ImGui: guard ShowCursor against a torn-down context

[Kheartz]

Summary

ImGUI::Wrapper::ShowCursor() was the only public entry point on the wrapper
that did not check isContextInitialized before touching ImGui state. A client
teardown path (disconnect handling → release control locks → ShowCursor(false))
ran after Wrapper::Shutdown() and dereferenced a destroyed ImGui context.

Found while symbolizing a user-submitted crash dump from a Release build of a
UE 4.27.2 title on the D3D12 backend.

Crash signature

ExceptionCode:  c0000005 (Access violation)
Operation:      write to 0x0000000000000078
Faulting insn:  mov byte ptr [rax+50h], bl
Registers:      rax=0x28  rbx=0  rcx=0  rbp=0
Thread:         GameThread

ImGui::GetIO() resolves to GImGui->IO. With a null context that is
nullptr + offsetof(ImGuiContext, IO) = 0x28 (matches rax), and
ImGuiIO::MouseDrawCursor at +0x50 puts the store at 0x78 (matches the
faulting address). rbx=0 is the false being written.

Release-only in practice: ImGui::GetIO() normally carries
IM_ASSERT(GImGui != NULL), which /DNDEBUG removes, degrading the guard into
a raw null dereference.

Two defects fixed

1. ShowCursor never checked the flag. Every other gated entry point
   (Init, Shutdown, Render, OnDeviceLost, OnDeviceReset) does.

2. Shutdown() cleared isContextInitialized after DestroyContext().
   That left a window in which the flag claimed the context was live while the
   backends were already shut down and the context destroyed — so the guard in
   (1) alone would not cover a caller landing inside it. The flag is now cleared
   up front, immediately after the early-out.

Investigation notes: GPU was ruled out, not a factor

Recording this so it is not re-chased. The reporting machine is a hybrid-graphics
laptop, and the crash context showed the title rendering on the integrated
GPU (Misc.PrimaryGPUBrand = Intel iGPU; only the Intel D3D12 UMD was loaded in
the process, the NVIDIA D3D UMD never was) despite a discrete GPU being present.

That initially looked causal — a renderer hook failing to attach would leave
isContextInitialized == false and reach the same defect. It was ruled out:

• Forcing the discrete GPU (Windows per-app graphics preference and the NVIDIA
  control panel) did not stop the crash.
• The reporter confirmed the mod UI was visible before the disconnect, so
  Init had succeeded and the context genuinely existed.

Conclusion: this is a teardown-ordering bug, not a renderer-init bug. The GPU
selection oddity is real but separate, and is being tracked downstream.

Secondary observation (not addressed here)

CMAKE_CXX_FLAGS_RELEASE is /O2 /Ob2 /DNDEBUG with no /DEBUG at link, so
Release builds ship without a PDB and user crash dumps cannot be symbolized
without reproducing the exact toolchain. Adding /Zi plus
/DEBUG:FULL /OPT:REF /OPT:ICF /PDBALTPATH:%_PDB% was measured on a real build
of this framework to cost exactly one 4 KB page of SizeOfImage while leaving
.text byte-identical. /OPT:REF /OPT:ICF must be explicit — /DEBUG silently
defaults them off, which would change code layout. Happy to open a separate PR
if that is wanted.

Testing

Not reproduced locally — the crash requires the teardown ordering seen on the
reporting machine. The fix was verified by inspection against the symbolized
stack, and the register arithmetic above accounts for the faulting address
exactly. No behavioural change when the context is live; when it is not,
ShowCursor becomes a no-op, which is correct (no cursor to draw).

Summary by CodeRabbit

• Bug Fixes
  • Improved shutdown handling to prevent invalid context state during teardown.
  • Prevented cursor display operations from accessing unavailable UI resources when initialization has not completed.

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}