Open Cobol程序突发Stack overflow错误，回滚版本仍无法解决

Hey there, sorry to hear you're stuck with this stack overflow error—especially since your program was stable for weeks and rolling back to a working version didn't fix it. Let's walk through the most likely causes and fixes tailored to your scenario:

1. First, Investigate the Input Data #

Since rolling back your code didn't resolve the issue, the problem is almost certainly data-related. Even identical code can hit stack limits if the input triggers far deeper loops or nested execution than usual.

• Look directly at line 46 in MERRILL_MAX_AMOUNTS.COB: Is this a PERFORM statement (either a loop or a paragraph call)?
• Check if recent input data has changed:
  • Did a record come in with a much higher iteration count (e.g., a variable controlling a PERFORM ... VARYING loop is now 10x larger than normal)?
  • Is there a nested data structure that's deeper than expected, causing recursive or nested PERFORM calls to stack up?
• Quick debug tip: Add temporary logging before line 46 to print the values of variables that control the PERFORM (like loop counters or flags). This will tell you if the loop is running way more times than intended.

2. Adjust libcob Stack Size Limits #

OpenCOBOL's libcob runtime has a configurable stack limit. Even if other programs work, your program might be hitting this limit with the new data.

• Try increasing the stack size via the COB_STACK_SIZE environment variable before running the program:

  export COB_STACK_SIZE=131072  # Sets stack size to 128KB; adjust as needed
  ./your-program
  

• If you compile the program yourself, check if you can pass a stack size flag during compilation (e.g., -fstack-size=131072—verify exact syntax with your OpenCOBOL version's docs).

3. Check for Unintended Recursion or Deep Nested PERFORMs #

Double-check the execution flow starting at line 46 to ensure there's no accidental recursion or overly nested paragraph calls:

• Does the paragraph at line 46 call another paragraph that eventually calls back to itself? This creates a recursive loop that can explode with certain data.
• Count the depth of nested PERFORM calls in the path starting at line 46. If it's significantly deeper than your usual execution, that's probably the issue.

4. Verify System Environment Changes #

Even if other programs are working, recent system changes could affect your program's stack behavior:

• Check if the system's process stack limit was lowered. Run ulimit -s to see the current stack size (in KB). If it's lower than when your program was stable, increasing it might help.
• Did the OpenCOBOL/libcob runtime get updated recently? Sometimes runtime changes can alter stack handling even if your code is the same.

Start with the data investigation—it's the most common culprit when code rollbacks don't fix stack overflow issues. Let me know if you spot anything unusual in line 46 or your input data!

内容的提问来源于stack exchange，提问作者JUNIEMAN