fix: ExplainNode guard against NULL planstate on QE

[roaldm153]

If subtree on QE lives in another slice there is not guards and it crashes with segfault, so I fixed it

Fixes #ISSUE_Number

What does this PR do?

This change fixes a crash that occurs on the Query Executor (QE) when a subtree lives in a different slice.
The original code assumed that planstate is always non‑NULL and directly dereferenced planstate->plan.
When the subtree is not instantiated in the current slice, planstate can be NULL, leading to a segmentation fault.
The patch adds:

A NULL guard for planstate at the start of ExplainNode.
Moves the assignment plan = planstate->plan after the guard.
Additional guard when recursing into the outer plan to avoid calling ExplainNode with a NULL outerPlanState.
These changes make EXPLAIN robust across slice boundaries without altering any existing functionality.

Type of Change

• Bug fix (non-breaking change)
• New feature (non-breaking change)
• Breaking change (fix or feature with breaking changes)
• Documentation update

Breaking Changes

No breaking changes. The public API and plan execution behavior remain unchanged; only additional safety checks were added.

Test Plan

Unit tests: No new unit tests were added because the fix is defensive and does not change the logical output of EXPLAIN.
Integration tests: Ran the full regression test suite, including parallel tests, and verified that the previously failing scenario no longer crashes.

• Passed make installcheck
• Passed make -C src/test installcheck-cbdb-parallel

Impact

Performance:
Negligible impact – the added NULL checks are inexpensive branch predictions and do not affect the typical execution path.

User-facing changes:
ExplainPrintPlan now works on QE

Dependencies:
Nope

Checklist

• Followed contribution guide
• Added/updated documentation
• Reviewed code for security implications
• This PR contains AI-assisted code generation
• Requested review from cloudberry committers

Additional Context

ExplainNode was called from the QE side with a planstate that had not been instantiated because the subtree resides entirely in another slice.

[github-actions]

Hi, @roaldm153 welcome!🎊 Thanks for taking the effort to make our project better! 🙌 Keep making such awesome contributions!

[Copilot]

Pull request overview

Fixes a QE-side crash in ExplainNode() when a plan subtree lives in a different slice and the corresponding PlanState * is NULL, by adding a defensive NULL guard and avoiding recursive descent into a NULL outer child.

Changes:

• Add a planstate == NULL early return before dereferencing planstate->plan.
• Move plan = planstate->plan assignment after the new NULL guard.
• Guard recursion into outerPlanState(planstate) to avoid calling ExplainNode() with a NULL child.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Alena0704]

LGTM. I think you should rephrase the commit message. Maybe like that

Fix segfault in ExplainNode() on a QE

A receiving Motion has no child PlanState when its child slice runs on
another gang.  ExplainNode() recursed into it anyway and crashed.
Return early when planstate is NULL.

As far as I see this code is unreachable in-tree. The only callers of ExplainPrintPlan() are ExplainOnePlan(), which runs on the QD (for example, in auto_explain - it returns from _PG_init when Gp_role != GP_ROLE_DISPATCH (contrib/auto_explain/auto_explain.c:101)) - so it never installs its hooks on a segment. Only an out-of-tree extension loaded on the segments can hit this path or debuggers today.

That's fine as a hardening fix, but I think you should state it explicitly in the commit message.

[Alena0704]

The comment doesn't say where the decision not to instantiate the subtree is made, so a reader has no starting point. I think you should add a pointer to ExecInitMotion here:

/*
 * Guard against the case where a subtree on the QE lives in another slice
 * and is not instantiated in this slice (see ExecInitMotion).
 */