From dd175b10fd8a0f98f95225d166e360d235d0f638 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Henrik=20Rydg=C3=A5rd?= <hrydgard@gmail.com>
Date: Wed, 4 Dec 2024 12:42:10 +0100
Subject: [PATCH] Warning fix

---
 GPU/Debugger/Playback.cpp | 1 -
 GPU/GPUCommon.cpp         | 8 ++++++--
 GPU/Software/SoftGpu.cpp  | 5 +++--
 headless/HeadlessHost.h   | 2 ++
 4 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/GPU/Debugger/Playback.cpp b/GPU/Debugger/Playback.cpp
index 1edc63a7bc..81288e6b77 100644
--- a/GPU/Debugger/Playback.cpp
+++ b/GPU/Debugger/Playback.cpp
@@ -432,7 +432,6 @@ bool DumpExecute::SubmitCmds(const void *p, u32 sz) {
 	}
 
 	execListQueue.clear();
-
 	return true;
 }
 
diff --git a/GPU/GPUCommon.cpp b/GPU/GPUCommon.cpp
index c79b053124..bac8d0c3a7 100644
--- a/GPU/GPUCommon.cpp
+++ b/GPU/GPUCommon.cpp
@@ -561,6 +561,8 @@ void GPUCommon::RunGe(DeferredAction action, bool forceRunDirect) {
 	RunDeferredAction();
 #else
 	// No sense in switching core mode if there's nothing to do, just maybe some bookkeeping.
+	// Also, in some situations like dump playback, we run DLs directly too (yes, this means that
+	// we can't step through display lists in a dump, which should be fixed later).
 	if (dlQueue.empty() || forceRunDirect) {
 		ProcessDLQueue(false);
 		RunDeferredAction();
@@ -580,7 +582,9 @@ void GPUCommon::RunDeferredAction() {
 		CoreTiming::ForceCheck();
 	}
 	if (deferredAction_.reschedule) {
-		hleReSchedule("ge deferred action");
+		// NOTE: hleReSchedule doesn't work if we're not in a syscall. We just reschedule, implementing the parts of hleFinishSyscall that are relevant.
+		__KernelForceCallbacks();
+		__KernelReSchedule(true, "ge deferred action");
 	}
 }
 
@@ -879,7 +883,7 @@ DLResult GPUCommon::ProcessDLQueue(bool fromCore) {
 
 	for (int listIndex = GetNextListIndex(); listIndex != -1; listIndex = GetNextListIndex()) {
 		DisplayList &l = dls[listIndex];
-		DEBUG_LOG(Log::G3D, "Starting DL execution at %08x - stall = %08x (startingTicks=%d)", l.pc, l.stall, startingTicks);
+		DEBUG_LOG(Log::G3D, "Starting DL execution at %08x - stall = %08x (startingTicks=%lld)", l.pc, l.stall, startingTicks);
 		if (!InterpretList(l)) {
 			return DLResult::Error;
 		}
diff --git a/GPU/Software/SoftGpu.cpp b/GPU/Software/SoftGpu.cpp
index d0b0b8d035..f64c81d633 100644
--- a/GPU/Software/SoftGpu.cpp
+++ b/GPU/Software/SoftGpu.cpp
@@ -1168,8 +1168,9 @@ void SoftGPU::Execute_TgenMtxData(u32 op, u32 diff) {
 	}
 
 	// Doesn't wrap to any other matrix.
-	if ((num & 0xF) < 12) {
-		matrixVisible.tgen[num & 0xF] = op & 0x00FFFFFF;
+	int wrappedNum = num & 0xF;
+	if (wrappedNum < 12) {
+		matrixVisible.tgen[wrappedNum] = op & 0x00FFFFFF;
 	}
 
 	num++;
diff --git a/headless/HeadlessHost.h b/headless/HeadlessHost.h
index 9f61bf08b3..5ef80d8c98 100644
--- a/headless/HeadlessHost.h
+++ b/headless/HeadlessHost.h
@@ -17,6 +17,7 @@
 
 #pragma once
 
+#include "Common/Log.h"
 #include "Common/CommonTypes.h"
 #include "Common/File/Path.h"
 
@@ -31,6 +32,7 @@ public:
 	virtual void SendDebugOutput(const std::string &output) {
 		if (!writeDebugOutput_)
 			return;
+		OutputDebugStringUTF8(output.c_str());
 		if (output.find('\n') != output.npos) {
 			FlushDebugOutput();
 			fwrite(output.data(), sizeof(char), output.length(), stdout);