BF6 BSOD on launch PAGE_FAULT_IN_NONPAGED_AREA eaanticheat.sys
Battlefield 6 has BSOD'd on launch every attempt since May 22, 2026 — one day after the May 21 Javelin anti-cheat depot update. No software, hardware, BIOS, or driver changes were made on my end before onset. The crash is deterministic, reproducible 100% of the time, and the dump analysis points to a regression inside the current Javelin build, not a system configuration issue.
Crash signature (identical across 8 minidumps)
- Process: bf6.exe
- Bug check: 0x50 PAGE_FAULT_IN_NONPAGED_AREA
- Parameter 2 (access type): 0 (read)
- Faulting instruction: inside eaanticheat.sys at the same fixed module-relative offset +0x80BA2C in every crash (raw RIP varies only by KASLR base)
- Faulting address (P1): varies per crash, always page-aligned, always in kernel pool (unmapped)
- R12 register at fault: ntoskrnl.exe + 0xE38040, identical across all dumps — a fixed kernel global serving as the anchor for a list-traversal
- OS: Windows 11 24H2 (build 26100)
Interpretation: a single instruction in eaanticheat.sys reads through a pointer that points to unmapped kernel pool memory. The instruction is constant, the data it loads is garbage that differs per crash. This is the classic signature of a deterministic code defect — most likely a use-after-free or stale pointer in a list-walk anchored at a fixed kernel global. Not random memory corruption; not hardware.
What I ruled out (so you don't have to ask)
- Hardware / RAM: EXPO/XMP disabled the entire time the system has existed. No overclock anywhere on the system. System ran BF6 and every other workload flawlessly for ~11 months prior. The deterministic same-instruction fault is mathematically inconsistent with random memory corruption.
- BF6 install: full uninstall/reinstall, file verification, and a complete reinstall to a separate physical drive — no change. Reinstalling pulls the same Javelin build, which is the broken one.
- Anti-cheat reinstall: EAAntiCheat.Installer.exe uninstall + reinstall — no change.
- Conflicting drivers (driver uninstall, kernel-level verification of removal):
- Radmin VPN (RvNetMP60.sys) — uninstalled, confirmed gone via pnputil/sc query, still crashes
- FACEIT Anti-Cheat (FACEIT_AC.sys) — uninstalled, confirmed gone, still crashes
- VirtualBox kernel drivers — not loaded in subsequent crashes either, still crashes
- Driver Verifier on every non-Microsoft driver with Special Pool + Pool tracking + Force IRQL + I/O verification + Misc checks: caught nothing. Bug check remained 0x50 in eaanticheat.sys at the same +0x80BA2C offset. No third-party driver implicated.
- Recently unloaded drivers list in the dumps: only one non-Microsoft transient driver appears (CorsairVHidDriver, normal HID re-enumeration), and the faulting address never falls in any unloaded driver's range. The fault is on freed/invalid pool, not on unloaded driver code.
Correlation with the May 21 Javelin push
- First crash: May 22, 2026. SteamDB shows a Javelin depot update on May 21, 2026.
- Pre-May 22: ~11 months of stable BF6 play, no changes to system.
- Pattern matches the previously documented April 16, 2026 Javelin regression that caused identical kernel-mode failures across multiple Battlefield titles.
- Multiple players in the EA forums are reporting the same onset date with similar symptoms.
What would actually help
- Revert the Javelin client to the pre-May 21 build, or push a hotfix targeting the list-traversal at module offset +0x80BA2C.
- A way for affected players to roll Javelin back to the prior depot while a fix is prepared.
System specs
- OS: Windows 11 24H2 (build 26100)
- CPU / platform: AMD AM5 (EXPO disabled)
- Memory Integrity / Core Isolation: OFF
- BF6 install: tested on two separate physical drives, same result
