Figure 1 illustrates a C program that contains a divide-by-zero error. The code was compiled with RENT so static and external variables need to be calculated from the WSA field. The code was compiled with LP64, GONUM (to produce statement numbers) and XREF, LIST and OFFSET to generate a listing, which is used to calculate addresses of functions and data. The code was processed by the binder with MAP to generate a binder map, which is used to calculate the addresses of static and external variables. The program was created with the option TERMTHDACT(UADUMP) which produced both a Language environment dump and a system dump.
Figure 1. C
routine with a divide-by-zero error (AMODE 64)
/* C Routine with a Divide-by-Zero Error */
#pragma options(noinline)
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
int statint = 73;
int fa;
int funcb(int *pp);
int main(void) {
int aa, bb=1;
aa = bb;
aa = funcb(&aa);
return(aa);
}
int funcb(int *pp) {
int result;
fa = *pp;
result = fa/(statint-73);
printf("Result = %d\n",result);
return result;
}To debug this routine, use the following steps:
- Locate the Original Condition message in the Condition Information
for Active Routines section of the dump. In this example, the message
is CEE3209S. The system detected a fixed-point divide
exception. This message indicates the error was caused by an attempt
to divide by zero. For additional information about CEE3209S, see z/OS Language Environment Runtime Messages.
The traceback section of the dump indicates that the exception occurred at offset X'52' within function funcb. This information is used along with the compiler-generated Pseudo Assembly Listing to determine where the problem occurred.
If the GONUMBER compiler option is specified, statement number information is in the dump. Figure 2 shows the generated traceback from the dump.
Figure 2. Sections of the dump from example C/C++ routine (AMODE 64) (Part 1 of 2)CEE3DMP V1 R8.0: Condition processing resulted in the unhandled condition. Tue Feb 21 16:50:50 2006 Page: 1 Information for enclave main Information for thread 25AC70A000000000 Traceback: DSA Entry E Offset Statement Load Mod Program Unit Service Status 00000001 CEEHDSP +00000000 CELQLIB CEEHDSP HLE7730 Call 00000002 CEEOSIGJ +0000094E CELQLIB CEEOSIGJ HLE7730 Call 00000003 CELQHROD +0000024E CELQLIB CELQHROD HLE7730 Call 00000004 CEEOSIGG -1D58AE60 CELQLIB CEEOSIGG HLE7730 Call 00000005 CELQHROD +0000024E CELQLIB CELQHROD HLE7730 Call 00000006 funcb +00000052 18 CDIVZERO CDIVZERO Exception 00000007 main +00000026 12 CDIVZERO CDIVZERO Call 00000008 CELQINIT +00001340 CELQLIB CELQINIT HLE7730 Call DSA DSA Addr E Addr PU Addr PU Offset Comp Date Attributes 00000001 00000001082FAAC0 0000000025793E88 0000000025793E88 00000000 20060109 XPLINK EBCDIC POSIX Floating Point 00000002 00000001082FD3E0 000000002588FE28 000000002588FE28 0000094E 20051214 XPLINK EBCDIC POSIX Floating Point 00000003 00000001082FDDE0 00000000257B50D8 00000000257B50D8 0000024E 20051214 XPLINK EBCDIC POSIX Floating Point 00000004 00000001082FE020 00000000258894E0 00000000258894E0 1D58AE60 20051214 XPLINK EBCDIC POSIX Floating Point 00000005 00000001082FEE40 00000000257B50D8 00000000257B50D8 0000024E 20051214 XPLINK EBCDIC POSIX Floating Point 00000006 00000001082FF080 00000000257080D0 0000000000000000 ******** 20060221 XPLINK EBCDIC POSIX IEEE 00000007 00000001082FF180 0000000025708170 0000000000000000 ******** 20060221 XPLINK EBCDIC POSIX IEEE 00000008 00000001082FF280 000000002570B010 000000002570B010 00001340 20051214 XPLINK EBCDIC POSIX Floating Point Fully Qualified Names DSA Entry Program Unit Load Module 00000006 funcb PLPSC://'POSIX.CRTL.C(CDIVZERO)' /u/alfcar/tools/CDIVZERO 00000007 main PLPSC://'POSIX.CRTL.C(CDIVZERO)' /u/alfcar/tools/CDIVZERO Condition Information for Active Routines Condition Information for PLPSC://'POSIX.CRTL.C(CDIVZERO)' (DSA address 00000001082FF080) CIB Address: 00000001082FBE00 Current Condition: CEE0198S The termination of a thread was signaled due to an unhandled condition. Original Condition: CEE3209S The system detected a fixed-point divide exception (System Completion Code=0C9). Location: Program Unit: PLPSC://'POSIX.CRTL.C(CDIVZERO)' Entry: funcb Statement: 18 Offset: +00000052 Machine State: ILC..... 0002 Interruption Code..... 0009 PSW..... 0785240180000000 0000000025708124 GPR0..... 0000000000000000 GPR1..... 00000001082FFA40 GPR2..... 0000000108414340 GPR3..... 0000000108413430 GPR4..... 00000001082FF080 GPR5..... 0000000108300060 GPR6..... 0000000000000000 GPR7..... 0000000100000001 GPR8..... 0000000025700CC8 GPR9..... 00000000257081B8 GPR10.... 0000000025708278 GPR11.... 0000000108414330 GPR12.... 0000000100005300 GPR13.... 0000000000006F60 GPR14.... 00000000257BF218 GPR15.... 000000000000001F FPC...... 00000000 FPR0..... 4327CCCA 93BCEEF1 FPR1..... 00000000 00000000 FPR2..... 00000000 00000000 FPR3..... 00000000 00000000 FPR4..... 3C100000 00000000 FPR5..... 00000000 00000000 FPR6..... 34100000 00000000 FPR7..... 00000000 00000000 FPR8..... 00000000 00000000 FPR9..... 00000000 00000000 FPR10.... 00000000 00000000 FPR11.... 00000000 00000000 FPR12.... 00000000 00000000 FPR13.... 00000000 00000000 FPR14.... 00000000 00000000 FPR15.... 00000000 00000000
VR0...... 4327CCCA 93BCEEF1 00000000 00000000 VR1...... 00000000 00000000 00000000 00000000
VR2...... 00000000 00000000 00000000 00000000 VR3...... 00000000 00000000 00000000 00000000
VR4...... 3C100000 00000000 00000000 00000000 VR5...... 00000000 00000000 00000000 00000000
VR6...... 3C100000 00000000 00000000 00000000 VR7...... 00000000 00000000 00000000 00000000
VR8...... 00000000 00000000 00000000 00000000 VR9...... 00000000 00000000 00000000 00000000
VR10..... 00000000 00000000 00000000 00000000 VR11..... 00000000 00000000 00000000 00000000
VR12..... 00000000 00000000 00000000 00000000 VR13..... 00000000 00000000 00000000 00000000
VR14..... 00000000 00000000 00000000 00000000 VR15..... 00000000 00000000 00000000 00000000
VR16..... 00000000 00000000 00000000 00000000 VR17..... 00000000 00000000 00000000 00000000
VR18..... 00000000 00000000 00000000 00000000 VR19..... 00000000 00000000 00000000 00000000
VR20..... 00000000 00000000 00000000 00000000 VR21..... 00000000 00000000 00000000 00000000
VR22..... 00000000 00000000 00000000 00000000 VR23..... 00000000 00000000 00000000 00000000
VR24..... 00000000 00000000 00000000 00000000 VR25..... 00000000 00000000 00000000 00000000
VR26..... 00000000 00000000 00000000 00000000 VR27..... 00000000 00000000 00000000 00000000
VR28..... 00000000 00000000 00000000 00000000 VR29..... 00000000 00000000 00000000 00000000
VR30..... 00000000 00000000 00000000 00000000 VR31..... 00000000 00000000 00000000 00000000
Storage dump near condition, beginning at location(0000000025708112)
+0000 0000000025708112 0004E300 70000014 A70BFFB7 8E600020 |..T.....x....-..|
+0010 0000000025708122 1D60B904 00075000 48C0E320 48C00014 |.-....&...T.....|
Parameters, Registers, and Variables for Active Routines:
funcb (DSA address 00000001082FF080):
DOWNSTACK DSA
Saved Registers:
GPR0..... **************** GPR1..... **************** GPR2..... **************** GPR3..... ****************
GPR4..... 00000001082FF080 GPR5..... BBBBBBBBBBBBBBBB GPR6..... 00000000257B50D8 GPR7..... 0000000100000001
GPR8..... 0000000025700CC8 GPR9..... 00000000257081B8 GPR10.... 0000000025708278 GPR11.... 0000000108414330
GPR12.... 4040404040404040 GPR13.... 4040404040404040 GPR14.... 4040404040404040 GPR15.... 4040404040404040Figure 3. Sections of the dump from example C/C++ routine (AMODE 64) (Part 2 of 2)GPREG STORAGE: Storage around GPR0 is invalid. Storage around GPR1 is invalid. Storage around GPR2 is invalid. Storage around GPR3 is invalid. Storage around GPR4 (00000001082FF080) +0800 00000001082FF880 00000001 082FF180 00000001 08300060 |......1........-| +0810 00000001082FF890 00000000 25708170 00000000 2570819A |......a.......a.| +0820 00000001082FF8A0 00000000 25700CC8 00000000 25709138 |.......H......j.| +0830 00000001082FF8B0 00000000 25708278 00000001 08414330 |......b.........| +0840 00000001082FF8C0 00000001 00005300 00000000 00006F60 |..............?-| +0850 00000001082FF8D0 00000000 257BF218 00000000 0000001F |.....#2.........| - In the traceback, statement number 12, corresponding to DSA 7, refers to line: aa = funcb(&aa); in the listing. This is were entry funcb is called. Similarly, statement number 18, corresponding to DSA 6, points to line: result = fa/(statint-73); in the listing. This line is where the divide by zero exception takes place.
- Locate the instruction with the divide-by-zero error in the Pseudo
Assembly Listing in Figure 4.
The offset (within funcb) of the exception from the traceback (X'52') reveals the divide instruction: DR R6,R0 at that location. Instructions at offsets X'32' through X'58' refer to the result = fa/(statint-73); line of the C/C++ routine.
Figure 4. Pseudo assembly listing (AMODE 64) (Part 1 of 3)OFFSET OBJECT CODE LINE# FILE# P S E U D O A S S E M B L Y L I S T I N G Timestamp and Version Information 000010 F2F0 F0F6 =C'2006' Compiled Year 000014 F0F2 F2F1 =C'0221' Compiled Date MMDD 000018 F1F5 F3F3 F1F5 =C'153315' Compiled Time HHMMSS 00001E F0F1 F0F8 F0F0 =C'010800' Compiler Version Timestamp and Version End 15694A01 V1.8 z/OS XL C 'POSIX.CRTL.C(CDIVZERO)': funcb 02/21/06 15:33:15 Page 148 OFFSET OBJECT CODE LINE# FILE# P S E U D O A S S E M B L Y L I S T I N G 000015 | * int funcb(int *pp) { 000028 @2L0 DS 0D 000028 00C300C5 =F'12779717' XPLink entrypoint marker 00002C 00C500F1 =F'12910833' 000030 00000108 =F'264' 000034 00000100 =F'256' 000000 000015 | funcb DS 0D 000000 EB59 4708 0024 000015 | STMG r5,r9,1800(r4) 000006 A74B FF00 000015 | AGHI r4,H'-256' 00000A End of Prolog 00000A C090 0000 006F 000000 | LARL r9,F'111' 000010 E310 4980 0024 000015 | STG r1,pp(,r4,2432) 000016 | * int result; 000017 | * fa = *pp; 000016 E360 4980 0004 000017 | LG r6,pp(,r4,2432) 00001C E300 6000 0014 000017 | LGF r0,(*)int(,r6,0) 000022 E360 4808 0004 000017 | LG r6,#Save_ADA_Ptr_2(,r4,2056) 000028 E360 6000 0004 000017 | LG r6,=A(fa)(,r6,0) 00002E 5000 6000 000017 | ST r0,fa(,r6,0) 000018 | * result = fa/(statint-73); 000032 E360 6000 0014 000018 | LGF r6,fa(,r6,0) 000038 E370 4808 0004 000018 | LG r7,#Save_ADA_Ptr_2(,r4,2056) 00003E E370 7008 0004 000018 | LG r7,=A(statint)(,r7,8) 000044 E300 7000 0014 000018 | LGF r0,statint(,r7,0) 00004A A70B FFB7 000018 | AGHI r0,H'-73' 00004E 8E60 0020 000018 | SRDA r6,32 000052 1D60 000018 | DR r6,r0 000054 B904 0007 000018 | LGR r0,r7 000058 5000 48C0 000018 | ST r0,result(,r4,2240) 000019 | * printf("Result = %d\n",result); 00005C E320 48C0 0014 000019 | LGF r2,result(,r4,2240) 000062 E360 4808 0004 000019 | LG r6,#Save_ADA_Ptr_2(,r4,2056) 000068 EB56 6010 0004 000019 | LMG r5,r6,=A(printf)(r6,16) 00006E B904 0019 000019 | LGR r1,r9 000072 0D76 000019 | BASR r7,r6 000074 0700 000019 | NOPR 0 000020 | * return result; 000076 E330 48C0 0014 000020 | LGF r3,result(,r4,2240) 000021 | * } 00007C 000021 | @2L3 DS 0H 00007C Start of Epilog 00007C E370 4818 0004 000021 | LG r7,2072(,r4) 000082 EB89 4820 0004 000021 | LMG r8,r9,2080(r4) 000088 4140 4100 000021 | LA r4,256(,r4) 00008C 47F0 7002 000021 | B 2(,r7)Figure 5. Pseudo assembly listing (AMODE 64) (Part 2 of 3)*** General purpose registers used: 1111011101000000 *** Floating point registers used: 1111111100000000 *** Size of register spill area: 128(max) 0(used) *** Size of dynamic storage: 0 *** Size of executable code: 144 000001 | * /* C Routine with a Divide-by-Zero Error from LE Debugging Guide */ 000002 | * #pragma options(noinline) 000003 | * #include <stdio.h> 000004 | * #include <stdlib.h> 000005 | * #include <errno.h> 000006 | * int statint = 73; 000007 | * int fa; 000008 | * int funcb(int *pp); 000009 | * int main(void) { 0000C8 @1L0 DS 0D 0000C8 00C300C5 =F'12779717' XPLink entrypoint marker 0000CC 00C500F1 =F'12910833' 0000D0 00000090 =F'144' 0000D4 00000100 =F'256' 000000 000009 | main DS 0D 000000 EB57 4708 0024 000009 | STMG r5,r7,1800(r4) 000006 A74B FF00 000009 | AGHI r4,H'-256' 00000A End of Prolog 000010 | * int aa, bb=1; 00000A A709 0001 000010 | LGHI r0,H'1' 00000E 5000 48C4 000010 | ST r0,bb(,r4,2244) 000011 | * aa = bb; 000012 E300 48C4 0014 000011 | LGF r0,bb(,r4,2244) 000018 5000 48C0 000011 | ST r0,aa(,r4,2240) 000012 | * aa = funcb(á); 00001C 4110 48C0 000012 | LA r1,aa(,r4,2240) 000020 E350 4808 0004 000012 | LG r5,#Save_ADA_Ptr_1(,r4,2056) 000026 A775 FF9D 000012 | BRAS r7,funcb 00002A 0701 000012 | NOPR 1 00002C B904 0003 000012 | LGR r0,r3 000030 5000 48C0 000012 | ST r0,aa(,r4,2240) 000013 | * return(aa); 000034 E330 48C0 0014 000013 | LGF r3,aa(,r4,2240) 000014 | * } 00003A 000014 | @1L2 DS 0H 00003A Start of Epilog 00003A E370 4818 0004 000014 | LG r7,2072(,r4) 000040 4140 4100 000014 | LA r4,256(,r4) 000044 47F0 7002 000014 | B 2(,r7) *** General purpose registers used: 1111011100000000 *** Floating point registers used: 1111111100000000 *** Size of register spill area: 128(max) 0(used) *** Size of dynamic storage: 0 *** Size of executable code: 72 Constant Area 000000 D985A2A4 93A3407E 406C8415 00 |Result = %d.. | PPA1: Entry Point Constants 000000 02 =AL1(2) Version 000001 CE =AL1(206) CEL signature 000002 07C0 =H'1984' GPR save mask 000004 00000090 =A(PPA2-PPA1) 000008 80800281 =F'-2139094399' Flags 00000C 0002 =H'2' Parm length/4 00000E 0503 =H'1283' Prol len/2; alloca reg; R4 change offset/2 000010 00000090 =F'144' Code length 000014 01000000 =F'16777216' Interface mapping flags 000018 0005 **** AL2(5),C'funcb' 000020 FFFFFEF8 =F'-264' Offset to Entry Point Marker PPA1 EndFigure 6. Pseudo assembly listing (AMODE 64) (Part 3 of 3)PPA1: Entry Point Constants 000000 02 =AL1(2) Version 000001 CE =AL1(206) CEL signature 000002 0700 =H'1792' GPR save mask 000004 00000068 =A(PPA2-PPA1) 000008 80800281 =F'-2139094399' Flags 00000C 0000 =H'0' Parm length/4 00000E 0503 =H'1283' Prol len/2; alloca reg; R4 change offset/2 000010 00000048 =F'72' Code length 000014 01000000 =F'16777216' Interface mapping flags 000018 0004 **** AL2(4),C'main' 000020 FFFFFF70 =F'-144' Offset to Entry Point Marker PPA1 End PPA4: Compile Unit Debug Block 000000 80000000 =F'-2147483648' Additional Flags 000004 84060238 =F'-2079981000' Flags 000008 0000000000000000 =D'0' R/O static Offset 000010 0000000000000000 =D'0' R/W static Offset 000018 0000000000000000 =D'0' Symbol Offset Table Offset 000020 FFFFFFFFFFFFFE80 =D'-384' CSECT Start Offset 000028 0000000000000000 =D'0' Code CSECT Size 000030 FFFFFFFFFFFFFE88 =D'-376' No program region 000038 00000000 =F'0' DWARF File Name 00003C **** C'' PPA4 End PPA2: Compile Unit Block 000000 0300 2204 =F'50340356' Flags 000004 FFFF FE40 =A(CELQSTRT-PPA2) 000008 FFFF FFC0 =A(PPA4-PPA2) 00000C FFFF FE50 =A(TIMESTMP-PPA2) 000010 0000 0000 =F'0' No primary 000014 9140 0000 =F'-1858076672' Flags PPA2 End - Verify the value of the divisor statint. The
procedure specified below is to be used for determining the value
of static variables only. If the divisor is an automatic variable,
there is a different procedure for finding the value of the variable.
Because this routine was compiled with the RENT option, find the WSA address in the Enclave Control Blocks section of the dump. In this example, this address is X'108300050'. Figure 7 shows the WSA address.
Figure 7. C/C++ CAA information in dump (AMODE 64)Enclave Control Blocks: . . . DLL Information: WSA Addr Module Addr Thread ID Use Count Name 0000000108300050 00000001 main - Routines compiled with the RENT option must also be processed
by the binder. The binder produces the Writable Static Map. Find the
offset of statint in the Writable Static Map in Figure 8. In this example, the offset
is X'30'. Figure 8. Writable static map (AMODE 64)
--------------- CLASS C_WSA64 LENGTH = 38 ATTRIBUTES = MRG, DEFER , RMODE= 64 OFFSET = 0 IN SEGMENT 002 ALIGN = QDWORD --------------- CLASS OFFSET NAME TYPE LENGTH SECTION 0 $PRIV000012 PART 10 10 CDIVZERO#S PART 20 CDIVZERO#C 30 statint PART 4 statint 34 fa PART 4 fa - Add the WSA address of X'108300050' to the offset of statint. The result is X'108300080'. This is the address of the variable statint, which is in the writable static area.
- Use IPCS to display the writeable static area in the system dump.
The value at location X'108300080' is X'49' (that
is, statint is 73), and hence the fixed-point divide
exception. Figure 9. IPCS storage display of the writeable static area (AMODE 64)
LIST 0000000108300050 LEN(X'00000100') LIST 01_08300050. ASID(X'0015') LENGTH(X'0100') AREA _8300050. C36DE6E2 C1F6F440 40404040 40404040 |C_WSA64 | _8300060. 00000001 08300084 00000001 08300080 |.......d........| _8300070. 00000000 000000C0 00000000 2548D200 |.......{......K.| _8300080. 00000049 00000001 00000000 00000000 |................| _8300090 LENGTH(X'10')==>All bytes contain X'00' _83000A0. 00000001 08300000 00000000 00000220 |................| _83000B0. 00000001 083002D0 00000001 083004B8 |.......}........| _83000C0. 00000001 083004F5 00000001 08300532 |.......5........| _83000D0. 00000001 0830056F 00000001 083005AC |.......?........| _83000E0. 00000001 083005E9 00000001 08300626 |.......Z........| _83000F0. 00000001 08300663 00000001 08300A70 |................| _8300100. 00000001 08300AAD 00000000 00000000 |................| _8300110 LENGTH(X'40')==>All bytes contain X'00'