|
View previous topic :: View next topic
|
| Author |
Message |
Catz
New User
Joined: 22 May 2015
Posts: 6
Location: Latvia
|
|
|
|
Good day.
I'm looking to understand the ABI used by zOS C compilers, so that I can reconstruct the backtrace.
Presently Im looking at something like:
| Code: |
Called:
001145 | * dump_x(512, &ima, FORWARD);
0008A6 5820 3142 001145 | L r2,=A(IMA)(,r3,322)
0008AA 58F0 3146 001145 | L r15,=V(DUMP@X)(,r3,326)
0008AE 4110 D098 001145 | LA r1,#MX_TEMP9(,r13,152)
0008B2 4140 0200 001145 | LA r4,512
0008B6 5040 D098 001145 | ST r4,#MX_TEMP9(,r13,152)
0008BA 5020 D09C 001145 | ST r2,#MX_TEMP9(,r13,156)
0008BE 5000 D0A0 001145 | ST r0,#MX_TEMP9(,r13,160)
0008C2 0DEF 001145 | BASR r14,r15
Prolog:
000148 000402 | dump_x DS 0D
000148 47F0 F022 000402 | B 34(,r15)
00014C 01C3C5C5 CEE eyecatcher
000150 000000C0 DSA size
000154 00001B60 =A(PPA1-dump_x)
000158 47F0 F001 000402 | B 1(,r15)
00015C 58F0 C31C 000402 | L r15,796(,r12)
000160 184E 000402 | LR r4,r14
000162 05EF 000402 | BALR r14,r15
000164 00000000 =F'0'
000168 07F3 000402 | BR r3
00016A 90E7 D00C 000402 | STM r14,r7,12(r13)
00016E 58E0 D04C 000402 | L r14,76(,r13)
000172 4100 E0C0 000402 | LA r0,192(,r14)
000176 5500 C314 000402 | CL r0,788(,r12)
00017A 4130 F03A 000402 | LA r3,58(,r15)
00017E 4720 F014 000402 | BH 20(,r15)
000182 58F0 C280 000402 | L r15,640(,r12)
000186 90F0 E048 000402 | STM r15,r0,72(r14)
00018A 9210 E000 000402 | MVI 0(r14),16
00018E 50D0 E004 000402 | ST r13,4(,r14)
000192 18DE 000402 | LR r13,r14
Epiolog:
0009A0 58D0 D004 001178 | L r13,4(,r13)
0009A4 58E0 D00C 001178 | L r14,12(,r13)
0009A8 9826 D01C 001178 | LM r2,r6,28(r13)
0009AC 051E 001178 | BALR r1,r14
0009AE 0707 001178 | NOPR 7 |
It loosely looks like a savearea chain with some additional detail, expanded on stack.
Presently Im looking into XL documentation. If someone knows a resource/doc, where ABI is specified, or some comments on the subject, please share.
Respectfully,
Catz. |
|
| Back to top |
|
 |
steve-myers
Active Member
Joined: 30 Nov 2013
Posts: 917
Location: The Universe
|
|
|
|
It appears to me to be the Language Environment model. In turn, this model goes back to 1960s common practice for reenterable programs. In this model, reg 13 points to a "standard" 72 byte register save area followed by a work area.
For better or worse, many working Assembler programmers have abandoned this model. Reg 13 still points to a save area, but a second register points to a work area shared by, at least, all the internal functions in a single program.
Yes, I know any CS101 alumni will point out this model is flawed, but those CS101 professors are seldom working Assembler programmers, and they look down on us as vermin. |
|
| Back to top |
|
|
Catz
New User
Joined: 22 May 2015
Posts: 6
Location: Latvia
|
|
|
|
| Thank you so much for this - was finally able to find docs, explaining this. |
|
| Back to top |
|
|
|
|
  |
All times are GMT + 6 Hours |
|
