Traced the far end of the online gate: FIFA's GetAuthCodeT::Serialize
(+0x278d2c0) yields the exact LSX request it would send to start the
Nucleus->Blaze chain -- <GetAuthCode UserId ClientId Scope AppendAuthSource/>.
It is wrapped by a GetAuthCodeJob in the EASFC online-job module. The wall
is not the LSX wire or event delivery (events parse fine) but that FIFA's
EASFC controller never enqueues the job -- it polls cached online state and
our single isOnline flag doesn't satisfy the combined condition. Closes the
is-it-the-events question (no); next artifact is the controller poll guard.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Add an unsolicited event-push path to the LSX server: after the session
establishes, re-push (every 2s) a Login then OnlineStatusEvent, encrypted with
the live session seed and null-terminated like any session frame. Formats RE'd
from FIFA23.exe's OriginSDK deserializers (authoritative):
<Login UserIndex="0" IsLoggedIn="true" LoginReasonCode="0"/> (LoginT, +0x287a30)
<OnlineStatusEvent isOnline="true"/> (OnlineStatusEventT, +0x28a4d0)
Value encoding is the true/false string literal (FIFA's bool parser at
FIFA23.exe+0x28fb50 compares to "false"); isOnline="1" is parsed as not-online.
Result (verified with in-process probes): FIFA parses both events every cycle and
returns success, and stops its impatient GetProfile polling / sets INGAME presence
— the first time FIFA has been driven "online" at the EbisuSDK level (anadius, being
offline DRM, never does this). But FIFA does not proceed to the GetAuthCode→Nucleus
→Blaze chain; the gate is its game-side event consumer (virtual dispatch at
FIFA23.exe+0x274d4e2), not delivery. Full trail in docs/connection-gate-findings.md.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Implement the LSX (EA App bootstrap, TCP :3216→:3217) server in the bridge
and clear the loading-screen crash gate. FIFA 23 now completes the full LSX
handshake + session against OpenFUT and advances to the online/Blaze phase
("<persona> is connecting to the EA Servers…") — verified live, not a cached
session (the displayed persona "fun" comes from our GetProfile).
Root cause of the invariant loading-screen crash: get_config() omitted the
PROGRESSIVE_INSTALLATION and PROGRESSIVE_INSTALLATION_EVENT services (Name="PI").
FIFA builds its service registry from GetConfig and resolves that handle during
online-init; the missing service left a null handle it dereferenced at
FIFA23.exe+0x133dfc5 (mov rax,[r13+0x18], r13=NULL) ~2s after LSX — identical
across every prior run. Found via FIFA's own CrashReport_*.json (stable RIP) +
diffing anadius64.dll's service-registration table against ours.
Also in this change:
- process_frame(): handle FIFA's pipelined (batched) LSX messages per read,
eliminating the ~15s handshake hang.
- Real anadius values (RE'd from anadius64.dll + anadius.cfg): PersonaId/UserId/
persona, locale list, GetGameInfo fixed `GameInfo="<value>"` attribute,
GetAllGameInfo 14-attr schema, GetSetting `Setting=` attribute.
- IsProgressiveInstallationAvailable served from sender="PI".
- docs/connection-gate-findings.md: full RE trail; relocate stale openfut-hook
copy under docs/ (canonical hook lives in openfut-launcher).
Next gate: Blaze/online (M2), ridden over ProtoSSL in-process.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
protossl-scan: add `jmpscan` (find function-entry E9 detours leaving a module).
Used to try to locate EbisuSDK::GoOnline in FIFA23.exe, but the 505MB image is
mostly embedded data => ~3875 false positives, no clean detour cluster. Combined
with the no-string and worker-thread blocks, the FIFA-side online-flow is not
cleanly reachable with the live-memory toolkit. Documented the verdict in
connection-gate-findings.md: playable FUT is research-grade (needs interactive
IDA/Ghidra GUI + full EA-online/Blaze emulator); the clean-room spec is the
finished deliverable.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add a bounded manual stack-scan in the GoOnline detour (FIFA-frame finder).
Result: zero FIFA23.exe frames, sp ~2.4KB below stack top => GoOnline runs on an
anadius worker thread (queued), not FIFA's thread. Three-way corroboration that
the gate is an async "online established" event anadius (offline-only) never
pushes; FIFA waits/retries. No handler-response flip can cross it. Documented in
connection-gate-findings.md.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Hook: force GetInternetConnectedState->connected (flags +0xCAB1A/+0xCAB1B) and
flip GoOnline to report "1" (+0xAF530) instead of "0" (+0xADE64).
- protossl-scan: add `read` mode (hex/ascii dump at addr|module+off).
- Finding: neither flip unblocks the game; it keeps retrying GoOnline every ~7s.
The FUT-online flow is event-driven -- the game waits for an async "online
established" event anadius (offline-only) never pushes. Documented in
connection-gate-findings.md. Next: trace FIFA-side flow (Ghidra) / anadius
event-send to inject the online event.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Found without Ghidra, via anadius's LSX command-registration table (handler list
is offset-by-one from the names; verified by +0x27060 = GetProfile). The gate is
GetInternetConnectedState @ anadius64.dll+0x27790: its LSX "connected" attribute
= (byte[+0xCAB1B]||byte[+0xCAB1A]) ? str(+0xAF530) : str(+0xADE64); both flags
default 0 -> reports offline. That is why the client aborts at "connecting".
M1 answers: (1) ProtoSSLConnect not reached (gate upstream); (2) connection-state
function named; (3) single actionable gate (token already provided by anadius
FakeAuth / GoOnline passes). M2 (next session) = make GetInternetConnectedState
report connected, then watch for the real Blaze connect.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add a read-only detour on anadius's GoOnline handler (anadius64.dll+0x2BB90).
Result: the game calls GoOnline during the "connecting" attempt and anadius
returns success, yet no Blaze connection follows. The gate is therefore
downstream of GoOnline -- the auth-token (GetAuthCode) and/or the "online
established" status callback. Recorded in connection-gate-findings.md.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Read-only M1 investigation results appended to connection-gate-findings.md:
- Confirmed ProtoSSLConnect is never reached on the "connecting" abort (gate is
upstream/in-process) via the existing connect/DNS hooks.
- Mapped the surface: redirector host table (gosredirector.* per env), the
redirector request config (X-BLAZE-ERRORCODE, Authorization:, <errorCode>),
and the enum-name serialization tables (ONLINE_ACCESS, ONLINE_STATUS_EVENT).
- Key answer: the online-connect path requires a Nucleus auth token (the
redirector request carries an Authorization header) => TWO gates (state +
token), not a single boolean flip.
- The exact connection-state function is behind C++ vtable indirection; pinning
it needs Ghidra. protossl-scan stays the live-process bridge.
protossl-scan: add `callers` mode (find call/jmp sites to a function) and
restrict xref/callers scans to the app modules (FIFA23.exe/anadius64.dll) for
speed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Stop fighting anadius's offline verdict. Document the clean-room findings
(the gate is an upstream in-process Ebisu connection-state check, not socket
traffic) and a milestone roadmap with two first-class outcomes: full playable
AI FUT (A) and a clean-room spec deliverable (B). Name M3 (first ProtoSSL
plaintext on the real Blaze connection) as the smallest end-to-end proof.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
FIFA 23 reverse-engineering proxy and integration scaffold.
- Catch-all HTTP proxy that captures all incoming FIFA 23 traffic
- Known-route mapper (speculative FUT paths → Core API calls)
- Placeholder JSON responses for unmapped endpoints
- Admin endpoints: GET /_bridge/captures, GET /_bridge/unknown
- Capture persistence to captures/*.json for RE analysis
- 4 unit tests passing
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
funman300