fix(wallet): reliable process detection + drop the harmful start-retry

Two issues the latest live log exposed:
- The switch's start-retry spawned a SECOND dragonxd while the first was still
  shutting down, so the two held wallet.dat against each other (BDB "Failed to
  rename wallet-savings.dat … Error"), and it then span on a stale "Daemon
  already running". Revert to a single start — the stopDaemonForWalletSwitch()
  wait already ensures the old process is gone, so a valid wallet opens cleanly
  and a bad one exits during init and reverts, without overlapping spawns.
- findProcessByName() used the non-suffixed PROCESSENTRY32/Process32First with an
  ANSI _stricmp; if UNICODE is defined those map to the wide variants, so the
  compare comparing garbage would NEVER match — silently making the process-gone
  wait a no-op. Rewritten with the explicit wide Toolhelp API + lstrcmpiW so it's
  correct either way (verified to compile under mingw with and without -DUNICODE).

Note: a corrupt wallet.dat (BDB recovery failing) still can't be opened by any
node — that's a data issue needing a clean reset, not a switch-flow bug.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

src/app_network.cpp

@@ -1137,21 +1137,18 @@ void App::switchToWallet(const std::string& walletFile, bool stopDaemonConfirmed
                 // as owned (stop/isRunning/exit behave normally for the next switch and app exit).
                 wallet_switch_phase_.store(static_cast<int>(WalletSwitchPhase::Starting));
                 if (daemon_controller_) daemon_controller_->clearExternalDaemonDetected();
-                // Retry the start. Right after the old node exits, its datadir lock can linger for a few
+                if (needRescan && daemon_controller_) daemon_controller_->setRescanOnNextStart(true);
-                // seconds (the process is gone but the OS/BDB lock isn't released yet), so the first start
+                // Start ONCE. Do NOT retry-spawn: a second start while the first is still shutting down leaves
-                // hits "Cannot obtain a lock on data directory" and exits within the grace below. Retry —
+                // two dragonxd holding wallet.dat against each other (BDB "Failed to rename … Error"). The
-                // with the CORRECT -wallet, since active_wallet_file isn't reverted until we give up — until
+                // stopDaemonForWalletSwitch() wait already ensured the old node's process is gone, so a valid
-                // one survives (valid launch stays alive while it loads the block index/rescans, ~30-60s).
+                // wallet opens cleanly; a bad/corrupt wallet exits during init and we revert.
-                for (int attempt = 0; attempt < 6 && !ok && !shutting_down_; ++attempt) {
+                ok = shutting_down_ ? true : startEmbeddedDaemon();
-                    if (attempt > 0) std::this_thread::sleep_for(std::chrono::milliseconds(2000));  // let the lock release
+                // A missing/corrupt wallet makes dragonxd exit during init — confirm the process survived a
-                    if (daemon_controller_) daemon_controller_->resetCrashCount();                  // retry lock-crashes aren't real
+                // moment; if not, the wallet is bad. (A valid launch keeps the process alive while it syncs.)
-                    if (needRescan && daemon_controller_) daemon_controller_->setRescanOnNextStart(true);  // one-shot; re-arm each try
+                if (ok && !shutting_down_) {
-                    if (shutting_down_) { ok = true; break; }
+                    std::this_thread::sleep_for(std::chrono::milliseconds(1500));
-                    if (!startEmbeddedDaemon()) continue;                          // couldn't spawn (port busy) — back off + retry
+                    if (!isEmbeddedDaemonRunning()) ok = false;
-                    std::this_thread::sleep_for(std::chrono::milliseconds(1500));  // grace: a datadir-lock crash exits within this
-                    ok = shutting_down_ || isEmbeddedDaemonRunning();
                 }
-                if (ok && daemon_controller_) daemon_controller_->resetCrashCount();  // clean slate for the survivor
                 // Process is up and staying up — now waiting for RPC to answer (onConnected closes the modal).
                 if (ok) wallet_switch_phase_.store(static_cast<int>(WalletSwitchPhase::Reconnecting));
             }

src/daemon/embedded_daemon.cpp

@@ -683,18 +683,25 @@ static DWORD findProcessByName(const char* name)
 {
     HANDLE snap = CreateToolhelp32Snapshot(TH32CS_SNAPPROCESS, 0);
     if (snap == INVALID_HANDLE_VALUE) return 0;
-    
+
-    PROCESSENTRY32 entry;
+    // Use the explicit WIDE Toolhelp API + a wide compare so this is correct regardless of the UNICODE
+    // macro. (The non-suffixed PROCESSENTRY32/Process32First map to the wide variants when UNICODE is
+    // defined, in which case szExeFile is WCHAR[] and an ANSI _stricmp would compare garbage and NEVER
+    // match — silently making findProcessByName a no-op that returns 0 for a running process.)
+    wchar_t wname[MAX_PATH];
+    if (MultiByteToWideChar(CP_ACP, 0, name, -1, wname, MAX_PATH) == 0) { CloseHandle(snap); return 0; }
+
+    PROCESSENTRY32W entry;
     entry.dwSize = sizeof(entry);
-    
+
     DWORD pid = 0;
-    if (Process32First(snap, &entry)) {
+    if (Process32FirstW(snap, &entry)) {
         do {
-            if (_stricmp(entry.szExeFile, name) == 0) {
+            if (lstrcmpiW(entry.szExeFile, wname) == 0) {   // Win32 case-insensitive wide compare
                 pid = entry.th32ProcessID;
                 break;
             }
-        } while (Process32Next(snap, &entry));
+        } while (Process32NextW(snap, &entry));
     }
     CloseHandle(snap);
     return pid;