IBM 1620 Manual page 84

The following additional subroutines are known as
relocatable subroutines and are found at the end of
the object program. The address of the first instruction
of these subrountes will be variable depending on
object program length. However, as the relocatable
subroutines are entered by branching to certain fixed
locations which contain further branch instructions to
the relocatable subroutine, it will normally not be
necessary to treat them any differently than fixed
subroutines.
At tines, these relocatable subroutines are entered
(two places) by means of an unconditional branch
(Op Code 49) when the subroutine is to be repeated
using the same data as used previously.
FORMAT FORTRAN
Relocatable Subroutine Name
Floating Natural Logarithm
Floating Exponent
Floating Square Root
Floating Sine
Floating Cosine
Floating Arctangent
Abbreviation
FLN
FEXP
FSQR
FSIN
FCOS
FATN
Figure 6
It is a characteristic of a branch and transmit Op code
that the next address in memory is always stored in
IR-2. As the subroutine is always entered by means
of a branch and transmit immediate Op code, the next
object program instruction is saved in IR-2. The last
instruction of all subroutines is always a branch back
Op code to return to the object program for the next
subroutine to be executed.
To break down failures on Fortran object programs that
result in a machine stoppage, use the following steps:
1. Display IR-1 and IR-2.
2. If IR-2 is blank, the fai lure occurred in the object
program and on the instruction at the IR-1 minus
12 address. Dump as follows: 35 (lR-1 minus 12)
00100.
3. If IR-2 contains an address, the failure occurred in
a subroutine and on the instruction at the IR-1 minus
12 address. Dump as follows:
35 (lR-1 minus 12) 00100
35 (JR-2 minus 12) 00100
The "P" address of the instruction at the IR-2 minus
12 address is the address of the first instruction of
the failing subroutine (see Figures 5 and 6). The
instruction at the IR-1 minus 12 address is the fail ing
i nstruc ti on.
7.8
No. of
Factors
Required
Address of
1st Instruction
19910
19930
19890
19990
19970
19950
4. To loop the failing subroutine, it is necessary to
dump from addresses IR-2 minus 12, IR-2 minus 24,
IR-2 minus 36, etc. , successively, until an instruc-
tion is found with the address of the liTo FAC"
subroutine (01238) in the
"p-"
field. This is the
first instruction of the object program group of
instructions in which the failure occurred. A
branch (Op code 49) may be inserted into the
IR-2 address branching to the first instruction of
the failing object program group of instructions as
follows:
OBJECT PROGRAM
07596 17 00408 19xxx
07608 17 01306 (FmFAC) 19xxx
07620 17 01238 (To FAC)19xxx
07632 17 00518 19xxx
07644 17 03270 19xxx (Failing Instruction)
07656 17 01862 19xxx
07668 17 01306 (FmFAC)19xxx
07689 17 01238 (To FAC)l9xxx
IR-2 contains 07656 Insert - 36 07656 00100 -
Release, Start - Type in 49 07620 - Insert
49 07620 - Release Start. The program is now
looped between 07620 and 07656.