Analyzing program logic for COBOL

This section describes the built-in dynamic analysis features that let you identify program structure, trace the flow of control, monitor execution coverage, and review the execution path. These features assist you in understanding what the program does and how to reach a certain location in the program. The Code Debug TSO commands covered here are FIND, TRACE, SHOW PREVIOUS, COUNT, MONITOR, REVERSE, and RESUME.

If you have Code Debug Db2 Extension and File-AID for Db2, you can analyze how SQL statements execute with the EXPLAIN command. See Using Code Debug for Db2 Extension for more information.

Identifying Program Structure

The Code Debug TSO FIND command allows you to search data relationships and program structures, in addition to locating character strings. For instance, the FIND command can process data names and identify COBOL statements that directly or indirectly affect or see  the data names. COBOL-structure keywords such as ALTER, CONDITION, I/O, etc. are processed to query COBOL statements that have the potential to modify data, conditional constructs, and I/O statements. With the highlighting effect and the capability to suppress statements that do not qualify for the search category, the source display screen can turn into a representation of “data flow cross reference” and a “high-level structure.”

One example of the COBOL sensitivity of FIND is the ability to find data names, aliases and the use of the data name. Some of the keywords related to finding data names are:

DEFine

Data name is defined.

MODify

Value of the data name has changed or has the potential for change.

USE

Value of the data name is used, but not modified.

REFerence

Data name is defined, modified, or used.

The default is REFERENCE. When FIND is issued on a data name with no additional keywords, all references to the data name are found.

Finding All References for a Data Name

Enter the following command to find the data name SUBS:

The response to the FIND SUBS command is illustrated in Result of Entering FIND SUBS. The message line indicates the number of times the data name SUBS is referenced in the program. There are 49 data references for SUBS: It is defined once, used 30 times, and modified 18 times. Each reference for the data name SUBS is highlighted and one of the following messages appear in the message area to the right of the found line: DEF, USE, or MOD.

To find the next occurrence of SUBS, press PF5 or type FIND on the command line as shown in the following figure, move the cursor down past the line in which SUBS is defined, and press Enter.

Result of Entering FIND SUBS

The following figure shows the result of the repeat FIND. The DEF, MOD, and USE messages remain on the display until execution begins or a new FIND command is issued. Note that entry of a repeat FIND does not remove these messages.

Result of a Repeat FIND

The remaining keywords that relate to finding data names are the following two groups—ALIAS, NOREDEFINE, NOALIAS, and DIRECT and INDIRECT. They are defined here and illustrated in examples on the following pages.

ALIas

Other references (such as redefined or group level names) to the same storage location are found.

NORedefine

Other references (except redefines) to the same storage location are found.

NOAlias

Alias names are not found; default.

DIRect

Only direct references to the data name are found; default.

INDirect

All references to the data name, its aliases (if indicated), and all places a data value is passed to or from the data name and its aliases are found.

Finding Aliases of a Data Name

The data name N-CNTR has the aliases N-N-C, N-N-C-TABLE, and NAME-N-CNTR-TABLE. Enter the following command to find all aliases of the data name N-CNTR:

The result of this FIND command is shown in following figure, where N-N-C-TABLE and N-N-C are aliases of N-CNTR because they are both group data names under which N-CNTR is defined. NAME-N-CNTR-TABLE is an alias of N-CNTR because N-N-C-TABLE redefines NAME-N-CNTR-TABLE. That is, N-N-C, N-N-C-TABLE, and NAME-N-CNTR-TABLE all have the same storage location as N-CNTR, as seen in the Data Division.

Enter the command DOWN;FIND, so the screen will be scrolled before a repeat FIND is issued.

Result of Finding N-CNTR With DIRECT and ALIAS

The display scrolls to show the uses and modifications of N-CNTR in the Procedure Division. When you look at the following figure, notice that the number of times N-CNTR is used and modified on the line is indicated on statement 42.

Scrolling to Modifications of N-CNTR

Finding Indirect References to a Data Name

In all of the examples that have been discussed so far, the FIND default of DIRECT was used. When INDIRECT is specified for a data name, all statements directly or indirectly affected by the data name are found. The following example illustrates how INDIRECT is used and the results.

A good way to view indirect references is to use the EXCLUDE keyword with the FIND command. The EXCLUDE keyword excludes from view all lines that were not found. For example, issue the following command for the data name IN-PASS1:

The results of this FIND command are shown in the following figure where all references to IN-PASS1 are displayed. To display the next level of indirect references, enter FIND INDIRECT or press the PF17 key.

Finding IN-PASS1 INDIRECT With EXCLUDE

Each time FIND INDIRECT is entered, a new level of indirect references is found. When you look at the following figure, notice that SE0ORDR1 references are highlighted.

FIND INDIRECT - First Level of Indirection

When FIND INDIRECT (PF17) is entered again, all references to IN-PASS1, SE0ORDR1, and SPA-PASS1 are found (See the following figure).

FIND INDIRECT - Second Level of Indirection

When no more levels of indirect references are found, the following message is displayed in the message line:

Enter END (PF3) to reset all excluded lines in your program.

Finding COBOL Structures

FIND is also able to find COBOL structures using keywords in place of a character string. When a COBOL-structure keyword is entered instead of a data name, Code Debug TSO highlights all lines in the program where the COBOL structure is used.

The COBOL-structure keywords are listed as follows. See COBOL-Structure Keywords for descriptions and source relationships.

Using FIND With the COBOL-Structure INPUT

Enter the following command to find all INPUT statements in the IQTEST program:

The first input statement found is a READ statement, as shown in the following figure. The line that contains the input verb is highlighted. If the verb statement extends over multiple lines, multiple lines are highlighted, as shown. To locate the next input statement, press PF5 (repeat FIND).

Result of FIND With COBOL-Structure Keyword INPUT

The next input statement is also a READ statement as shown in the following figure.

Result of a Repeat FIND for INPUT Keyword

When you enter FIND and no more INPUT statements are found, the following message is displayed in the message area:

The input statements remain highlighted until execution begins or a new FIND command is issued.

Using FIND With the COBOL-Structure DLI and EXCLUDE Keywords

In this example, the FIND command is entered with the DLI and EXCLUDE keywords. The DLI keyword finds not only all lines in the OSDLI program that say EXEC DLI, but in each case, the entire DLI statement. The EXCLUDE keyword displays only the lines containing the requested information; all other lines are excluded from display. Enter the following command:

When you look at the following figure, you can see at a glance where the DLI statements are, what types they are, and all parameters on each statement.

Result of FIND DLI With EXCLUDE

Now, you must enter a NOLINES command to eliminate the message line xx LINES NOT DISPLAYED from the display. As illustrated in the following figure, the resulting display can now accommodate additional found lines.

Result of Issuing the NOLINES Command

Finding a String IN COBOL Structures

COBOL-structure keywords can also be used with the IN keyword. They are used to find a string or a data name IN a COBOL structure. For example, the following FIND command is entered with the string B010, part of a performed paragraph label:

The FIND string IN COBOL-structure lets you focus on the statement that is of concern, rather than issue several repeat FIND commands. See the results illustrated in the following figure.

Result of FIND Data String With IN PARAGRAPH

The NOLINES Keyword and Command

In the previous example, the NOLINES command was entered to suppress the xxx LINES NOT DISPLAYED message line that appears when the EXCLUDE keyword is used with FIND. A NOLINES keyword is also available with FIND. Like the NOLINES command, the NOLINES keyword eliminates the message line that appears with the use of the EXCLUDE keyword. However, it is effective only when used in conjunction with the EXCLUDE keyword.

Using the EXCLUDE Command With FIND

The EXCLUDE command includes the keyword parameter ALL, which excludes all lines in a program. The EXCLUDE ALL command can be used effectively with FIND.Issuing the EXCLUDE ALL command results in the removal of all the source lines in the display. You can also use the NOLINES command to suppress the xxx LINES NOT DISPLAYED message line.

Entering the following commands results in the display shown in Result of following figure:

After all lines in the program have been excluded, the FIND command can be issued for multiple data names, to make a cumulative search for the source of a problem.

Result of Excluding ALL Lines

Note that you can also concatenate the EXCLUDE ALL command along with a FIND command. For example, entering the following commands results in the display shown in the following figure:

Result of FIND SUBS Command After EXCLUDE ALL Command

Using the FIND CSR Command

The CSR keyword issues a FIND command for the data name or string under the cursor. It functions like PEEK CSR, searching for data names as well as strings. IN and OF qualifications are automatically picked up.

When you look at the following figure, the cursor is positioned on the first M on line 553. Press the PF14 key or enter the FIND CSR command as shown.

Using the FIND CSR Command

FIND CSR searches for the first instance of the data name under the cursor and highlights it, as shown in the following figure.

Result of Entering FIND CSR

Logging the Results of a FIND Command

Results of Data Name and COBOL Structure FINDs can optionally be written to the session log. Since this option defaults to OFF, a user must enter the SET command as follows to activate this feature:

Information written to the session log includes the FIND command entered, the name of the program being searched, the number of occurrences detected, and the source lines containing the argument. For all indirect FINDs and enhanced FINDs, the level of indirection is also logged.

All source statements containing the argument are displayed after the initial command has been issued, regardless whether the NEXT, PREV, LAST, or FIRST keywords are used. No additional logging will occur when a repeat FIND command is issued. Caution should be used when the FIND command is frequently issued, with explicit attention given to the amount of space allocated to the session log file.

The general format of the logged data can be seen in the following example (see the following figure):

Data Format When Logging the Results of a FIND Command

Tracing the Flow of Control

You can trace the flow of control with the TRACE and the SHOW PREVIOUS commands.

Using the TRACE Command

The TRACE command traces the flow of control as your program executes and lets you view it on the Source display screen. The specified statements or paragraph names are highlighted as they are executed, until a breakpoint is reached, an abend is intercepted, a terminal I/O is issued, a keyboard interrupt is detected, or the end of the program is encountered.

In the case of tracing module calls, however, the tracing is not visible on the source screen. The calling module, the called module, and the number of times the calls are made are written to the session log during program execution. The call activities can be viewed by entering LOG and reviewing the session log.

For example, suppose you enter the following TRACE command for the TRIMAIN program:

TRACE MODULES

After you have executed the program by entering GO, you can access the session log using the LOG command. As seen in the following figure, you can trace the flow of control as the various modules are called.

Session Log for the TRACE MODULES Command

When the TRACE command is used with the MAX keyword, the trace function pauses when execution reaches the preset limit. The default value for the maximum limit is 25. For additional information regarding the TRACE command see Using the TRACE Command.

Using the SHOW PREVIOUS Command

The SHOW PREVIOUS command lists, in logical sequence, the previous 100 statements along with the executed breakpoints. The list presented by the SHOW PREVIOUS command can be useful in reviewing the execution path to understand how you got to the present location. Since any implied breakpoints are recognized with the SHOW PREVIOUS command, a program that was run with the TRACE ALL PARAGRAPHS command presents a list like that shown in the following figure as the result of entering SHOW PREVIOUS.

Result of Entering the SHOW PREVIOUS Command

Monitoring Execution Coverage

You can monitor execution coverage with the COUNT and SHOW COUNT commands.

Using the COUNT Command

The COUNT command maintains execution counts and lets you analyze statement level execution coverage after running the program. The following figure shows the result of setting counters at every paragraph by entering the following command, then pressing PF12 (GO) to resume execution:

Result of Entering the COUNT ALL PARAGRAPHS and GO Commands

A 7-digit counter is displayed at the right of the screen for each statement or paragraph that is counted. You can globally monitor execution coverage by using the ALL keyword or selectively monitor by specifying the statement numbers. When the ALL keyword is issued, however, it only applies to the current module.

Using the SHOW COUNT Command

The statements that are monitored with the COUNT command can be listed by entering the SHOW COUNT command. The source lines without counters are excluded from the screen display. The following figure illustrates the result of the SHOW COUNT command after monitoring execution coverage at the paragraph level.

Result of Entering the SHOW COUNT Command

When the SHOW COUNT command is issued, the entire Procedure Division is written to the session log with the 7-digit counters displayed.

If you do not want to write the entire Procedure Division to the session log, enter the SHOW COUNT command with the NOLOG keyword. The results of execution coverage will not be recorded.

Monitoring and Reviewing the Execution Path

The MONITOR and REVERSE commands are used to activate review mode. Review mode lets you monitor and review the execution path by stepping backwards through your program. You can view the statements that were executed during normal (forward) execution. You can trace backwards through the actual sequence of instructions that led to the current breakpoint and see the data values as they were at the time. There is no guesswork about which of the possible paths the program took; the actual path that was taken during forward execution of the program is displayed.

The MONITOR command records the execution history and the REVERSE command enables you to review the execution history.

To activate review mode, enter the MONITOR command from the primary command line. MONITOR without a module name records history for the current module—not necessarily the active module that is currently executing, but the module indicated by the program field on the third line.

After issuing the MONITOR command, execute the statements you want to review. Then, when your program pauses during logical execution, enter the following primary command:

The REVERSE command places the execution arrow on the last statement that was executed, and changes the execution direction of your program—from forward to reverse (backward). From this point on, the REVERSE command acts as a toggle that changes the direction in which your program is executed. During review mode, the execution status message on the fourth line of the screen indicates the execution direction and the statement where execution is paused.

While review mode is activated, you can set and remove breakpoints, perform tracing (in either direction), and display data. For example, you may want to open a Keep window to view the data values as they are restored to their original state as reverse execution is performed, as shown in the following figure for the data value WORK-REC.

Review Mode with a Keep Window Opened

To exit from review mode when you have finished doing your analysis, use the RESUME command.

The message AT CURRENT EXECUTION POSITION is displayed, and review mode is automatically ended. Normal forward execution occurs until you again enter the REVERSE command. You can also terminate review mode while in forward execution by entering GO or GO n until you see the message AT CURRENT EXECUTION POSITION.

Kept items are logged at each breakpoint while in review mode, just as in normal execution mode. The logged items are all logged independently of review mode or normal execution mode, except for the following:

• Encountering the beginning of the program while in review mode
• Encountering the current execution location while in review mode.