FIX: Debug: Single-Step Fails with Emulated Float Intructions

ID: Q106561

The information in this article applies to:

Microsoft FORTRAN PowerStation for MS-DOS, version 1.0

SYMPTOMS

When using the debugger, attempting to use the Step Into (F8) or Step Over (F10) function causes execution to the next breakpoint (or to the end of the program if no breakpoints are encountered).

CAUSE

The debugger treats the Step commands similar to the Go (F5) command when stepping into or over an assembler floating-point instruction. This error occurs only when using floating-point emulation and does not occur when using a math coprocessor.

STATUS

Microsoft has confirmed this to be a problem in FORTRAN PowerStation version 1.0 for MS-DOS. This problem has been resolved with FORTRAN PowerStation maintenance release version 1.0a for MS-DOS.

MORE INFORMATION

FORTRAN PowerStation version 1.0 can be differentiated from the maintenance release version 1.0a by invoking the linker. Typing link32 | more from \F32\BIN directory will show version 2.8 for FORTRAN PowerStation version 1.0, and it will show version 1.0f for the maintenance release version 1.0a.

To demonstrate the problem:

Disable the PowerStation's use of the coprocessor (if present). This is done by typing the following command before starting Windows:
set tnt=-fpem

Start Windows and the FORTRAN Visual Workbench.

Build a project with F32\DEBUGME

Set breakpoints at lines 22 and 26 of F32\DEBUGME\DEBUGME.FOR:


      MyReal4 = MyInt4 * 3.14159

-and-


      If ((MyInt4 * MyReal4) .EQ. MyReal8) Then

Press F5 to "GO"; the current line will be 22.

Single-step (press F8 or F10); the next line (24) is skipped and the debugger will go to the breakpoint on line 26.

To see what is happening in more detail, after step 5, turn on display of mixed source and assembly from the View menu (or with the command CTRL+F7).

Additional query words: 1.00 buglist1.00 fixlist1.00a

Keywords :
Version : :1.0
Platform :
Issue type :