diff --git a/rpcs3/Emu/ARMv7/ARMv7Decoder.cpp b/rpcs3/Emu/ARMv7/ARMv7Decoder.cpp
index fac6873a33..ce2dbb6914 100644
--- a/rpcs3/Emu/ARMv7/ARMv7Decoder.cpp
+++ b/rpcs3/Emu/ARMv7/ARMv7Decoder.cpp
@@ -1,6 +1,7 @@
 #include "stdafx.h"
 #include <unordered_map>
 #include "Utilities/Log.h"
+#include "Emu/Memory/Memory.h"
 #include "ARMv7Thread.h"
 #include "ARMv7Interpreter.h"
 #include "ARMv7Opcodes.h"
@@ -1262,7 +1263,7 @@ void armv7_decoder_initialize(u32 addr, u32 end_addr, bool dump)
 		addr += 16;
 	}
 
-	LOG_NOTICE(ARMv7, "armv7_decoder_initialize() finished, g_opct.size() = %lld", g_opct.size());
+	LOG_NOTICE(ARMv7, "armv7_decoder_initialize() finished, g_opct.size() = %lld", (u64)g_opct.size());
 }
 
 u32 ARMv7Decoder::DecodeMemory(const u32 address)
diff --git a/rpcs3/Emu/ARMv7/ARMv7Interpreter.cpp b/rpcs3/Emu/ARMv7/ARMv7Interpreter.cpp
index 6ea768832b..c204351616 100644
--- a/rpcs3/Emu/ARMv7/ARMv7Interpreter.cpp
+++ b/rpcs3/Emu/ARMv7/ARMv7Interpreter.cpp
@@ -191,7 +191,7 @@ namespace ARMv7_instrs
 
 	template<typename T> T AddWithCarry(T x, T y, bool carry_in, bool& carry_out, bool& overflow)
 	{
-		const T sign_mask = (T)1 << (sizeof(T) - 1);
+		const T sign_mask = (T)1 << (sizeof(T) * 8 - 1);
 
 		T result = x + y;
 		carry_out = ((x & y) | ((x ^ y) & ~result)) & sign_mask;
@@ -823,6 +823,7 @@ void ARMv7_instrs::B(ARMv7Context& context, const ARMv7Code code, const ARMv7_en
 
 	if (ConditionPassed(context, cond))
 	{
+		//LOG_NOTICE(ARMv7, "Branch to 0x%x (cond=0x%x)", context.thread.PC + jump, cond);
 		context.thread.SetBranch(context.thread.PC + jump);
 	}
 }
@@ -1174,6 +1175,7 @@ void ARMv7_instrs::CMP_REG(ARMv7Context& context, const ARMv7Code code, const AR
 	}
 	case T3:
 	{
+		cond = context.ITSTATE.advance();
 		n = (code.data & 0xf0000) >> 16;
 		m = (code.data & 0xf);
 		shift_t = DecodeImmShift((code.data & 0x30) >> 4, (code.data & 0x7000) >> 10 | (code.data & 0xc0) >> 6, &shift_n);
@@ -1187,13 +1189,17 @@ void ARMv7_instrs::CMP_REG(ARMv7Context& context, const ARMv7Code code, const AR
 
 	if (ConditionPassed(context, cond))
 	{
+		const u32 m_value = context.read_gpr(m);
+		const u32 n_value = context.read_gpr(n);
 		bool carry, overflow;
-		const u32 shifted = Shift(context.read_gpr(m), shift_t, shift_n, true);
-		const u32 res = AddWithCarry(context.read_gpr(n), ~shifted, true, carry, overflow);
+		const u32 shifted = Shift(m_value, shift_t, shift_n, true);
+		const u32 res = AddWithCarry(n_value, ~shifted, true, carry, overflow);
 		context.APSR.N = res >> 31;
 		context.APSR.Z = res == 0;
 		context.APSR.C = carry;
 		context.APSR.V = overflow;
+
+		//LOG_NOTICE(ARMv7, "CMP: r%d=0x%08x <> r%d=0x%08x, shifted=0x%08x, res=0x%08x", n, n_value, m, m_value, shifted, res);
 	}
 }
 
@@ -1619,11 +1625,60 @@ void ARMv7_instrs::LDRH_REG(ARMv7Context& context, const ARMv7Code code, const A
 
 void ARMv7_instrs::LDRSB_IMM(ARMv7Context& context, const ARMv7Code code, const ARMv7_encoding type)
 {
+	u32 cond, t, n, imm32;
+	bool index, add, wback;
+
 	switch (type)
 	{
+	case T1:
+	{
+		cond = context.ITSTATE.advance();
+		t = (code.data & 0xf000) >> 12;
+		n = (code.data & 0xf0000) >> 16;
+		imm32 = (code.data & 0xfff);
+		index = true;
+		add = true;
+		wback = false;
+
+		reject(t == 15, "PLI");
+		reject(n == 15, "LDRSB (literal)");
+		reject(t == 13, "UNPREDICTABLE");
+		break;
+	}
+	case T2:
+	{
+		cond = context.ITSTATE.advance();
+		t = (code.data & 0xf000) >> 12;
+		n = (code.data & 0xf0000) >> 16;
+		imm32 = (code.data & 0xff);
+		index = (code.data & 0x400);
+		add = (code.data & 0x200);
+		wback = (code.data & 0x100);
+
+		reject(t == 15 && index && !add && !wback, "PLI");
+		reject(n == 15, "LDRSB (literal)");
+		reject(index && add && !wback, "LDRSBT");
+		reject(!index && !wback, "UNDEFINED");
+		reject(t == 13 || t == 15 || (wback && n == t), "UNPREDICTABLE");
+		break;
+	}
 	case A1: throw __FUNCTION__;
 	default: throw __FUNCTION__;
 	}
+
+	if (ConditionPassed(context, cond))
+	{
+		const u32 offset_addr = add ? context.read_gpr(n) + imm32 : context.read_gpr(n) - imm32;
+		const u32 addr = index ? offset_addr : context.read_gpr(n);
+		const s8 value = vm::psv::read8(addr);
+
+		context.write_gpr(t, value); // sign-extend
+
+		if (wback)
+		{
+			context.write_gpr(n, offset_addr);
+		}
+	}
 }
 
 void ARMv7_instrs::LDRSB_LIT(ARMv7Context& context, const ARMv7Code code, const ARMv7_encoding type)
diff --git a/rpcs3/Emu/ARMv7/ARMv7Thread.cpp b/rpcs3/Emu/ARMv7/ARMv7Thread.cpp
index 1a968dd858..f58acfd0d3 100644
--- a/rpcs3/Emu/ARMv7/ARMv7Thread.cpp
+++ b/rpcs3/Emu/ARMv7/ARMv7Thread.cpp
@@ -120,10 +120,19 @@ void ARMv7Thread::InitRegs()
 
 void ARMv7Thread::InitStack()
 {
-	if(!m_stack_addr)
+	if (!m_stack_addr)
 	{
-		m_stack_size = 0x10000;
-		m_stack_addr = (u32)Memory.Alloc(0x10000, 1);
+		assert(m_stack_size);
+		m_stack_addr = vm::cast(Memory.Alloc(m_stack_size, 4096));
+	}
+}
+
+void ARMv7Thread::CloseStack()
+{
+	if (m_stack_addr)
+	{
+		Memory.Free(m_stack_addr);
+		m_stack_addr = 0;
 	}
 }
 
@@ -218,7 +227,7 @@ void ARMv7Thread::FastStop()
 	m_status = Stopped;
 }
 
-arm7_thread::arm7_thread(u32 entry, const std::string& name, u32 stack_size, u32 prio)
+armv7_thread::armv7_thread(u32 entry, const std::string& name, u32 stack_size, u32 prio)
 {
 	thread = &Emu.GetCPU().AddThread(CPU_THREAD_ARMv7);
 
@@ -229,3 +238,47 @@ arm7_thread::arm7_thread(u32 entry, const std::string& name, u32 stack_size, u32
 
 	argc = 0;
 }
+
+cpu_thread& armv7_thread::args(std::initializer_list<std::string> values)
+{
+	assert(argc == 0);
+
+	if (!values.size())
+	{
+		return *this;
+	}
+
+	std::vector<char> argv_data;
+	u32 argv_size = 0;
+
+	for (auto& arg : values)
+	{
+		const u32 arg_size = vm::cast(arg.size(), "arg.size()"); // get arg size
+
+		for (char c : arg)
+		{
+			argv_data.push_back(c); // append characters
+		}
+
+		argv_data.push_back('\0'); // append null terminator
+
+		argv_size += arg_size + 1;
+		argc++;
+	}
+
+	argv = vm::cast(Memory.PSV.RAM.AllocAlign(argv_size, 4096)); // allocate arg list
+	memcpy(vm::get_ptr(argv), argv_data.data(), argv_size); // copy arg list
+	
+	return *this;
+}
+
+cpu_thread& armv7_thread::run()
+{
+	thread->Run();
+
+	// set arguments
+	static_cast<ARMv7Thread*>(thread)->context.GPR[0] = argc;
+	static_cast<ARMv7Thread*>(thread)->context.GPR[1] = argv;
+
+	return *this;
+}
diff --git a/rpcs3/Emu/ARMv7/ARMv7Thread.h b/rpcs3/Emu/ARMv7/ARMv7Thread.h
index c806f4b6c4..a8a0f75311 100644
--- a/rpcs3/Emu/ARMv7/ARMv7Thread.h
+++ b/rpcs3/Emu/ARMv7/ARMv7Thread.h
@@ -1,29 +1,19 @@
 #pragma once
 #include "Emu/CPU/CPUThread.h"
-#include "Emu/Memory/Memory.h"
 #include "ARMv7Context.h"
 
 class ARMv7Thread : public CPUThread
 {
 public:
 	ARMv7Context context;
-	//u32 m_arg;
-	//u8 m_last_instr_size;
-	//const char* m_last_instr_name;
 
 	ARMv7Thread();
 	~ARMv7Thread();
 
-	//void update_code(const u32 address)
-	//{
-	//	code.code0 = vm::psv::read16(address & ~1);
-	//	code.code1 = vm::psv::read16(address + 2 & ~1);
-	//	m_arg = address & 0x1 ? code.code1 << 16 | code.code0 : code.data;
-	//}
-
 public:
 	virtual void InitRegs(); 
 	virtual void InitStack();
+	virtual void CloseStack();
 	u32 GetStackArg(u32 pos);
 	void FastCall(u32 addr);
 	void FastStop();
@@ -42,48 +32,16 @@ protected:
 
 	virtual void DoCode();
 };
-class arm7_thread : cpu_thread
+
+class armv7_thread : cpu_thread
 {
-	static const u32 stack_align = 0x10;
-	vm::ptr<u64> argv;
+	u32 argv;
 	u32 argc;
-	vm::ptr<u64> envp;
 
 public:
-	arm7_thread(u32 entry, const std::string& name = "", u32 stack_size = 0, u32 prio = 0);
+	armv7_thread(u32 entry, const std::string& name = "", u32 stack_size = 0, u32 prio = 0);
 
-	cpu_thread& args(std::initializer_list<std::string> values) override
-	{
-		if (!values.size())
-			return *this;
+	cpu_thread& args(std::initializer_list<std::string> values) override;
 
-		//assert(argc == 0);
-
-		//envp.set(vm::alloc((u32)sizeof(envp), stack_align, vm::main));
-		//*envp = 0;
-		//argv.set(vm::alloc(u32(sizeof(argv)* values.size()), stack_align, vm::main));
-
-		for (auto &arg : values)
-		{
-			//u32 arg_size = align(u32(arg.size() + 1), stack_align);
-			//u32 arg_addr = vm::alloc(arg_size, stack_align, vm::main);
-
-			//std::strcpy(vm::get_ptr<char>(arg_addr), arg.c_str());
-
-			//argv[argc++] = arg_addr;
-		}
-
-		return *this;
-	}
-
-	cpu_thread& run() override
-	{
-		thread->Run();
-
-		//static_cast<ARMv7Thread*>(thread)->GPR[0] = argc;
-		//static_cast<ARMv7Thread*>(thread)->GPR[1] = argv.addr();
-		//static_cast<ARMv7Thread*>(thread)->GPR[2] = envp.addr();
-
-		return *this;
-	}
+	cpu_thread& run() override;
 };
diff --git a/rpcs3/Emu/CPU/CPUThread.cpp b/rpcs3/Emu/CPU/CPUThread.cpp
index a69ec8338a..b2224af9c0 100644
--- a/rpcs3/Emu/CPU/CPUThread.cpp
+++ b/rpcs3/Emu/CPU/CPUThread.cpp
@@ -64,17 +64,6 @@ void CPUThread::Reset()
 	DoReset();
 }
 
-void CPUThread::CloseStack()
-{
-	if(m_stack_addr)
-	{
-		Memory.StackMem.Free(m_stack_addr);
-		m_stack_addr = 0;
-	}
-
-	m_stack_size = 0;
-}
-
 void CPUThread::SetId(const u32 id)
 {
 	m_id = id;
diff --git a/rpcs3/Emu/CPU/CPUThread.h b/rpcs3/Emu/CPU/CPUThread.h
index dd79700c65..35010483cc 100644
--- a/rpcs3/Emu/CPU/CPUThread.h
+++ b/rpcs3/Emu/CPU/CPUThread.h
@@ -1,8 +1,7 @@
 #pragma once
-
 #include "Utilities/Thread.h"
 
-enum CPUThreadType :unsigned char
+enum CPUThreadType : unsigned char
 {
 	CPU_THREAD_PPU,
 	CPU_THREAD_SPU,
@@ -46,10 +45,10 @@ protected:
 	bool m_trace_call_stack;
 
 public:
-	virtual void InitRegs()=0;
+	virtual void InitRegs() = 0;
 
-	virtual void InitStack()=0;
-	virtual void CloseStack();
+	virtual void InitStack() = 0;
+	virtual void CloseStack() = 0;
 
 	u32 GetStackAddr() const { return m_stack_addr; }
 	u32 GetStackSize() const { return m_stack_size; }
@@ -271,4 +270,4 @@ public:
 	{
 		return thread->GetId();
 	}
-};
\ No newline at end of file
+};
diff --git a/rpcs3/Emu/Cell/PPCThread.cpp b/rpcs3/Emu/Cell/PPCThread.cpp
index 4248fd58a8..4d77900d90 100644
--- a/rpcs3/Emu/Cell/PPCThread.cpp
+++ b/rpcs3/Emu/Cell/PPCThread.cpp
@@ -25,18 +25,3 @@ PPCThread::~PPCThread()
 void PPCThread::DoReset()
 {
 }
-
-void PPCThread::InitStack()
-{
-	if(m_stack_addr) return;
-	if(m_stack_size == 0) m_stack_size = 0x10000;
-	m_stack_addr = (u32)Memory.StackMem.AllocAlign(m_stack_size, 0x100);
-	/*
-	m_stack_point += m_stack_size - 0x10;
-	m_stack_point &= -0x10;
-	vm::write64(m_stack_point, 0);
-	m_stack_point -= 0x60;
-	vm::write64(m_stack_point, m_stack_point + 0x60);
-	*/
-}
-
diff --git a/rpcs3/Emu/Cell/PPCThread.h b/rpcs3/Emu/Cell/PPCThread.h
index 99471f7a80..2c3fa5c05b 100644
--- a/rpcs3/Emu/Cell/PPCThread.h
+++ b/rpcs3/Emu/Cell/PPCThread.h
@@ -4,9 +4,6 @@
 class PPCThread : public CPUThread
 {
 public:
-	virtual void InitRegs()=0;
-	virtual void InitStack();
-
 	virtual std::string GetThreadName() const
 	{
 		return fmt::format("%s[0x%08x]", GetFName(), PC);
diff --git a/rpcs3/Emu/Cell/PPUThread.cpp b/rpcs3/Emu/Cell/PPUThread.cpp
index e064a54feb..e6d229583a 100644
--- a/rpcs3/Emu/Cell/PPUThread.cpp
+++ b/rpcs3/Emu/Cell/PPUThread.cpp
@@ -80,6 +80,24 @@ void PPUThread::InitRegs()
 	TB = 0;
 }
 
+void PPUThread::InitStack()
+{
+	if (!m_stack_addr)
+	{
+		assert(m_stack_size);
+		m_stack_addr = vm::cast(Memory.StackMem.AllocAlign(m_stack_size, 4096));
+	}
+}
+
+void PPUThread::CloseStack()
+{
+	if (m_stack_addr)
+	{
+		Memory.StackMem.Free(m_stack_addr);
+		m_stack_addr = 0;
+	}
+}
+
 void PPUThread::DoRun()
 {
 	switch(Ini.CPUDecoderMode.GetValue())
diff --git a/rpcs3/Emu/Cell/PPUThread.h b/rpcs3/Emu/Cell/PPUThread.h
index 51722f1645..820b244c2b 100644
--- a/rpcs3/Emu/Cell/PPUThread.h
+++ b/rpcs3/Emu/Cell/PPUThread.h
@@ -799,8 +799,10 @@ public:
 	}
 
 public:
-	virtual void InitRegs();
-	virtual void Task();
+	virtual void InitRegs() override;
+	virtual void InitStack() override;
+	virtual void CloseStack() override;
+	virtual void Task() override;
 	u64 GetStackArg(s32 i);
 	u64 FastCall2(u32 addr, u32 rtoc);
 	void FastStop();
diff --git a/rpcs3/Emu/Cell/SPUThread.cpp b/rpcs3/Emu/Cell/SPUThread.cpp
index 1c861da43c..f5faf9fff1 100644
--- a/rpcs3/Emu/Cell/SPUThread.cpp
+++ b/rpcs3/Emu/Cell/SPUThread.cpp
@@ -103,6 +103,17 @@ void SPUThread::InitRegs()
 	R_ADDR = 0;
 }
 
+void SPUThread::InitStack()
+{
+	m_stack_size = 0x1000; // this value is wrong
+	m_stack_addr = m_offset + 0x40000 - m_stack_size; // stack is the part of SPU Local Storage
+}
+
+void SPUThread::CloseStack()
+{
+	// nothing to do here
+}
+
 void SPUThread::DoRun()
 {
 	switch(Ini.SPUDecoderMode.GetValue())
diff --git a/rpcs3/Emu/Cell/SPUThread.h b/rpcs3/Emu/Cell/SPUThread.h
index 50d70b1c7f..c2e2cc2f03 100644
--- a/rpcs3/Emu/Cell/SPUThread.h
+++ b/rpcs3/Emu/Cell/SPUThread.h
@@ -568,6 +568,8 @@ public:
 
 public:
 	virtual void InitRegs();
+	virtual void InitStack();
+	virtual void CloseStack();
 	virtual void Task();
 	void FastCall(u32 ls_addr);
 	void FastStop();
diff --git a/rpcs3/Emu/Memory/vm.cpp b/rpcs3/Emu/Memory/vm.cpp
index 2c23ecdb66..69e42e62cb 100644
--- a/rpcs3/Emu/Memory/vm.cpp
+++ b/rpcs3/Emu/Memory/vm.cpp
@@ -2,6 +2,7 @@
 #include "Memory.h"
 #include "Emu/CPU/CPUThread.h"
 #include "Emu/Cell/PPUThread.h"
+#include "Emu/ARMv7/ARMv7Thread.h"
 
 namespace vm
 {
@@ -167,13 +168,14 @@ namespace vm
 
 	u32 stack_push(CPUThread& CPU, u32 size, u32 align_v, u32& old_pos)
 	{
+		assert(align_v);
+
 		switch (CPU.GetType())
 		{
 		case CPU_THREAD_PPU:
 		{
 			PPUThread& PPU = static_cast<PPUThread&>(CPU);
 
-			assert(align_v);
 			old_pos = (u32)PPU.GPR[1];
 			PPU.GPR[1] -= align(size, 8); // room minimal possible size
 			PPU.GPR[1] &= ~(align_v - 1); // fix stack alignment
@@ -199,8 +201,22 @@ namespace vm
 
 		case CPU_THREAD_ARMv7:
 		{
-			assert(!"stack_push(): ARMv7 not supported");
-			return 0;
+			ARMv7Context& context = static_cast<ARMv7Thread&>(CPU).context;
+
+			old_pos = context.SP;
+			context.SP -= align(size, 4); // room minimal possible size
+			context.SP &= ~(align_v - 1); // fix stack alignment
+
+			if (context.SP < CPU.GetStackAddr())
+			{
+				// stack overflow
+				context.SP = old_pos;
+				return 0;
+			}
+			else
+			{
+				return context.SP;
+			}
 		}
 
 		default:
diff --git a/rpcs3/Loader/ELF32.cpp b/rpcs3/Loader/ELF32.cpp
index 9c7f19d0c2..c622308f3d 100644
--- a/rpcs3/Loader/ELF32.cpp
+++ b/rpcs3/Loader/ELF32.cpp
@@ -323,7 +323,7 @@ namespace loader
 				armv7_init_tls();
 				armv7_decoder_initialize(code_start, code_end);
 
-				arm7_thread(entry & ~1 /* TODO: Thumb/ARM encoding selection */, "main_thread").args({ Emu.GetPath()/*, "-emu"*/ }).run();
+				armv7_thread(entry & ~1 /* TODO: Thumb/ARM encoding selection */, "main_thread").args({ Emu.GetPath(), "-emu" }).run();
 				break;
 			}
 			case MACHINE_SPU: spu_thread(m_ehdr.is_le() ? m_ehdr.data_le.e_entry : m_ehdr.data_be.e_entry, "main_thread").args({ Emu.GetPath()/*, "-emu"*/ }).run(); break;