Debugging a sourceless program for C

Code Debug TSO allows you to interactively debug programs that do not have the source available (no source listing in the DDIO). You can choose either of two available methods.

If requested, Code Debug TSO can create a “pseudo-assembler” view of the object code. With this method, a temporary source member will be created. This allows you to debug main programs and subprograms while employing the same strategy you would use if a DDIO source listing had been created during assembly. See Creating Pseudo-Assembler Source for more information.

You can also choose to use Code Debug’s traditional support for programs without source. With this support, you can debug “sourceless” main programs, subprograms, or GETMAINed areas in the same manner as interactive source level debugging, with a few differences. Test session setup and startup is the same as setting up a session to interactively debug at the source level. The differences occur during the debugging session:

• Code Debug TSO debugging commands such as AFTER, BEFORE, COUNT, ONETIME, SKIP, TRACE, WHEN, PEEK, KEEP, GOTO, and DELETE cannot be used.
• To set a breakpoint, you must use the AT command, which sets a before breakpoint.

When the AT breakpoint is reached, the AT Display screen shows your sourceless program in dump format. You can enter Code Debug TSO debugging commands such as GPREGS, MEMORY, SHOW, and so on.

Accessing a sourceless main program

When program execution begins and the main program does not have source, the session log is automatically displayed with the message,

NO SOURCE LISTING DATA SET MEMBER FOR program-name.

See the following figure.

Showing No Source Message for a Main Procedure

You can use the AT command on the Log screen to set a breakpoint in the specified program if you are familiar with the program. However, if you are not familiar with the program, you can use the MEMORY command to display the program in dump format. For example:

displays the Memory screen showing a dump of TRIMAINC.

The AT command can then be entered to set breakpoints in the program. Issuing the AT command will also load the module into memory if it is not loaded. The full syntax of the AT command is provided in the Command-and-syntax-reference space.

For example, entering the following on the Memory screen displays the AT Display screen shown in the following figure.

AT Display Screen

If the address specified with the AT command is not a valid instruction (data or privileged), a message is returned. Also, if the specified program has source, no breakpoint is set and the message MODULE HAS SOURCE; USE Xpediter BREAKPOINT COMMANDS is displayed.

On the AT Display screen, the offset and the instruction pointer are highlighted, indicating where the breakpoint is set. The message area contains the breakpoint address and the instruction in hexadecimal and mnemonic form. If the AT command is used on an address and not in a module loaded by Code Debug TSO, the compile date and time displayed on the third line will be blank.

Accessing a Sourceless Subprogram

When your debugging session begins and your main program has source, the Source display screen is displayed. You can set breakpoints in the regular manner on this screen.

Normally, breakpoints can be set in a subprogram by entering a fully qualified breakpoint command or using the SOURCE or INTERCEPT commands to access the subprogram from this screen. However, if the subprogram does not have source, the message NO SOURCE LISTING INFORMATION FOUND FOR MODULE is displayed at the top of the source display.

If you are not familiar with the program, you can use the MEMORY command to access main storage and display the module in dump format on the Memory screen. For example:

With the program displayed on the Memory screen, you can decide exactly where you want to set breakpoints using the AT command.

If you are familiar with the application you are debugging, you can use the AT command with the module/external function. For example, to set a breakpoint at displacement 0 into an external function, enter:

Using Code Debug TSO Commands for Sourceless Debugging

The rules for using Code Debug TSO commands for sourceless debugging are as follows:

Primary Commands

None of the debugging commands, such as AFTER, BEFORE, COUNT, ONETIME, SKIP, TRACE, WHEN, PEEK, KEEP, GOTO, GO n, and DELETE are valid. The GO command is valid. All other commands can be used.

Line Commands

The only valid line commands are X (eXclude) and T (Template). To insure proper results, the rest of the line command area should be blanked out after entering the line command.

The following are examples of using some Code Debug TSO commands to debug a sourceless program.

Using the UP command on the AT Display screen scrolls the screen toward the beginning of the module/procedure. Scrolling beyond the beginning of the module/procedure displays negative offsets. Using the LOCATE * command scrolls to the current execution point.

The GPREGS command can be used to open a modifiable window at the bottom of your screen. The register values displayed can be changed by typing over the values. An example of the GPREGS window is shown in the following figure.

GPREGS Window on the AT Display Screen

Data and instruction areas can be typed over, as long as a breakpoint is not set on the instruction being typed over.

The SHOW ACTIVE command can be entered to display a summary of general-purpose register contents, PSW, and some control block information. The result of the SHOW ACTIVE command is shown in the following figure.

SHOW ACTIVE Display Screen

The END command returns you to the previously displayed GPREGS Window on the AT Display screen as shown in the figure above.

The SHOW AT command can also be entered to display all the outstanding breakpoints set by the AT command.

Use the DELETE AT command to remove all outstanding AT breakpoints.