENTIONS LA 3 CARD PTR TO CARD IMAGE READ IN XREAD CARD 80 READ DATA CARD XPRNT CARD 80 ECHO PRINT XDECI 4 0 3 CONVERT DECIMAL TO INTERNAL HEX XDECI 5 0 1 CONVERT NEXT ON CARD THE NEXT STMT PRINTS CONTENTS OF USERS REGISTERS NOTE REG 4 5 XDUMP B 4000 ABEND BRANCH OUT OF PROGRAM CARD DS 80C END DATA INPUT DD 100 1024 2 This next program is a batch run of 5 jobs each terminating abnormally The program is stored on RJE file Punch up the follwoing cards EXACTLY to run the program YOUR JOB CARD EXEC ASACG PARM BATCH SYSIN DD INCLUDE RABO1 BATCH 3 To merely get a listing of the prog in 2 use the following cards YOUR JOB CARD INCLUDE RABO1 PRINT INCLUDE RABO1 BATCH CS102M1 02 A GUIDE TO S 360 MNEMONIC OPERATION COD T E a l INTRODUCTION The beginning programmer facing the varietv of operations available on a modern large computer is often overwhelmed bv the large number of operations and complexitv thereof In some cases a few hints can be helpful in learning and remembering the names purposes and usage of the various operations In particular certain properties of S 360 menemonics can help the learner a great deal Some of them are A REGULAR SCHEME FOR NAMING OPCODES In general a fairly coherent and regular method has been used in naming operations In s
183. NNING ADDR ONLY END MAIN Following is the output from this example KKK AIN ENTERED KKK NEXT ENTERED KKK AST ENTERED KKK KKK TRACE ENTERED KKK BACK TRACE OF SAVE AREAS TRACE LAST NEXT MAIN IEWLCTRL BACK TRACE COMPLETED TRACE EXITED LAST EXITED KKK NEX EXITED KKK MAIN EXITED xxx Fol NAMECONV LOC 000300 000000 000000 000386 00038A 00038C 000390 000394 00039A 00039E 0003A2 0003A4 0003CE 0003D4 0003D8 0003DC 000406 0004D2 0004D8 000500 000000 000000 000004 000008 00000C 000010 000000 000000 000004 000005 DS dsects ECE 3 lowing is the actual assembler listing of the TRACE Notice those instructions which referenc Look at the object code and see what the bas a esects labels from the SAVEAREA and register and displacement by which they were assembled is OBJECT CODE ADDR1 ADDR2 STMT SOURCE STATEMENT 192 TRACE CSECT 193 XSAVE SA TRACESA 220 USING SAVEAREA R13 221 USING NAMECONV R15 222 XPRNT CL15 BACK TRACE 58DD 0004 00004 232 L R13 4 R13 12DD 233 LOOP LTR R13 R13 4780 COTE 003E0 234 BZ DONE 58FO D010 00010 235 L R15 REG15SAV D503 F000 C1C6 00000 00528 236 CLC BRANCH 2 X 47F0 4770 C13A 0049C 237 B
184. NOT text material i e sample programs test data for some projects etc NAME CARDS SOURC DESCRIPTION SBRTEST 178 ATTACH 222 BDAM1 BDAM2 2 BPAM 322 BSAM 157 CS411FPJ 74 CS411FPK 110 DUMPTEST 84 EXCP 102 FLOTLINK 152 GETMAIN 22 LINKLOAD 92722 BAT BAT BAT test data for the ASSIST base register module replacement assignment see assignment ASBROPS2 multi tasking example shows ATTACH CHAP DETACH etc run ot illustrate Basic Direct Access Method an entire run with JCL to illustrate use of BPAM macros for accessing pPartitioned Data Sets Reads selected macros from macro libraries and prints them Shows FIND BLDL READ CHECK etc entire run to illustrate BASM macros shows READ WRITE CHECK etc reads cards writes on disk reads from disk prints test deck of SSJOB cards for Final Project see assignment CS411FP1 2 3 4 5 same as CS411FPJ but with different SSJOB cards contains 4 runs set up to produce dumps for students to look at see DUMPSJCL writeup note that each run is preceded by BAT file comma card so that they may be extracted by listing with the PAGE option to find their starting sequenc s contains complete run set up to read cards and print them using EXCP macro and CCW strings contains a complete run set up to show various combinations of FORTRAN and assembler linkage
185. NPK WORK 9 FULL5S 5 convert TR WORK TAB3 make printable C ODUMPO PRINT LABELED REGISTER VALUE IN DECIMAL OR HEX REQUIRED FUNCTION OF MACRO label QDUMPO register message typ label is optional statement label register is name or number of register to be printed message is an OPTIONAL message to be used to identify output type Dor H D gt print in decimal H gt hexadecimal if omitted default to D If anything else used issue severity 4 warning message then default to D anyway Calling QDUMPO should request the following actions a If a GBLB set variable named amp QDUMPO currently has the value 1 NOTHING IS GENERATED EXCEPT POSSIBLY label DS OH the desired register to b If amp QDUMPO has the value 0 an which is always enclosed in quotes Bi identifying messag preceding c The identifying message noted i REGIST ER AT LABEL labe register d Neither registers nor condition macro expansion WHAT TO HAND IN decimal or hexadecimal and must be usable under any USING base from 2 CS411MC1 08 then code is generated to convert and print it out with it Add blank carriage control n b is either the message operand or the message 1 code may be changed by the to 13
186. NUMBER OF INPUT OUTPUT CHANNELS Performing input output for a user JOB requires the use of an IO CHANNEL The following are the possible options CHN1 the system is a small one with only 1 IO channel which all user jobs must use when they need IO CHNx the number of I O channels is fixed at the number x Channels are numbered from 1 to x amp NUMCHNS gives the number of channels with amp NUMCHNS varying from 1 to some given upper limit and OS 411 must be written t o handle any possible value Channels are numbered 1 to NUMCHNS in this case No system has more than 15 channels CS411FP1 05 ILE INPUT JOB STREAM OS 411 reads a deck of input cards which contain cards which describe JOBS and may contain other kinds of cards also The kinds of cards possible are as follows A JOB CARD DESCRIBES CHARACTERISTICS OF ONE JOB Each BATCH of JOBS consists of from 0 to some maximum number of JOBS This maximum number may be a constant or be given by amp MAXJOBS Each card is of the following format starting in column 1 SSJOB jobname T x SP x IOIN x IORL x PRIO x CHAN x CAT x jobname is a 1 8 character name separated from other items by or more blanks on each side It is a unique name which is used to identify the specific job in any messages The rest of the JOB card consists of the parameters shown WHICH MAY OCCUR IN ANY ORDER The following
187. OGEN SLL R1 8 EFT JUSTIFY TH ST R1 WORD PUT DAY AND YEA FOR UNPK UNPK OUT 5 WORD 3 GET OUTPUT VC PUT1 3 OUT 2 PUT YEAR IN OUTE VC PUT 2 OUT PUT JULIAN E WHEN TIME IS ISSUED WITH DEC IT RETURNS IN RO THE E IN RS Ge Me b sb e De de af E LOOP LOOP1 A A A A A A F F THE FOLLOWING FORM HHMMSSTH WHERE HH IS THE HOUR ON A 24 HOUR CLOCK WHERE MM IS THE MINUTES HHERE SS IS THE SECONDS WHERE THE T IS THE TENTHS OF SECONDS AND WHERE THE SINGLE IS THE HUNDREDS OF SECONDS DISCARD THE T AND SINGLE H FIELD AND THEN PLACE THE HH MM SS IN CORE AND UNPACK IT AND THEN OUTPUT THE TIME AND DATE LA R11 240 SET RIO TO JEX 000000F0 OR R11 RO PLACE F IN BITS 24 TO 27 R11 SRL R11 4 RIGHT JUSTIFY THE TIME ST R11 WORD PLACE TIME I
188. ON ECONDS TI THE CONT MAC IS DITIONS R EPLY Y FOLLOWING EGINNING OF PROGRAMM ci TT NDIX tr CJ PRINT NOGEN XRETURN SA EXB DC F O EPLYAD DC C END
189. ON WILL BEGIN AT ROUTINE NAMED MAIN STEP2 EXEC ASGCG PARM SOURCE NOXREF PARM DATA MAP EP MAIN SOURCE INPUT DD TITLE MODULE A2 OF FLOTLINK A2 CSECT WITH ARGUMENTS A B C I A2 COMPUTES SAME AS F2A F2A A 2 B C I 10 2 RESULT IS RETURNED IN FLOATING POINT REGISTER 0 PRINT NOGEN A2NOMUL A2RI A2MSG A2LEN Al AA F OR x E SULT EQUREGS EQUREGS L F DO 0 6 2 FLOATING POINT EQUATES SPACE XSAVE R1 CONTAINS ADDRESS OF ADDRESS LIST OF ARGUMENTS L R2 R5 0 R1 GET ADDRESSES OF A B C I E FO 0 R2 LOAD VALUE OF A HER FO FO A 2 USING HALVE SHORT INSTR E FO 0 R3 A 2 B MULTIPLY SHORT NOW COMPUTE C I 10 PART LD F2 D 1 INITIALIZE WILL MULITPLEY IN HERE E F4 0 R4 F4 C WILL SAVE IN REGISTER L R6 0 R5 R6 I FOR LOOP COUNTER XSNAP T FL LABEL REGISTERS IN A2 LPR RO R6 RO ABS I BZ A2NOMUL IF I 0 C I 1 0 ALWAYS MER F2 F4 F2 C LOOPING BCT RO 2 LOOP MULT
190. ONS FLOTLINK CONSISTS OF A FORTRAN MAIN PROGRAM A FORTRAN FUNCTION F2A AND A FORTRAN SUBROUTINE F3A PLUS SEVERAL ASSEMBLER LANGUAGE SUBROUTINES Al A2 THE EXEC FGC REQUESTS A FORTRAN G COMPILATION EXEC FGC SOURCE INPUT DD AIN PROGRAM READS VALUES CALLS Al WRITE 6 9000 1000 READ 5 9020 END 8000 A B C I WRITE 6 9040 A B C I c EXAMPLE CALL ASSEMBLER SUBROUTINE FROM FORTRAN CALL A1 A B C L GO TO 1000 8000 WRITE 6 9060 STOP 20 9000 FORMAT 1 FLOTLINK FORTRAN ASSEMBLER FLOATING PT LINKAGE 9020 FORMAT 3F10 O I10 9040 FORMAT 0 INPUT VALUES A B C I 3F20 10 110 9060 FORMAT 0 END EXECUTION OF FLOTLINK END FUNCTION F2A A B C L THIS FORTRAN FUNCTION COMPUTES THE FOLLOWING VALUES F2A A 2 B C 1 10 2 C IT GENERATES CODE RETURNING RESULT IN FLOATING PT REGISTER 0 E IF RESULT WERE FIXED PT IT WOULD BE IN GP REGISTER 0 WRITE 6 9000 F2A A 2 B C I 10 2 WRITE 6 9020 F2A RETURN 9000 FORMAT FUNCTION F2A ENTERED 9020 FORMAT RETURN FROM F2A RESULT E15 8 END SUBROUTINE F3A VALUE TITLE NTITLE C THIS SUBROUTINE IS USED TO PRINT THE SINGLE PRECISION FLOATING POINT VALUE FOLLOWED BY TITLE HAVING NTITLE CHARACTERS LOGICAL 1 TITLE NTITLE WRITE 6 9000 VALUE TITLE RETURN 9000 FORMAT E15 8 60A1 END fh THE EXEC ASGCG REQESUTS FIRST AN ASSEMBLY AND THEN EXECUTION EXECUTI
191. OT CS411FP3 01 08 3rd part of final project debugging machine level 1 0 overview of operating systm debug technique XSNAP IF option similair methods I O interrupts adn handling CHANNEL cicuitry shared or separate CCW s CAW CSW and how they interact Expand I O process note command and data chanining SCATTER READ GATHE WRITE use of protection keys like main memory OPERATING SYSTEM FINAL PROJECT overview RDR INT INIT DISPATCHER I OHADNLING half period quiz next time on hardware general concepts especially for final project NO XREAD and XPRNT will not be used on final project CS411TPA 05 17 02 10 simulation concepts and implementation continuous versus discrete simulations clocks event queues follow entire process of final project simulation outlining lists modified and programs whihc manage them with overall scheduling structure options to be done for this project 1024K bytes memory CPU 1 CHANv 3 15 JOB card PARMI simple one etc through all options desired HANDOUT CS411FP4 01 08 specific implementation on FP 18 02 15 module management final project discussion implementation of global table csect dsect source program gt translator gt object program object modules gt LOADER gt executable program object modules gt LINK EDITOR gt load module load module
192. PRINT GEN EJEC TCB DESCRIPTION HANNA A A A A F F A F BYTE 1 x BYTE 2 x BYTE 3 x BYTE 4 x x x x x x ANSWER AREA ADDRESS x x x x ADDRESS GENERAL PURPOSE x GRS x SAVE AREA FOR TASKS REG S x x x WHEN TASK NOT ACTIVE x x x ADDRESS GENERAL PURPOSE x FRS x x SAVE AREA FOR TASKS x x x FLOATING POINT REGISTERS x x x WHEN TASK NOT ACTIVE x RESERVED SET TO 0 x x ADDRESS OF END OF TASK x AETX x ROUTINE SPECIFIED IN x x x ATTACH MACRO EXTR x x x TASK x TASK x PRI x x LIMIT x DISPATCHx x x PRIORITYxPRIORITY x x x COMPLETION CODE CMC x 1 IF NOT COMPLETE 0 x ADDRESS OF TASK INPUT AND x TIOT x OUTPUT TABLES x COMM x ADDRESS OF COMMAND x x SCHEDULER COMMUNIATIONSS x x LIST x x ADDRESS OF TIME SHARING x TSO x x FLAGS FIELD IN TCB x x ADDRESS OF PROTECTED x PSB x x STORAGE CONTROL BLOCK x THE TERMINAL JOB x TIID x IDENTIFIER x EJECT OBTAIN DISPATCH PRIORITY FOR MAIN ATTACH SECOND SNAP THE TCB S FOR MAIN AND SECOND SET A WIAT MACRO FOR THE ECB OUT OF SECOND DETACH SECOND A copo Me cao sb ACA ACA CACA ACA F F F F
193. R 8 HAS MAX KEY VALUE 9 HAS THE MIN KEY VALUE AND REGISTER 7 HAS ADDRESS OF THE COMPARE INSTRUCTION OPEN INPT INPUT OTPT OUTPUT PRINT NOGEN R7 COMPARE R8 1020 SET R8 TO LAST KEY VAUUE R9 1001 SET R9 TO MIN KEY VALUE READ IN THE CARD AND CONVERT THE KEY TO BINARY FORM IN REGISTER 10 THEN CHECK TO SEE IF THIS IS THE NEXT KEY IN THE LIST OF KEYS LOOP ET INPT AREA
194. RED L R2 0 R1 LOAD THE ADDRESS OF THE MAXIMUM OF THE RANDOM NO GENERATOR L R8 4 R1 LOAD THE ADDRESS OF 2ND ARG L R8 0 R8 LOAD THE MIN VALUE L R2 0 R2 LOAD THE MAX OF RANDOM NO GENERATOR x lt FIND THE NO OF RANDOM BITS TO BE SET OVERFL A R4 0 PUT A 0 IN R4 LR R3 R2 AKE A COPY OF MAX OVER SRL R3 1 FIND THE NO OF BITS TO BE SET LA R4 1 R4 R4 IS THE NO OF BITS TO BE SET LTR R3 R3 DETERMINE IF R3 IS 0 BNE OVER IF NOT 0 THEN SHIFT AGAIN PLACE A 1 IN THE ZERO BIT OF R5 LA R5 2048 PUT A 1 IN THE O BIT OF R5 SLL R5 20 MOVE THE BIT TO BIT 0 SR R6 R6 ZERO REG 6 THE NEXT SECTION GENERATES A RANDOM BIT AGAIN L RO SAVEODD GET THE ODD NO LTR RO RO DETERMINE IF SAVEODD 0 BNE BEGIN IF SAVEODD NO 0 THEN BEGIN COMPUTING PRINT GEN TIME BIN PRINT NOGEN LOADODD ALR RO RO MAKE RO JUST BELOW OVERFLOW BNO LOADODD IF NOT OVERFLOW MAKE
195. REFOR ISSUING THE XCTL MACRO IS DELETED BY THE XCTL MACRO HEREFORE THE REGISTERS ARE RELOADED IN THE ISSUING STEP SO HAT WHEN THE STEP SHICH IS XCTLED TO RETURNS IT RETURNS TO THE PROPER POINT PRINT NOGEN FOURTH CSEC XSAVE TR NO OPEN ETPT OUTPUT PUT ETPT FHEAD CLOSE ETPT XRETURN SA TR NO FHEAD DC CL80 OFOURTH LOADED AND ENTERED VIA XCTL MACRO ETPT DCB DSORG PS MACRF PM LRECL 80 RECFM FA BLKSIZE 80 DDNAME FT06F001 EROPT ACC PRINT GEN END LOG CJ OBJECT SXSLMOD DD DSNAM EXEC ASGCLG SOURCE SXSGO DD DISP OLD PASS SOURCE INPUT DD amp amp LOADMOD FOURTH DISP OLD PASS DELETE THIS IS THE MAIN JOB STEP WHEN IT RECEIVES CONTROL IT FIRST LOADD SECOND AND CALLS SECOND THEN WHEN CONTRO IS RETURNED T IT PASSES CONTROL TO THIRD VIA THE LINK MACRO FINALLY IT XCTL TO FOURTH AND CONTROL IS NOT RETURNED e Me Oe
196. RESS OF AN AREA THAT CONTAINES DEBUGGNNG INFORMATION SO PRINT IT AND RETURN HE SYNADRLS MACRO RESTORES THE REGISTERS THAT EXISTED WHEN CHECKER RECEIVED CONTROL IT RESTORES THEM FOR YOU HE SYNAD ROUTINE CAN NOT SAVE IN CALLING PROGRAM SAVEAREA x lt CHECKER SYNADAF ACSMETH BDAM PRINT NOGEN XPRNT O 1 PRINT GEN SPACE 5 SYNADRLS BR 14 ARFA DC 10F O DC 10F 0 DUMAREA DS CL5 x lt E THIS IS DCB FOR BDAM OUTPUT TO CREATE BDAM DATA SET WHEN CREATING THE DATA SET THE DSORG MUST BE PS THE MACRF MUST BE WL DEVD MUST BE DA DDNAME IS STANDARD SYNAD IS ADDRESS OF ROUTINE TO BE GIVEN CONTROL WHEN ERROR BURING EXECUTION OF THE I O OCCURS OTPT DCB DSORG PS MACRF WL DDNAME DAOUTPUT DEVD DA SYNAD CHECKER PRINT NOGEN INPT DCB DSORG PS MACRF GM LRECL 80 BLKSIZE 80 RECFM F DDNAME INPUT EROPT ACC EODAD EODADD LTORG PRINT Gi CJ N END DATA DAOUTPUT DD DSNAME amp amp TEMP UNIT SYSDA DISP NEW PASS DCB DSORG DA BLKSIZE 40 KEYLEN 4 RECFM F SPACE 44 21 DATA SYSUDUMP DD SYSOUT A DATA XSNAPOUT DD SYSOUT A DATA INPUT DD 1001FIRST RECOR 1003THIRD RECORD C El nj tA 1005RECORD FI 1007SEVENTH RE 1009NINETH RECOR 1011ELEVENTH RECORD H R 1013THIRTEEN 1015FIFTEENTH RE 1017SEVENTEENTH RECORD
197. RUC RACK ADDRI GIN ON RD OF CON ION TO BU ESS FOR A HA RO 4 CON 12 AINS my p HE COUNT LF WORD BO OF LIST IONS ILD A LIST THAT EACH USER CREATED UNDRY WITH TION RIES L INFORMAT EN CONTAINS HE LENGTH OF EACH EN RY PLIES DISC AT R EAS FU 14 CONTIG IN BYT MAX ES MAX 76 US BYTES FO EACH LIST ENI LOW TRY RIPTO HE EMB LL WORD FO ER NAME D R WITH BI KS ED THE A N IST TRY AND IF NECESSA JUSTIFII ES CO FJ D A ACRO INST IES THE RE RAC ESS ON TRACK OF T A NAME R ES HI Ho m E R ry ER R I RY E FOUND NAME IS A IN NA K OR JOB E OR AN TYPE R FIND AC E OF ATION IS RO US DATA ACED IN E p IN T DIRECTORY ES 8 13 OF EACH e PDS BY INSTRUCTION Pp IS ISSUED WHICH CONV CJ CJ ATIVE A DDR ESSES IN T ABSO SSES AN ALLOWS ENT REA WITH THE E D MEMB Fi XNOTE D INSI DS WRIT IR HE E BLDL LIST INTO ERTS T
198. S AN INNER MACRO amp RND specifies register to be set to address of NODE amp HDR usually specifies the name or other RX address of the header cell of a list amp END if specified requests code to test the LINK just loaded into amp RND and branch to the label specified if it 0 LOOPA QPOP R6 HDR1 END ENDA LOOPA Ti R6 HDRI LTR R6 R6 BZ END T MVC LINK HDR1 4 LINK R6 E QPUSH PUSH NODE ONTO BEGINNING OF LIST amp LABEL QPUSH amp RND amp HDR amp RND contains the address of a NODE which is pushed onto the list begun at amp HDR amp RND specifies a register name or number amp HDR is a name of a header cell PUSHX QPUSH R8 HDR1 PUSHX MVC LINK 4 R8 HDR1 ST R8 HDR1 F QSRCH CS411MC2 05 LIST SEARCH AND INSERT Qh amp LABE Ti OSRCH RND amp HDR amp RGW RWK1 RWK2 amp KO 4 amp KL KEYL This macro searches the list begun by header amp HDR which is linked in ascending order by KEY fields for the correct place to insert the NODE addre are amp KL by needed fo amp RND Qh HDR amp RGW amp KO amp KL have Note ssed by register amp RND The KEY fields are at offset amp KO and tes long amp RGW gives two registers which may be used if r temporary work registers without disturbing anything gives address of the NODE to be inserted is the name of the list HEADER CELL gives t
199. S THESE TABLES AND FLOWCHARTS ARE USUALLY RE COMPLEX THAN NEEDED FOR THE VERSIONS REQUIRED TO BE TURNED N FOR GRADE LEAVING MUCH ROOM FOR CHANGES KK KK KK x k KK KK KK KK SS KR SS KK k kt k k k k k k k k k O T H lt Q H G gt gt A NAA AA RC A x CS411FP1 04 11 COMPUTER CONFIGURATION The OS 411 operating system is to be designed for use with a paticular model of the famous or infamous depending on viewpoint System 411 series of computers This section lists the various sizes and nature of systems which may occur A MEMORY SIZ CJ amp MEMSIZE gives the size of the S 411 computer in K 1024 bytes amp MEMSIZE may range from 1 to 1024 and OS 411 may or may not be required to run on different sized machines Some versions of 0S 411 may be written to run only with one memory size thus allowing for more fficient special case programming techniques B NUMBER OF CENTRAL PROCESSING UNITS CPU A CPU is one of the most important resources to be needed for the execution of a JOB A basic S 411 system contains only 1 CPU while the bigger models may contain more than one The possible options are CPU1 Only 1 CPU exists CPUv amp NUMCPUS gives the number of CPU S in the system amp NUMCPUS may range from 1 to an upper limit to be given CJ C
200. SFERRED TO IOEOF1 WHENEVER THERE ARE NO MORE CARDS LEFT IOREAD GE IOCRDDCB IOCARD OVE NEXT CARD TO IOCARD PUT IOPRTDCB IOCARD PRINT IT IT HAS CARRIAGE CONTROL SPACE 1 NEXT PUT ILLUSTRATES PUT LOCATE 0S 360 RETURNS IN R1 THE ADDRESS OF NEXT BUFFER IN WHICH TO PLACE OUTPUT CARD WE MOVE IT THERE OURSELVES PUT IODSKDCB 2ND OP OMITTED SINCE PL VC 0 80 R1 IOCARD OVE THE CARD THERE B IOREAD GO BACK FOR MORE SPACE 1 IOEOF1 EQU BRANCH HERE SEE EODAD IOEOF1 USE CLOSE MACRO TO CLEAN UP AT END OF PROCESSING WILL ALSO WRITE OUT LAST BUFFER CREATED CLOSE IOCRDDCB IODSKDCB NOTE EXTRA COMMA REQUIRED WE NOW REOPEN IODSKDCB FOR INPUT THIS TIME NOTE THAT THE NAME ONLY IS INCLUDED IF THE OPTION IS OMITTED INPUT IS IMPLIED OPEN IODSKDCB OPEN IT FOR INPUT PUT IOPRTDCB IOTITLE2 PRINT SECOND TITLE SPACE 1 AT THIS POINT WE CLOSE PRINT DCB DYNAMICALLY CHANGE LRECL AND BLKSIZE TO 80 SO WE DON T HAVE TO PAD WITH BLANKS THIS TIME THEN RE OPEN CLOSE IOPRTDCB CLOSE IT FLUSH BUFFERS MVC IOPRTDCB X 3E 2 H 80 MAKE BLKSIZE 80 MVC IOPRIDCB X 52 2 H 80 MAKE LRECL 80 ALSO OPEN IOPRTDCB OUTPUT REOPEN NOW THE ABOVE KLUDGE ONLY NECESSARY BECAUSE LAZY PROGRAMMER DOESN T WANT TO PAD WITH BLANKS AND MOVE CARDS AROUND SPACE 1 SPACE 1 FOLLOWING LOOP READS THE RECORDS BACK FROM DISK THEN PRINTS THEM OUT AGAIN
201. ST and run and test it using the ASSIST Replace Monitor The programmer should first consult the following writeup for general information ASSIST conventions and use of the Replace Monitor ASSIST REPLACEM CJ NT USER S GUIDI CJ ASREPLGD I S 360 BASE REGISTER ASSIGNMENT This section brieflv describes the conversion of program addresses to base displacement form as done by S 360 assemblers particularly ASSIST The following manual should also be consulted IBM S 360 OS Assembler Language GC28 6514 pp 19 21 A Each control section and each dummy section in an assembly is assigned a unique number or section identification ID and every label in a given section has that same section ID associated with it B When a register is specified in a USING statement it is assumed to contain the specified location counter value and is also flagged with the section ID of the first expression in the USING C When a value used in an instruction must be converted to base displacement form the only possible registers which are usable are the ones if any which hav th sam section ID as the value to be converted D If two or more registers are usable as base registers and have the same section ID the register used is that one having a value which results in the smallest displacement 0 4095 E If two or more registers have the same ID and value
202. ST LENGTH CALCULATIONS Whenever a JOB gains control of a CHANNEL the time it will use it must be calculated and an 1 0 INTERRUPT scheduled to occur at that time when the job will give up control of the channel This length of time using the channel may be found in ways analogous to the IOR options for determining the intervals between requests IORC Constant interval from SYSGEN For ALL jobs amp IOREQL gives the number of milliseconds a job uses a Channel IORJ Fixed interval from JOB card For each job the length of tim a Channel is used is constant and taken from the IORL parameter on the JOB card IORR Random lengths using JOB card information For example TORRU UNIFORM the actual length is computed from a UNIFORM distribution using IORL from JOB card as the MEAN Thus any value from 0 to 2 IORL is equally likely CS411FP3 02 VII REPORTS At various times during a simulation run various reports may be printed In some cases these may be useful for debugging and others are most useful for obtaining statistics for comparison of different versions of OS 411 Unlike most of the other option groups the ones in this set are NOT mutually exclusive R1 SYSGEN report This report is printed 1 time only at the beginning of an entire run even if multiple batches of jobs are run It describes which options are being used and also prints the values of various SET variables used to genera
203. STE3 FOR MODULES MODO MOD3 THESE ARE IN OUR JOBLIB SO WE CAN JUST CALL THEM OUT WITH EXEC STMTS STEPO EXEC PGM MODO NO OVERLAY XSNAPOUT DD SYSOUT STEP1 EXEC PGM MOD1 3 SEGMENT OVERLAY XSNAPOUT DD SYSOUT A fe STEP2 EXEC PGM MOD2 10 SEGMENT 1 REGION OVERLAY XSNAPOUT DD SYSOUT STEP3 EXEC PGM MOD3 14 SEGMENT 3 REGION OVERLAY XSNAPOUT DD SYSOUT A fa END OF EXAMPLE ON OVERLAY PROGRAMS 7 8 Z 1 SAMPLE PROGRAM EXAMPLE OF IEBPTPCH ON PDS SHOWING HOW TO PRINT SEVERAL MACROS FROM SYS1 MACLIB CALL SAVE RETURN Lp PRINT SEVERAL MACROS USING PTPCH VIA PRINTMAC EXEC PGM IEBPTPCH SYSPRINT DD SYSOUT A SYSUT2 DD SYSOUT A SYSUT1 DD DSN SYS1 MACLIB DISP SHR ZE DSN CMACLIB DISP SHR FOR PSU LOCAL MACROS SYSIN DD PRINT TYPORG PO MAXFLDS 6 MAXNAME 3 ITLE ITEM MACRO LISTINGS 40 ITLE ITEM 40 EMBER NAME CALL RECORD FIELD 72 10 FIELD 8 73 90 EMBER NAME SAVE RECORD FIELD 72 10 EMBER NAME RETURN RECORD FIELD 72 10 T LT LT LT Lt
204. STEP READS IN THE CURRENT RECORDS AND PRINTS THEM GEN ae ke cd sP OPEN INPUT AND OUTPUT DATA SETS SET R11 TO NUMBER OF BLOCKS SET R3 TO 0 THE RELATVVE ADDRESS OF FIRST BLOCK STORE R3 AT REF OPEN INPT INPUT OTPT OUTPUT LA R11 20 SR R3 R3 ST R3 REF PRINT GEN SPACE 5 THE POSITIONAL PARAMETERS FOR THE READ MACRO ARE 1 DECB NAME TO BE CREATED BY READ MACRO 2 TYPE DIRECT ADDRESS ING BY RELATIVE BLOCK 3 DCB ADDRESS NNPT 4 AREA ADDRESS AREA WHERE RETRIEVED DATA TO BE PUT 5 LENGTH S TO BE TAKEN FROM DCB 6 KEY ADDRESS 0 INDICATES KEY NOT TO BE READ 7 BLOCK ADDRESS ADDRESS OF THREE BYTES OF CORE CONTAINING THE RLLATIVE KEY ADDRESS LOOP READ DECB DI INPT INPTT S 0 REF 1 SPACE 5 CHECK DECB INCREASE RELATIVE BLOCK ADDRESS THEN OUTPUT THE RECORD RETURN FOR NEXT INPUT L R3 REF LA R3 1 R3 ST R3 REF PRINT NOGEN PU OTPT INPTT 1 BCT R11 LOOP CLOSE DATA SETS AND RETURN TO OS CLOSE INPT OTPT XRETURN SA ERROR REOUTINE FOR USE WHEN ERROR OCCURS CHECKER SYNADAF ACSMETH
205. TINES External subroutines are used for major program segments and can usually be assembled separately from the rest of the program In fact they can be written in a different language i e FORTRAN and ASSEMBLER combinations 1 DEFINITION An external subroutine may be written in either of two ways in assembly language as a CSECT or as an ENTRY within a CSECT In the first case the subroutine is entered at the CSECT statements and return at one or more places depending on the desired code In the second case each entrypoint may be given control and may share code or be totally separate from the other entries This form is often used for a group of related routines like SIN and COS which are both entries in a CSECT or for a routine requiring initialization or termination functions different from the normal calling function ltiple e NAM CS ntry CSI ECT ECT is typically s ENTRY ENTRY1 ENTRY2 1 ENTR ENTRY1 for e ntered o LINKAGE ntry at CSECTNAM nly at the entry E CODE R executable code w ETURN LIN SAVE XSAVE hen called at E KAGE entry CSECTS ining entrypoint nal subroutines data areas used by CODE R names and various of he following are important points to remember when using E HE DIFE ERI ENT ENTRY PO
206. TPT IS THE DCB NAME DUMAREA IS ADDRESS OF 5 BYTES OF CORE FOR USE IN DUMMY OUTPUT INCREASE THE MIN KEY VALUE AND RETURN TO CHECK THE CURRENT KEY VALUE AGAIN CHECK IS USED TO CHECK THE DECB x lt WRITE DECB2 SD OTPT DUMAREA SPACE 5 CHECK DECB2 PRINT NOGEN LA R9 1 R9 INCREASE R9 BY ONE BR R7 GO TO DO NEXT COMPARE T AT THIS POINT WE HAVE JUST READ IN THE LAST INPUT SO GO TO FINISH OUTPUTTING DUMMY RECORDS UNTIL THE DATA SET IS FULL F F F EODADD LA R7 DUMMY BR R7 RETURN TO CONTINUE DUMMY OUTPUT PRINT GEN SPACE 5 x x NOW WE HAVE FILLED THE DATA SET SO CLOSE THE DATA SETS AND RE NOW WE HAVE FILLED THE DATA SET SO CLOSE THE DATA SETS AND RETURN TO OPERATING SYSTEM EOJ CLOSE INPT OTPT PRINT NOGEN XRETURN SA PRINT GEN SPACE 5 x lt X WHEN AN UNCORRECTABLE ERROR HAS OCCURED IN THE WRITE STATEMEN THIS ROUTINE IS GIVEN COOTROL BY THE CONTROL PROGRAM SYNADAF RETURNS IN REGISTER 1 THE ADD
207. TTACH SECOND WITH EXIT ROUTINE TO DETACH S SNAP TCB FOR MAIN AND SECOND LPR 10 10 PRINT GEN RESORE PRIORITY FOR MAIN 10 INDICATES THE VALUE TO BE ADDED TO IN REG 10 THE S INDICATES THE CURRENT LOAD MODULE THEN SET WAIT MACRO IN MAIN BEFORE CONTINUING THE TCBADD IS THE ADDRESS OF THE TCB FOR LOAD MODULE TO ECOND EXTR IS THE ONYL NEW PARM IT GIVES THE ADDR BE GIVEN CONTROL WHEN SECOND FINISHES ATTACH EP SECOND PARAM OTPT ZERO ECB MECB ETXR M DISPATCH FOR MAIN IS ESS OF A ROUTINE TO EXTR VL 1 ST 1 TCBADD PUT ADD OF TCB FOR DE AGAIN SNAP TCB S FRO MAIN AND SECOND EXTRACT SE VEN TCBADD FIELDS ALL EN A T xK mono uN Z D rg ep H O HOE x H bu D Q H n EN S FIELDS ALL SEVEN SEVEN 28 T NOREGS CB FOR MAIN WITH ONLY ALL X DE a Z D FU n O HO E GE SEVEN SEVEN 28 T NOREGS ABEL THIS IS TCB FOR SECOND ATTACH OF SECOND SPI ECIFTI TACH ED FOR FI ELDS U hg H pa H Q ti WIAT FOR SECOND TO COMPLETE BEFORE ATTACHING THIRD WAI ECB MECB A xXx A A A A A F F F xXx FK
208. VE E THE CORR ry f E CT VALUES IN T EX a WHEN THEY AR ry CALLI ECUTION TIM Bo IF THEY DON T HE E E ED FROM DIFF CJ UPS HEY WILL ASSE ECTIONS HAVING DIFFERENT USING RI BLE PROPERLY F F F E PARTICULA EX operatio R n THI WITH ah EN BL UP A He H ACE US PROGRAMMER SHOUL THE SECTION OF COD E Pp U Houpssoa ATA AREAS IF HEY T ER EXECUTION IM TO BE EXECU NOT AT WITH F H INS ING IONS AND Ho PERFORM ANY SYMBOLIC A all entry poi subroutines a save area just ab since it will alway conditions across a ENTRYX L USIN Note that the above can be accomplished with the XSAV XSAV The problems described above are typically nt cod segm ove these cod s have that same nts set up the same USING setting a specific register to point to the EXecuted instructions and data sections value DDRI handled either by making conditions or by beginning of the internal If register 13 points to it can be used this way Getting the same USING ECT can be done n entire multi entry CS Li BAS G CS EREG A CSECTNAM CTNAM BASEREG
209. VERALL FLOW OF THIS PROGRAM IS AS FOLLOWS 1 THE FOLLOWING CSECTS ARE ASSEMBLED AND LINK EDITED SECOND THIRD AND FOURTH 2 MAIN IS ASSEMBLED AND LINKEDITED THEN IT IS GIVEN CONTRLL AND EXECUTES 3 DURING EXECUTION OF MAIN SECOND IS LOADED USING THE LOAD MACRO THEN SECOND IS CALLED USING CALL MACRO CONTROL IS THEN RETURNED TO MAIN AND SECOND IS DELTED USING THE DELTTE MACRO 4 THIRD IS LOADED AND CONTROL PASSED TO IT USING THE INK MACRO CONTROL IS RETURNED TO MAIN THROUGH THE LINK MACRO CONTROL PROGRAM AND THE OONTROL PROGRAM DELETES THIRD 5 FOURTH IS LOADED AND GIVEN OONTROL THROUGH THE XCTL MACRO THE XCTL MACFO DELETES MAIN AND THEN CONTROL IS RETURNED TO THE OPERATING SYSTEM USIGN IS RETURNED TO THE OPERATING SYSTE EJEC PRINT NOGEN THE PURPOSE OF THIS CSECT IS TO BE LOADED USING THE LOAD MACRO AND THEN TO BE CALLED USING CALL MACRO THEN IT PRINTS A MESSAGE AND RETURNS TO MAIN NOTE THE NORMAL SAVE AND RETURN CONVENTIONS SECOND CSECT XSAVE TR NO OPEN OTPT OUTPUT PUT OTPT SHEAD CLOSE OTPT XRETURN SA TR NO SHEAD DC CL80 OSECOND HAS BEEN LOADED AND CALLED RETURN TO MAIN OTPT DCB DSORG PS MACRF PM LRECL 80 BLKSIZE 80 RECFM FA DDNAME FTO6F001 EROPT ACC PRINT GEN END OBJECT SYSLMOD DD DSNAME 66LOADMOD SECOND EXEC ASGCL SOURCE SYSGO DD DISP OLD PASS X
210. VIDE THE NUMBER OF BYTES BY 4 TO GET THE NUMBER OF WORDS TO BE READ LA R3 1 PLACE A 1 IN BIT 31 OF R3 SLL R3 24 OVE THE BIT TO BIT 7 OF R3 OR RO R3 PLACE SUBPOOL IN RO PRINT GEN USING A GETMAIN MACRO OBTAIN THE NUMBER OF BYTES REQUESTED THE REGISTER CONVENTION REQUIRES THAT THE SUBPOOL NUMBER BE PLACED IN THE FIRST BYTE OF REG 0 GETMAIN R LV 0 PRINT NOGEN MAKE TWO COPIES OF THE ADDRESS OF THE AREA OBTAINED FROM THE OPERATING SYSTEM THEN USING R4 FOR LOOP READ IN THOSE NUMBERS AND PLACE THEM IN THE AREA OBTAINED LR R6 R1 AKE A COPY OF ADD OF NEW STORAGE LR R7 R1 AKE A COPY OF ADD OF NEW STORAGE LOOP XREAD WORD READ IN THE NUMBERS XPRNT WORD 1 81 ECHO PRINT THE NUMBERS XDECI R5 WORD CONVERT THE NUMBERS TO INTERNAL FORM ST R5 0 R6 PUT NEW NUMBER IN STORAGE LA R6 4 R6 INCREASE POINTER TO NEXT NEW WORD BCT R4 LOOP IF NOT LAST NUMBER RETURN FOR NEXT FINALLY SNAP THE AREA OBTAINED AND THEN FREE THE AREA OBTAINED a l A a not OR F O O Bm VA OSS NE OF NWO 0 lt Y Ot EST G IN Aa
211. WORD like S 360 S BDR 64 RX YES BRANCH ON DECREMENTING REGISTER Modify reg R by subtracting 1 Set CC according to result and branch to EA location if result gt 0 AWM 66 RX YES ADD WORD TO MEMORY add contents of register R to word at location EA BCS 69 RX NO BRANCH CONDITIONS SET uses R as a Mask field to branch or not same as S 360 BC RD 6C RX YES READ DIRECT read a card set CC 0 if no more remain CC 1 A single signed number anywhere on the card is converted to binary and placed in the register specified by R EA is totally ignored and may have any value WD 6D RX NO WRITE DIRECT the number in register R is converted to decimal and printed SSJOB LEGALOPS1 0 kokok x AS A RA RR k x k x x x x K k RR RR RR AAA kkkkkkkxxx ATTENTION USE LARGE PARAMETER IF YOU ARE GETTING AS999 kkkkkkkkkk SIMILIAR MESSAGE ABOUT EXCEEDING SPACE SEE ASSIST MANUAL wk ee KKK NOTE EXTENSION EXTENSION EXTENSION FINAL PROJECTS DUE AT kkkkkkxxx 11AM WEDNESDAY NOT MONDAY ki kok x kk kk kk K K k k x KKK KKK K k k k x kk x x lt lt lt k x k KKK x k k lt k KKK KKK k k k k k k amp AAA AAA AAA THIS PROGRAM CONTAINS NO ERRORS IT IS A TEWST FOR THE RECOGNITION OF ALL LEGAL OPCODES THE PROGRAM WILL NOT BE EXECUTED AI 1 2 AW 1 3 2 LI 12 EGAL OP CW 172 LABEL MI 15 2 LW 1
212. X1 01 03 programming exercises hand assembly interrupts 02 16 more on assembly process location counter control use of ORG to set up tables timetable for gettting final project done program design process and debugging 02 18 quiz TR TRT 30 minute quiz hand assembly BXLE loop setup TR uses setup workings TRT uses setup xamples READINGS STRUBLE CH 15 pp 342 345 350 352 prob 1 3 4 ASSIGN write TRT table for scanning over hex digits 02 21 programming techniques use of TR TRT conversions use of global table pointer examples on TR TRT decimal input conversion using two TRT s EX PACK CVB hexadecimal output conversion using UNPK TR ASSIGN write code to perform conversions also to read in names place in table then search table for later names READINGS STRUBLE CH 15 ED EDMK start 02 23 conversions hexadecimal input decimal output ED go through hexadecimal input but not in detail TRT TRT EX of MVC right justified TR PACK 9 into 5 ignoring extra byte decimal output CVD UNPK OI for plus number with leading zeroes decimal output begin on ED EDMK doing parts with basic workings of ED and standard pattern for integer numbers CS411AS1 01 CMPSC 411 ASSIGNMENT 1 LINKAGE HANDLING FORTRAN ASSEMBLER AND OBJECT DECKS DUE This writeup pages 01 02 T
213. XO ODUMPO These macros cover the following topics macro module linkage some macro processing techniques and hexadecimal conversions The following instructions may be useful TRT PACK TR UNPK A QHEXI MACRO TO SCAN AND CONVERT HEXADECIMAL NUMBERS TO BINARY 1 REQUIRED FUNCTION OF MACRO QHEXI is to function somewhat like XDECI but for hexadecimal input rather than decimal It is to be called as follows label QHEXI reg address label is an optional statement label reg is the name or number of a register address is an RX type address i e anything legal in an LA instr Calling QHEXI should cause the following actions to be done a The address should be evaluated and used as a scan pointer to some character string in memory A scan will be made starting at that address until the first character is found which is a hexadecimal digit 0 9 A F b Starting at the first hex digit found a hexadecimal number of 1 8 digits followed by any character NOT a hex digit is to be scanned and converted to a 32 bit binary value right justified and filled on the left with leading zeroes c Register 1 is set to contain the address of the first non hex digit following the hexadecimal number d The register specified in the macro call is loaded with the value given by the hex number Just converted this may be any register e The condition code must end up being
214. Y CHI1 END OF SUBIB NSERT SUBI PUT SUBIC AT END OF SUBIB VERLAY CHI1 BACK TO END OF SUBIB NSERT SUBI PUT SUB1D BEGIN LIKE SUBIC VERLAY ALPHA BACK TO WHERE SUBI BEGAN NSERT SUB2 BEGIN SUB2 WHERE SUBI DID VERLAY BETA2 ORIGIN FOR SUB2 S SUBROUTINI NSER SUB2A PUT SUB2A AT END OF SUB2 VERLAY BETA2 BACK EVEN WITH SUB2A NSER SUB2B PUT SUB2B IN SAME VERLAY BETA2 ONCE MORE NSERT SUB2C PUT C IN SAME AS A AND B VERLAY REGION2 RI NEW REGION NSERT MSUB1 TAKE OUT OF ROOT SEGMENT VERLAY REGION2 REPOSITION DON T REGION E AS SUB2A CJ E AS SUB2A INSER SUB2 PUT MSUB2 SAME AS MSUB1 OVERLAY REGION3 REGION NEED ANOTHER REGION FOR MSUB3 4 INSERT SUB3 PUT IN THIS REGION OVERLAY REGION3 BACK TO BEGINNING OF REGION INSERT SUB4 PUT MSUB4 SAME AS MSUB3 NAME 0D3 3 REGION 14 SEGMENT OVERLAY Z 7 THE FOLLOWING SECTIONS MODIFY THE SYSLMOD CORRECTLY AND DEFINE amp amp STUPID SO THAT IT CAN BE INCLUDED Pipe OBJECT SYSLMOD DD VOL REF JOBLIB DSN amp amp LOADMOD DISP OLD PASS OBJECT OBJ DD DSN amp amp STUPID DISP OLD PASS AE EJE NOW EXECUTE THE LOAD MODULES PRODUCED BY ASMLINK STEPS STEPO
215. age with initial value DC setting the location counter ORG and START instruction ILL tt CS102FP1 02 C Machine Code or Object Code 1 Machine code is the output of the assembler The format of the code varies from machine to machine but in general consists of the following OPCODE coded form of the operation to be performed REGISTER part of an operand ADDRESS part of the operand It can consist of base and displacement 10 2 an actual address 100 an indirect address an indexed register form 0 11 12 reg 11 index shift counter SLA 2 4 immediate operand CLI 2 x FO Ho OQ oo IMPLEMENTATION A Monitor 1 The monitor examines the control cards to determine the start and end of a job 2 The monitor usually examines switches to determine what to do e g If the assembler sets the switch for valid assembly done the monitor then calls for execution of the program 3 The monitor skips to the next job if the current job can not be executed as shown by flags may skip data cards B Assembler Assemblers vary as to the number of passes it performs The number is a function of the size of the machine the speed wanted and the complexity if the source language We will be interested in a 2 pass assembly which is most common 1 The first pass a In the first pass the opcodes are checked for validity e g an opcode may not exist or may not be impleme
216. and also takes are of checking against time remaining if user being charged for channel time also i e if the SYSG EN specified amp ECCH 1 CHANALC must make sure that CINTR is called whenever the I O interrupt is to occur possibly to call JOBTERM if there is no more 0 time left and channel time is being charged his module implements IOR options 7 CHANNEL INTERRUPT HANDLER CINTR The channel This module is called whenever a job finishes with a channel is freed marked idle and if any other job is waiting for that channel it calls CHANALC to give the channel to the job needing BEG The job s status is changed from I O to READY it is placed back in the AJL and a request made that the DISPATCHER be called so that the job may effectively compete for a CPU again CS411FP4 02 8 JOB TERMINATOR JOBTERM This module is called when a job finally consumes the time allotted to it on the JOB card It removes the job from memory returns any control elements to free list of such and also notes that the JBINIT routine be called since it may now be possible to initiate another Job 9 EVENT EVNQUEUER EVNQ This module or possibly just a macro is called whenever it is necessary to schedule some type of event to occur at a certain time It inserts or constructs and inserts an EQ node into the EQ list ac
217. and is issued This not only clearly labels your listing but it saves time in looking through a listing which is more than a few pages long The command EJECT skips the listing to a new page and is useful in blocking off major parts of a program he command SPACE n inserts n blank lines into the listing at that point This is useful for blocking off smaller sections of a program particularly small loops register equates etc Not only do listing control instructions aid to the readability of a program but they also save the programmer time in debugging DON T merely restate an instruction when you place a comment on it Of the following two examples which is more explanatory A 1 VAR ADD VAR TO REGISTER 1 A 1 VAR R1 SUMMATION OF ODD PRIME NUMBERS DON T put several single comments between statements in an unreadable manner It is often useful to indent a single comment to column 16 This keeps it from interfering with the reading of labels and opcodes and thus distinguishes it from the machine instructions DOCUMENT 4 CJ DO use comment card blocks which list useful information For example a list of register allocation and usage is extremely helpful not only in debugging but also in revising a program Such a list should appear as part of the preface to the appropriate section of code Another example is a l
218. arrays of 128K bytes with physical word size of doublewords i e each has 2 arrays of 16K doublewords and is thus 2 way interleaved at this level Each 2365 is independent of the others CYCLE TIME 750 nanosec ACCESS TIME 375 nanosec 2861 TI l unit Large Core Storage LCS 2048K bytes organized physically of 2 way interleaved doublewords CYCLE TIME 8000 nanosec 8 microsec ACCESS TIME 3 2 mic Of the two types of storage the first contains user programs and heavily used parts of system programs while the LCS contains less used system programs tables and buffer areas HARDWARI 02 CHANNI ti m a 2870 ULTIPLEXOR CHANNEL includes 2 SELECTOR SUBCHANNELS used for magnetic tape drives generally handles LOW SPEED devices card readers printers etc AXIMUM TOTAL TRANSFER RATE 426 KB kilobytes per second 2860 SELECTOR CHANNELS 5 total 2 in 2860 II 3 in 2860 III used for HIGH SPEED devices disk drum etc AXIMUM DATA TRANSFER UNIT EACH SELECTOR 1250KB LT All CHANNELS and the CPU contend for use of memory modules The BCU arbitrates among them using a simple priority scheme in following order SERVED EARLIER gt SERVED LATER CHANNEL 1 2 0 3 4 5 CPU drums disk mx disk disk ADAGE The above order is used
219. bed below and diagram A shows the connections References are made to DEVICE ADDRESSES Each individually addressable devic such as a single disk drive card reader etc has a 3 digit hexadecimal number which uniquely identifies it to the system and is used in all input output operations The DEVICE ADDRESS is of the following form abc where a gives the CHANNEL NUMBER from 0 up b specifies a CONTROL UNIT attached to that channel c notes which device attached to a given control unit Since each digit can have the value 0 F theoreticallv it would be possible to attach 16 devices to each of 16 control units attached to 16 channels for a maximum of 4096 separate devices In practice this number is much less since most S 360 s allow a MAXIMUM of 7 channels or less The devices follow more or less in order from the CPU outward CENTRAL PROCESSING UNIT 2067 1 a single 360 67 CPU uses 200 nanosec 2 microsec cycle Read Only Storage ROS of 88 bits word to implement S 360 instruction set Universal plus special model 67 instructions includes a HIGH RESOLUTION TIMER 13 microsec cycle includes a BCU Bus Control Unit which is connected to all memory modules and determines which channel or CPU gets to use a given memory module PRIMARY STORAGE 2365 III 4 units each unit contains 256K bytes Physically each 2365 contains 2
220. bles to generate the system is referred to as a SYSGEN 5 It shoul CS411FP1 02 Ld be noted that some of the terminology employed in this writeup is not exactly standard in particular the word QUEUE is used This usually is taken to mean a list usually ordered on quite often some key from which onl only added to the end ly the first item can be removed and new items Although this may be the most common case for this writeup the items in some queues may be inspected modified added and deleted in any position For example it may be necessary to scan a queue in order for the first item meeting a specified test then removing that item C A BRI EF DESCRIPTION OF THE PROGRAM AND WHAT IT DOES The program to be written will read an INPUT STREAM which is made of one or more BATCHes of JOB cards Each JOB card descibes the needs and characteristics of a single JOB The program will then simulate the handling of OS 411 First JOBS at all their desir this job by a typical handles each BATCH of operating system one or more JOB cards as follows OS 411 initializes itself to a status in which there are no in the system and all immediately counters flags etc are set to d beginning values The JOB cards will then be read either or at various interval JOBS is finished executing
221. bytes card 1 3 _ _ 2540 PUNCH 80 bytes card 0 4 2671 PAPER TAPE 10 REFERENCES GA22 6810 IBM S 360 SYSTEM SUMMARY GA27 2719 IBM S 360 MODEL 67 FUNCTIONAL CHARACTERISTICS OSHASP 01 OVERVIEW OF 0S 360 WITH HASP This writeup gives a quick overview of the process by which any 0S 360 sytstem is initialized how storage is used particularly in 0S 360 MVT and describes how OS 360 is modified by the use of HASP Houston Automatic Spooling Priority system The storage layout is described for the PSU CC 360 67 system I INITIALIZATION getting a system up and running Consider a computer with no operating system currently in it The first necessity is to get a workable operating system in it so that jobs can be run This is NOT a trivial process note that there is no Program Fetch resident in the machine no I O Access Method routines and not even a correct set of PSW s in low core for directing interrupt actions For OS 360 the initialization process is composed of two parts IPL and NIP IPL Initial Program Loader initializes memory and some other things and brings the NUCLEUS the core of the OS into memory NIP Nucleus Initialization Program performs the remaining actions required to set up a specific NUCLEUS to be ready to execute A IPL Initial Program Loader The process of getting an 0S 360 system running is called IPLing and includes the following main steps 1
222. ccur after failure to do so It is a good idea to mark a save area upon exit This is usually done be moving X FF into the first byte of the fourth word of the save area the place register 14 was stored Although this technique does not seriously affect the contents of the save area for reading in a dump this technique quickly shows what save areas are active and which are not active when reading a dump Register 13 must be kept as the save area pointer however by careful programming it can also double as a base register Consult the appropriate section from XSAVE and XRETURN documentation for the coding sequence using these macros You may set up your own save area for this purpose by setting it high in a program and following it by a USING on register 13 referencing the name of the save area For reserving the 18 fullwords of storage for a save area use DC instead of DS A constant of F O or F 1 will quickly show in a dump if the save area was ever used SAVE and RETURN are two system macros which will eliminate much of the coding for saving and returning conventions SAVE generates th cod necessary to save a specified series of registers The registers are specified as they would be for a STM instruction In addition the operand T will cause registers 14 and 15 to be stored regardless of what other registers may also be saved from the pair specified The
223. cording to time of event 10 REPORT MODULE RPORT T This module may be called to produce the various types of reports possible at various times In some cases the R options may be included as sections of code in other routines like R3 in RDR R4 in JBINIT This module might be written in FORTRAN at least partially 11 MAIN PROGRAM his module first prints the SYSGEN report Rl if required It then consists of two nested loops he outer loop is executed once for each BATCH of jobs and is made of the following steps INITIALIZATION reset flags counters put all list nodes into the respective fr lists and also request that the RDR be called at time 0 set clock GNOW 0 etc INNER LOOP the actual simulation and COMPLETION print report R9 if exists If it is discovered at the end of this loop that a SSQUIT card has been found or end file the loop ends and the entire program terminates The inner loop is executed one time for each EVENT which occurs and thus traces the system actions at those times during which the system actually changes status ignoring times in between Each loop performs the following actions a POP s the first node from the Event Queu EQ removes the data item s from the node and returns the node to the free list of such nodes b Sets the global clock GNOW time from the node i e th current time is
224. ctions hand assembly etc 17 18 19 20 21 CS102TPA 05 02 14 on midterm and final project review of midterm results and problem areas final project overall structure useful modules and how to set them up decimal scan and output conversions symbol scan symbol table manager opcode lookup hexadecimal output etc review of BXLE loop control HANDOUT CS102PX1 01 03 programming exercises hand assembly interrupts 02 16 more on assembly process location counter control use of ORG to set up tables timetable for gettting final project done program design process and debugging 02 18 quiz TR TRT 30 minute quiz hand assembly BXLE loop setup TR uses setup workings TRT uses setup xamples READINGS STRUBLE CH 15 pp 342 345 350 352 prob 1 3 4 ASSIGN write TRT table for scanning over hex digits 02 21 programming techniques use of TR TRT conversions use of global table pointer examples on TR TRT decimal input conversion using two TRT s EX PACK CVB hexadecimal output conversion using UNPK TR ASSIGN write code to perform conversions also to read in names place in table then search table for later names READINGS STRUBLE CH 15 ED EDMK start 02 23 conversions hexadecimal input decimal output ED go through hexadecimal input but not in detail TRT TRT EX of MVC right justified TR PACK 9 into 5 ignoring e
225. d ONLY THEN it is now possible to modify registers L 15 label 4 V type adcon for routine CNOP 2 4 make sure next inst not on F boundry BALR 14 15 call routine also point 14 at the argument list following DC V subroutine entry point adcon to get there label DS 3F 3 words for saving 14 15 0 DS F space for arguments after first DS OR DC space here for any remaining arguments the subroutine will return control to next instruction LM 14 0 4 14 reload registers Note that this is only safe way since 15 might have current base register ASPRGTCI 05 The following shows the typical code used to enter and exit the supporting module used with the previous macro expansion entry point of the routine might be either a CSECT name Note that the or an ENTRY name i e one CSECT might contain several entrypoints one for each supporting subroutine needed entrypoint label definition CSECT or label DS 0H USING entrypoint 15 initial base register save all registers which may be modified by code Save into THIS CSECT unlike normal OS 360 conventions DO NO SAVE INTO CALLER S SAVE AREA since it may not exist especially if caller is a lowest level routine initialization code if this routine performs I O or calls any others or requests any supervisor services it is a good idea to set up another base register than 15 set up a save area and put its addre
226. d and proven to be effective aids in programming assembler language CMPSC 411 DSECT Example PRINT NOGEN EQUREGS MAIN CSECT XSAVE ESTAB H STANDARD LINKAGE CALL NEXT CALL LOWER ROUTINE XRETURN SA ESTABLISH SAVE AREA LTORG NEXT CSECT XSAVE ESTAB H STANDARD LINKAGE CALL LAST CALL LOWEST ROUTINE XRETURN SA ESTAB H SAVE AREA LTORG LAST CSECT XSAVE ESTAB H STANDARD LINKAGE CA TRACE CALL TRACE RTN TO PRNT S A XRETURN SA GENERATE SAVE AREA LTORG THE ABOVE ROUTINES DO NOTHING BUT ESTABLISH LINKS TO TRACE THROUGH THE SAVE AREAS kud x ROUTINE TRACE PROVIDES A PRINTED TRACE OF THE NAMES OF THE CSECTS OF ACTIVE S A S IT USES DSECTS SAVEAREA AND NAMECONV TO FORMAT THE SAVEAREA AND FIRST FEW BYTES OF THE PROGRAM TRACE CSEC XSAVE SA TRACESA ESTABLISH LINKS USING SAVEAREA R13 USING NAMECONV R15 XPRNT CL25 OBACK TRACE OF SAVE AREAS 25 L R13 4 R13 CONNE O FIRST ACTIVE S A LOOP LTR R13 R13 C
227. d in all input output operations The DEVICE ADDRESS is of the following form abc where a gives the CHANNEL NUMBER from 0 up b specifies a CONTROL UNIT attached to that channel E notes which device attached to a given control unit Since each digit can have the value 0 F theoretically it would be possible to attach 16 devices to each of 16 control units attached to 16 channels for a maximum of 4096 separate devices In practice this number is much less since most S 360 s allow a MAXIMUM of 7 channels or less The devices follow more or less in order from the CPU outward CENTRAL PROCESSING UNIT 2067 1 a single 360 67 CPU uses 200 nanosec 2 microsec cycle Read Only Storage ROS of 88 bits word to implement S 360 instruction set Universal plus special model 67 instructions includes a HIGH RESOLUTION TIMER 13 microsec cycle includes a BCU Bus Control Unit which is connected to all memory modules and determines which channel or CPU gets to use a given memory module PRIMARY STORAGE 2365 III 4 units each unit contains 256K bytes Physically each 2365 contains 2 arrays of 128K bytes with physical word size of doublewords i e each has 2 arrays of 16K doublewords and is thus 2 way interleaved at this level Each 2365 is independent of the others CYCLE TIME 750 nanosec ACCESS TIME 375
228. d interpreters It examines the JCL to determine the beginning and end of a job The function of the translator or assembler is to accept as input a source language e g IBM 360 Assembler Language and generate equivalent code in some target language e g machine language This code is called object code In addition to translating the assembler should have the capability of detecting syntactic errors in the source listing The interpreter however executes the intructions of the target language LANGUAGE FORMAT A Control Cards 1 Control cards will be used in TRIVIAL These cards direct the monitor about how to process a job On the IBM s 360 these cards are called Job Control Language cards JCL Also included with the JCL cards will be control cards used by the assembler 2 These cards indicate the start of a program end of assembly and start and end of data B Assembly Language There are two types of instructions 1 Machine instructions can b xecuted by the computer hardware The instruction consists of a mnemonic opcode followed by one or more operands A label field is optional All are in free format i e one or more blanks in between ex LABEL OPCODI CJ OPERAND S COMMENT separated by commas E 2 Assembler Instructions or pseudo instructions or pseudo ops These are instructions to the assembler Examples are reserving storage DS reserving stor
229. data management two types with SVC calls regs 0 1 15 destroyed often and without 0 1 14 15 usually wiped need save area JOB MANAGEMENT communication with operator WTL write to log WTO write to operator WTO ROUTCDE 11 write to programmer Program CAL WTOR write to operator and get reply PROGRAM TASK MANAGEMENT Linkage inside one load module 1 RETURN SAVE Linkage inside load module for overlay modules SEGLD begin loading a segment SEGWT make sure segment in Program Linkage between load modules note responsibility count TASKS currently using a load module if 0 not needed any more LOAD bring to memory CNT CNT 1 DELETE CNT CNT 1 remove if desired LINK bring to memoy CNT CNT 1 pass control XCT ask Cre Task is Control CALL between load modules L CNT calling module CNT 1 bring called module to memory if needed its CNT CNT 1 pass control to it no return BRANCH between modules ation and Management basic resource allocation unit each has a Task Block and can compete for resources EAch job has JOB S EP TASK and 0 or moreSUBTASKS in tree structure ATTACH create a task DETACH destroy task terminate processing ABEND abnormal end of task CHAP change priority of a task EXTR
230. deck setup is JOB CARD INCLUDE JRMO2 PRINT substitute PUNCH if desired INCLUDE userid filename 1 or more cards like this aa to terminate input The BAT files generallv have about 50 lines per page with a limit of 500 lines in any one file Some writeups consist of several BAT files together while other BAT files contain several distinct writeups CS411GI1 06 NAME PAGES SOURCE DESCRIPTION ASBROPS2 3 BAT assignment using the ASSIST REplace Monitor each student replaces the base register table portion of ASSIST exercies in table search or linked list manipulation USING DROP address conversions ASPRGTCI 8 BAT gives S 360 Assembler programming hints how to use modules macros and combined forms how to set up safe nondestructive linkage to a module ASREPLGD 11 BAT ASSIST REPLACEMENT USER S GUIDE describes in general terms how to use the ASSIST Replace Monitor CS411AS1 2 BAT assignment linkage between FORTRAN and Assembler linkage conventions argument passing PARM field access CS411GI1 10 BAT the writeup you are looking at CS411GI2 BAT CS411MC1 8 BAT two assignments on macro writing pages 01 03 have one to write own version of CALL SAVE RETURN macros and test them pages 04 08 have combined macro module writing hexadecimal conversions and dumping various macro features illustrated CS411MC2 6 BAT assignment write package of macros to manipulat
231. ded in one contiguous block it may be moved around in memory The large S 411 s may contain special page translation hardware which permits the program to be loaded as many noncontiguous pieces and moved around as desired It should be noted that OS 411 uses a STATIC allocation scheme i e a JOB requests memory only 1 time when it is initiated It does NOT request and return memory areas while executing Some systems us a DYNAMIC allocation scheme in which a program can request an area of a given size and be supplied with the address of such an area use it then return it to the operating system later Some systems combine both of these methods i e they allocate an area of memory when the JOB is INITIATED but allow programs to allocate de allocate space within that area OS 360 is an example of such a method Finally various methods exist for determining which of several blocks of unused memory should be allocated to a requestor Each of the methods has advantages and disadvantages both in implementation difficulties and in statistical properties je The next pages describe SOME of the possible options CS411FP2 02 MEMORY DESCRIPTION VALUES The following SET variables describe the various parameters of memory allocation amp MEMSIZE amp MEMOS amp MEMPGR amp JOBLIM1 amp JOBLIM2 amp JOBLIM3 2 1024 also mentioned previously
232. e 2 If required print out a report giving statistics describing the entire previous BATCH of JOBS watch out for case where there were 0 JOBS in the BATCH as when SSCLEAR card is first one in deck 3 Reset all necessary counters lists work queues to such status as to allow acceptance and simulation of another BATCH of cards I e OS 411 must be SERIALLY REUSABLE Es SSQUIT CARD FINISH ENTIRE RUN This card indicates that no more BATCHes of JOBS follow i e so that this functions like the SSCLEAR except that no more BATCHes are processed NOTE an END FILE INDICATION SHOULD BE TREATED AS A SSQUIT CARD i e if a read finds nothing there it should take this action D SSDEBUG CARD DEBUG ACTION CARD This card has the following format SSDEBUG number number This card is mainly for debugging use and supplies the various numbers to whatever debug counters flags you may wish to use This may be particularly useful to turn on off trace and debug output CS411FP1 07 E a IV JOB INPUT AND SCHEDULING TECHNIOU Briefly when OS 411 reads a JOB card it scans it evaluates the various parameters on it and makes up an entry for it on the list of JOBS waiting on disk until they can be executed The list is generally kept in the order in which the JOBS should be executed if possible The options following describe HOW O
233. e one way linked lists each macro is fairly easy CS411FP1 8 BAT assignment write simulation of typical batch CS411FP2 8 BAT multiprogramming computer system Many possible CS411GP3 8 BAT combinations of scheduling allocation resources are CS411FP4 7 BAT available Uses almost all features in BAL CS411FP5 6 DTO example flowcharts for this project DOCUMENT 4 BAT S 360 Assembler Language documentation hints and good practices DSECT 3 BAT example of use of dsects to trace save areas also showing assembly listing of program DUMPSJCL 3 BAT gives basic Job Control Language cards for running most typical assembler programs gives several useful hints on special JCL available at PSU gives programs to be run to obtain representative system completion dumps NAME PAG ES SOURCI CJ CS411GI1 07 DESCRIPTION DUMP1 DUMP 2 HARDWAR1 LINKAGI E OSHASP XDUMP XGET XHEXI XREAD DTO DTO BAT BAT BAT BAT BAT BAT BAT debugging and dump reading hints for assembler programmers concentrates more on use with Link Editor but has some material with Loader Gives examples of hunting down causes of errors should be upgraded and rewritten describes most devices currently part of PSU CC 360 67 system gives device addresses speeds capacities channel priorities Pa
234. e illustrates how to use the address list passed through register 1 LA 1 ARGLIST get argument list address L 15 V CALLRIN get entry address BALR 14 15 call routine ARGLIST DC A ARG1 DC A ARG2 DC X 80 AL3 ARGn Note the length factor does not provide auto matic alignment CALLRTN CSECT L 2 0 T get addr of next arg LTR 1 1 last arg in list BM RETURN if yes return LA 1 4 1 else get addr of next arg When a programmer receives control from the system information from the PARM field of his EXEC card is passed via register 1 Register 1 points to a fullword of storage Bit 0 of this fullword is set to 1 to indicate the last only argument of the list This fullword contains the address of a halfword The halfword is a count of the number of characters in the parm field message and these characters follow immediately after the halfword count field The contents of the halfword may be picked up to use as a length count in an execute instruction and the address of the halfword may be used as a base to move the information characters of the PARM field REGISTER 13 is called the save area register It contains the add ress of an 18 fullword area on a f w boundary within the calling routine The routine called will use this area to save the contents of registers to be able to return the registers intact to the calling program This save area has a set format Word 1 Used by PL I and FORTRAN Word 2 address of the save area used
235. e job s INPUT QUEUE ntry is enqueued in priority order with other jobs awaiting execution DIO O tabs When all of the cards for a job have been read it has in effect been split up into the following 1 INPUT QUEUE ENTRIES in priority order in a special system data set used only for work queu ntries referred to as SYSJOBQUE 2 INPUT STREAM DATA SETS placed on DASD using normal OS 360 Direct Access Device Storage Management DADSM routines NOTE DADSM routines are themselves kept on DASD nonresident and allocating disk space often requires a fair number of accesses to disk to look for free space on one and to allocate the space appropriately The DADSM routines are quite general and powerful but also create some overhead B INITIATING JOB STEPS The operator may START one or more INITIATORS each of which can initiate jobs from one or more classes categories of Jobs Each initiator will then attempt to initiate the highest priority job from the first class of jobs which has a ready job If there are no jobs awaiting execution in its allowed classes it WAITs for one to become available Note that it essentially removes input queue entries from SYSJOBQUE Like every RDR each INITIATOR is executed as a separate task INITIATOR may be abbreviated INIT OSHASP 04 When an allowable job becomes available the in
236. e should LINKAGE 5 RN can provide added B Both can generate exit from a module gister or to print a a save area NOTE r 13 still points to used to save regis XRETURN SA will suf nt however XRETURN one and at that one his is that SA will be created per module For further details on the parameters involved in these two macros see the appropriate PSU documentation The following example causes register 12 to base register causes all registers to be saved trace messages to be printed on entry or On exit loaded with the return code value 8 MAIN CSECT be established as a on entry cuases no and causes R15 to be XSAVE BR 12 TR NO Note default is for all XRETURN SA TR NO RC 8 registers to be saved CMPSC 411 DSECT Example PRINT NOGEN EQUREGS MAIN CSECT XSAVE ESTAB H STANDARD LINKAGE CALL NEXT CALL LOWER ROUTINE XRETURN SA ESTABLISH SAVE AREA LTORG
237. e user program on the simulated SIGMA 4 5 There may be any number of blanks before and after jobname i e these cards are FREE FORMAT 2 CARD INDICATES END OF A JOB This card is of following format beginning in column 1 SR rest of card is ignored 3 OVERALL DECK SETUP The input to TRIVIAL is made up of 1 or more JOBS each as follows SSJOB jobname 1 instructions limit for job 1 poe TRIVIAL assembly program END card showing end of assembly program al t e 0 or more data cards to be read by program Note that any user program cannot be allowed to read beyond a or JOB card into the next user s program Test decks will be supplied to the students CS102FP2 02 TRIVIAL ASSEMBLY LANGUAG CJ The following describes the format of a TRIVIAL assemblv program giving in detail the forms of mnemonic opcodes operands and labels NOTE the reader should probably first consult Part C since it gives the machine code formats used bv the SIGMA 4 5 1 MACHINE T INSTRUCTIONS There are essentially two formats for machine instructions RX format which is similiar to S 360 RX format and RI format operates on a register and an immediate operand field which a RI INSTRUCTIONS These instructions follow the format below label opcod register immediat label is an optional statement label opcode is one of the immedia
238. ed a queue and such as when it finishes rather than in FO1 this method is somewhat rather than giving the le Compared to jobs in memory advantage FOR JOB hese methods compare actual times rather than the T form EO84S SET SH ORT EST E APS amount of tim JOB s EO84R SHORT SRT e a job having lesser time EST R ED TIME order jobs according to the had control of CPU and give CPU to This is very fair nas E ah EMAINING TIM the time give CPU jobs out jobs sitti to of m EO86 ENTS I O REQUIREM For each Job CPU and I O as in random generation systems EO861 EO86C EO86I is preferred estimated to be remaining jobs having emory quick ng there forever since order jobs according to i e T lapsed and least time left This favors getting ly but may result in long running maintain totals of the amount of time used for necessarv onlv if I O activitv is non constant Use this ratio to order favors I O bound jobs favors CPU bound jobs it utilizes CPU and CHANNELS better CS411FP2 07 EO90 TIME SLICING his family of methods attempts to allow short requests to finish quickly without having to know what the actual time needed is Dbefore the job is run It also attempts to allocate CPU time to ALL of the jobs in memory in a fair way It is most o
239. ed from someone else since it aids one in understanding the program quickly It also makes other people much more willing to look at a program PROGRAM MODIFICATIONS Clear and complete documentation is absolutely invaluable when a program must be modified especially if anyone but the original programmer is making the changes It may be noted that useful programs tend to be modified often DOCUMENT 2 CJ ASSEMBLY LANGUAGE DOCUMENTATION The following advantages apply to any computer language However they are most important for assembly language for the following reasons 1 Assembly language programs typically require many more statements than do high level language programs for the same task 2 Assembly language programs are not usually self documenting Without good documentation not even the programmer who wrote the code will be able to understand it several months later 3 Assembly language programs are often very sensitive to minor changes much more so than higher level languages The remainder of this paper describes a well documented assembly program and notes the various techniques which can be used to achieve this result Briefly a well documented program has the following characteristics 1 The documentation structure mirrors the program structure and it leads from the general to the specific Thus the program begins with a block of comments which describes th
240. ed on the left with character 0 s hexadecimal X FO Thus this 8 byte field is the hexadecimal number with leading zeroes The bytes values then can b A E 1059 06 XVCIV X 06 7 XUFOLSX FIE Then TR WORK TAB3 convert Cl C6 to FA FF PACK FULL5 5 WORK 9 do funnv pack WORK DS CL8 C area with number right just d FULL5 DS F C fullword plus wiped out byte TAB3A DC X FAFBFCFDFEFF for converting C A gt X FA etc TAB3 EQU TAB3 C A put theoretical table origin ORG TAB3 C 0 to position for character 0 TAB4 DC C 0123456789 leave digits alone Note what S OCCURRING ABOVE since UNPK generally converts each pair of bytes into one byte it is likely we would want to use it for a conversion in this direction Unfortunately it also reverses the last two nibbl the extra nint then each of t the second nib NOTE if the statements ru es from the source field But if we pack 9 h byte is reversed and placed into the extra he other bytes has the first nibble X F bles packed together into a 4 byte area ORG amp EQU games above make little sense try nning them with ASSIST B QHEXO CONVERT VALUE IN REGISTER TO HEXADECIMAL 1 REQUIRED FUNCTION OF MACRO This is like XDECO but for hexadecimal instead label QHEXO register address label register and address are as described above Calling QHEXO should cause the following to occur a The value in the regi
241. el routine will have no called routines It is important to know the conventions on save areas but the use of XSAVE AND XRETURN consult appropriate documentation can reduce the problems in coding and linking save areas THE NAME CONVENTION is a means of having the EBCDIC form of the name of a routine appear at certain key places on dumps To use this convention the first four bytes of a routine must be a branch on 15 as a base register which passes over a series of bytes These bytes contain the EBCDIC form of the name of a routine and also a length count for this name area This example shows how to code a name field name CSECT B m 1 4 15 DC X m DC CLm name next instruction The value of m must be odd in order to have the next instruction properly aligned An alternate approach uses the convention on register 15 name CSECT USING name 15 B NEXTINST DC X m DC CLm name NEXTINST next instruction LINKAGI ti l ws Notes O S follows these conventions strongly In particular the system often destroys the contents of registers 0 1 14 and 15 when it returns control from a system macro an SVC or another system function One must SAVE THE CONTENTS of these registers BEFORE exec uting one of these functions hard to locat rrors will frequently occur after failure to do so It is a good idea to mark a save area upon exit This is usually done be moving X FF into the first byte
242. empty lists LISTA of which contains KEY and DATA for 20 After each card is read obtain an empty ill it with the KEY and DATA just read and use a DSECT to re STA Dump lists FREE L r the same KEY val 5 the NODE 6 T back onto the FRE fer to these fields at this point then enter ISTA and Read in 5 cards each with a KI EY value on columns 1 8 Search ue If not found print a message If found Peform same action 7 Show macro expansi to display Demonstrate t USE THE FOLLOWING SEO CCCCCCCC ZZZZZZZZ set NOTFOUND as in 5 hat they UENCE OF t POP each node of LISTA pri E list u ons for a hen push it onto beginning of LISTB nt each as it is obtained then place ntil LISTA is empty but for LISTB ny other special cases you may wish if you feel like it AAAAAAAA DDDDDD XXXXXXXX GGGGGG to be moved from ZZZZZZZZ AAAAAAAA AAAAAAAA DD FFFFFFFF GG KKKKK ES AS TEST DATA BBBBB y LISTA to CS411TPA 01 COMPUTER SCIENCE 411 TOPICS COVERED HANDOUTS WINTER TERM 1972 MASHEY The handouts given out are are described in file CS411HN DATE topics handouts 72 01 06 introduction to course covering prerequisistes 101 102 404 or eq
243. equiring initialization or termination functions different from the normal calling function ltiple e NAM CS ntry CSI ECT ECT is typically s ENTRY ENTRY1 ENTRY2 1 ENTRY1 for e ntered o LINKAGE ntry at CSECTNAM nly at the entry E CODE R executable code w ETURN LIN SAVE XSAVE hen called at KAGE CODE RI E CSECTS entry E HE DIFE ERI ENT ining entrypoint nal subroutines data areas used by names and needed by various of he following are important points to remember when using ENTRY POINTS N EV ER CALL of the routi ETUR ASPRGTC1 03 et up as follows ENTRYN multiple entry routines at the often points not CSECT etc ENTRY1 N XR ETURN etc code more than one entry point the entry point routines multipl EACH OTH ER In essence all level in cal ONE F SAVE ONLY nes represented by the various entry points are at the ling structure of an entire program ARI EED EA IS ACTUALLY N the end can refer to it for the later ones constants mu the SA operand is CSEC ah never call each other of the LAST section of code note that if placed o st be used instead ED same Since the routin
244. erforms various housekeeping actions such as checking and setting the timer to make sure it is working correctly initializing some pointers for storage management initializing the SVC table which gives a pointer to each routine associated with a defined SVC number It also sets up to be able to obtain modules from the SYS1 LINKLIB which contains the heaviest used load modules for the system and also establishes communications with the operator 5 For any system having one NIP loads reentrant modules into the LINK PACK These modules can be used during following execution and are located at the high end of memory In a system with fast core LCS the LINK PACK can be split residing at both the high end of fast core and the high end of LCS 6 With the addition of various other miscellaneous operations NIP prepares a REGION which will contain the MASTER SCHEDULER which is the program doing overall job scheduling and operator communication It then can pass control LINK or XCTL to the MASTER SCHEDULER and the system is finally ready to run jobs At this point memory layout fast core only is as follows HIGH ADDRESS LINK PACK reentrant modules MS MASTER SCHEDULER FREE AREA dynamic for problem programs SOS SYSTEM QUEU SPACE contains space for system control blocks TCB s etc LOW ADDRESS NUCLEUS NOTE in systems w
245. ero length Either print the PARM field or the message NO PARM EXISTS Assume first character of PARM is a legitamate carriage control B Make same two calls on MAXIM The first must be hand coded no macros while the second uses the IBM macro CALL IV JOB CONTROL LANGUAGE Do not punch an object deck until you are sure the MAXIM program is correct While testing you can use the following JCL EXEC ASGC SOURCE INPUT DD MAXIM program EXEC FGCG PARM DATA MAP SOURCE INPUT DD Ge ies main program ed should do a CALL LETDMP before using MAXIM DATA SYSUDUMP DD SYSOUT A may need XSNAPOUT card following also CS411AS1 02 An alternate form of the preceding is to use EXEC FGC SOURCE INPUT DD ve FORTRAN main program EXEC ASGCG PARM DATA MAP SOURCE INPUT DD intron MAXIM DATA SYSUDUMP DD SYSOUT A When testing MAXIM with the assembler program you may either run it and MAXIM as one assembly 1 ASGCG or as two 1 ASGC and 1 ASGCG V WHAT TO HAND IN A Run a job which produces the object deck for MAXIM and also does the test using the assembler main program note that the END card in the second assembly should specify the name of the main program on it so that execution will start there Use the following deck setup EXEC ASGC PARM DECK SOURCE INPUT DD ve
246. ers to the data into the INPUT UEUE entries so that it can be found later The job s INPUT QUEUE ntry is enqueued in priority order with other jobs awaiting execution DIO O tabs When all of the cards for a job have been read it has in effect been split up into the following 1 INPUT QUEUE ENTRIES in priority order in a special system data set used only for work queu ntries referred to as SYSJOBQUE 2 INPUT STREAM DATA SETS placed on DASD using normal 0S 360 Direct Access Device Storage Management DADSM routines NOTE DADSM routines are themselves kept on DASD nonresident and allocating disk space often requires a fair number of accesses to disk to look for free space on one and to allocate the space appropriately The DADSM routines are quite general and powerful but also create some overhead B INITIATING JOB STEPS The operator may START one or more INITIATORs each of which can initiate jobs from one or more classes categories of Jobs Each initiator will then attempt to initiate the highest priority job from the first class of jobs which has a ready job If there are no jobs awaiting execution in its allowed classes it WAITs for one to become available Note that it essentially removes input queue entries from SYSJOBQUE Like every RDR each INITIATOR is executed as a separate task INITIATOR may be abbreviated INIT
247. es inside the so t user hat a n th using With n the to access it coded only o LAS E ST BE CARE MU dE AKEN WITH AD DRI ESSIBILIT can of course addr ss the data er must th he programm rites becaus be very careful with BAS REGISTERS USED E TO XSAVE XRETURN PE areas at the end of the CS any can code th area at the 11 of the previous sections e first it would be difficult a LA instruction address this means that sav XRI E TURN All of the code sections ECT However subroutines he RNAL SUBROUTINES internal ASSEMBL IN ry E THE UST HAVE CORR ry f E CT VALUES IN T EX a WHEN THEY AR ry CALLI ECUTION TIM He IF THEY DON T HE E E ED FROM DIFF CJ UPS HEY WILL ASSE ECTIONS HAVING DIFFERENT USING RI BLE PROPERLY F F F E PARTICULA EX operatio THE R n U Hap lt z C WITH T EN BL UP A H ACE US PROGRAMMER SHOUL HE SECTION OF COD CJ ah ATA AREAS IF HEY T ING ER EXECUTION IM TO BE EXECU NOT AT WITH HEHE INS IONS AND ar E PERFORM ANY SYMBOLIC A all entry poi
248. es under what condition each i EFINITION AND CALLING TRUCTIONS and either placed at the b a USER macro or entered into When called it generates 0 or more nvocation and th cod 01 coding o write y parts of doing This and best ges s eginning a macro assembly nerated g the P ob th comma whic de are n begins with MACRO the name of the macro The body EL STATEMENTS which are assembler nerated and macro operations n processing of the macro The macro END statement The following steps followed by MOD in defining a macro DETERMINE a a good idea to write at least some of the together as a code b DECIDE ON E ERATED CODE statements to get some feel for EN Tt to what BASIC PURPOSE AND Gi is g be g is segment first ENTS AND THEIR USAGE N EC ESSARY ARGUMI it may idea to write the purpose of each argument in the operand list and include it in a block of comme nts at the beginning of the This hel Use POSITIONA argument MUS of positionals it is much more co Use KEYWORD argume ones which are conveniently supplied with default values whic often used Use S c WRIT AROUND THE MO E DE lps the macro to be done L operands for heavily used T be supplied every time place the most heavily used ones near th
249. ffers somewhat better setup facilities for optimizing use of tapes and non SPOOL disks A disadvantage is the requirement of two CPU s either of which may have problems and thus stop th ntire system B LASP LOCAL ASP or CLASP CLOSELY LINKED ASP These are versions of ASP in which the code from the smaller computer is moved over into a region on the larger machine This allows an ASP system to be run on one processor If the system is also run under straight ASP it requires switches to switch the unit record devices over to the bigger machine It also requires more memory than HASP but does allow the system to run even with one CPU down C TUCC HYPERDISK his method uses LCS plus part of a 2314 disk pack to simulate the entire disk pack containing heavy used systems programs The most recently used tracks of this disk are kept in LCS thus making the disk effectively faster without changing the internals of 0S 360 Mio PSU CE OSHASP 09 360 67 SYSTEM OS MVT WITH HASP The following tables gi e the current layout with no guarantee of future appearance as of 6 12 72 for the 360 67 at the PSU CC The system has both fast core 1024K and Large Core Storage 2048K LOW HIGH K LOW HIGH S 2928 3072 144 2DC000 300000 HASP 1968 2928 960 1EC000 2DC000 FMGR 1628 1968 340 197000 1EC000 RJE 1346 1628 282 150800 197000 WATFOR 1336 1346 10 14E
250. fficult to read as a text but REFERENCE This manual completely defines JCL examples of JCL usage It is is valuable as a reference This manual is a rewritten version of the combination of the two fol redundant information although this is the best acceptable OR D lt JCLR1 gt Llowing manuals Therefore GC28 6704 1 IBM S 360 OS JOB CONTROL LANGUAG REFERENCE which together contain much any one of these three manuals are CJ Although mostly like lt JCLR gt this contains fewer examples OR D lt JCLUG gt This R lt GREEN gt 8 This a compact assembler radix conversions assignments form R lt ASSIST gt 9 instructions formats of most common CCW opcodes GC28 6703 GX20 1703 9 IBM 8 360 R is the GREEN CARD and contains including information on condition code setting PSW CAW CCW CSW permanent storage instructions and IB USER S GUIDE S 360 OS JOB CONTROL LANGUAG CJ version leans more towards examples rather than a reference EFERENCE ED LT ASSIST INTRODUCTORY ASSEMBLER USER S GUIDE DATA much useful information in DMK patterns interrupt codes contants EBCDIC bytes This describes u version 1 0 is aval sage of t lable ne in the an updated version may be availabl CMPSC office e on BAT files ASSIST a
251. following example will cause registers 5 6 10 and 14 and 15 to be saved SAVE 5 10 T The RETURN macro will generate code to restore registers insert a ret urn code in register 15 flag the save area X FF in wd 4 and branch back via register 14 The registers to b restored ar coded as with SAVE If 15 already has a return code in it and should not be restored it is coded as RC 15 else RC n may be coded where n is some value to insert into register 15 The operand T causes the flag X FF to be inserted in the save area The following code will restore registers 5 6 10 to be reloaded the save area to be flagged and 15 to be loaded with a value 16 RETURN 5 10 T RC 16 NOTE Both of these macros expect that register 13 will already be loaded with the address of the appropriate save area The use of the PSU macros XSAVE and XRETU flexibility in saving and restoring registers code to print a trace message showing entry and XSAVE can be used to establish and load a base re snap of the registers saved XRETURN can create that as with RETURN XRETURN assumes that registe the relevant save area For most uses the code XSAVE alone can be ters For a routine with only one return point fice if a routine has more than one return poi alone should be coded at all return points except XRETURN SA should be coded The reason for t cause a save area to be created only on
252. ften used for interactive and Time Sharing systems where many users all wish to obtain very fast or short computations but occasionaly have longer requests EO9ORR ROUND ROBIN when a job reaches the head of the RJQ it is given control of a CPU for a maximum of amp TIMSLIC millisec or until it waits for I O In either case the next job in the list is given control of the CPU and the job just removed from control of it enters the RJQ at the end When a job completes it is removed from memory Each slice of time is called a QUANTUM Note that a job requiring only 1 QUANTUM will enter execute and be completed quickly while those needing more must stay longer This is very commonly used paticularly in Time Sharing operating systems is simple but can be catastrophically bad if many jobs want to use the CPU and if the jobs are only partially core resident Many other variations exist on this scheme Che PRI Ej EMPTION When a job first enters the RJQ or when it completes I O and thus becomes eligible again for use of a CPU it enters the RJQ in the appropriate place If the job enters the RJQ in the earliest place either one of two actions may occur 1 The job may wait until a currently executing job releases a CPU by either requesting I O or terminating execution entirely 2 If the job has effective priority or order such that it would rank ahead of a currently executing job if
253. g up the usual save area linkage This name will be saved in a GBLC variable for later use by QRETURN Defaults to SA The OSAVE field as described in macro wil the LINKAG ry WRIT 1 also automatically generate an identification EUP This identification field will us ither the lab current CSECT name the minimum storage needed i e B 14 15 DC X 9 CLJ name on the B TURN return from a subr Q T argume his macro is similiar to XR nts REGS regl reg2 SA REGS regl reg2 register SA value a save area SA NO SA and works SA name exactly QSAVI if there is no label on the OSAVE E statement or will use the It will generate E it cannot just generate outine value s to b as ETURN or RETURN will and accept as Default REGS 14 12 restored controls save area linkage and generation of does and omitted operand XRI it ETURN Thus accepts CS411MC1 02 C QCALL call a subroutine This macro can be written in any of the following forms 1 label QCALL argl arg2 2 label QCALL entryname 3 label QCALL entryname address list name 4 label QCALL entry name argl arg2 Version I merely generates a list of addresses of each of the arguments with the last one flagged appropirately to show that fact Version 2 loads
254. ge 5 is only DTO and it contains diagram of 360 67 layout explanation of OS 360 standard linkage conventions used by FORTRAN Assembler etc programs for entry exit argument passing 0S 360 with HASP explains how OS 360 is loaded into memory IPL NIP how it runs and how it is modified by the use of HASP describes use of XDUMP and XSNAP debugging macros XDUMP is simple form used in ASSIST XSNAP is more complex IRMO4 XDUMP describes use of generalized XGET XPUT 1 0 macros describes use of XHEXI XHEXO hexadecimal conversion macros available in ASSIST JRMO4 XHEXI describes XREAD XPRNT XPNCH I O macros available in ASSIST JRMO4 XREAD describes XSAVE and XRETURN linkage macros available in ASSIST only if using the macro library JRMO4 XSAVE In addition to the above writeups the students will normally purchase a copy of the ASSIST Introductory Assembler User s Guide The very latest copy of this manual is available in the following JRMO4 ASSIST ASSIST ASSIST ASSIST ASSIST ASSIST ASSIST ASSIST 1 TIA 2 T2A 3 T3A 4 T4A BAT files 14 BAT 8 BAT BAT 9 BAT 17 BAT Part I describes language availabel in ASSIST Part II describes debug and I O instructions Part III control cards parameter options Part IV output format error messages loader IV BAT FILES CS411GI1 08 The following lists BAT files which are
255. grams ASPRGTCI 01 S 360 Assembler Language Programming Techniques John R Mashey Winter 1972 Topic Program Modularity and Parametrization Methods Using Macros Internal Subroutine External Subroutines This topic pages 01 08 It is generally important in any computer program to avoid coding any procedure more times than necessary It is generally best to write something one time then have it available for later use in many parts of a program In assembler language there are thr main ways of doing this macros internal subroutines and external subroutines This writeup describes each of these techniques gives the advantages and disadvantages of each and notes under what condition each is best 1 DESCRIPTION DEFINITION AND CALLING A MACRO INSTRUCTIONS A macro instruction is defined and either placed at the beginning of an assembly language program a USER macro or entered into a macro library a SYSTEM macro When called it generates 0 or more assembly language statements at the point of invocation and th cod generated may vary greatly from call to call 1 DEFINITION A macro definition begins with MACRO followed by the PROTOTYPE STATEMENT which gives the name of the macro The body of the macro includes 0 or more MODEL STATEMENTS which are assembler commands and machine instructions to be generated and macro operations which serve to direct the expansion
256. gt fetch gt executable program object module parts ESD TXT RLD END and purposes of each usage of loader and waht it does HANDOUTS CS411FP5 01 04 FP flowcharts 19 02 15 20 02 17 21 02 22 MODULES 01 09 19 object load module management link editor load modules overlays differences between loader and link editor object modules and load modules advantages and disadvantages overlays concepts trees commands to set them up briefly on implementation of them options to be used on loader and link editor user overview of OS 360 services program managment and design data managment job management task management program structures simple planned overlay dynamic serial amp dynamic parallel management of resources job task and data data management types of data sets SDS Direct PDS IS and what they are used for overlay methods input output concepts and record formats go over OVLY1 and OVLY2 programs pointing out size reductions possible and the control cards used note PROC on OVLY2 I O concepts buffer groups flip flop buffers record formats F FB V VB VBS U and comparisons of LRECL BLKSIZE records block fficiency ase of use outline of rest of term READINGS first 10 pages in DATA MANAGEMENT SERVICES ASSIGN run QSAM and EXCP files for next time CS411TPA 06 22 02 22 0S 360 macros all except
257. h programming and computing time Good documentation is especially necessary for programming projects requiring either a long period of time by one programmer any period of time by more than one programmer or modifications to any code by anyone other than the original author Good documentation techniques can be helpful in the following ways PROGRAM DESIGN Many beginning programmers seem to write programs in haphazard and unplanned ways and often add comments only after the program is running This method not only leads to poorly structured programs but usually results in wasted time and is not feasibl xcept for relatively trivial problems A much better method is to write most of the overall comments with a flow chart first specifying the structure and conventions of the program and then writing the program to fit This usually leads to cleaner coded well structured programs which are produced in less time than those written by most novice programmers PROGRAM DEBUGGING Program debugging is aided by documenting a program before and during its creation rather than afterward Many mistakes can be avoided by having programming conventions well specified befor writing the code The very act of adding a comment to a statement often helps identify errors in the statement because it forces the programmer to think about the function of the statement Finally good documentation is useful if help is requir
258. hat time from JOB card IOIN the job actually gives up the CPU if the time remaining in which case it must obtain a channel for this action When away channel if it is free for a desired channel at which time an I O INT is the time at which the I O is is ever allowed for use of a CHANN the 1 0 REQU EST occurs or else must enter a queue if requests waiting it will be able to use the RRUPT must b completed Sooner or later the job lated system are not actually performing ve control of a CPU by OS 411 ED In general Pime Slicing methods 0 B T EL as and an I O the interval or computed from it it either TERMINATES and or else makes an ither uses a channel right scheduled 1 ep this In general no PRE T EMPTION it would be disastrous to interrupt the use of a CHANNEL When job releas an I O INTER es the C RUPT HANNEL being used occurs more compete with other jobs for use of a CPU A The 1 just befor IOIC IOIJ IOIR IOIRU I O INTE E ength of p e the CE Constant interval RVAL CALCULATIONS he job SYSGEN from the number of mil Fixed interval from JOB card the next I O REQU lliseconds unti F the JOB card Random Intervals using Job Card information as follows
259. he names numbers of 2 registers which can safely be used as temporary work registers and destroyed by the macro gives a number or EQU symbol of the offset in bytes from the beginning of the NODE to the KEY field in the NODE is the length number or EQU value of the KEY field that the NODE to be inserted is inserted AFTER any NODES which the SAME KEY value SEARCH QSRCH R10 HDR1 SEARCH LA RWK1 HDRI 000002AA LR RWK2 RWK1 L RWK1 LINK RWK1 LTR RWK1 RWK1 BZ 14 CLC A KEYL RWK1 4 R10 BNH Q00002AA ST RWK1 LINK R10 ii ST R10 LINK RWK2 OSNAP PRINT CONTENTS OF LIST amp LABEL QSNAP amp HDR amp MSG amp RGW RWK1 RWK2 amp COUNT 4095 amp LEN 20 This macro dumps the list beginning at the HEADER amp HDR using XSNAP if desired with a message amp MSG printed The work registers needed ar given by amp RGW and up to amp COUNT NODES are printed amp HDR amp MSG amp RGW amp COUNT amp LI EN DUMP DUMP 0000 0000 is name of LIST HEADER is quoted string used as title for list output specifies names numbers of 2 registers which may be erased specifies the maximum number of nodes to be printed is an absolute expression giving node length in bytes QSNAP HDRI LIST 1 COUNT 4 LEN NODELEN LA RWK1 4 LA RWK2 HDR1
260. holds 4 once each 17 5 milliseconds time for 83 TRANSF hig ER Jobs disk usage 8 units etc of 8 6 milliseconds h transfer rate bytes maximum size RATE THIS These hol 2314 2319 disk drives SYSTEM DIRECT ACCESS STORAGE DEVICES DRUMS 2 drums 09 megabytes million bytes 1 2 megabytes second HARDWAR1 FAST 03 printed here requiring and they can enter commands these display current and also are used to et DASDs attached to channel 1 via 2820 of data with average rotational del Records data 4 bits i Has 200 conceptual TRAC EACH DRUM IS UN rotates ay n KS REMOVABLE EST DEVICES ld most heavily used compilers and system programs MAGNETIC DISK STORAGE including 2 spare ones 2316 DISK PACK OTE ne ne time without moving t ecord at most unlike drums mounted in is place if desired each Each CY contains the 20 TRACKS accessible at he REA 7294 bytes of i DISK PACK ca FACI ITI ES 29 17 megabytes maximum 11 plates outside ones not used to access information usable at any one time INDER D WRITE H Each Uses of w nformatio O ATION K TIM E N O R S E LT ES TIME time t IN AX IMUM NOTE 230 237 25
261. if code must be used many times by a CSECT then the faster linkage of internal subroutines usually makes it worth writing it that way CODE NEEDED ONLY BY ONE CSECT if not too long it is fairly logical to incorporate it as part of that CSECT It will probably be much more efficient since it will have access to the internal variables of the CSECT and be able to communicate via register values easily rather than requiring long operand lists Ex EXTERNAL SUBROUTINES LONG CODE if something is long and complex enough it may be a good idea to make a separate module of it test it get an object deck then leave it along thereafter CODE OF GENERAL USE NEEDED MANY PLACES in this case it is practically necessary to make code an external subroutine so that it can be accessed where needed D COMBINED FORMS These are useful anywhere the others ar The nondestructive form is specially useful if it is to De used by beginning programmers DOCUMENT 1 CJ S 360 Assembler Language Documentation and Listing Techniques bv John R Mashev and Andrea Rhodes Goals of Good Documentation 1 Aid in designing good programs 2 Aid in debugging programs 3 Make programs clear and understandable once written 4 Make structure of program well organized Good documentation is a great aid to producing clear well written and understandable programs and can save muc
262. ing on the needs of the JOB it either gains control of a CHANNEL and uses it to perform its IO immediately or else it enters a queue of JOBS competing for the use of CHANNELS In the latter case it may have to wait a while until it is selected to use the CHANNEL it needs It is said to be in WAIT STATE at any time during which it is READY to use a CPU or CHANN EL but cannot obtain what it needs CS411FP1 03 When a JOB is finished using a CHANNEL an IO INTERRUPT is said to occur This means that the JOB relinquishes control of the CHANNEL becomes READY to use a CPU and thus enters the list of such JOBS once more competing with the others The CHANNEL thus released of course becomes available for use by other JOBS again and may be used to fill a request from some previous JOB which is WAITING for the CHANNEL Among the values found on the JOB card is a time limit for the execution of the JOB When this limit is reached the JOB is considered to be finished It is then JOBTERMED i e it is removed from the 0 list of JOBS competing for CPU usage the memory it occupies is freed for possible allocation to other JOB S i e the ones waiting on disk to obtain memory At various times calculations are made to determine some values indicating the performance and nature of the particular operating system version being generated Typical ti
263. ion entry point of the routine might be either a CSECT name name i e one CSECT might contain several entrypoints supporting subroutine needed entrypoint label definition CSECT or label DS 0H USING entrypoint 15 initial base register and exit the Note that the or an ENTRY one for each save all registers which may be modified by code Save into THIS CSECT unlike normal 0S 360 conventions DO NOT SAVE INTO CALLER S SAVE AREA since it may not exist especially if caller is a lowest level routine initialization code if this routine performs I O or calls any others or requests any supervisor services it is a good idea to set up another base register than 15 set up a save area and put its address into register 13 since any of the above actions may result in registers being saved at wherever 13 points processing code to perform required actions result return code result may be left in register 0 in which case it should not be restored neither here nor in generated code before i e change LM 14 0 4 14 to LM 14 15 4 14 and STM likewise register restoration restore all registers modified in this routine Especially restore 14 SPM 14 restore original condition code note that calling BALR 14 15 saved it B number 14 branch to displacement number beyond address in 14 enough to pass control to statement LM 14 0 4 14 It
264. irable to incl end of the CS ECT in orde s may modify inspect the same lists It may ude the actual list nodes themselves near th r to get all important items in one area In addition to the above CS411FP4 04 the following are GLOBAL VARIABLES which might be included in the GCTB INCORE LASTIT LAGS EBUGS ULTIP HINTS XSNAP they are C 2 This free list should be dsect for EQSECT EQLINK EQTIME EQDATA EQTYPE 3 A UJ There mig DS DS DS DS DS DS quick F F F F clock current time initialized to 0 number of jobs currently in memory init 0 time last initiation or termination occurred useful flag byte for such items as whether SSCLEAR found SSQUIT etc debug control byte could be used with XSNAP IF to control debug output from input SSDEBUG cards multiprogramming statistic see report R9 group important variables so they can be dumped with a single nsider bracketing them with character DC constants so that ly located in a dump EV EN QU FUE is the most important single queue header of empty nodes generated having header cell EQHD EQFREE and of which amp EQ total nodes 100 unless otherwise necessary A use amp EQ each node might look like DSECT DS DS DS DS 33 jp link to
265. irements in class Note that all options are listed with underlines where things must be filled in Also note that not all options may be applicable since some are needed only if some particular previous choice was made The heading numbers correspond to the numbers of sections in this writeup Space is left for comments LE amp MEMSIZE K bytes CPU amp NUMCPUS only if variable CHN amp NUMCHNS only if variabl III amp MAXJOBS jobs at most in system PARM _ if PARM2 then need values for following amp TDFT amp SPDFT amp IOINDFT amp IORLDFT amp PRIODFT amp CHANDFT amp CATDFT IV RDR amp RDRINT millisec if RDR1 JO JSR amp JSRANGE if variable amp MEMSIZE K bytes repeated from II amp MEMOS K bytes amp MEMPGR K bytes amp JOBLIMI jobs amp JOBLIM2 amp JOBLIM3 calculated by SETA arithmetic MAA _ MAC MAM amp MAMOVE 1f MAC2 etc amp INITDEL milliseconds CS411FP4 07 RQ CJ Q Q E lI Il charge for channel use 0 don t ta La Qh td O HU ll a II allow preemption of CPU VI IOI amp IOINTVL if IOLC 100 IOR amp IOREQL if IORC VII Since reports are not mutually exclusive just list ones required below amp RDRINTV millisec if report R2 is required KKKKKKKK NOTES KKKKKKKK
266. is not necessary as there ar often only minor changes between one version of a manual and the next In addition there are combinations of differently named manuals which are equivalent to others In the list below the most desirable manuals are given but equivalents are noted where possible Items coded R are definitely required items coded D are desirable while the remaining ones are useful but can be done without and may not even be available Abbreviations to be used later are given in in brackets for each one NOTE for IBM manuals first six digits show the specific manual while the remaining one s indicate the version Normally but not always manuals having close version numbers are not very different D lt STRUBLE gt 1 STRUBLE ASSEMBLER LANGUAGE PROGRAMMING THE SYSTEM 360 This text will be referenced at most occasionally but contains some more readabl xplanations than some of the manuals below It is also good for review of CMPSC 102 410 R lt ASM gt 2 GC28 6514 8 IBM S 360 OS ASSEMBLER LANGUAG This is heavily used throughout the course By the 5th week of term students will be expected to understand almost everything in this manual while knowing offhand much of it Besides the few sections in the first half not already known to the students the entire second half MACROs will be covered CJ R lt POP gt 3 GC
267. ist of calling conventions for subroutines For extensive programs lists of the following might be kept at the beginning of each subroutine MACROS USED SUBROUTINES H T CALLED BY THIS SUBROUTINE SUBROUTINES WHICH CALL THIS SUBROUTINE VARIABLES USED BY THIS SUBROUTINE VARIABLES CHANGED BY THIS SUBROUTINE etc DO block off large sections of comment cards Large blocks of comments can begin in whatever column is appropriate but in general should use most of the card since they will otherwise add a great deal of length to a program For the sake of appearance comments should be blocked off by blank lines SPACE n or lines of continuous characters The most common characters used for this purpose are asterisks in columns 1 71 or in just the odd columns An esthetic appearance can be obtained by placing an asterisk in column 71 of each comment card in a major block with lines of asterisks before and after the entire block of documentation DO flag instructions which will be modified during execution in order to make programming logic obvious This may be accomplished by using or the latter EQU ed to zero for any modified field For example EQU 0 S gt INST MODIFIED IN EX sein Preus other statements STE 2 MVC 1 SET BUFFER LEN FOR LATER x USE do other statements MVC MVC OUTPUT
268. ith HIARCHY SUPPORT FR LINK PACK would also have areas in LCS LT E AREA MS and OSHASP 03 II RUNNING JOBS IN AN OS 360 SYSTEM This section describes how jobs are run in a standard 0S 360 system using either OS MFT or OS MVT Note that OS PCP runs jobs one at a time sequential scheduling uniprogramming with no SPOOLing of jobs to and from disk before and after execution OS MFT and OS MVT are generally similiar in that they both can SPOOL input onto DASD execute jobs in priority order and write the output out later The main difference is in the handling of storage in which OS MVT is much more dynamic Note that all scheduling of jobs and communication with the operator is effectively under the control of the MASTER SCHEDULER A READING INPUT STREAMS For each existing input stream card reader or input on tape the operator can issue a START RDR command This causes a copy of the READER INTERPRETER program referred to herafter as a RDR to read cardimages from the requested input device During its operation a RDR reads an input stream scans JCL cards nd converts them to a standard internal text form and also obtains ataloged procedure definitions from the procedure library PROCLIB rom the internal text it builds INPUT QUEUE entries representing he information on the user JCL cards It also writes any input data ards onto disk while placing point
269. itiator obtains a REGION for the job from the FREE area also called the DYNAMIC area uses the information from the RDR to allocate DASD storage tape drives and other I O devices It then ATTACHes the first module of the program to be executed thus creating the JOB STEP TASK and WAITs until the job step completes T When a job step is finished the TERMINATOR part of the INITIATOR really so that the whole unit is called an INITIATOR TERMINATOR ffectively cleans up performing disposition of I O devices DISP parameter in JCL and releasing the REGION which had been acquired for the job step During this process job steps are essentially independent i e they could require different sizes of regions and might execute in different locations Note that the INITIATOR TERMINATOR must also control the skipping of steps as controlled by the JCL COND option During execution SYSOUT datasets are written to DASD to be printed punched later When the last job step of a job completes the INIT creates a work queue entry calling for the job s output to be printed punched C WRITING SYSTEM OUTPUT A program called a SYSTEM OUTPUT WRITER WTR can be STARTed by the operator to transcribe output from DASD to printers or punches or even tapes to be printed punched later Output can originally be grouped into CLASSes which can be written according to priority or otherwi
270. ival there In this case they are order according to the FIRST time they entered the RJO i e just after they were INITIATED and thus this order does not change This method favors those jobs which were loaded earliest and thus tends to minimize the longest time a job must remain in memory EO2 STRAIGHT PRIORITV JOBS are ordered from highest prioritv to lowest lowest PRIO to highest values FO3 CATEGORY PROGRESSION ordered by CATEGORY numbers this of course favors the lowest numbered categories EO4 LENGTH OF TIME REQUIRED BY JOB either EO4S SJF SHORTEST JOB FIRST or EO4L LJF ONGEST JOB FIRST Note EO4S tends to expedite the short jobs and thus move jobs in and out faster while EO4L tends to minimize the longest time anything ever has to remain in memory EO5 MEMORY REQUIRED BY JOB EO5S SMF SMALLEST MEMORY FIRST EO5L LMF LARGEST MEMORY FIRST Note that of these two EO5L is usually desirable since it will attempt to get rid of large memory jobs quicker EO6 INPUT OUTPUT REQUIREMENT These orderings are computed according to IOIN and IORL from JOB cards and remain constant regardless of what JOB s actually do EO6I I O BOUND JOBS FIRST EO6C CPU BOUND JOBS FIRST Of these FOGI is usually desirable since it allows JOBs performing much I O to get what little CPU they need then let others use CPU EO7 STATIC ORDI An order number can be initiated and that ERING COMPUTA CS411FP2
271. l element showing which one this job is using if any status byte shows condition of job ready but not executing executing waiting for 1 0 doing 1 0 UJLNODE a typical setup might be AJLNODE DSECT AJLLOAD DS F AJLTLAST DS F AJLTRE DS E AJLTINTR DS F AJLC DS A AJLSTATS DS B 5 INPUT OUTPUT QUEUE Each node represents a request for a channel which cannot be filled until job using it releases it Maximum of amp JOBLIM3 1 of these needed IORNODE DSECT TORLINK DS A IORORDER DS F IORAJL DS A 6 CHANNI Two li one node usage of each one addr of next one in list possible ordering field depending on IOO option address of AJLNODE showing job making request EL CPU ELEMENTS sts or tables depending on convenience might exist with for each CPU CHANNEL mainly used for keeping statistics for Of course they might just be variables in systems having either 1 CPU or 1 CHANNEL only Typical dsect for CHANNEL CPU node would look the same CHNNODE DSECT CHNAJL DS A CHNTLAST DS F CHNUSE DS F addr of AJL of job using this device 0 if IDL last time status of this device changed total accumulated usage time during run CJ CS411FP4 06 X SYSTEM GENERATION OPTIONS LIST The following lists the various options possible for a SYSGEN of OS 411 This chart may be filled in when the instructor goes over the project requ
272. location counter to value specified in operand which may be any of the forms symbol symbol n symbol n n Stn n allowed by first section of RX operand format EXAMPLES ORG ABEL 2 T NOTE may not have a LABE be previously defined field specified and any symbols used must b END signifies end of TRIVIAL program may have no label c RES REServe similiar to S 360 DC and DS reserves storage and may fill in constants It may have any r of the following forms label RES number The statement above reserves number words of storage number being a positive integer from O to 1023 label RES W number This statement causes the decimal number signed unsigned to be converted to binary and assembled at the given location corresponds to S 360 DC F number label RES numberW number The first number gives a duplication factor from 1 1023 and causes that many copies of the constant to be assembled like S 360 DC numberF number 3 MISCELLANEOUS The label field may contain any label usable on S 360 except that are excluded i e a label begins with a letter and then containues with 0 7 letters digits NOTE THAT ALL STATEMENTS ARE FREE FORMAT they may have 1 or more blanks between fields but no blanks inside each field CJ CS102FP2 04 C DESCRIPTION OF THE SIGMA 4 5 COMPUTER SYSTEM
273. locking off major parts of a program The command SPACE n inserts n blank lines into the listing at that point This is useful for blocking off smaller sections of a program particularly small loops register equates etc Not only do listing control instructions aid to the readability of a program but they also save the programmer time in debugging DON T merely restate an instruction when you place a comment on it Of the following two examples which is more explanatory A 1 VAR ADD VAR TO REGISTER 1 A 1 VAR R1 SUMMATION OF ODD PRIME NUMBERS DON T put several single comments between statements in an unreadable manner It is often useful to indent a single comment to column 16 This keeps it from interfering with the reading of labels and opcodes and thus distinguishes it from the machine instructions DOCUMENT 4 DO use comment card blocks which list useful information For example a list of register allocation and usage is extremely helpful not only in debugging but also in revising a program Such a list should appear as part of the preface to the appropriate section of code Another example is a list of calling conventions for subroutines For extensive programs lists of the following might be kept at the beginning of each subroutine MACROS USED SUBROUTINES CALLED BY THIS SUBROUTINE SUBROUTINES WHICH CALL THIS SUBROUTINE
274. mats EQU DC DS TITLE EJECT SPACE PRINT ORG LTORG CNOP END Some students may have done something with DSECTs MACROs and linkage of FORTRAN and Assembler modules but this is not necessarily required It is expected that most incoming students have done most of their programming under ASSIST and are thus not yet proficient in debugging programs and reading completion dumps under 0S 360 directly Students are expected to have written at least half a dozen or more programs in Assembler including typically a small two pass assembler for a simple assembly language B CMPSC 404 or equivalent data structures arrays linked lists tree structures queues stacks perhaps a little on searching and sorting methods hash tables etc Anyone who has taken equivalent courses elsewhere or wishes to substitute other programming experience for the above should contact the instructor immediately to make sure their background is adequate for the course CS411GI1 02 LI H EXTBOOKS AND MANUALS The text materials for the course are listed below Many of the text items are IBM manuals which are often in a continual state of change Particulariv each time IBM offers an updated version of 0S 360 many manuals are modified somewhat It is generally desirable to have the manuals appropriate for the current version of 0S 360 which is Release _ at this time However this
275. mes are the end of a JOB and the end of a BATCH of JOBS The process described above occurs for every JOB in a BATCH Each BATCH is processed until none remain CS411FP5 01 has job flow chart The remaining portions of the writeup describe the configuration of the computer system begin run by OS 411 the major options possible SET VARIABLES which may be needed and many hints on the implementation of this program The last includes general methods ideas on good ways to divide the program into modules sample data structures and DSECTS and overall flowcharts which might be useful Every option is given a mnemonic name of some sort and SET VARIABLES are of course named as they might be inside OS 411 Ki SRS AA SA A SK SK ONS E STARS ie E eR SE AS RE SA SA OR A NKU OK U AK SE O SS A OR AER NOTE HINTS DSECTS ESTIONS H E WAYS OUTLINED IN PARTICULAR HESE THINGS ARI EN TO COVER ALMOST ALL OF THE POSSIB OPTIONS GIVEN N THIS WRITEU AS A RESULT THEY ARE MUCH MORE GENERAL THAN EVER BE NEEDED TO COVER ONLY ONE TWO OPTIONS FROM EACH O E T AND FLOWCHARTS GIVEN IN THIS WRITEUP STUDENT MAY BE ABLE TO DO THEM MUCH Es po ta gt Z TE E El Hu UV Q IL ROUP OF OPTIONS THU
276. mputer operators messages ar to the system at this location ALPHAMERIC DISPLAY SCOPES system status display requests for magnetic tapes to be mounted i STORAGE AGNI ETIC latency parallel each of 20 4 MAXIMUM 1 USED ON 1 Each holds 4 once each 17 5 milliseconds time for 83 TRANSF hig ER Jobs RATE of 8 6 milliseconds h transfer rate bytes maximum size THIS These hol 2314 2319 disk drives SYSTEM disk usage 8 units etc DIRECT ACCESS STORAGE DEVICES DRUMS 2 drums 09 megabytes million bytes 1 2 megabytes second HARDWAR1 03 printed here requiring and they can enter commands these display current and also are used to ete DASDS attached to channel 1 via 2820 of data with average rotational del Records data 4 bits i Has 200 conceptual EACH DRUM IS UN rotates ay n TRACKS EMOVABLE RI FASTEST DEVICES ld most heavilv used compilers and system programs MAGNETIC DISK STORAGE including 2 spare ones 2316 DISK PACK OTE ne ne time without moving t ecord at most unlike drums mounted in is place if desired each Each CY contains the 20 TRACKS accessible at he REA 7294 bytes of i DISK PACK ca FACI ITI ES 29 17 megabytes maximum 11 plates outside ones n
277. my E F AD operand ASPRGTCI 04 D COMBINED FORMS In some cases it is convenient to combine the ease of use of the macro with the small size of internal or external subroutines In this case the macro expansion sets up any needed arguments saves registers etc then generates code to invoke the subroutine The subroutine then provides the major portion of the processing code any needed large data areas etc Examples of the combined form are the following macros XDECI XDECO XPRNT XSNAP which call XXXXDECI XXXXDECO XXXXPRNT and XXXXSNAP respectively Two different extremes exist in writing combined forms 1 COMBINED FORM STANDARD LINKAG CJ In some case the calling sequence to invoke an external subroutine essentially includes the CALL macro or equivalent code i e it uses standard conventions It typically assumes that registers 0 1 14 15 may be modified without causing trouble This method is efficient and general but can cause trouble if used improperly 2 COMBINED FORM SPECIAL NONDESTRUCTIVE LINKAGE In some cases it may be useful to define a macro instruction which invokes a subroutine but can be used ANYWHERE without disturbing any registers changing the condition code or requiring that certain of the registers not be the ones being used as base registers in particular register 15 This is the kind of linkage
278. n pad parameter name to a standard length then use a CLC BXLE loop to find the position of the parameter in a parameter table which may also contain flags or jump codes to control further processing of the parameter values When finished processing a card this module may request that it be called at some time in the future and may also set various global flags his module essentially implements the following option groups RDR JO 2 JOB INITIATOR This module is c from the queue of job memory AJL This p a A new JOB ha b A JOB in the JOBTERM routine In either case not previous It removes th ly present or could not fit JBINIT alled whenever it MIGHI s waiting on disk UJL CS411FP3 08 be possible to move JOB s to the queue ready in ossiblity occurs normal s just arrived and AJL has just been it may be possible AJL element from the times and values from adds the AJ it to be added at a 1 initiate any more job EO als probably needs JBINIT probably must request that DISP b there may be a change 3 DISPATCH ER Whenever a CPU b or a termination t be assigned the free determine whether to L element to the AJL immediately list of such elements the UJL marks the el ater time i e NOW 4 Sr O in the AJL DISP ecomes availabl ly immediately after to INITIATI in memory selected
279. ndex to BAT files available for CMPSC 102 Unless otherwise specified these files are kept under the following RJE ID JRMO2 CS102AS1 first assignment mainly for arithmetic operations CS102FP1 final project writeup write assembler interpreter for CS102FP2 what is basically an XDS SIGMA 5 subset computer CS102M1 two writeups beginning run setup explanation of the conventions used to make up S 360 opcode mnemonics CS102TPA day by day outline of most of CS 102 course FLOTLINK sample program illustrating FORTRAN ASSEMBLER LINKAGE and floating point instructions A number of files used in CMPSC 411 may also be suitable for CMPSC 102 410 See also file JRMO2 CS411GI1 Some of the files mentioned in CS411GI1 include DOCUMENT hints on good documentation for assembler DSECT sample DSECT usage DUMPSJCL typical dump setups common ASM G JCL setups LINKAGE explanation of 0S 360 linkage conventions 01 09 73 INDEX411 01 CMPSC 411 INDEX OF BAT FILES J R MASHEY The following provides a brief index to materials useful for CMPSC 411 CS411GI1 CS411GI2 CS411TPA INDEX102 NOTE Ti 11 contains general information about CMPSC 411 text materials and also has further index to BAT files contained in CS411GIl Approximately 45 files of sample programs assignments writeups etc are listed here
280. ned in such a way as to handle any of the combinations of options given previously As a result they may D xtremely inefficient for a particular set of options In this case you should throw out code or variables which are not at all needed However be careful about departing too far from the methods shown in the direction of INCREASED COMPLEXITY because you will NOT get the project finished if you go out of your way to make it extra complicated A PROGRAM MODULES The required functions must be performed by sections of OS 411 some of which may be implemented as CSECTs or perhaps as internal subroutines of CSECTs he functions described will are not generally suitable to be written as MACROS although it may be useful to write macros wherever necessary for support of their functions 1 JOB CARD SCANNER also called READER INTERPRE H ER abbrev RDR When this module is called it a Determines the type of command on the next input card if any and then branches to code sections to handle the individual types of cards possible b If command was a SSJOB it decodes the various parameters obtains a free UJL element block for job waiting on disk places it it in the UJL Un initiated Job List in the appropriate position Lookup of parameter options should be done using clean table driven code i e for parameters scan to an sig
281. ng one way linked list handling techniques and some macro instruction character scan and operand methods This assignment involves writing and testing a group of macros to handle one way linked lists to be used heavily in the final project Each LIST to be processed is of the following form A HEADER WORD or HEADER CELL or just HEADER is a fullword in length and contains the address of the first NODE in the list Each NODE is a block of storage aligned on a fullword boundary with a total length a multiple of 4 with 3 basic subfields LINK the first fullword of the NODE offset 0 It contains ither the address of the next following NODE if any exist or else a fullword binary zero to indicate that no more NODES follow i e that this NODE is the LAST NODE in the LIST KEY 0 or more bytes most commonly 4 bytes at offset 0 i e the second fullword of the NODE If present this is used to ORDER a IST for insertion and retrieval of NODES In this assignment any ORDERED LIST is kept in ASCENDING LOGICAL ORDER by the KEY field For example a NODE having a key value of AAAA would precede that having KEY BBBB and so forth DATA 0 or more bytes most commonly at offset 8 divided into as many subfields as needed and convenient to store the various values associated with the KEY if it exists An example of such a list is an alphabetical symbol table used in an assembler in which the KEY of each NODE is the symbol itself and the D
282. ng it INDEX Te PREREQUISITES te e dra e E aaa a ae da a a a delete ette 1 04 II TEXTBOOKS AND MANUALS a e ua a ara ana tna a a aa ae el abs diw 02 III MRIETEUPS 2 2 DIA AA AS LAR AROSA SAS sei Qu Tu 05 IV BAT ETERS Et o GT 5 Sb batt set A a Be eS S 1 08 Vi ASSIGNMENTS DUE DATES usas tese mel there a o pe eie die Je 0 VI COURSE OUTLINE READINGS uste eee give eselet VII ISCELLANEOUS INFORMATION oooooooooooooooooo o 1 PREREQUISITES A CMPSC 102 or 410 or equivalent effectively a fairly complete introduction to much of System 360 computer structure and assembler language programming The following should have been covered S 360 structure registers PSW memory organization common interrrupts two s complement arithmetic Programming experience with most of standard instruction set possibly some with decimal opcodes and conversions Privileged operations are not expected to be well known and floating point operations will not be used in the course S 360 Assembler Language familiarity with most of the things covered in the first half of the OS Assembler Language manual sections 1 5 Most of the following terms or operations should be familiar self defining terms location counter literals absolute versus relocatable expressions USING DROP START CSECT ENTRY various instruction for
283. nt one of the 3 messages H IS LESS THAN ZERO H IS GREATER THAN ZERO H IS ZERO 11 EXTENDED VERSION OF PREVIOUS PROGRAM Modify the previous program which only had to read 1 card to read cards and follow the actions above for each card until there are no more cards END OF FILE Keep a count of the number of cards read and print out this total number before ending the program III HALFWORD VERSION OF PROGRAM IL Modify program II to use halfwords wherever possible i e store A K as halfwords use AH instead of A etc Watch out for divides since no DH instruction exists How much storage is saved IV REGISTER VERSION OF PROGRAM II Change program II by saving all values A F in registers then use RR instructions rather than RX instructions Do XDECI commands directly into registers where the values are saved A useful trick may be to NAME the registers symbolicly RA EQU 3 REGISTER WHERE VALUE A KEPT XDECT RA CARD CONVERT VALUE A INTO REG RA This technique will make it clear which value you are using note that any register reference can be symbolic to an EQU symbol LT CS102AS1 02 V WHAT TO HAND IN By using the BATCH feature in ASSIST you can run several programs in one run Turn in one run with each of the programs II III and IV shown in execution with results and
284. nted on a certain machine b Symbols statement labels variabl names ar ntered in the symbol table c Storage is reserved for constants Note that some storage requested must be on a specified boundary as in the 360 fullword CS102FP1 03 d The location counter is incremented according to the length of the instruction or storage allocated Note instructions may vary in length for some computers like the 360 e The source statements are saved so that a listing may be printed at a later pass Information added to each source statement would be the opcode type statement number the location counter value or anything else f Assembler instructions are recognized and appropriate action is taken g A copy of these source statements may reside in core if there is enough space or may be put on disk or magnetic tape 2 The second pass first it must retreive the source statements fo further processing a The operands are then processed according to their type amp and type of op code they are with b Object code is then created for the instructions and data definitions It then resides in core or is put on disk or tape c A listing is printed It contains the location counter generated code and source statment for each source statement Error messages are also printed on the listing d The errors to check for assembly time errors multiply defined symbols undefined symbols invalid opc
285. nts should study the various options performed they will NOT write the code to implement all of the options One option or several to be compared will be chosen by the instructor to be coded by the students THE STUDENTS WILL DEFINITELY NOT BE REQUIRED TO WRITE EVERYTHING DESCRIBED HERE ALTHOUGH THEY SHOULD ATTEMPT TO BECOME FAMILIAR WITH THE VARIOUS OPTIONS 2 In some cases there are variables which are to be given certain values for test purposes but which are to be represented by GLOBAL SET VARIABLES in the student program These SET VARIABLES may be referrred to in various parts of this writeup The values to be used will be given by the instructor to fill in chart on page CS411FP4 06 3 The student will write ONLY ONE source program to cover all of the different options desired Alternate versions of a particular method will be selected or deleted using conditional assembly in open code i e using GBLA GBLB SET VARIABLES and AIF AGO commands 4 The instructor will supply test decks and request that they be run given 1 or more sets of options parameters The student will then generate the required version s of his program by changing ONLY a few SETA SETB SETC statements at the beginning of his program which will create the specific version needed This process of using SET varia
286. nuniformity and lack of generality D COMBINED FORMS 1 ADVANTAGES In general the combined forms can possess all the advantages of the separate forms especially since the macro portions can generate different code depending on circumstances thus the code for the same macro might expand in line in one case and generate an out of line call to a routine in another 2 DISADVANTAGES COMPLEXITY it of course requires somewhat more planning and code to set up a good combined form system since both a macro and module must be created and meshed together properly ASPRGTC1 08 III CIRCUMSTANCES FAVORING USE OF THE VARIOUS FORMS A MACRO INSTRUCTIONS In general a pure macro instruction is used as follows VARYING CODE the required cod varies radically from call to call For example XSAVE and XRETURN SHORT CODE if a macro can generate less in line cod to perform the required function than is needed to generate a call to the routine then it should be written as a macro In some cases it takes as much work to set up the arguments as it does just to perform the operations For example the code to obtain the minimum or maximum of several arguments is probably most efficiently written as a in line macro LINKAGE CODE code for linking to routines is almost
287. oad module into memory The first csect loaded is the NIP loaded just below IPL followed by the I O Interrupt Handler at 0 which thus defines all of the special PSW s in low core IPL then passes control to NIP OSHASP 02 B NIP Nucleus Initialization Program The IPL process described above applies to all versions of 0S 360 The NIP is generated in different ways depending on the specific type of system and choice of options desired Note NIP is a csect which is link edited with the nucleus so that it can refer to sections of the nucleus via address constants and provide efficient and specific initialization services It includes the following steps 1 The CVT Communications Vector TAble is initialized and its location placed at location 16 so that it can be accessed from any routine whether part of the nucleus or not 2 NIP determines whether the computer has Large Core Storage LCS attached to it or not This is particularly necessary for those systems which include HIARCHY SUPPORT i e the ability to usefully distinguish between main core and LCS perhaps splitting programs into heavily used and lesser used sections 3 NIP checks the workability of operator console s and also checks the workability of ready direct access devices using TIO instructions It particularly checks that the SYSRES volume is mounted and contains certain datasets needed by the system 4 NIP p
288. ode invalid values e g if a machine has 16 registers and program uses register 17 5 invalid labels in Fortran a label of 9 letters BONDE 3 During assembly a table which will remain fixed will contain all of the legal opcodes Information included could be the name of the opcode mnemonic and type machine or pseudo op 4 A symbol table is also formed It contains the symbol the location counter value and maybe other flags Each symbol is entere when it appears and when a reference is made in an operand the symbol is looked up and the value obtained tt CS 102FP1 04 C Interpreter Execution monitor 1 If the target language is executable of the 360 then merely branch to the program 2 Otherwise initialize the program counter location counter to BEGIN and execute the object code using pseudo registers and pseudo storage You will also need a Program Status Word PSW containing necessary flags Execution is as follows 1 Fetch the next instruction amp increment location counter according to the length of the instruction 2 Decode the instruction amp evaluates address es 3 Execute the instruction 4 Go to 1 3 During execution there is to be error checking for a reading beyond end of file b executing too many instructions a limit will be set on the job card c invalid op code by branching into data area d address out of range of program
289. ode when finished with it This may use less total space but ating a node of given size whenever anyone is slower and more complex to program In addition to the list elements the following are tables and variables which might bD HINTS ON POSSIBLE WAYS TO YOUR CHOICE THE FOLLOWI useful necessary AS USUAL THESE ARE ONLY DO THINGS ACTUAL IMPLEMENTATION METHODS ARI NG ARE JUST TO HELP YOU GET STARTED E 1 GLOBAL CONTROL TABLE GCTB This CSECT effective to hold variables and use the entire OS 411 and le Typically the main program would do the following L ly acts like FORTRAN COMMON and is used ful constants to be addressed by any section of ts them communicate easily with each other RGCTB V GCTB address of the csect USING GCTB RGCTB note pointer Then each other csect would include the USING stmt and no one would ever modify register RGCTB so that it would always point at GCTB and make its variables available Among the typical items contained in this common CSECT would be USEFUL CONSTANTS su patterns for conversion routines Translate tables if needed several places useful Mask words etc etc HEADER CI ELLS and FRE ch as big blocks of zeroes blanks Edit ELIST HEADER CELLS for all global lists since several different routine also be des
290. ome cases it is possible to determine the bit pattern and operation of a mnemonic just from looking at it Related operations usually have related mnemonics B COMMONLY USED MNEMONICS The designers apparently went to some effort to make the most often used mnemonics the shortest and easiest to remember Most of these have 1 or 2 letter mnemonics 11 NAMING OF MNEMONIC OPCOD E S A VERB MODIFIER DATA TYPE MACHINE FORMAT T The mnemonics generally follow the format given above with the VERB always present while the others may be omitted The general meanings of the fields are given below CJ 1 VERB specifies a general type of action performed such as addition subtraction comparison data movement 2 MODIFIER specifies a modification of the general action given by the verb such as logical addition rather than algebraic moving multiple registers rather than single ones and performing different actions while loading one register into another 3 DATA TYPE specifies the type of data being operated on and is usually the same letter as that used to define a constant of the given type such as H halfword P packed decimal etc 4 MACHINE FORMAT gives the type of machine instruction being used This is most typically done by adding R or I to an RX mnemonic to obtain a similiar RR or SI instruction In general the RX instruc
291. on scheme used is quite efficient and is described later 2 Execution Stage HASP has the ability to control which jobs may execute in which portions of the OS FREE area and using the various priority and storage requirements it selects jobs from its queue to be executed One OS RDR exists permanently STARTed to a card reader This card reader does not actually exist i e it has a device address which does not correspond to a real card reader Since SVC O s are intercepted by HASP anyway HASP effectively selects a job and feeds it to the OS RDR which thinks the job is coming across a real card reader The OS RDR includes an EXIT LIST which allows it to call some routine after it has scanned each JCL Card but before the JCL card s data is actually recorded HASP is entered and takes this opportunity to modify any JCL that it wishes to for example removing any REGION requests on JOB or EXEC cards HASP has special treatment for any system input or output data sets as described below XXXXXXXX DD or DATA the OS RDR would normally expect data to follow such a card and would normally thus SPOOL such to disk itself HASP does not want this to occur since it has already SPOOLed the data It happens that there are large number of UCB s for pseudo card readers already in the system HASP selects one of these UCB s which is not being used and effectively changes the tables for this
292. only had to read 1 card to read cards and follow the actions above for each card until there are no more cards END OF FILE Keep a count of the number of cards read and print out this total number before ending the program III HALFWORD VERSION OF PROGRAM IL Modify program II to use halfwords wherever possible i e store A K as halfwords use AH instead of A etc Watch out for divides since no DH instruction exists How much storage is saved IV REGISTER VERSION OF PROGRAM II Change program II by saving all values A F in registers then use RR instructions rather than RX instructions Do XDECI commands directly into registers where the values are saved A useful trick may be to NAME the registers symbolicly RA EQU 3 REGISTER WHERE VALUE A KEPT XDECT RA CARD CONVERT VALUE A INTO REG RA This technique will make it clear which value you are using note that any register reference can be symbolic to an EQU symbol in o show will SYSIN DD CS102AS1 02 HAT TO HAND IN By using the BATCH feature in ASSIST you can run several programs ne run Turn in one run with each of the programs II III and IV n in execution with results and output as requested The run use control cards like EXEC ASACG PARM BATCH SJOB ASSIST PROGRAM VERSION II program II SENTRY pk teste test data repeat abo
293. ood habits READINGS STRUBLE Chapter 11 01 28 logical algebraic arithmetic shifts 20 minute quiz on previous instructions differences between condition code setting aroverflow in algebraic arithmetic and logical arithmetic examples shift instructions and how they are used READINGS STRUBLE Chapter 11 begin on chapter 10 11 12 1 3 14 15 16 CS102TPA 04 01 31 bit manipulation and uses review on branching bit manipulation instructions NR XR OR N X O NI XI OI NC XC OC plus TM what they do and how to use them EQU trick for SI instructions and how to use it review prototypes on loop control advantages disadvantages READINGS STRUBLE Chapter 10 first 3 sections 02 02 assembler housekeeping misc areas go over all of DC DS operand formats in detail showing what can exist as duplication factor type length constant including multiple operands and constants expressions as duplication factors and length modifiers also cover TITLE EJECT SPACE READINGS STRUBLE CHAPTER 6 pp 110 121 problems 7 9 10 ASM LANG 3 7 9 10 18 except variable symbols sequence symbols 19 21 29 33 section 5 EQU DC all except Bit Length Modifier Scale Modifier Exponent Modifier all types except E D L P Z Y S Q complex relocatability DS ORG LTORG END SPACE EJECT TITLE POP pp 24 34 except CVB CVD Logical instruc
294. opriate number is approximately 6 12 CSECTS At least SOME of them can be assembled and debugged fairly quickly object decks obtained for them and included with the other modules yet to be debugged Inside each CSECT there may be INTERNAL SUBROUTINES where useful rather than complex flag testing setting code B PROGRAM DESIGN FLOWCHARTING DATA STRUCTURING The program should be fairly well laid out before MUCH code is written for it In some cases it is definitely possible to write a complete section of code without having designed the next one This can be a good procedure since it allows the programmer to be running and debugging one section while writing the next However it is a good idea to be SURE that the current section s design does NOT depend very mcuh on what the next one does since this will require unnecessary changes AVOID CONTINUALLY CHANGING CODE AROUND Flowcharting is useful because it helps eliminate unnecesaary or redundant code aids determinging the dependence independence of code segments and helps in the designing of good data structures Heavy use should be made of various forms of subroutines and macros Big assembler programs are totally unreasonable without them C NAMING CONVENTIONS Very early in the design process and definitely BEFORE any code is written the programer should make up some simple and consistent con
295. options are possible PARM1 all of the parameters will exist on any JOB card and no errors need be tested for in them PARM2 some parameters may be omitted or in error in which case DEFAULT values are to be set at SYSGEN time and used instead The corresponding SET variables are then named by amp the name of the parameter and DFT i e default T value is given by amp TDFT etc So value of amp PRIODFT is used if PRIO is omitted PARM3 This includes PARM2 above except that the defaults are done by JOB CATEGORY the CAT value so that omitted options from JOBS in different categories will have different defaults All the values of x above are unsigned decimal numbers and the specific meaning of each parameter is given below T a number from 1 to 32767 giving the number of milliseconds which is a time limit on the execution of the JOB As will be given in a later section this always counts use of the CPU but may or may not count use of a CHANNEL SP SPace a number from 1 1024 giving the number of K byte blocks of storage required to execute a JOB IOIN I O INterval a number from 1 32767 which describes the interval between I O requests from a job value is in millisecs IORL I O Request Length number from 1 32767 describing the duration in milliseconds for one I O request by the JOB PRIO PRIOrity a number from 0 255 specifying the relative
296. or Good documentation techniques can be helpful in the following ways PROGRAM DESIGN Many beginning programmers seem to write programs in haphazard and unplanned ways and often add comments only after the program is running This method not only leads to poorly structured programs but usually results in wasted time and is not feasibl xcept for relatively trivial problems A much better method is to write most of the overall comments with a flow chart first specifying the structure and conventions of the program and then writing the program to fit This usually leads to cleaner coded well structured programs which are produced in less time than those written by most novice programmers PROGRAM DEBUGGING Program debugging is aided by documenting a program before and during its creation rather than afterward Many mistakes can be avoided by having programming conventions well specified befor writing the code The very act of adding a comment to a statement often helps identify errors in the statement because it forces the programmer to think about the function of the statement Finally good documentation is useful if help is required from someone else since it aids one in understanding the program quickly It also makes other people much more willing to look at a program PROGRAM MODIFICATIONS Clear and complete documentation is absolutely invaluable when a program must be modified es
297. orage use program with START to check relocatable vs absolutes modify one of batch programs to get 0C6 rather than 0C4 5 O1 17 introduction to arithmetic and data movement instructions introduce idea of instruction families and regularity of mnemonics Go thru following instructions LR LPR LCR LNR LTR L LH LA AR ALR A AL AH SR SLR 8 SL SH mention M and D also briefly note existence of Condition Code and show how to test it without worrying about encoding 20 minute question answer and review questions occurred on differences between literals and self defining terms and on use of symbolic register equates READINGS STRUBLE Chapter 5 HANDOUTS CS102AS1 pages 01 02 first assignment input output of numbers calculations in binary CS102M1 pages 02 05 S 360 mnemonic construction 6 10 CS102TPA 03 01 19 quiz and finish up data movement and binary arithemetic Twenty minute quiz diagnostic mainly base conversions 2 8 10 16 negative numbers base index displacement addrs relocatable vs absolute n what to nd obtain Instructions LM STM MVC MVI M MR D DR MH and hints watch for XREAD XPRNT XDECI XDECO method for building messages Programming techniques review input output amp conversions a ing length for XPRNT via MSGL EQU MSG ASSIGNMENT indexing and comparison assignment CS102AS1 03
298. ot used to access information usable at any one time INDER D WRITE H Each Uses of w nformatio O ATION EEK TIM ie N R S La ES TIME time t IN 25 AXIMUM NOTE 230 237 r DATA TRANSE each of t control unit AVERAGE 25 millisec H o move 60 or 330 337 430 433 IS AS FOLLOWS 2314 8 drv 233 megabytes 2319 8 drv 233 megabytes 2314 4 drv 116 megabytes 2301 2 drums 8 megabytes TOTAL 590 megabytes approx 13 ER RATE AVERAGE AT MAX on channels 2 EADS n n be removed ENCY EADS to correct cylinder 130 or 135 312 000 bytes sec 3 DRIVE on 20 OVABLE hich t total of 22 holds one disk surfaces HEADS here are 200 Each track can and another 12 5 millisee millisec he three storage facilities contains its own and each drive is numbered accordingly Tee 4 S ECONDARY STORAGE SEQUENTIAL DEVICES 240x 2402 III 2403 III MAGNETIC TAPE at maximum density of 800 BPI bits inch 2 of the drives also read write 7 track tapes of 4 drives is connected to one SELECTO MULTIPLEXOR CHANNEL contained in the 2403 un MAXIMUM TRANSFER RATE blocks of data its 90 000 bytes sec speed of 112 5 inches per second
299. ouped into CLASSes which can be written according to priority or otherwise treated differently as desired COMMENTS ON THE PROCESS ABOVE The process described above is quite flexible and general However it does require a fair amount of time to set up any job even a small one As such it is quite satisfactory for any installation which runs jobs which require a fair amount of time since then the setup time is negligible However due to the use of OS DADSM for PSPOOLed input and output DASD space can become fragmented disk head movement can becom xcessive and much time can be used up allocating and deallocating disk space Although 0S 360 is quite reasonable in a commercial installation or in one running a few large jobs it seems to have too much overhead for university or other installations which often run many small jobs For this reason most larger S 360 computers i e models 75 67 65 and larger 50 s typically use some method to reduce the overhead in running small jobs All of the methods involve faking out OS 360 in some aspect or other The method emphasized here which happens to be the most popular one is HASP Houston Automatic Spooling Priority system OSHASP 05 III RUNNING AN OS 360 HASP SYSTEM In any OS 360 system it is fairly typical to have one or more special jobs in the system which are loaded before normal user jobs and typically remain re
300. p RLK before amp RND is changed useful for list search and insert operations If END is specified a test is made and branch taken if there are NO MORE NODES in the list CS411MC2 04 The operands for QNEXT are described as follows amp RND is a register EQU symbol or number into which will be loaded the address of the next NODE amp ADDR if specified at all is an RX type address of the LINK field which addresses the next NODE The word at this address is to be loaded into amp RND If omitted entirely it is to be assumed that amp RND contains the address of the LINK already amp RLK if specified gives the name or number of a register into which amp RND should be saved before it is changed amp END if specified gives a statement label in which case the value newly loaded into amp RND is to be tested and if found 0 a branch taken to the given statement label LOOP QNEXT 5 RGL RWK1 END ENDLOOP LOOP LR RWK1 5 L 5 LINK 5 LTR 55 T BZ ENDLOOP D QPOP POP FIRST ELEMENT OF LIST amp LABEL QPOP amp RND amp HDR amp END QPOP sets register amp RND to the address of the first NODE in the LIST begun by HEADER at address amp HDR taking branch to amp END if the list is empty IT SHOULD BE OBVIOUS TO USE QNEXT A
301. pecially if anyone but the original programmer is making the changes It may be noted that useful programs tend to be modified often DOCUMENT 2 Gl ASSEMBLY LANGUAGE DOCUMENTATION The following advantages apply to any computer language However they are most important for assembly language for the following reasons 1 Assembly language programs typically require many more statements than do high level language programs for the same task 2 Assembly language programs are not usually self documenting Without good documentation not even the programmer who wrote the code will be able to understand it several months later 3 Assembly language programs are often very sensitive to minor changes much more so than higher level languages The remainder of this paper describes a well documented assembly program and notes the various techniques which can be used to achieve this result Briefly a well documented program has the following characteristics 1 The documentation structure mirrors the program structure and it leads from the general to the specific Thus the program begins with a block of comments which describes th overall purpose of the program and gives some indication of the general structure Each major section has a block of comments describing it as does each of the section s subsets 2 At least 95 of machine instruction statements have comments 3 The program is easy to
302. priority of either single jobs or entire groups of jobs The disk allocation scheme used is quite efficient and is described later 2 Execution Stage HASP has the ability to control which jobs may execute in which portions of the OS FREE area and using the various priority and storage requirements it selects jobs from its queue to be executed One OS RDR exists permanently STARTed to a card reader This card reader does not actually exist i e it has a device address which does not correspond to a real card reader Since SVC O s are intercepted by HASP anyway HASP effectively selects a job and feeds it to the OS RDR which thinks the job is coming across a real card reader The OS RDR includes an EXIT LIST which allows it to call some routine after it has scanned each JCL Card but before the JCL card s data is actually recorded HASP is entered and takes this opportunity to modify any JCL that it wishes to for example removing any REGION requests on JOB or EXEC cards HASP has special treatment for any system input or output data sets as described below XXXXXXXX DD or DATA the OS RDR would normally expect data to follow such a card and would normally thus SPOOL such to disk itself HASP does not want this to occur since it has already SPOOLed the data It happens that there are large number of UCB s for pseudo card readers already in the system HASP selects one of
303. processing of the macro The macro definition is terminated by the MEND statement The following steps are typical in defining a macro a DETERMINE BASIC PURPOSE AND GENERATED CODE It is generally a good idea to write at least some of the statements to be generated together as a code segment first to get some feel for what is needed b DECIDE ON NECESSARY ARGUMENTS AND THEIR USAGE it may be a goo idea to write the purpose of each argument in the operand list punch it and include it in a block of comments at the beginning of the macro This helps the macro to be done to do what it is supposed to do Use POSITIONAL operands for heavily used arguments i e if an argument MUST be supplied every time make it positional In a group of positionals place the most heavily used ones near th front since it is much more convenient to omit th later ones than the earlier Use KEYWORD arguments for values which may not be needed always or for ones which are conveniently supplied with default values which are most often used Use SUBLISTS or amp SYSLIST for variable numbers of arguments c WRITE ACTUAL BODY OF MACRO BUILDING MACRO TYPE COMMANDS AROUND THE MODEL STATEMENTS TO BE GENERATED CJ ASPRGTCI 02 2 INVOCATION A macro can be called merely by writing its name and supplying i
304. r G the most common causes of error termination on the completion dumps printed by 0S 360 0S 360 DUMP ASSIGNMI a NT DUMPSJCL 03 This assignment should help familiarize you with the typical cards Run each of the following programs and show several of with the effects they have using appropriate JCL cards Use MSGLEVEL 1 and LOG cards for all of them Use procedure ASGCG for all programs and ASGCLG for part C in addition total 4 runs A TYPICAL INTERRUPT DUMP PRECISE INTERRUPT 0C6 DUMP 1 CSECT XSAVE ID NO SPACE 2 L 042 cause 0C6 XRETURN SA return create save area END DUMP 1 B ABEND DUMP CAUSED BY USER PROGRAM ABEND DUMP2 CSECT XSAVE TR NO SPACE 2 ABEND 400 DUMP U0400 completion code XRETURN TR NO SA no trace generat save area END C INTERRUPT DUMP WITH MULTIPLE CSECTS IMPRECISE INTERRUPT 0C4 DUMP 3 CSECT XSAVE SR 0 0 0 for main program CALL SUBI CALL SUBROUTINE XRETURN SA SUB1 CSECT XSAVE LA 0 1 set to 1 for lst level sub CALL SUB2 CNOP 0 8 line up for max overrun ST 0 20 OC4 store into protected core AR 0 0 get value of 2 AR 0 0 get value of 4 XRETURN SA SUB2 CSECT XSAVE LA 0 2 set to another value XSNAP XRETURN SA END DUMP3 J BRK KKK k K k k k k k k K K K k K K K K K KOK K K KOK OK KOK K K K KOK KOK KOK KOK KOK KOK KOK OK KOK K K K A K
305. r at the current position of any read write head i e where it can read or write without even moving a head 2 It then searches for a free cylinder at 1 from current head positions then 1 from each followed by 2 2 etc up to 8 8 cylinders away from current head position 3 If the above fail it searches sequentially through all cylinders in the master bit map When a cylinder is found its bit is turned off in the master bit map and turned on in the appropriate job bit map The overall effect of this process is to minimize head movement When disk storage for a job is to be released the deallocation scheme is extremely fast and efficient the job bit maps are Just OR d into the master bit map thus returning all of the cylinders to free storage IV OTHER PSEUDO DEVICE SYSTEMS FOR USE WITH 0S 360 he following are other systems which are based on 08 360 but use some kind of pseudo devices to make it run faster A ASP ATTACHED SUPPORT PROCESSOR In this system 2 computers are used All unit record devices are attached to the multiplexor channel of a medium sized 360 along with some disk It performs all SPOOLing control of remote terminals etc It is connected to a larger system via a chnnel 0S 360 is in the large system and it reads its input and sends its output along the channel channel hookup between the two CPU s A typical set
306. rag number of jobs in memory usually gt 1 0 under multiprogramming CS411FP3 04 The following other report items might be printed at the end of a batch under some circumstances R9TU Average TURNAROUND time for all jobs in a batch This option could usefully be generalized to show turnaround by groups of jobs for example by category CAT or by splitting jobs into several groups according to T SP etc The reason for this is to determine just which types of jobs are being favored and if so how much RDBG DEBUGGING REPORTS these may be added as desired Things which might be useful debug output to build in are RDWJ A dump of the list of jobs waitng on disk RDMJ A DUMP OF THE LIST IF JOBS IN MEMORY AJL plus others RDST a dump of all the important system variables such as the clock or current time the status of CPU s and CHANNEL s important flags etc etc Note these reports are not meant to be pretty but they are probably the MOST IMPORTANT items for getting the program running Note that an excellent technique is to embed such debugging reports in a program from the beginning allowing them to be removed in one of two ways ASSEMBLY TIME if absolute sure that a given section of code is correct debug output can be temporarily canceled using SET variables When the inevitable bugs then creep out of hiding they can be caught by changing the SET variables again i e
307. ring enclosed in quotes If omitted the list only is created but if present the message string is generated as a C type constant rounded up to a 4 multiple and placed on a fullword boundary preceding the list It is used for debugging i e to locate a list in memory in a dump amp LABEL is generated on or before the first node generated LAVS1 QDEFL 2 NODELEN LIST OF FREE NODES FOR NODE1 DC OF O CL28 LIST OF FREE NODES FOR NODE1 LAVS1 DC A NODELEN 3 4 4 CL NODELEN 4 DC A 0 CL NODELEN 4 B QHED DEFINE LIST HEADER amp LABEL QHED amp LISTNAM If amp LISTNAM is coded this macro defines a HEADER containing the address OF aes r defining an empty list f this list otherwise it is a header cell having a value of For debug purposes it may be very useful to let this macro generat message amp LABEL LIST HEADER or similiar thing immediately preceding the header word HDRI QHED LAVS1 DC 0F 0 CL16 HDR1 HEADER HDR1 DC A LAVS1 LIST1 QHED LIST1 DC A O C ONEXT GET ADDRESS OF NEXT NODE IN LIST amp LABEL QNEXT amp RND amp ADDR amp RLK amp END This macro sets register amp RND to the address of the next NODE in a list given that amp ADDR indicates the address of a NODE whose LINK field points to the NEXT NODE If desired the valu in amp RND may be saved into register am
308. rmation times addresses completion code listing of JCL cards supplied note manual Messages Codes Assembly listing External Symbol Dictionary listing Cross Reference and uses of each follow linkage code 0S 360 Loader MAP addresses information in the dump how to find where error occurred COMPLETION CODE APSW SA TRACE REGISTERS STORAGE SP 000 HANDOUTS CS411AS1 pages 01 02 1st assignment linkage decks DUE IN ONE WEEK DINGS ASSEMBLER LANGUAGE Sections VI to end FERENCE Computing Surveys 1 4 Dec 1969 183 196 T ne bi tA D CS411TPA 02 5 01 13 complete dump reading follow step by step procedure for locating errors in dump system completions precise and imprecise user completion differences between link editor and loader dumps interrupts 1 2 3 4 5 6 and their meanings 6 01 18 begin macro instruction writing basic concepts of macros comparison with FORTRAN subroutine HANDOUTS CS411MC1 pages 01 03 MCALL MSAVE MRETURN ASSIGNMEN 7 01 18 details on macro statements most of macro statement types with FORTRAN counterparts various examples MACRO PROTOTYPE LCLx GBLx SETx S VARIABLES AIF AGO CONCATENATION SUBSTRING amp SYSECT amp SYSNDX EH Ho 8 01 20 completion of S 360 macro language Operand processing postional keyword sublists attributes
309. rt an address ID to base dsiplacement ENTRY CONDITIONS RA address to be converted to base displacement form 0 2 24 1 RB section ID of the address to be converted 1 255 EXIT CONDITIONS RA base displacement form of address if there was one in low order halfword of register bits 16 31 Bits 0 15 should be zeroes RB 0 if the address was properly converted nonzero value if an addressibility error occurred i e if there was no register with the proper section ID and a value from 0 to 4095 less than the value to be converted NOTE the above rules must be followed exactly If they are not error messages will be given by the ASSIST Replace Monitor ASBROPS2 03 ILLE IMPLEMENTATION METHODS This section outlines several different methods of implementing the BROPS2 module The following are ways in which the assignment may be handled with the instructor specifying in class which one is to be followed Write the module in 1 specific way Write the module in several ways and compare their performance Write the module any way at all Write the module any way but optimizing for one of several goals A REGISTER TABLE ORDERED BY REGISTER NUMBER A fairly simple way to handle the problem is to just keep a table ordered by register number which can easily be indexed into to change the values and can be searched by a fairly simple loop x B REG
310. s relocatable vs absolute n what to nd obtain Instructions LM STM MVC MVI M MR D DR MH and hints watch for XREAD XPRNT XDECI XDECO method for building messages Programming techniques review input output amp conversions a ing length for XPRNT via MSGL EQU MSG ASSIGNMENT indexing and comparison assignment CS102AS1 03 que 02 02 pwan HANDOUTS CS102AS1 03 labeled CS 102 AS2 also indexing READINGS STRUBLE CHAPTER 5 start on STRUBLE CHAPTER 7 01 21 condition code branching instructions loops condition code values and encoding BCR BC Extended Mnemonics subroutine example of flowcharti READINGS STRUB 01 24 finis Explain BX forward BX show need recommended for use over BC 4 BALR BAL and s BCT BCTR usage including decrementing regs basic loop to sum array of numbers ng and good design versus kludge programming E Chapters 7 8 5 h loop control begin on USING DROP linkage H BXLE instructions give typical setups a E loop backwards BXH loop BXH scan loop for USING command give rules for computation of base displacements minimum base displacement for those which are available higher numberred register if several have same begin conventions exaplin registers 15 14 usage on entry HANDOUTS LINKAGE 0S 360 linkage conventions READINGS STRUB
311. s a few hints can be helpful in learning and remembering the names purposes and usage of the various operations In particular certain properties of S 360 menemonics can help the learner a great deal Some of them are A REGULAR SCHEME FOR NAMING OPCODES In general a fairly coherent and regular method has been used in naming operations In some cases it is possible to determine the bit pattern and operation of a mnemonic just from looking at it Related operations usually have related mnemonics B COMMONLY USED MNEMONICS The designers apparently went to some effort to make the most often used mnemonics the shortest and easiest to remember Most of these have 1 or 2 letter mnemonics 11 NAMING OF MNEMONIC OPCOD E S A VERB MODIFIER DATA TYPE MACHINE FORMAT T The mnemonics generally follow the format given above with the VERB always present while the others may be omitted The general meanings of the fields are given below CJ 1 VERB specifies a general type of action performed such as addition subtraction comparison data movement 2 MODIFIER specifies a modification of the general action given by the verb such as logical addition rather than algebraic moving multiple registers rather than single ones and performing different actions while loading one register into another 3 DATA TYPE specifies the
312. s first usage after an IPL UCB s Unit Control Blocks exist for every device connected to the computer system HASP now scans these and essentially allocates to itself 1 All real unit record devices readers punches printers 2 All disk packs which have volume label names beginning SPOOL It also obtains effective control of the operator s console s plus remote terminals if any Finally HASP modifies the SVC table which contains pointers to the routines which are called for each specific SVC number so that the following ones go into HASP rather than to the original routines also saving these addresses for later use for itself SVC 0 EXCP all input output SVC 34 WTL write to log SVC 35 WTO WTOR write to operator with withput reply OSHASP 06 B RUNNING NORMAL USER JOBS UNDER 0S 360 WITH HASP 1 Input Stage HASP continually reads cards from whatever card readers are active in the system It checks for JOB cards performs various accounting checks on input jobs and transcribes the jobs to disk In this stage each job is split up into two sections the JCL cards with certain modifications and the input data cards It enqueues the jobs according to a priority scheme which can be found from many different sources of information These include category time output storage requirements originating site of job and commands from the operator to change
313. s timesharing typewriter terminals or provide any other service which the installation desires Such jobs are normally placed into the high address sections of the FREE area or of the two FREE areas if the system has both main core plus LCS When HASP is used it is normally the first job submitted to OS 360 and it essentially takes over the system even though it appears to 0S 360 as just another job A HASP INITIALIZATION There are two possible cases when starting HASP up after an IPL A COLD START occurs when the system is completely empty i e there are no jobs already enqueued on disk which can be executed If there are disk packs on the system containing previously read jobs the start is called a WARM START A WARM START normally occurs if the system was previously taken down on purpose such as for systems programming or if enough information had been saved previous to a crash A COLD START only occurs when the system crashes badly and destroys records of jobs already SPOOLed onto disk In this case the jobs must be read in again When HASP first gains control it issues a special SVC call which returns to HASP with protect key 0 and supervisor state also supplyin HASP with some useful pointers to control blocks in the nucleus NOTE this special SVC can only be called 1 time since it locks after its first usage after an IPL UCB s Unit Control Blocks exi
314. se treated differently as desired COMMENTS ON THE PROCESS ABOVE The process described above is quite flexible and general However it does require a fair amount of time to set up any job even a small one As such it is quite satisfactory for any installation which runs jobs which require a fair amount of time since then the setup time is negligible However due to the use of OS DADSM for PSPOOLed input and output DASD space can become fragmented disk head movement can becom xcessive and much time can be used up allocating and deallocating disk space Although 0S 360 is quite reasonable in a commercial installation or in one running a few large jobs it seems to have too much overhead for university or other installations which often run many small jobs For this reason most larger S 360 computers i e models 75 67 65 and larger 50 s typically use some method to reduce the overhead in running small jobs All of the methods involve faking out OS 360 in some aspect or other The method emphasized here which happens to be the most popular one is HASP Houston Automatic Spooling Priority system OSHASP 05 III RUNNING AN 0S 360 HASP SYSTEM In any OS 360 system it is fairly typical to have one or more special jobs in the system which are loaded before normal user jobs and typically remain resident from one IPL to the next Such jobs may control remote batch terminal
315. set according to the sign of the number placed into the register 0 gt 0 lt 0 gt 1 gt 0 gt 2 2 ASSUMPTIONS PERMITTED The implementation of this macro depends strongly on the what assumptions are made about error condtions modifications of registers etc The following assumptions might be made Consider their ffects on the code to be generated CS411MC1 05 a Assume registers 0 1 14 15 may be completely destroyed by the execution of the macro b Assume that registers 0 14 15 may be changed by the macro but must be restored to their original values by the macro c Assume that neither register 14 nor 15 is the current base reg but that registers 0 14 15 must all be restored by the macro if changed d Assume that input hexadecimal numbers definitely contain no more than 8 digits e Assume that errors may exist in the input i e more than 8 condition code to 3 to show error el digits Then either set th convert the first 8 hex digits normally e2 f Assume that R13 contains the address of a usable s 360 savearea Always assume the macro may not create code changing regs 2 13 FOR THIS ASSIGNMENT TH ar el E ASSUMPTIONS FROM ABOVE TO BE USED ARE and f 3 USEFUL HINTS ON INSTRUCTIONS TO BE USED a SCANNING the TRT instruction can usefully be employed to do several tricks exist for se
316. sident from one IPL to the next Such jobs may control remote batch terminals timesharing typewriter terminals or provide any other service which the installation desires Such jobs are normally placed into the high address sections of the FREE area or of the two FREE areas if the system has both main core plus LCS When HASP is used it is normally the first job submitted to OS 360 and it essentially takes over the system even though it appears to 0S 360 as just another job A HASP INITIALIZATION There are two possible cases when starting HASP up after an IPL A COLD START occurs when the system is completely empty i e there are no jobs already enqueued on disk which can be executed If there are disk packs on the system containing previously read jobs the start is called a WARM START A WARM START normally occurs if the system was previously taken down on purpose such as for systems programming or if enough information had been saved previous to a crash A COLD START only occurs when the system crashes badly and destroys records of jobs already SPOOLed onto disk In this case the jobs must be read in again When HASP first gains control it issues a special SVC call which returns to HASP with protect key 0 and supervisor state also supplyin HASP with some useful pointers to control blocks in the nucleus NOTE this special SVC can only be called 1 time since it locks after it
317. ss into register 13 since any of the above actions may result in registers being saved at wherever 13 points processing code to perform required actions result return code result may be left in register 0 in which case it should not be restored neither here nor in generated code before i e change LM 14 0 4 14 to LM 14 15 4 14 and STM likewise register restoration restore all registers modified in this routine Especially restore 14 SPM 14 restore original condition code note that calling BALR 14 15 saved it B number 14 branch to displacement number beyond address in 14 enough to pass control to statement LM 14 0 4 14 It may be useful for the programmer to create a DSECT which describes the control block generated by the macro expansion This would permit the module to refer to arguments and return points using symbols rather than absolute displacements A typical DSECT might be dsectnam DSECT DS V routine space for adcon DS 3F space for regs 14 15 0 argument DS F argument value placed here if any A further arugment DS statements follow return LM 14 0 4 14 return label YES THIS IS LEGAL it does NOT generate code but it makes the point clear as to description of block If such a DSECT were used the routine code would include USING dsectnam 14 to set up DSECT addressibility B return return instead of B number 14 ASPRGTCI
318. ssembler 426 McAllister A slightly outdated while ask instructor CS411GI1 05 III WRITEUPS This section provides an alphabetic list of various explanatory writeups and assignment handouts which are available for use in CMPSC 411 or other courses Some of these may be handed out while others can be accessed by anyone who is interested Names number of pages source and description are given for each E pi he following notations may appear for the SOURCE of each writeup BAT The writeup exists as as PSU CC BAT file accessible from RJE or batch terminals An otherwise blank card with a comma in column indicates the beginning of each page including the first Note that these normally contain both upper and lower case letters DTO The writeup exists only as a dittoed handout TAP The writeup exists on a tape in which case the tape name fil number and file name are given CRD The writeup is in a punched card deck MTS The writeup is on an MT ST tape If the description of a BAT file begins with userid that is the userid under which the file is saved If not mentioned the file is saved under the following userid JRM02 Several utility programs are available for printing or punching any BAT file s JRM02 PRINT prints any input converting lower case letters to upper case while JRM02 PUNCH punches its input Both will run in any category including category W The
319. st for every device connected to the computer system HASP now scans these and essentially allocates to itself 1 All real unit record devices readers punches printers 2 All disk packs which have volume label names beginning SPOOL It also obtains effective control of the operator s console s plus remote terminals if any Finally HASP modifies the SVC table which contains pointers to the routines which are called for each specific SVC number so that the following ones go into HASP rather than to the original routines also saving these addresses for later use for itself SVC 0 EXCP all input output SVC 34 WTL write to log SVC 35 WTO WTOR write to operator with withput reply OSHASP 06 B RUNNING NORMAL USER JOBS UNDER 0S 360 WITH HASP 1 Input Stage HASP continually reads cards from whatever card readers are active in the system It checks for JOB cards performs various accounting checks on input jobs and transcribes the jobs to disk In this stage each job is split up into two sections the JCL cards with certain modifications and the input data cards It enqueues the jobs according to a priority scheme which can be found from many different sources of information These include category time output storage requirements originating site of job and commands from the operator to change priority of either single jobs or entire groups of jobs The disk allocati
320. ster should be converted hexadecimal fo b rmat The condition code is unchanged by the execution of this and placed in the 8 byte area at the address bytes into 5 fifth byte removed and as desired punching the and checking the location values of decimal under QHEXI to printable given macro CS411MC1 07 2 ASSUMPTIONS PERMITTED T a Assume registers 0 1 14 15 may be completely destroyed b Regs 0 1 14 15 can safely be destroyed but restored by the end c Assume that the current base register may be ANY register 1 15 Always assume that registers 2 13 may not be changed in the code generated by the macro FOR THIS ASSIGNMENT MAKE THE FOLLOWING ASSUMPTIONS FROM ABOVE e c consider using this macro in lowest level routime with br 15 3 USEFUL HINTS ON INSTRUCTIONS TO BE USED a CONVERSION the output conversion is much easier than the input one Briefly consider the effects of storing a register into a fullword in memorv then unpacking 5 FIVE bvtes from that area into a 9 byte workarea translating the first 8 bytes of that area then moving 8 bvtes where desired Assume the data areas from the previous example that the value to be converted is in first 4 bytes of FULL5 and that the following has been added immediately after statement labeled TAB4 TAB4A DC C ABCDEF convert X FA gt C A etc Then U
321. store registers insert a ret urn code in register 15 flag the save area X FF in wd 4 and branch back via register 14 The registers to b restored ar coded as with SAVE If 15 already has a return code in it and should not be restored it is coded as RC 15 else RC n may De coded where n is some value to insert into register 15 The operand T causes the flag X FF to be inserted in the save area The following code will restore registers 5 6 10 to be reloaded the save area to be flagged and 15 to be loaded with a value 16 RETURN 5 10 T RC 16 NOTE Both of these macros expect that register 13 will already be loaded with the address of the appropriate save area The use of the PSU macros XSAVE and XRETU flexibility in saving and restoring registers code to print a trace message showing entry and XSAVE can be used to establish and load a base re snap of the registers saved XRETURN can create that as with RETURN XRETURN assumes that registe the relevant save area For most uses the code XSAVE alone can be ters For a routine with only one return point fice if a routine has more than one return poi alone should be coded at all return points except XRETURN SA should be coded The reason for t Cause a save area to be created only one should LINKAGE 5 RN can provide added B Both can generate exit from a module gister or to print a a save area NOTE r 13 still points to
322. structions coding forms for these All 4 of the above modules are definitely necessary but they are combined in various ways with any combination providing all parts being generally acceptable although the first combination is preferred Each manual notes which information a b c d it contains Later references may refer to lt S amp DM gt x where x is a b c or d In such cases the information can be found in any of the manuals which have that section of information R lt S amp DM gt 5 XXXX XXXX X SUPERVISOR AND DATA MANAGEMENT SERVICES AND MACRO INSTRUCTIONS R lt SS amp M gt GC28 6646 6 IBM S 360 OS SUPERVISOR SERVICES AND MACRO INSTRUCTIONS a b This is recommended version and contains the general methods used for management of programs including linkage conventions tasks and main storage allocation with the macros for these R lt DMSG gt GC26 3746 1 OS DATA MANAGEMENT SERVICES GUIDE c This is recommended form and gives the general methods and many examples of processing each of the different types of datasets in different ways Good explanatory material is given on various characteristics of data sets record formats control characters etc direct access devices magnetic tapes and general procedures of data management OPEN CLOSE DCB GET PUT READ WRITE etc R lt DMM gt GC26 3794
323. t FORTRAN REAL 8 EX AD SD LTDR LD IMPLIES RR OR RX INSTRUCTION FORMAT E Exponent single precision floating point fullword 32 bit FORTRAN REAL 4 EX AE LER ME IMPLIES RR OR RX INSTRUCTION FORMAT H Halfword 16 bit binary number FORTRAN INTEGER 2 EX AH H STH CH IMPLIES RX FORMAT P Packed decimal format 2 decimal digits per byte EX AP SP CP IMPLIES SS INSTRUCTION FORMAT OF TWO LENGTH TYPE 4 MACHINE FORMATS Several characters are used to denote the specific type of operand format being used note that the data types can also imply specific formats If they imply one of several the last character distinguishes among them FORMAT MEANING EXAMPLES I Immediate IMPLIES SI FORMAT EX MVI CLI Ol R Register IMPLIES RR FORMAT EX AR BCR DDR LIE EXAMPLE This section lists all OF FAMILY OF RELATED OPCOD the members of the menemonics nam The letters V machine Format CS102M1 05 S Compare family of showing their relationships adn the elements present in each D F stand for Verb Modifier Data type and OP CODE VM DF TYPE COMMENTS E C RX fullword algebraic compare the basic one CL C L RX fullword logical comparison logical modifier CD E D RX compare double precision floating numbers CE G E RX
324. t with any needed arguments Note that a label on a macro call is never generated and is thus UNDEFINED unless the macro definition is made to generate it on some model statement B INTERNAL SUBROUTINES Internal subroutines are sections of code written as parts of a given control section CSECT and are only used inside that CSECT Like external subroutines internal subroutines can of course call others They are typically used for small to medium sections of code which are needed at several places in a CSECT but are not needed by any other or are not big enough to warrant the overhead in making them external subroutines 1 DEFINITION It is often typical to place a group of internal subroutines near the end of the code section of a program just before the data areas It is a good idea to set up conventions for the use of internal subroutines before writing any The following ar often needed return register either one standard one or several different ones argument registers and work registers which can be used without saving In general internal subroutines should not need to do much saving and restoring of registers They should be able to return via BR REG 2 INVOCATION Calling an internal subroutine is usually done by first filling any argument registers with needed values then coding BAL REG INSUB This type of linkage can be fast and small C EXTERNAL SUBROU
325. t even moving a head 2 It then searches for a free cylinder at 1 from current head positions then 1 from each followed by 2 2 etc up to 8 8 cylinders away from current head position 3 If the above fail it searches sequentially through all cylinders in the master bit map When a cylinder is found its bit is turned off in the master bit map and turned on in the appropriate job bit map The overall effect of this process is to minimize head movement When disk storage for a job is to be released the deallocation scheme is extremely fast and efficient the job bit maps are just OR d into the master bit map thus returning all of the cylinders to free storage IV OTHER PSEUDO DEVICE SYSTEMS FOR USE WITH 0S 360 he following are other systems which are based on 08 360 but use some kind of pseudo devices to make it run faster A ASP ATTACHED SUPPORT PROCESSOR In this system 2 computers are used All unit record devices are attached to the multiplexor channel of a medium sized 360 along with some disk It performs all SPOOLing control of remote terminals etc It is connected to a larger system via a chnnel 0S 360 is in the large system and it reads its input and sends its output along the channel channel hookup between the two CPU s A typical setup would use a 360 50 hooked to a 360 75 An advantage over HASP is that ASP o
326. tach to a number of devices and it is used to control greatly different devices in a such a way as to make them appear more alike as far as the channels are concerned Each device must be attached to a particular type of control unit and each control unit normally can control a group of related devices 2820 STORAGE CONTROL UNIT controls the 2301 drum units attached to channel 2821 CONTROL UNIT controls UNIT RECORD devices card readers printers punches attached to multiplexor channel 2848 DISPLAY CONTROL controls the 8 2260 scopes which display system status to the operators 2701 DATA ADAPTOR controls a small number of high speed transmission lines i e high speed terminals 4800 bits sec transmit rate such as 360 20 s at various locations 2703 TRANSMISSION CONTROL controls a larger number of lower speed terminals including typewriter teletype terminals and read print punch terminals at Commonwealth Campuses such as IBM 2780 DCS CP 4 etc DISPLAY DI EVICES 1052 2260 SECONDARY 2301 231x TOTAL DASD STORAGE CONSOLE TYPEWRITER action by computer operators messages ar to the system at this location ALPHAMERIC DISPLAY SCOPES system status display requests for magnetic tapes to be mounted E STORAGE AGN ETIC latency parallel each of 20 4 MAXIMUM 1 USED ON 1 Each
327. te the particular version of OS 411 being used R2 CURRENT STATUS REPORT Every amp RDRINTV milliseconds this report is printed for the prupose of showing the overall status of the system It might include R2T The current time 0 1 amp RDRINTV 2 amp RDRINTV etc R2N The current number of jobs waiting on disk for execution and the current number in memory R2SM The current status of each job in memory with following codes and other information if applicable R Ready ready to execute but does not have CPU Cc Cpu executing now add number of CPU used in multiple CPU systems W Wait waiting for channel to become free E 1 0 performing I O add channel number being used in multiple channel systems R2MJ The current memory status for each job i e which blocks of memory are currently allocated to the jobs R2MM Memory Map a map of memory showing which jobs are located in which areas R2MT The total amounts of memory in use and unused R3 JOB READ REPORT printed whenever a JOB card is read and JOB thus enters system Is essentially an ECHO of JOB card preceded by the time it enters system R4 JOB INITIATION REPORT printed whenever a job is selected to be loaded into memory Gives time and the name of the job from the JOB card An additional feature might be R4 memory locations in which job is loaded NOTE the
328. te operand instructions like AT register is a decimal number from 0 15 no leading zeroes immediate is a signed or unsigned decimal number from 1048576 to 1048575 to be used as an immediate value b RX INSTRUCTIONS This format is basically like S 360 RX format with some restrictions plus an addition for indirect addressing label opcode register address label is optional statment label opcode is one of the RX opcodes like AW register is decimal number from 0 15 like register in RI above address represents an address in memory and may have any of the forms given below n represents any unsigned number and index represents any decimal number from 0 to 7 Note that symbol represents any legal label representable in 17 bits giving an index register The formats are then symbol implying direct addressing index 0 Ss ymb e ME n ymb e JE n n any of the above followed by index i e symbol n index implying index register of given value any of the above preceded by indicating INDIRECT ADDRESS Any place symbol appears above the character S can appear which refers to the location counter value like S 360 EXAMPLES AI 0 1000 7 AW AREAH12 7 A BCS 8 LOOP1 H BDR 2 LOOP E LW 1 VECTOR 5 2 BCS 15 5 2 CS102FP2 03 2 ASSEMBLER INSTRUCTIONS PSEUDO OPS a ORG sets
329. ted n I O request after the This module contains code for option group and ti 4 I O This m before bein CS411FP4 01 T INTERVAL LENGTH ROUTINE IOILENG odule determines how long a job will be allowed to execute g halted for an 1 0 request This might reasonably be an internal subroutine of DISPATCHER This m with amp EOP v odule implements option group IOI and may be involved ariable 5 CHANNEL REQUEST PROCESSOR CRPROC This m odule is called whenever a job is to relinquish control of CPU and attempt to perform I O It removes the job from control of he CPU and markes the CPU idle also requesting that the DISPATCHER called immediately to try to schedule a job into the CPU If the he channel hannels a E b required CHANNEL is currently idle CHANALC can be called to allocate t e c I to the job and the job marked as in I O status If the hannel is not available it must enter a queue of jobs waiting for IOR and wait here until the needed channel becomes fr n this case the job is marked as in WAIT state This module typically implements the option group IOO 6 CHANNEL ALLOCATOR CHANALC This routine is supplied with a job and a channel and essentially gives the c hannel to the job determines how long it will be until the I O interrupt occurs which ends the I O
330. tely DOCUMENT 3 DON T Place a comment card before every statement This bad habit makes programs absolutely unreadable Embedded comments should be used to block programs into logical sections not to explain the function of individual statements DON T bury code with too many interspersed comments If so many comments are necessary place them in blocks ahead of the program segments and not in the middle DO put a comment on nearly every machine instruction Comments are also helpful for explanations of variables and flags Each comment should describe the function of its statement and generally it alone If a comment is needed to describe the function of a block of half a dozen cards it probably should be placed on a comment card preceding the block of code These comments should be punched when the program is originally punched A good technique is to add these comments while keypunching the program Often this results in catching many mistakes at that point It is noted that few novice programmers do this while most experts do It is also noted that many programmers who do this wish they had started doing so earlier since they realize how much time they had wasted by not commenting the original deck DO use TITLE SPACE and EJECT commands The command I E A HEADING ESSAGE skips the listing to a new page and prints the heading message at the top of every page until another TITLE comm
331. th of DSECT using NL EQU NAME and how this can be used in DC and other instructions for real ease of change note importance of DROP especially with different registers and over what BROPS2 does 11 12 13 01 27 HANDOUTS ASSIGN types of assemblers answer questions about assignments bas CS411TPA large 4 pass assemblers assembler classification by number of passes Two pass Pass 1 opcodes table literal table location counter ORG START EQU registers 2 allocate storage 74 1 03 tc etc symbol Pass 2 produces object code for machine ops DC s etc base register table ITL listing PRINT Four Pass NOT Fil F2 initialization Pass 1 differences betw E POINT operation hash opcodes source program scan into gl E n Assemblers F and G phases versus passes obal get system macros from libraries F3 pass 2 F7 pass 3 location FI ESD written out F8 pass 4 FPP Global Dictionary Local Dictionary symbols NOTE addrs parameters for macros set variables Values macro expansion final assembly post processor opcodes counter etc TXT cards XREF macro names NOTE addrs use of NOTE POINT stack object code dictionary build macro dictionaries GBLx 1 for each macro open code
332. the following are all comparison operations C CD CE CH CL CP CR 16 2 DR CER CLC CIR MODIFIE RS The followi ng lists verbs and their common modifiers VERBS MODIFIERS MEANING EXAMPLES A C S L Logical addition comparison or subtract is used rather than algebraic EX AL CL CLC SIR B AL And Link form of branch for doing linkage to subroutine so it can return EX BAL BALR E Condition branch or not depending on a previously set condtion IF GOTO EX BC BCR GT Count branch form used to decrement a register and branch if value not zero DO LOOP EX BCT BCTR X indeX branch form for incrementing and testing index quantities DO LOOP EX BXH BXLE L Complement used to set a register to complement itself or other Y ABS X EX LNR LNDR P Positiv set register to positive value from self or other Y ABS X EX LPR LPER E Test set register to value from self or other L ST M Multipl several registers are loaded or stored in one operation EX LM STM CS102M1 04 3 DATA TYPES As noted previously a data type character is usually the same as that used in a DC or DS statement to obtain a given type of data If a type character is omitted it usually implies that the instruction operates on 32 bit fullword binary quantities such as A C S etc
333. the higher numbered register is used F ASSIST only in ASSIST all values for USING statements are relocatable and register 0 is handled exactly the same as any other which is slightly different from the standard handling Also the ASSIST section ID s range from 1 to 255 only EL ASBROPS2 02 ASSIST INTERFACE REQUIREMENTS FOR BROPS2 General register conventions are given in ASREPLGD This section describes the module to be written with the specific requirements for each used of the entry points of the BROPS2 module The register notation is that from ASREPLGD CSECT NAME BROPS2 A T ENTRY POINTS BRINIT BRUSIN BRDROP BRDISP ENTRY AND EXIT CONVENTIONS BRINIT is called one time at beginning of asembly to perform any initialization required by BROPS2 Must be serially reusable and so cannot just DC any tables to required values B BRUSIN is called whenever a USING is processed ENTRY CONDITIONS RA number of register which can be used 0 15 RB address declared for base register 0 2 24 1 RC section ID of the address 1 255 C BRDROP is called when a DROP is found ENTRY CONDITIONS RA number of register to be dropped 0 15 EXIT CONDITIONS RB 0 if register was an active base register nonzero value if register was not usable at the time D BRDISP is called to conve
334. tine essentially includes the CALL macro or equivalent code i e it uses standard conventions It typically assumes that registers 0 1 14 15 may be modified without causing trouble This method is efficient and general but can cause trouble if used improperly 2 COMBINED FORM SPECIAL NONDESTRUCTIVE LINKAGE In some cases it may be useful to define a macro instruction which invokes a subroutine but can be used ANYWHERE without disturbing any registers changing the condition code or requiring that certain of the registers not b register 1535 th etos ones being used as base registers This is the kind of linkage used from XDECO XPRNT to XXXXPRNT in to particular XXXXDECO The following shows the general form of such a linkage setup giving first the kind of code to be generated by the macro part then the entry and exit code for the associated routine NOTE label is typically an amp SYSNDX generated unique label STM 14 0 label save registers to be changed evaluate arguments of macro any required Load Addresses must be done using LA 0 argument since doing LA into any other register could destroy a base register If more than one argument is needed the remaining ones can be stored into control block after label Examples LA 0 argument ST 0 label 12 2nd argument one arg left in RO after all arguments are evaluated and saved an
335. tine the one which calls no others need not set up a save area The reason for this is the save area is for the use of any called routines but that the lowest level routine will have no called routines It is important to know the conventions on save areas but the use of XSAVE AND XRETURN consult appropriate documentation can reduce the problems in coding and linking save areas THE NAME CONVENTION is a means of having the EBCDIC form of the name of a routine appear at certain key places on dumps To use this convention the first four bytes of a routine must be a branch on 15 as a base register which passes over a series of bytes These bytes contain the EBCDIC form of the name of a routine and also a length count for this name area This example shows how to code a name field name CSECT B m 1 4 15 DC X m DC CLm name next instruction The value of m must be odd in order to have the next instruction properly aligned An alternate approach uses the convention on register 15 name CSECT USING name 15 B NEXTINST DC X m DC CLm name NEXTINST next instruction LINKAGI ti l ws Notes O S follows these conventions strongly In particular the system often destroys the contents of registers 0 1 14 and 15 when it returns control from a system macro an SVC or another system function One must SAVE THE CONTENTS of these registers BEFORE exec uting one of these functions hard to locat rrors will frequently o
336. tions which are the heaviest used have the shortest mnemonics and most of the other instructions can be built from them by adding more letters CS102M1 03 EXAMPLES OF COMMONLY USED MNEMONIC ELEMENTS The following sections explain the common mnemonic elements VERBS MEANING COMMENTS Add two numbers which may be binary decimal or floating Branch to another instruction like GOTO Compare two fields numbers or character strings ConVert a number from one base to another Divide one number by another Load a quantity into a register from another or from storage Multiply one number by another MoVe information from one area in storage to another aNd information together logical AND Or information together logical OR Subtract one number from another STore a register or part of one into storage exclusive or information together logical exclusive OR For example note that a given VERB may begin many instructions which immediately shows they are related to each other For example the following are all comparison operations C CD CE CH CL CP CR 16 2 DR CER CLC CIR MODIFIE RS The followi ng lists verbs and their common modifiers VERBS MODIFIERS MEANING EXAMPLES A C S L Logical addition comparison or subtract is used rather than algebraic EX AL
337. tions except TR TRT ED EDMK Branching except EX 02 04 give out final project discuss assembler interpreters concepts of assemblers 2 pass assemblers how to set up opcode and symbol tables indexed jump methods output desired go over structure of SIGMA 4 5 computer and its interpreter noting indirect addressing in particular HANDOUTS CS102FP1 01 08 general assembler interpreter descr CS102FP2 01 06 specific material for final project ASSIGN Final project due 13 March described in CS102FPx 02 07 decimal numbers and conversions zoned packed decimal to and from binary PACK UNPK CVB CVD equivalent codes using M D loops for decimal binary decimal examples of various formats conversions T READINGS STRUBLE Chapter 5 106 110 Chapter 218 228 228 233 02 09 misc review misc instructions program mask SPM instruction use of program mask review BXLE BXH etc 02 11 MIDTERM covered data representations most standard instructions hand assembly etc 17 18 19 20 21 CS102TPA 05 02 14 on midterm and final project review of midterm results and problem areas final project overall structure useful modules and how to set them up decimal scan and output conversions symbol scan symbol table manager opcode lookup hexadecimal output etc review of BXLE loop control HANDOUT CS102P
338. tting up the 256 byte table needed to use a TRT instr The following is an example of code which scans until it finds the first instance of any letter fast general scanning Also fROM A to R in a string LA 1 STRING init address of string LOOP TRT 0 256 1 TABI scan BNZ 12 skip if any found LA 12 000 EL increment B LOOP go back for next wales at this pt Rl gt 1st letter NAN remaining code TAB1 DC 256X 00 define whole tabl ORG TAB1 C A back up location counter to position of A in table DC 9X 04 nonzero stop on AIJ ORG TABI C J now to position J DC 9X 04 nonzero stop on J R ORG 7 set loccntr to end again The above table is used to STOP on any letter A R The table below stops on any character EXC EPT letters A R TAB2 DC C A X 01 stop on anvthing before A DC 9x 00 skip over A I DC CII SECT O A stop on ones between I J DC 9x 00 skip over J R DC 255 C R X 01 stop on everything after R CS411MC1 06 b HEX CONVERSION USING TR PACK A simple trick exists for obtaining conversion from hexadecimal to binary as long as the hexadecimal number ahas no more than 14 digits As an example the code below can be used to convert hex numbers of up to 8 digits Assume first that the 1 8 digit hexadecimal number has been moved to an 8 byte work area right justified NOT LEFT which is easier and fill
339. type of card so that it appears as XXXXXXXX DD UNIT xxx As a result the OS RDR thinks that the data set will be read from unit xxx so that it does not try to SPOOL the input In any case the input no longer follows that JCL card because HASP feeds the RDR only the JCL cards of a user job During this process HASP connects up the device address xxx to the specific input data set which had been previously SPOOLed XXXXXXXX DD SYSOUT x HASP also has a large number of UCB s for nonexistent pseudo printers punches It does the same thing to this kind of card as it does to DD cards except that it only allocates the pseudo devices and will later save the output which is written to them As soon as the RDR finishes reading a job an initiator can immediately initiate it since HASP chooses jobs appropriately When the initiator chooses i o devices it finds that it can always allocate devices for unit record i o since HASP had already checked to make sure a pseudo reader printer punch was available for each SYSIN or SYSOUT data set OSHASP 07 Finally a job step of the user job executes When it wishes to read cards or print lines it acts as though it were using a real device attached to the system and 0S 360 accepts this Whenever an SVC 0 is issued to request such I 0 HASP intercepts it HASP may be entered for any of the following reasons 1 WTO WTOR WTL HASP adds own processing as desired 2
340. type of data being operated on and is usually the same letter as that used to define a constant of the given type such as H halfword P packed decimal etc 4 MACHINE FORMAT gives the type of machine instruction being used This is most typically done by adding R or I to an RX mnemonic to obtain a similiar RR or SI instruction In general the RX instructions which are the heaviest used have the shortest mnemonics and most of the other instructions can be built from them by adding more letters CS102M1 03 EXAMPLES OF COMMONLY USED MNEMONIC ELEMENTS The following sections explain the common mnemonic elements VERBS MEANING COMMENTS Add two numbers which may be binary decimal or floating Branch to another instruction like GOTO Compare two fields numbers or character strings ConVert a number from one base to another Divide one number by another Load a quantity into a register from another or from storage Multiply one number by another MoVe information from one area in storage to another aNd information together logical AND Or information together logical OR Subtract one number from another STore a register or part of one into storage exclusive or information together logical exclusive OR For example note that a given VERB may begin many instructions which immediately shows they are related to each other For example
341. uivalent outline of topics to be covered later macros operating systems input output large machines assemblers loaders memory organization JCL OS 360 etc useful macros some review XSNAP XDUMP XSTOP XSET XREAD XPRNT XPNCH XDECI XDECO HANDOUTS INFOR411 basic information amp what to expect 01 08 OMITTED 01 11 administrative details linkage conventions and macros Registers on entry to subroutine 1 13 14 15 Parameter list format PARM field setup as special case Linkage actions expected of subroutine save area format Code for return from subroutine function return return code XSAVE RGS SA BR TR operands XRETURN RGS SA TR operands use of XSET with them ASSIGNMENT run programs as described by writeup DUMPSJCL bring dumps to next class READING ASM LANGUAGE MANUAL Sections 1 2 3 except on External Dummy Sections CXD DXD COM 4 5 except OPSYN CCW ICTL ISEQ PUNCH REPRO HANDOUTS LINKAGE linkage conventions for S 360 DOCUMENT documentation techniques INST hints on machine and assembler instructions XREAD XREAD XPRNT XPNCH macro descriptions XSNAP XSNAP XSTOP XSET macro descriptions XSAVE XSAVE XRETURN macro descriptions DUMPSJCL simple JCL sample runs to get dumps 01 13 interpreting output dumps debug techniques go through entire run of ASGCG using MSGLEVEL 1 LOG System log info
342. up would use a 360 50 hooked to a 360 75 An advantage over HASP is that ASP offers somewhat better setup facilities for optimizing use of tapes and non SPOOL disks A disadvantage is the requirement of two CPU s either of which may have problems and thus stop th ntire system B LASP LOCAL ASP or CLASP CLOSELY LINKED ASP These are versions of ASP in which the code from the smaller computer is moved over into a region on the larger machine This allows an ASP system to be run on one processor If the system is also run under straight ASP it requires switches to switch the unit record devices over to the bigger machine It also requires more memory than HASP but does allow the system to run even with one CPU down C TUCC HYPERDISK This method uses LCS plus part of a 2314 disk pack to simulate the entire disk pack containing heavy used systems programs The most recently used tracks of this disk are kept in LCS thus making the disk effectively faster without changing the internals of 0S 360 V PS OSHASP 09 U CC 360 67 SYSTEM OS MVT WITH HASP The following tables gise the current layout with no guarantee of future appearance as of 6 12 72 for the 360 67 at the PSU CC The svstem has both fast core 1024K and Large Core Storage 2048K LOW HIGH K LOW HIGH S 2928 3072 144 2DC000 300000 HASP 1968 2928 960 IECOOO 2DC000 FMGR 1628 1968 340 197000 1ECOOO RJE 1346 16
343. used from XDECO to XXXXDECO XPRNT to XXXXPRNT etc The following shows the general form of such a linkage setup giving first the kind of code to be generated by the macro part then the entry and exit code for the associated routine NOTE label is typically an amp SYSNDX generated unique label STM 14 0 label save registers to be changed evaluate arguments of macro any required Load Addresses must be done using LA 0 argument since doing LA into any other register could destroy a base register If more than one argument is needed the remaining ones can be stored into control block after label Examples LA 0 argument ST 0 label 12 2nd argument one arg left in RO after all arguments are evaluated and saved and ONLY THEN it is now possible to modify registers L 15 label 4 V type adcon for routine CNOP 2 4 make sure next inst not on F boundry BALR 14 15 call routine also point 14 at the argument list following DC V subroutine entry point adcon to get there label DS 3F 3 words for saving 14 15 O DS F space for arguments after first DS OR DC space here for any remaining arguments the subroutine will return control to next instruction LM 14 0 4 14 reload registers Note that this is only safe way since 15 might have current base register ASPRGTCI 05 The following shows the tvpical code used to enter supporting module used with the previous macro expans
344. ve starting at JOB for programs III and IV The following test data should be used for each program A B D E 5 2 4 2 2 2 I 10 1 1 4096 1 1 f Note that the columns they are punched in should not matter CS102AS1 03 COMPUTER SCIENCE 102 ASSIGNMENT 2 This assignment uses the concept of indexing into an array of elements Es BASIC PROGRAM A Read a card and echo print containing a maximum of 20 numbers Convert the numbers to hex XDECI and store them in successive fullwords in memory Use a loop to eliminate redundant coding Then for each card find the maximum value and the minimum value printing out these numbers with appropriate labels B Form of data 1 Each card contains a maximum of 21 numbers where the first number the number of numbers on the card You will need the first number for a counter in the loop in part A 2 There are an unspecified of data cards i e make your program general to accept any of data cards L DATA FOR YOUR PROGRAM 56 76 76 11 123 432 123 748 9087 0 33 33 45 10 6 90 145 1024 6698 1024 345 Ur O Ju Iris tt CS102FP1 01 COMPUTER SCIENCE 102 FINAL PROJECT DUE Pages 1 8 INTRODUCTION One of the concerns of a programmer is to design computer systems languages and translators for the languages The monitor is that part of the system which controls the running of the translators an
345. ventions for names Typical rules are 1 Every register EQU begins with R and symbolic registers are used EVERYWHERE 2 EAch CSECT and DSECT will have a name such that every label in the section begins with the first 1 2 or 3 choice letters of that section 3 Important symbols used across many CSECTS might be assigned to begin with particular characters Of these is a often chosen to mark such symbols CS411FP3 06 Why bother with rules such as are given on the previous page Briefly it is because such rules cost little effort but save much time and trouble Among other things 1 It is EXTREMELY annoying to lose a run with 1 STATEMENT FLAGGED because a statement label is duplicated very easy to do in a large program if no rules are followed 2 When looking for an error or at a listing for any reason the NAME of something shows immediately where it is located what section it belongs to and perhaps what its purpose is This is particularly useful for dsect symbols belonging to specific control blocks 3 The name methods given cause related symbols to appear close together in the CROSS REFERENCE listing which also saves time D SYMBOLIC CODING DUMMY SECTIONS DSECTS are an absolute necessity Flag values and switches are also easier set and tested if they have mnemonic EQU values rather than
346. wing related instructions INSTRUCTION ASSEMBLED ri 1901 2 3 4 49034002 4 5 6 59056004 The first teo bits give the the third and fourth give a Data Type The fifth eighth bits give the ther ssence is only 1 Compare which is branched to after the operands are obtained CS102TPA 01 LT COMPUTER SCIENCE 102 TOPICS COVERED HANDOUTS WINTER TERM 1972 MASHEY The handouts given are described in file CS102HN DATI CJ TOPICS HANDOUTS READING ASSIGNMENTS 1 2 01 07 READI 01 10 READI HANDO 72 introduction to course prerequistites 101 401 equiv listed text materials for course 1 STRUBLE ASSEMBLER LANGUAGE PROGRAMMING IBM SYSTEM 360 2 IBM SYSTEM 360 PRINCIPLES OF OPERATION POP 3 IBM S 360 OS ASSEMBLER LANGUAGE 4 PSU ASSIST INTRODUCTORY ASSEMBLER USER S MANUAL 25 cents at 426 McAllister 5 IBM S 360 REFERENCE CARD GREEN CARD BRING TO CLASS Gl tA T introduction to information representation in computer memorv addressing similiaritv to FORTRAN vector witl index beginning at O rather than 1 elements of memory in S 360 byte halfword fullword doubleword positional notation number systems binary octal decimal hexadecimal conversion between them uses representations of binary
347. x x kx x k KKK KKK SPACE 2 ROUND GLOBAL VARIABLE amp LNGTH TO EVEN NUMBER IF NECESSARY EN SETA amp LNGTH 1 2 2 ATTACH BLDL LIST TO amp NAME DS 0H ALIGN TO HALE WORD BOUNDRY AND STICK TO amp NAME DC H amp NUM NUMBER OF BLDL LIST ENTRIES DC H amp LNGTH DEFINE LENGTH IN BYTES OF EACH ENTRY END SPACE 10 ACRO LIST amp ENTRY GBLA amp LEN CLA amp T1 SPACE 2 KK x x x x x x x x III THIS MACRO CAUSES EACH LIST ENTRY TO BE PLACED INTO THE BLDL LIST AND TO ALLOCATE STORAGE FOR EACH ENTRY INTO WHICH THE CONTROL PROGRAM WILL PLACE ANY DIRECTORY INFORMATION IT CAN FIND ABOUT EACH ENTRY EACH ENTRY IS amp LNGTH BYTES LONG KK x x KK KIKKI SPACE 2 SET T1 TO amp LNGTH 8 FOR CORRECT FILL LENGTH AFTER ENTRY NAME DEFINITION SETA amp LEN 8 GET LENGTH 8 DC CL8 amp ENTRY ENTRY NAME IN CHARACTER LEFT JUSTIFIED AND RIGHT PADDED WTH BLANK DC XL amp T1 FF GENERATE CORRECT ENTRY LENGTH M li a T S D SE BPAMIO SE P BPAM x P 5 S RI E DCBADD E FNDADD E DECB E AREA E ONE E HEADING E NEXTHD E BLDLCNT E RCRDPT E END E BASE E SAVE E RETADD E ENTRY E S a T 5 B SE a N SE ii G x G SE SE OPDCB VONtwHO NONE
348. xecution Only when a job is finished punching is its disk space released This allows for jobs to be saved across system crashes and for such useful services as repeating output by operator control C DASD STORAGE MANAGEMENT IN HASP HASP manages its DASD storage quit fficiently not only needing NO accesses to DASD to allocate or deallocate space but also doing a good job of minimizing arm movement on moveable head devices HASP requires the use of entire volumes normally 2311 or 2314 disks For example the PSU CC s 360 67 has 3 2314 disk packs for HASP The management of this storage works as follows A MASTER CYCLINDER BIT MAP is maintained in HASP This is a string of bytes in which each bit represents 1 CYLINDER on the SPOOL disks for example 600 bits for the cylinders on 3 packs A one bit represents a FREE CYLINDER while a zero bit shows that the given cylinder is allocated to some job HASP also remembers for each disk which cylinder was the last referenced thus always noting the current position of the read write heads OSHASP 08 Two JOB BIT MAPS exist for each job one for SYSIN data and the other for SYSOUT data Whenever a cylinder is required for a job HASP searches for a free one in the following fashion 1 It first searches the master bit map for a free cylinder at the current position of any read write head i e where it can read or write withou
349. xists in one form or another on XDS SIGMA 7 RCA SPECTRA 70 46 BURROUGHS B5500 B6500 etc GE 645 and IBM 360 67 used as 67 not as 65 like PSU does This type of system may be used to allow programs to be written which are not totally in memory and which can appear to the user to be much larger than the actual physical memory available on the computer system 4 MEMORY MERGE FOR ALLOCATED MEMORY These options are only meaningful with option MAC2 above why MAM1 AUTOMATIC COMPACTION memory is allocated from one end of storage and whenever a job terminates any jobs above it are moved down to fill the empty space Thus at any point in time the memory consists of 1 region of contiguous jobs and 1 contiguous region of free space MAM2 COMPACTION WHEN NECESSARY rather than compacting always this method only moves the jobs around when necessary to initiate some desired job which is larger than any hole currently existing but smaller than the total available space With either option all CPU s must be stopped while the programs are being moved around since they cannot execute a job while in motion In this case ALL CPU s are stopped from executing any jobs for the following length of time time used amp MAMOVE K BYTES MOVED amp MUMCPUS i e the CPU s share the task of moving the jobs around as needed Note that the decision on whether to
350. xtra byte decimal output CVD UNPK OI for plus number with leading zeroes decimal output begin on ED EDMK doing parts with basic workings of ED and standard pattern for integer numbers CS102AS1 01 T T COMPUTER SCIENCE 102 ASSIGNMENT 1 DUE This assignment covers simple input output binary arithmetic for fullword and halfword numbers and basic data movement and testing codes for handling such numbers AI BASIC PROGRAM The basic program should do the following A Read a card XREAD and print it out immediately called an ECHO CHECK standard practice The card contains 5 numbers punched on it which are to be scanned and converted XDECI to binary form and placed in 5 consecutive fullwords in memory Print the hexadecimal values of these 5 words 20 bytes using XDUMP B Perform the following computations in a straightforward way storing each result in name given using RX instructions where you can 1 F A B C 2 G A B C LCR useful 3 H A BE 4 I A B be careful watch for negativ s 5 J MOD A B i e remainder from 4 6 K A E B C D C Print all of the above values F K in hexadecimal XDUMP then also print them in decimal using XDECO and XPRNT print their values an headings all on one line D According to the sign of result H pri
351. y contiguous free block s so that they can together be used to satisfy a larger request If this is not done memory becomes FRAGMENTED into small free blocks FIRST FIT a table of free blocks is usually kept in order by address The request is satisfied from the first area of size gt request size with the unneeded portion remaining in the free table This is usually the easiest to implement and has the best statistical properties for most applications BEST FIT like FIRST FIT except that storage is allocated from the smallest free area of size gt request size thus trying to maintain large free areas May be best under some conditions although not usually as good as FIRST FIT BUDDY SYSTEM see KNUTH Chapter 2 for this method CS411FP2 03 Se MEMORY ALLOCATION CONTIGUITY OF ALLOCATION MAC1 CONTIGUOUS NO MOVEMENT the memory for a job must be allocated in one contiguous unit and a job may never be moved from that area MAC2 CONTIGUOUS MOVEMENT the job must be allocated contiguous memory but may be moved around in memory if necessary NOTE a system must use bounds registers or more sophisticated setup in order to use this method MAC3 NONCONT IGUOUS the required number of pages are allocated anywhere at all in memory This method definitely requires special hardware is most commonly used for Time Sharing systems and e
352. y of showing their relationships adn the elements present in each D F stand for Verb Modifier Data type and OP CODE VM DF TYPE COMMENTS E C RX fullword algebraic compare the basic one CL C L RX fullword logical comparison logical modifier CD E D RX compare double precision floating numbers CE G E RX compare single precision floating numbers CH Cc H RX compare a register algebraicly with halfword from storage with sign extension CE E P ss compare two packed decimal numbers CR Cc R RR compare two fullword values algebriacly gotten from C by adding R CLC CEC ss compare logically character strings CLI C L I sI compare logical immediate a byte in memory with the one inside the instruction CDR CD R RR compare double precision in registers CER CE R RR compare single precision in registers The Sytem 370 computers have some additional opocdes CLM CL M RS compare logical masked from register to mem CLCL CLCL RR compare logical character strings long up to 16 million bvtes in one compare Consider the problem of writing a FORTRAN program which would simulate the operation of the instructions above Memorv GP Registers variables representing PSW the Fetch Instruction arrangement Decod r Fetch Operands of the opcodes would make would not be necessarv to code each i example maintain and go through Li E etc

asmintro - Jay Moseley

Contents

Download Pdf Manuals

Related Search

Related Contents