BuildSystem: Fix keymap in base sysroot

BAN/include/BAN/Math.h

@@ -2,7 +2,6 @@
 
 #include <BAN/Limits.h>
 #include <BAN/Numbers.h>
-#include <BAN/Swap.h>
 #include <BAN/Traits.h>
 
 #include <float.h>
@@ -82,37 +81,6 @@ namespace BAN::Math
 		return x + 1;
 	}
 
-	template<unsigned_integral T>
-	requires is_same_v<T, unsigned int> || is_same_v<T, unsigned long> || is_same_v<T, unsigned long long>
-	inline constexpr int clz(T x)
-	{
-		if constexpr (is_same_v<T, unsigned int>)
-			return __builtin_clz(x);
-		if constexpr (is_same_v<T, unsigned long>)
-			return __builtin_clzl(x);
-		return __builtin_clzll(x);
-	}
-
-	template<unsigned_integral T> requires(sizeof(T) <= sizeof(unsigned long long))
-	inline constexpr int ctz(T x)
-	{
-		if constexpr (sizeof(T) <= sizeof(unsigned int))
-			return __builtin_ctz(x);
-		if constexpr (sizeof(T) <= sizeof(unsigned long))
-			return __builtin_ctzl(x);
-		return __builtin_ctzll(x);
-	}
-
-	template<unsigned_integral T> requires(sizeof(T) <= sizeof(unsigned long long))
-	inline constexpr int popcount(T x)
-	{
-		if constexpr (sizeof(T) <= sizeof(unsigned int))
-			return __builtin_popcount(x);
-		if constexpr (sizeof(T) <= sizeof(unsigned long))
-			return __builtin_popcountl(x);
-		return __builtin_popcountll(x);
-	}
-
 	template<integral T>
 	__attribute__((always_inline))
 	inline constexpr bool will_multiplication_overflow(T a, T b)
@@ -133,7 +101,11 @@ namespace BAN::Math
 	requires is_same_v<T, unsigned int> || is_same_v<T, unsigned long> || is_same_v<T, unsigned long long>
 	inline constexpr T ilog2(T x)
 	{
-		return sizeof(T) * 8 - clz(x) - 1;
+		if constexpr(is_same_v<T, unsigned int>)
+			return sizeof(T) * 8 - __builtin_clz(x) - 1;
+		if constexpr(is_same_v<T, unsigned long>)
+			return sizeof(T) * 8 - __builtin_clzl(x) - 1;
+		return sizeof(T) * 8 - __builtin_clzll(x) - 1;
 	}
 
 // This is ugly but my clangd does not like including
@@ -336,6 +308,36 @@ namespace BAN::Math
 		return exp2<T>(y * log2<T>(x));
 	}
 
+	template<floating_point T>
+	inline constexpr T scalbn(T x, int n)
+	{
+		asm("fscale" : "+t"(x) : "u"(static_cast<T>(n)));
+		return x;
+	}
+
+	template<floating_point T>
+	inline constexpr T ldexp(T x, int y)
+	{
+		const bool exp_sign = y < 0;
+		if (exp_sign)
+			y = -y;
+
+		T exp = (T)1.0;
+		T mult = (T)2.0;
+		while (y)
+		{
+			if (y & 1)
+				exp *= mult;
+			mult *= mult;
+			y >>= 1;
+		}
+
+		if (exp_sign)
+			exp = (T)1.0 / exp;
+
+		return x * exp;
+	}
+
 	template<floating_point T>
 	inline constexpr T sqrt(T x)
 	{
@@ -468,17 +470,7 @@ namespace BAN::Math
 	template<floating_point T>
 	inline constexpr T hypot(T x, T y)
 	{
-		x = abs(x);
-		y = abs(y);
-
-		if (x < y)
-			swap(x, y);
-
-		if (y == (T)0.0)
-			return x;
-
-		const T r = y / x;
-		return x * sqrt<T>((T)1.0 + r * r);
+		return sqrt<T>(x * x + y * y);
 	}
 
 #ifdef BAN_MATH_POP_OPTIONS

CMakeLists.txt

@@ -27,12 +27,6 @@ set(BUILD_SHARED_LIBS True)
 
 set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
 
-find_program(CCACHE_PROGRAM "ccache")
-if (CCACHE_PROGRAM)
-	set(CMAKE_C_COMPILER_LAUNCHER "${CCACHE_PROGRAM}")
-	set(CMAKE_CXX_COMPILER_LAUNCHER "${CCACHE_PROGRAM}")
-endif()
-
 # include headers of ${library} to ${target}
 function(banan_include_headers target library)
 	target_include_directories(${target} PUBLIC $<TARGET_PROPERTY:${library},SOURCE_DIR>/include)

bootloader/bios/command_line.S

@@ -35,13 +35,13 @@ read_user_command_line:
 	cmpb $'\n', %al
 	je .read_user_command_line_done
 
-	movb %al, %bl
+	pushw %ax
 
 	call isprint
 	testb %al, %al
 	jz .read_user_command_line_loop
 
-	movb %bl, %al
+	popw %ax
 
 	# put byte to buffer
 	movb %al, (%di)

bootloader/bios/framebuffer.S

@@ -108,11 +108,9 @@ vesa_find_video_mode:
 	cmpb $0x4F, %al; jne .vesa_unsupported
 	cmpb $0x00, %ah; jne .vesa_error
 
-	# check whether in graphics mode and linear framebuffer
-	movb (vesa_mode_info_buffer + 0), %al
-	andb $(0x80 | 0x10), %al
-	cmpb $(0x80 | 0x10), %al
-	jne .vesa_find_video_mode_next_mode
+	# check whether in graphics mode
+	testb $0x10, (vesa_mode_info_buffer + 0)
+	jz .vesa_find_video_mode_next_mode
 
 	# compare mode's dimensions
 	movw -2(%ebp), %ax; cmpw %ax, (vesa_mode_info_buffer + 0x12)

bootloader/bios/memory_map.S

@@ -2,8 +2,6 @@
 
 .section .stage2
 
-.set mmap_entry_size, 24
-
 # fills memory map data structure
 # doesn't return on error
 # NO REGISTERS SAVED
@@ -14,7 +12,7 @@ get_memory_map:
 	movl $0x0000E820, %eax
 	movl $0x534D4150, %edx
 	xorl %ebx, %ebx
-	movl $mmap_entry_size, %ecx
+	movl $20, %ecx
 	movw $memory_map_entries, %di
 
 	clc
@@ -26,15 +24,15 @@ get_memory_map:
 	cmpl $0x534D4150, %eax
 	jne .get_memory_map_error
 
-	# confirm that we got at least 20 bytes
+	# FIXME: don't assume BIOS to always return 20 bytes
 	cmpl $20, %ecx
-	jl .get_memory_map_error
+	jne .get_memory_map_error
 
 	# increment entry count
 	incl (memory_map_entry_count)
 
 	# increment entry pointer
-	addw $mmap_entry_size, %di
+	addw %cx, %di
 
 	# BIOS can indicate end of list by 0 in ebx
 	testl %ebx, %ebx
@@ -109,7 +107,7 @@ print_memory_map:
 
 	call print_newline
 
-	addw $mmap_entry_size, %si
+	addw $20, %si
 
 	decl %edx
 	jnz .loop_memory_map
@@ -129,8 +127,6 @@ memory_map_error_msg:
 memory_map:
 memory_map_entry_count:
 	.skip 4
-memory_map_padding:
-	.skip 4
-# 128 entries should be enough...
+# 100 entries should be enough...
 memory_map_entries:
-	.skip mmap_entry_size * 128
+	.skip 20 * 100

kernel/CMakeLists.txt

@@ -51,8 +51,6 @@ set(KERNEL_SOURCES
 	kernel/Interruptable.cpp
 	kernel/InterruptController.cpp
 	kernel/kernel.cpp
-	kernel/Lock/Mutex.cpp
-	kernel/Lock/RWLock.cpp
 	kernel/Lock/SpinLock.cpp
 	kernel/Memory/ByteRingBuffer.cpp
 	kernel/Memory/DMARegion.cpp
@@ -85,7 +83,6 @@ set(KERNEL_SOURCES
 	kernel/Processor.cpp
 	kernel/Random.cpp
 	kernel/Scheduler.cpp
-	kernel/SchedulerThreadNode.cpp
 	kernel/SSP.cpp
 	kernel/Storage/ATA/AHCI/Controller.cpp
 	kernel/Storage/ATA/AHCI/Device.cpp

kernel/arch/i686/Signal.S

@@ -59,7 +59,7 @@ signal_trampoline:
 	addl $24, %esp
 
 	// restore sigmask
-	movl $79, %eax      // SYS_SIGPROCMASK
+	movl $78, %eax      // SYS_SIGPROCMASK
 	movl $3,  %ebx      // SIG_SETMASK
 	leal 72(%esp), %ecx // set
 	xorl %edx, %edx     // oset

kernel/arch/i686/Syscall.S

@@ -63,7 +63,7 @@ sys_fork_trampoline:
 
 	call read_ip
 	testl %eax, %eax
-	jz .Lsys_fork_trampoline_done
+	jz .done
 
 	movl %esp, %ebx
 
@@ -73,7 +73,7 @@ sys_fork_trampoline:
 	call sys_fork
 	addl $16, %esp
 
- .Lsys_fork_trampoline_done:
+ .done:
 	popl %edi
 	popl %esi
 	popl %ebx

kernel/arch/i686/User.S

@@ -28,26 +28,27 @@ safe_user_strncpy:
 	testl %ecx, %ecx
 	jz safe_user_strncpy_fault
 
- .Lsafe_user_strncpy_loop:
+ .safe_user_strncpy_loop:
 	movb (%esi), %al
 	movb %al, (%edi)
 	testb %al, %al
-	jz .Lsafe_user_strncpy_done
+	jz .safe_user_strncpy_done
 
 	incl %edi
 	incl %esi
 	decl %ecx
-	jnz .Lsafe_user_strncpy_loop
+	jnz .safe_user_strncpy_loop
 
  safe_user_strncpy_fault:
 	xorl %eax, %eax
-	jmp .Lsafe_user_strncpy_return
+	jmp .safe_user_strncpy_return
 
- .Lsafe_user_strncpy_done:
+ .safe_user_strncpy_done:
 	movl $1, %eax
 
- .Lsafe_user_strncpy_return:
+ .safe_user_strncpy_return:
 	movl 4(%esp), %edi
 	movl 8(%esp), %esi
 	ret
+
 safe_user_strncpy_end:

kernel/arch/i686/boot.S

@@ -168,12 +168,14 @@ has_sse:
 
 check_requirements:
 	call has_cpuid
-	jz system_halt
+	jz .exit
 	call has_pae
-	jz system_halt
+	jz .exit
 	call has_sse
-	jz system_halt
+	jz .exit
 	ret
+.exit:
+	jmp system_halt
 
 enable_sse:
 	movl %cr0, %eax
@@ -223,8 +225,8 @@ _start:
 
 	# load boot GDT
 	lgdt V2P(boot_gdtr)
-	ljmpl $0x08, $V2P(.Lgdt_flush)
-.Lgdt_flush:
+	ljmpl $0x08, $V2P(gdt_flush)
+gdt_flush:
 	# set correct segment registers
 	movw $0x10, %ax
 	movw %ax, %ds
@@ -241,10 +243,10 @@ _start:
 	movl $g_boot_stack_top, %esp
 
 	# jump to higher half
-	leal .Lhigher_half, %ecx
+	leal higher_half, %ecx
 	jmp *%ecx
 
-.Lhigher_half:
+higher_half:
 	# call global constuctors
 	call _init
 
@@ -296,18 +298,18 @@ ap_ready:
 	.skip 4
 
 1:	cli; cld
-	ljmpl $0x00, $AP_REL(.Lap_cs_clear)
+	ljmpl $0x00, $AP_REL(ap_cs_clear)
 
-.Lap_cs_clear:
+ap_cs_clear:
 	# load ap gdt and enter protected mode
 	lgdt AP_REL(ap_gdtr)
 	movl %cr0, %eax
 	orb $1, %al
 	movl %eax, %cr0
-	ljmpl $0x08, $AP_REL(.Lap_protected_mode)
+	ljmpl $0x08, $AP_REL(ap_protected_mode)
 
 .code32
-.Lap_protected_mode:
+ap_protected_mode:
 	movw $0x10, %ax
 	movw %ax, %ds
 	movw %ax, %ss
@@ -321,13 +323,13 @@ ap_ready:
 
 	# load boot gdt and enter long mode
 	lgdt V2P(boot_gdtr)
-	ljmpl $0x08, $AP_REL(.Lap_flush_gdt)
+	ljmpl $0x08, $AP_REL(ap_flush_gdt)
 
-.Lap_flush_gdt:
-	movl $.Lap_higher_half, %ecx
+ap_flush_gdt:
+	movl $ap_higher_half, %ecx
 	jmp *%ecx
 
-.Lap_higher_half:
+ap_higher_half:
 	movl AP_REL(ap_prepare_paging), %eax
 	call *%eax
 

kernel/arch/x86_64/Signal.S

@@ -61,7 +61,7 @@ signal_trampoline:
 	addq $40, %rsp
 
 	// restore sigmask
-	movq $79, %rdi       // SYS_SIGPROCMASK
+	movq $78, %rdi       // SYS_SIGPROCMASK
 	movq $3,  %rsi       // SIG_SETMASK
 	leaq 192(%rsp), %rdx // set
 	xorq %r10, %r10      // oset

kernel/arch/x86_64/Syscall.S

@@ -33,13 +33,13 @@ sys_fork_trampoline:
 
 	call read_ip
 	testq %rax, %rax
-	jz .Lsys_fork_trampoline_done
+	jz .done
 
 	movq %rax, %rsi
 	movq %rsp, %rdi
 	call sys_fork
 
- .Lsys_fork_trampoline_done:
+ .done:
 	popq %r15
 	popq %r14
 	popq %r13

kernel/arch/x86_64/User.S

@@ -20,28 +20,28 @@ safe_user_strncpy:
 	testq %rcx, %rcx
 	jz safe_user_strncpy_fault
 
- .Lsafe_user_strncpy_align_loop:
+ .safe_user_strncpy_align_loop:
 	testb $0x7, %sil
-	jz .Lsafe_user_strncpy_align_done
+	jz .safe_user_strncpy_align_done
 
 	movb (%rsi), %al
 	movb %al, (%rdi)
 	testb %al, %al
-	jz .Lsafe_user_strncpy_done
+	jz .safe_user_strncpy_done
 
 	incq %rdi
 	incq %rsi
 	decq %rcx
-	jnz .Lsafe_user_strncpy_align_loop
+	jnz .safe_user_strncpy_align_loop
 	jmp safe_user_strncpy_fault
 
- .Lsafe_user_strncpy_align_done:
+ .safe_user_strncpy_align_done:
 	movq $0x0101010101010101, %r8
 	movq $0x8080808080808080, %r9
 
- .Lsafe_user_strncpy_qword_loop:
+ .safe_user_strncpy_qword_loop:
 	cmpq $8, %rcx
-	jb .Lsafe_user_strncpy_qword_done
+	jb .safe_user_strncpy_qword_done
 
 	movq (%rsi), %rax
 	movq %rax, %r10
@@ -52,36 +52,36 @@ safe_user_strncpy:
 	notq %r11
 	andq %r11, %r10
 	andq %r9, %r10
-	jnz .Lsafe_user_strncpy_byte_loop
+	jnz .safe_user_strncpy_byte_loop
 
 	movq %rax, (%rdi)
 
 	addq $8, %rdi
 	addq $8, %rsi
 	subq $8, %rcx
-	jnz .Lsafe_user_strncpy_qword_loop
+	jnz .safe_user_strncpy_qword_loop
 	jmp safe_user_strncpy_fault
 
- .Lsafe_user_strncpy_qword_done:
+ .safe_user_strncpy_qword_done:
 	testq %rcx, %rcx
 	jz safe_user_strncpy_fault
 
- .Lsafe_user_strncpy_byte_loop:
+ .safe_user_strncpy_byte_loop:
 	movb (%rsi), %al
 	movb %al, (%rdi)
 	testb %al, %al
-	jz .Lsafe_user_strncpy_done
+	jz .safe_user_strncpy_done
 
 	incq %rdi
 	incq %rsi
 	decq %rcx
-	jnz .Lsafe_user_strncpy_byte_loop
+	jnz .safe_user_strncpy_byte_loop
 
  safe_user_strncpy_fault:
 	xorq %rax, %rax
 	ret
 
- .Lsafe_user_strncpy_done:
+ .safe_user_strncpy_done:
 	movb $1, %al
 	ret
 safe_user_strncpy_end:

kernel/arch/x86_64/boot.S

@@ -161,10 +161,12 @@ is_64_bit:
 
 check_requirements:
 	call has_cpuid
-	jz system_halt
+	jz .exit
 	call is_64_bit
-	jz system_halt
+	jz .exit
 	ret
+.exit:
+	jmp system_halt
 
 enable_sse:
 	movl %cr0, %eax
@@ -224,10 +226,10 @@ _start:
 
 	# flush gdt and jump to 64 bit
 	lgdt V2P(boot_gdtr)
-	ljmpl $0x08, $V2P(.Llong_mode)
+	ljmpl $0x08, $V2P(long_mode)
 
 .code64
-.Llong_mode:
+long_mode:
 	movw $0x10, %ax
 	movw %ax, %ds
 	movw %ax, %ss
@@ -238,10 +240,10 @@ _start:
 	addq $KERNEL_OFFSET, %rsp
 
 	# jump to higher half
-	movabsq $.Lhigher_half, %rcx
+	movabsq $higher_half, %rcx
 	jmp *%rcx
 
-.Lhigher_half:
+higher_half:
 	# call global constuctors
 	call _init
 
@@ -291,18 +293,18 @@ ap_ready:
 	.skip 8
 
 1:	cli; cld
-	ljmpl $0x00, $AP_REL(.Lap_cs_clear)
+	ljmpl $0x00, $AP_REL(ap_cs_clear)
 
-.Lap_cs_clear:
+ap_cs_clear:
 	# load ap gdt and enter protected mode
 	lgdt AP_REL(ap_gdtr)
 	movl %cr0, %eax
 	orb $1, %al
 	movl %eax, %cr0
-	ljmpl $0x08, $AP_REL(.Lap_protected_mode)
+	ljmpl $0x08, $AP_REL(ap_protected_mode)
 
 .code32
-.Lap_protected_mode:
+ap_protected_mode:
 	movw $0x10, %ax
 	movw %ax, %ds
 	movw %ax, %ss
@@ -316,14 +318,14 @@ ap_ready:
 
 	# load boot gdt and enter long mode
 	lgdt V2P(boot_gdtr)
-	ljmpl $0x08, $AP_REL(.Lap_long_mode)
+	ljmpl $0x08, $AP_REL(ap_long_mode)
 
 .code64
-.Lap_long_mode:
-	movq $.Lap_higher_half, %rax
+ap_long_mode:
+	movq $ap_higher_half, %rax
 	jmp *%rax
 
-.Lap_higher_half:
+ap_higher_half:
 	movq AP_REL(ap_prepare_paging), %rax
 	call *%rax
 

kernel/icxxabi.cpp

@@ -38,39 +38,25 @@ extern "C" void __cxa_finalize(void* f)
 	}
 };
 
-extern "C" int __cxa_guard_acquire(uint64_t* g)
+namespace __cxxabiv1
 {
-	uint8_t* guard = reinterpret_cast<uint8_t*>(g);
+	using __guard = uint64_t;
 
-	for (;;)
+	extern "C" int __cxa_guard_acquire (__guard* g)
 	{
-		if (BAN::atomic_load(guard[0], BAN::memory_order_acquire))
-			return 0;
+		uint8_t* byte = reinterpret_cast<uint8_t*>(g);
+		uint8_t zero = 0;
+		return __atomic_compare_exchange_n(byte, &zero, 1, false, __ATOMIC_ACQUIRE, __ATOMIC_ACQUIRE);
+	}
 
-		uint8_t expected = 0;
-		if (BAN::atomic_compare_exchange(guard[1], expected, 1, BAN::memory_order_acquire))
-		{
-			if (BAN::atomic_load(guard[0], BAN::memory_order_acquire))
-			{
-				BAN::atomic_store(guard[1], 0, BAN::memory_order_release);
-				return 0;
-			}
-			return 1;
-		}
+	extern "C" void __cxa_guard_release (__guard* g)
+	{
+		uint8_t* byte = reinterpret_cast<uint8_t*>(g);
+		__atomic_store_n(byte, 0, __ATOMIC_RELEASE);
+	}
 
-		while (BAN::atomic_load(guard[1], BAN::memory_order_acquire))
-			Kernel::Processor::pause();
+	extern "C" void __cxa_guard_abort (__guard*)
+	{
+		Kernel::panic("__cxa_guard_abort");
 	}
 }
-
-extern "C" void __cxa_guard_release(uint64_t* g)
-{
-	uint8_t* guard = reinterpret_cast<uint8_t*>(g);
-	BAN::atomic_store(guard[0], 1, BAN::memory_order_release);
-	BAN::atomic_store(guard[1], 0, BAN::memory_order_release);
-}
-
-extern "C" void __cxa_guard_abort(uint64_t*)
-{
-	Kernel::panic("__cxa_guard_abort");
-}

kernel/include/kernel/ACPI/ACPI.h

@@ -4,7 +4,6 @@
 #include <kernel/ACPI/AML/Namespace.h>
 #include <kernel/ACPI/EmbeddedController.h>
 #include <kernel/ACPI/Headers.h>
-#include <kernel/Interruptable.h>
 #include <kernel/Memory/Types.h>
 #include <kernel/ThreadBlocker.h>
 
@@ -24,7 +23,11 @@ namespace Kernel::ACPI
 
 		const SDTHeader* get_header(BAN::StringView signature, uint32_t index);
 
-		BAN::ErrorOr<void> enter_acpi_mode();
+		// mode
+		//   0: PIC
+		//   1: APIC
+		//   2: SAPIC
+		BAN::ErrorOr<void> enter_acpi_mode(uint8_t mode);
 
 		BAN::ErrorOr<void> initialize_acpi_devices();
 

kernel/include/kernel/ACPI/EmbeddedController.h

@@ -1,12 +1,11 @@
 #pragma once
 
 #include <BAN/Atomic.h>
-#include <BAN/Optional.h>
 #include <BAN/UniqPtr.h>
 
 #include <kernel/ACPI/AML/Scope.h>
 #include <kernel/Lock/Mutex.h>
-#include <kernel/ThreadBlocker.h>
+#include <kernel/Thread.h>
 
 namespace Kernel::ACPI
 {

kernel/include/kernel/APIC.h

@@ -2,7 +2,6 @@
 
 #include <BAN/Vector.h>
 #include <kernel/InterruptController.h>
-#include <kernel/InterruptNumbers.h>
 #include <kernel/Lock/SpinLock.h>
 #include <kernel/Memory/Types.h>
 
@@ -13,10 +12,10 @@ namespace Kernel
 	{
 	public:
 		virtual void eoi(uint8_t) override;
-		virtual void enable_irq(uint8_t, bool level_triggered) override;
+		virtual void enable_irq(uint8_t) override;
 		virtual bool is_in_service(uint8_t) override;
 
-		virtual BAN::ErrorOr<void> reserve_irq(uint8_t irq, bool shared) override;
+		virtual BAN::ErrorOr<void> reserve_irq(uint8_t irq) override;
 		virtual BAN::Optional<uint8_t> get_free_irq() override;
 
 		virtual void initialize_multiprocessor() override;
@@ -24,7 +23,7 @@ namespace Kernel
 		virtual void broadcast_ipi() override;
 		virtual void enable() override;
 
-		BAN::ErrorOr<uint8_t> reserve_gsi(uint32_t gsi, bool shared = false);
+		BAN::ErrorOr<uint8_t> reserve_gsi(uint32_t gsi);
 
 		void initialize_timer();
 		void set_timer_dealine(uint64_t ns);
@@ -72,8 +71,7 @@ namespace Kernel
 		Kernel::vaddr_t        m_local_apic_vaddr = 0;
 		BAN::Vector<IOAPIC>    m_io_apics;
 		uint8_t                m_irq_overrides[0x100] {};
-		uint8_t                m_used_gsis[m_irq_count / 8] {};
-		uint8_t                m_excl_gsis[m_irq_count / 8] {};
+		uint8_t                m_reserved_gsis[m_irq_count / 8] {};
 		uint64_t               m_lapic_timer_frequency_hz { 0 };
 	};
 

kernel/include/kernel/Audio/HDAudio/AudioFunctionGroup.h

@@ -67,7 +67,6 @@ namespace Kernel
 		BAN::Vector<BAN::Vector<const HDAudio::AFGWidget*>> m_output_paths;
 		BAN::Vector<const HDAudio::AFGWidget*> m_output_pins;
 		size_t m_output_path_index { SIZE_MAX };
-		size_t m_amplifier_idx { SIZE_MAX };
 
 		uint8_t m_stream_id    { 0xFF };
 		uint8_t m_stream_index { 0xFF };

kernel/include/kernel/Audio/HDAudio/Definitions.h

@@ -1,6 +1,5 @@
 #pragma once
 
-#include <BAN/Optional.h>
 #include <BAN/Vector.h>
 
 namespace Kernel::HDAudio
@@ -65,7 +64,6 @@ namespace Kernel::HDAudio
 		};
 
 		BAN::Optional<Amplifier> output_amplifier;
-		BAN::Optional<Amplifier> input_amplifier;
 
 		BAN::Vector<uint16_t> connections;
 	};

kernel/include/kernel/BananBootloader.h

@@ -21,17 +21,13 @@ struct BananBootloaderMemoryMapEntry
 	uint64_t address;
 	uint64_t length;
 	uint32_t type;
-	uint32_t unused;
-};
-static_assert(sizeof(BananBootloaderMemoryMapEntry) == 24);
+} __attribute__((packed));
 
 struct BananBootloaderMemoryMapInfo
 {
 	uint32_t entry_count;
-	uint32_t padding;
 	struct BananBootloaderMemoryMapEntry entries[];
-};
-static_assert(sizeof(BananBootloaderMemoryMapInfo) == 8);
+} __attribute__((packed));
 
 struct BananBootloaderInfo
 {
@@ -39,5 +35,4 @@ struct BananBootloaderInfo
 	uint32_t framebuffer_addr;
 	uint32_t memory_map_addr;
 	uint32_t kernel_paddr;
-};
-static_assert(sizeof(BananBootloaderInfo) == 16);
+} __attribute__((packed));

kernel/include/kernel/Epoll.h

@@ -3,7 +3,6 @@
 #include <BAN/HashMap.h>
 #include <kernel/FS/Inode.h>
 #include <kernel/Lock/Mutex.h>
-#include <kernel/ThreadBlocker.h>
 
 #include <sys/epoll.h>
 

kernel/include/kernel/FS/DevFS/FileSystem.h

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <BAN/Vector.h>
+#include <kernel/Device/Device.h>
 #include <kernel/FS/TmpFS/FileSystem.h>
 #include <kernel/Lock/Mutex.h>
 #include <kernel/ThreadBlocker.h>
@@ -8,8 +9,6 @@
 namespace Kernel
 {
 
-	class Device;
-
 	class DevFileSystem final : public TmpFileSystem
 	{
 	public:

kernel/include/kernel/FS/EventFD.h

@@ -2,7 +2,6 @@
 
 #include <kernel/FS/Inode.h>
 #include <kernel/Lock/Mutex.h>
-#include <kernel/ThreadBlocker.h>
 
 namespace Kernel
 {
@@ -23,16 +22,16 @@ namespace Kernel
 		BAN::ErrorOr<size_t> read_impl(off_t, BAN::ByteSpan) override;
 		BAN::ErrorOr<size_t> write_impl(off_t, BAN::ConstByteSpan) override;
 
-		bool can_read_impl() const override;
-		bool can_write_impl() const override;
+		bool can_read_impl() const override { return m_value > 0; }
+		bool can_write_impl() const override { return m_value < UINT64_MAX - 1; }
 		bool has_error_impl() const override { return false; }
 		bool has_hungup_impl() const override { return false; }
 
 	private:
 		const bool m_is_semaphore;
-		uint64_t m_value;
+		BAN::Atomic<uint64_t> m_value;
 
-		mutable Mutex m_mutex;
+		Mutex m_mutex;
 		ThreadBlocker m_thread_blocker;
 	};
 

kernel/include/kernel/FS/Ext2/FileSystem.h

@@ -2,9 +2,8 @@
 
 #include <BAN/HashMap.h>
 #include <kernel/Device/Device.h>
-#include <kernel/FS/Ext2/Inode.h>
 #include <kernel/FS/FileSystem.h>
-#include <kernel/Memory/VirtualRange.h>
+#include <kernel/FS/Ext2/Inode.h>
 
 namespace Kernel
 {
@@ -29,7 +28,7 @@ namespace Kernel
 			BAN_NON_COPYABLE(BlockBufferWrapper);
 
 		public:
-			BlockBufferWrapper(BAN::ByteSpan buffer, void (*callback)(void*, const uint8_t*), void* argument)
+			BlockBufferWrapper(BAN::Span<uint8_t> buffer, void (*callback)(void*, const uint8_t*), void* argument)
 				: m_buffer(buffer)
 				, m_callback(callback)
 				, m_argument(argument)
@@ -61,13 +60,13 @@ namespace Kernel
 			const uint8_t* data() const { return m_buffer.data(); }
 
 			BAN::ByteSpan span() { return m_buffer; }
-			BAN::ConstByteSpan span() const { return m_buffer; }
+			BAN::ConstByteSpan span() const { return m_buffer.as_const(); }
 
 			uint8_t& operator[](size_t index) { return m_buffer[index]; }
 			uint8_t operator[](size_t index) const { return m_buffer[index]; }
 
 		private:
-			BAN::ByteSpan m_buffer;
+			BAN::Span<uint8_t> m_buffer;
 			void (*m_callback)(void*, const uint8_t*);
 			void* m_argument;
 		};
@@ -129,7 +128,7 @@ namespace Kernel
 		private:
 			struct BlockBuffer
 			{
-				BAN::ByteSpan buffer;
+				BAN::Vector<uint8_t> buffer;
 				bool used { false };
 			};
 
@@ -145,7 +144,6 @@ namespace Kernel
 
 			Mutex m_buffer_mutex;
 			ThreadBlocker m_buffer_blocker;
-			BAN::UniqPtr<VirtualRange> m_buffer_data;
 			BAN::Array<BlockBuffer, max_threads * max_buffers_per_thread> m_buffers;
 			BAN::Array<ThreadInfo, max_threads> m_thread_infos;
 		};

kernel/include/kernel/FS/FAT/FileSystem.h

@@ -5,7 +5,6 @@
 #include <kernel/FS/FAT/Definitions.h>
 #include <kernel/FS/FAT/Inode.h>
 #include <kernel/FS/FileSystem.h>
-#include <kernel/Lock/Mutex.h>
 
 namespace Kernel
 {

kernel/include/kernel/FS/FileSystem.h

@@ -1,12 +1,11 @@
 #pragma once
 
+#include <kernel/Device/Device.h>
 #include <kernel/FS/Inode.h>
 
 namespace Kernel
 {
 
-	class BlockDevice;
-
 	class FileSystem : public BAN::RefCounted<FileSystem>
 	{
 	public:

kernel/include/kernel/FS/Inode.h

@@ -83,7 +83,6 @@ namespace Kernel
 	public:
 		virtual ~Inode() {}
 
-		static bool can_access(uid_t uid, gid_t gid, mode_t mode, const Credentials& credentials, int);
 		bool can_access(const Credentials&, int) const;
 
 		bool operator==(const Inode& other) const { return dev() == other.dev() && ino() == other.ino(); }

kernel/include/kernel/FS/ProcFS/FileSystem.h

@@ -2,12 +2,11 @@
 
 #include <kernel/FS/TmpFS/FileSystem.h>
 #include <kernel/FS/TmpFS/Inode.h>
+#include <kernel/Process.h>
 
 namespace Kernel
 {
 
-	class Process;
-
 	class ProcFileSystem final : public TmpFileSystem
 	{
 	public:

kernel/include/kernel/FS/ProcFS/Inode.h

@@ -2,12 +2,11 @@
 
 #include <kernel/FS/TmpFS/FileSystem.h>
 #include <kernel/FS/TmpFS/Inode.h>
+#include <kernel/Process.h>
 
 namespace Kernel
 {
 
-	class Process;
-
 	class ProcPidInode final : public TmpDirectoryInode
 	{
 		// FIXME: dynamically update ruid/rgid.

kernel/include/kernel/IDT.h

@@ -10,6 +10,21 @@
 namespace Kernel
 {
 
+	// IDT entries
+	//   0x00->0x1F (32):  ISR
+	//   0x20->0x7F (96):  PIC/IOAPIC
+	//   0x80->0xEF (112): MSI
+	//   0xF0->0xFE (15):  internal
+
+	constexpr uint8_t IRQ_VECTOR_BASE   = 0x20;
+	constexpr uint8_t IRQ_MSI_BASE      = 0x80;
+	constexpr uint8_t IRQ_MSI_END       = 0xF0;
+#if ARCH(i686)
+	constexpr uint8_t IRQ_SYSCALL       = 0xF0; // hard coded in kernel/API/Syscall.h
+#endif
+	constexpr uint8_t IRQ_IPI           = 0xF1;
+	constexpr uint8_t IRQ_TIMER         = 0xF2;
+
 #if ARCH(x86_64)
 	struct GateDescriptor
 	{

kernel/include/kernel/Input/PS2/Keyboard.h

@@ -36,7 +36,6 @@ namespace Kernel::Input
 		uint8_t m_byte_index { 0 };
 		bool m_basic { false };
 
-		bool m_scancode_set_uncertain { false };
 		uint8_t m_scancode_set { 0xFF };
 		uint16_t m_modifiers { 0 };
 

kernel/include/kernel/InterruptController.h

@@ -14,7 +14,7 @@ namespace Kernel
 		virtual ~InterruptController() {}
 
 		virtual void eoi(uint8_t) = 0;
-		virtual void enable_irq(uint8_t, bool level_triggered = false) = 0;
+		virtual void enable_irq(uint8_t) = 0;
 		virtual bool is_in_service(uint8_t) = 0;
 
 		static void initialize(bool force_pic);
@@ -26,7 +26,7 @@ namespace Kernel
 		virtual void broadcast_ipi() = 0;
 		virtual void enable() = 0;
 
-		virtual BAN::ErrorOr<void> reserve_irq(uint8_t irq, bool shared = false) = 0;
+		virtual BAN::ErrorOr<void> reserve_irq(uint8_t irq) = 0;
 		virtual BAN::Optional<uint8_t> get_free_irq() = 0;
 
 		bool is_using_apic() const { return m_using_apic; }

kernel/include/kernel/InterruptNumbers.h

@@ -1,23 +0,0 @@
-#pragma once
-
-#include <kernel/Arch.h>
-
-namespace Kernel
-{
-
-	// IDT entries
-	//   0x00->0x1F (32):  ISR
-	//   0x20->0x7F (96):  PIC/IOAPIC
-	//   0x80->0xEF (112): MSI
-	//   0xF0->0xFE (15):  internal
-
-	constexpr uint8_t IRQ_VECTOR_BASE   = 0x20;
-	constexpr uint8_t IRQ_MSI_BASE      = 0x80;
-	constexpr uint8_t IRQ_MSI_END       = 0xF0;
-#if ARCH(i686)
-	constexpr uint8_t IRQ_SYSCALL       = 0xF0; // hard coded in kernel/API/Syscall.h
-#endif
-	constexpr uint8_t IRQ_IPI           = 0xF1;
-	constexpr uint8_t IRQ_TIMER         = 0xF2;
-
-}

kernel/include/kernel/Lock/BlockableSpinLock.h

@@ -1,41 +0,0 @@
-#pragma once
-
-#include <kernel/Lock/SpinLock.h>
-#include <kernel/Lock/Mutex.h>
-
-namespace Kernel
-{
-
-	// FIXME: These classes are HACKS to allow passing spinlock
-	//        to unblock functions. Write a better API that either
-	//        allows passing spinlocks or do something cleaner that
-	//        whatever shit this is
-
-	template<typename Lock> requires requires (Lock& lock) { lock.lock(); lock.unlock(InterruptState::Disabled); lock.current_processor_has_lock(); }
-	class BlockableSpinLock : public BaseMutex
-	{
-	public:
-		BlockableSpinLock(Lock& lock)
-			: m_lock(lock)
-		{
-			ASSERT(m_lock.current_processor_has_lock());
-		}
-
-		void lock() override
-		{
-			m_lock.lock();
-		}
-
-		void unlock() override
-		{
-			m_lock.unlock(InterruptState::Disabled);
-		}
-
-		uint32_t lock_depth() const override { return m_lock.lock_depth(); }
-		bool is_locked_by_current_thread() const override { return m_lock.current_processor_has_lock(); }
-
-	private:
-		Lock& m_lock;
-	};
-
-}

kernel/include/kernel/Lock/Mutex.h

@@ -2,6 +2,7 @@
 
 #include <BAN/Atomic.h>
 #include <BAN/NoCopyMove.h>
+#include <kernel/Thread.h>
 
 #include <sys/types.h>
 
@@ -12,10 +13,12 @@ namespace Kernel
 	{
 	public:
 		virtual void lock() = 0;
+		virtual bool try_lock() = 0;
 		virtual void unlock() = 0;
 
+		virtual pid_t locker() const = 0;
+		virtual bool is_locked() const = 0;
 		virtual uint32_t lock_depth() const = 0;
-		virtual bool is_locked_by_current_thread() const = 0;
 	};
 
 	class Mutex final : public BaseMutex
@@ -26,14 +29,63 @@ namespace Kernel
 	public:
 		Mutex() = default;
 
-		bool try_lock();
-		void lock() override;
-		void unlock() override;
+		void lock() override
+		{
+			const auto tid = Thread::current_tid();
+			if (tid == m_locker)
+				ASSERT(m_lock_depth > 0);
+			else
+			{
+				ASSERT(!tid || !Thread::current().has_spinlock());
+				pid_t expected = -1;
+				while (!m_locker.compare_exchange(expected, tid))
+				{
+					ASSERT(Processor::get_interrupt_state() == InterruptState::Enabled);
+					Processor::yield();
+					expected = -1;
+				}
+				ASSERT(m_lock_depth == 0);
+				if (tid)
+					Thread::current().add_mutex();
+			}
+			m_lock_depth++;
+		}
 
-		pid_t locker() const { return m_locker; }
-		bool is_locked() const { return m_locker != -1; }
+		bool try_lock() override
+		{
+			const auto tid = Thread::current_tid();
+			if (tid == m_locker)
+				ASSERT(m_lock_depth > 0);
+			else
+			{
+				pid_t expected = -1;
+				if (!m_locker.compare_exchange(expected, tid))
+					return false;
+				ASSERT(m_lock_depth == 0);
+				if (tid)
+					Thread::current().add_mutex();
+			}
+			m_lock_depth++;
+			return true;
+		}
+
+		void unlock() override
+		{
+			const auto tid = Thread::current_tid();
+			ASSERT(m_locker == tid);
+			ASSERT(m_lock_depth > 0);
+			if (--m_lock_depth == 0)
+			{
+				m_locker = -1;
+				if (tid)
+					Thread::current().remove_mutex();
+			}
+		}
+
+		pid_t locker() const override { return m_locker; }
+		bool is_locked() const override { return m_locker != -1; }
 		uint32_t lock_depth() const override { return m_lock_depth; }
-		bool is_locked_by_current_thread() const override;
+		bool is_locked_by_current_thread() const { return m_locker == Thread::current_tid(); }
 
 	private:
 		BAN::Atomic<pid_t>	m_locker		{ -1 };
@@ -48,14 +100,74 @@ namespace Kernel
 	public:
 		PriorityMutex() = default;
 
-		bool try_lock();
-		void lock() override;
-		void unlock() override;
+		void lock() override
+		{
+			const auto tid = Thread::current_tid();
 
-		pid_t locker() const { return m_locker; }
-		bool is_locked() const { return m_locker != -1; }
+			if (tid == m_locker)
+				ASSERT(m_lock_depth > 0);
+			else
+			{
+				ASSERT(!tid || !Thread::current().has_spinlock());
+				bool has_priority = tid ? !Thread::current().is_userspace() : true;
+				if (has_priority)
+					m_queue_length++;
+				pid_t expected = -1;
+				while (!(has_priority || m_queue_length == 0) || !m_locker.compare_exchange(expected, tid))
+				{
+					ASSERT(Processor::get_interrupt_state() == InterruptState::Enabled);
+					Processor::yield();
+					expected = -1;
+				}
+				ASSERT(m_lock_depth == 0);
+				if (tid)
+					Thread::current().add_mutex();
+			}
+			m_lock_depth++;
+		}
+
+		bool try_lock() override
+		{
+			const auto tid = Thread::current_tid();
+
+			if (tid == m_locker)
+				ASSERT(m_lock_depth > 0);
+			else
+			{
+				bool has_priority = tid ? !Thread::current().is_userspace() : true;
+				pid_t expected = -1;
+				if (!(has_priority || m_queue_length == 0) || !m_locker.compare_exchange(expected, tid))
+					return false;
+				if (has_priority)
+					m_queue_length++;
+				ASSERT(m_lock_depth == 0);
+				if (tid)
+					Thread::current().add_mutex();
+			}
+			m_lock_depth++;
+			return true;
+		}
+
+		void unlock() override
+		{
+			const auto tid = Thread::current_tid();
+			ASSERT(m_locker == tid);
+			ASSERT(m_lock_depth > 0);
+			if (--m_lock_depth == 0)
+			{
+				bool has_priority = tid ? !Thread::current().is_userspace() : true;
+				if (has_priority)
+					m_queue_length--;
+				m_locker = -1;
+				if (tid)
+					Thread::current().remove_mutex();
+			}
+		}
+
+		pid_t locker() const override { return m_locker; }
+		bool is_locked() const override { return m_locker != -1; }
 		uint32_t lock_depth() const override { return m_lock_depth; }
-		bool is_locked_by_current_thread() const override;
+		bool is_locked_by_current_thread() const { return m_locker == Thread::current_tid(); }
 
 	private:
 		BAN::Atomic<pid_t>		m_locker		{ -1 };

kernel/include/kernel/Lock/RWLock.h

@@ -1,7 +1,7 @@
 #pragma once
 
 #include <kernel/Lock/SpinLock.h>
-#include <kernel/ThreadBlocker.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 
 namespace Kernel
 {
@@ -13,10 +13,54 @@ namespace Kernel
 	public:
 		RWLock() = default;
 
-		void rd_lock();
-		void rd_unlock();
-		void wr_lock();
-		void wr_unlock();
+		void rd_lock()
+		{
+			SpinLockGuard _(m_lock);
+			while (m_writers_waiting > 0 || m_writer != -1)
+			{
+				SpinLockGuardAsMutex smutex(_);
+				m_thread_blocker.block_indefinite(&smutex);
+			}
+			m_readers_active++;
+		}
+
+		void rd_unlock()
+		{
+			SpinLockGuard _(m_lock);
+			if (--m_readers_active == 0)
+				m_thread_blocker.unblock();
+		}
+
+		void wr_lock()
+		{
+			if (m_writer == Thread::current_tid())
+			{
+				m_writer_depth++;
+				return;
+			}
+
+			SpinLockGuard _(m_lock);
+
+			m_writers_waiting++;
+			while (m_readers_active > 0 || m_writer != -1)
+			{
+				SpinLockGuardAsMutex smutex(_);
+				m_thread_blocker.block_indefinite(&smutex);
+			}
+			m_writers_waiting--;
+
+			m_writer = Thread::current_tid();
+			m_writer_depth = 1;
+		}
+
+		void wr_unlock()
+		{
+			if (--m_writer_depth != 0)
+				return;
+			SpinLockGuard _(m_lock);
+			m_writer = -1;
+			m_thread_blocker.unblock();
+		}
 
 	private:
 		SpinLock m_lock;

kernel/include/kernel/Lock/SpinLockAsMutex.h

@@ -0,0 +1,64 @@
+#pragma once
+
+#include <kernel/Lock/SpinLock.h>
+#include <kernel/Lock/Mutex.h>
+
+namespace Kernel
+{
+
+	// FIXME: These classes are HACKS to allow passing spinlock
+	//        to unblock functions. Write a better API that either
+	//        allows passing spinlocks or do something cleaner that
+	//        whatever shit this is
+
+	template<typename Lock>
+	class SpinLockAsMutex : public BaseMutex
+	{
+	public:
+		SpinLockAsMutex(Lock& lock, InterruptState state)
+			: m_lock(lock)
+			, m_lock_depth(lock.lock_depth())
+			, m_state(state)
+			, m_locker(Thread::current_tid())
+		{
+			ASSERT(m_lock.current_processor_has_lock());
+		}
+
+		void lock() override
+		{
+			m_lock.lock();
+			m_lock_depth++;
+		}
+
+		bool try_lock() override
+		{
+			lock();
+			return true;
+		}
+
+		void unlock() override
+		{
+			m_lock.unlock(--m_lock_depth ? InterruptState::Disabled : m_state);
+		}
+
+		pid_t locker() const override { return is_locked() ? m_locker : -1; }
+		bool is_locked() const override { return m_lock_depth; }
+		uint32_t lock_depth() const override { return m_lock_depth; }
+
+	private:
+		Lock& m_lock;
+		uint32_t m_lock_depth { 0 };
+		InterruptState m_state;
+		const pid_t m_locker;
+	};
+
+	template<typename Lock>
+	class SpinLockGuardAsMutex : public SpinLockAsMutex<Lock>
+	{
+	public:
+		SpinLockGuardAsMutex(SpinLockGuard<Lock>& guard)
+			: SpinLockAsMutex<Lock>(guard.m_lock, guard.m_state)
+		{}
+	};
+
+}

kernel/include/kernel/Memory/MemoryBackedRegion.h

@@ -1,6 +1,5 @@
 #pragma once
 
-#include <BAN/Vector.h>
 #include <kernel/Memory/MemoryRegion.h>
 
 namespace Kernel

kernel/include/kernel/Memory/SharedMemoryObject.h

@@ -1,27 +1,30 @@
 #pragma once
 
 #include <BAN/HashMap.h>
-#include <BAN/RefPtr.h>
 #include <BAN/UniqPtr.h>
-#include <BAN/Vector.h>
 #include <kernel/Lock/Mutex.h>
 #include <kernel/Lock/SpinLock.h>
 #include <kernel/Memory/MemoryRegion.h>
 
-#include <sys/shm.h>
+#include <fcntl.h>
 
 namespace Kernel
 {
 
+	class SharedMemoryObject;
+
 	class SharedMemoryObjectManager
 	{
+	public:
+		using Key = size_t;
+
 	public:
 		static BAN::ErrorOr<void> initialize();
 		static SharedMemoryObjectManager& get();
 
-		BAN::ErrorOr<int> shmget(key_t key, size_t size, int shmflg);
-		BAN::ErrorOr<void> shmctl(int shmid, int cmd, struct shmid_ds* user_buf);
-		BAN::ErrorOr<BAN::UniqPtr<MemoryRegion>> shmat(int shmid, const void* shmaddr, int shmflg);
+		BAN::ErrorOr<Key> create_object(size_t size, PageTable::flags_t);
+		BAN::ErrorOr<void> delete_object(Key);
+		BAN::ErrorOr<BAN::UniqPtr<SharedMemoryObject>> map_object(Key, PageTable&, AddressRange);
 
 	private:
 		SharedMemoryObjectManager() {}
@@ -29,31 +32,49 @@ namespace Kernel
 	private:
 		struct Object : public BAN::RefCounted<Object>
 		{
-			Object(key_t key, int id, shmid_ds info)
-				: key(key)
-				, id(id)
-				, info(info)
-			{ }
 			~Object();
 
-			bool can_current_process_access(int flags) const;
-
-			const key_t key;
-			const int id;
-			shmid_ds info;
-
-			Mutex mutex;
+			Key key;
+			size_t size;
+			PageTable::flags_t flags;
 			BAN::Vector<paddr_t> paddrs;
-			bool marked_for_deletion { false };
+			SpinLock spin_lock;
 		};
 
 	private:
 		Mutex m_mutex;
-		BAN::HashMap<key_t, int> m_ids;
-		BAN::HashMap<int, BAN::RefPtr<Object>> m_objects;
+		BAN::HashMap<Key, BAN::RefPtr<Object>> m_objects;
 
 		friend class SharedMemoryObject;
 		friend class BAN::UniqPtr<SharedMemoryObjectManager>;
 	};
 
+	class SharedMemoryObject : public MemoryRegion
+	{
+		BAN_NON_COPYABLE(SharedMemoryObject);
+		BAN_NON_MOVABLE(SharedMemoryObject);
+
+	public:
+		static BAN::ErrorOr<BAN::UniqPtr<SharedMemoryObject>> create(BAN::RefPtr<SharedMemoryObjectManager::Object>, PageTable&, AddressRange);
+
+		BAN::ErrorOr<BAN::UniqPtr<MemoryRegion>> clone(PageTable& new_page_table) override;
+		BAN::ErrorOr<BAN::UniqPtr<MemoryRegion>> split(size_t offset) override;
+
+		BAN::ErrorOr<void> msync(vaddr_t, size_t, int) override { return {}; }
+
+	protected:
+		BAN::ErrorOr<bool> allocate_page_containing_impl(vaddr_t vaddr, bool wants_write) override;
+
+	private:
+		SharedMemoryObject(BAN::RefPtr<SharedMemoryObjectManager::Object> object, PageTable& page_table)
+			: MemoryRegion(page_table, object->size, MemoryRegion::Type::SHARED, object->flags, O_EXEC | O_RDWR)
+			, m_object(object)
+		{ }
+
+	private:
+		BAN::RefPtr<SharedMemoryObjectManager::Object> m_object;
+
+		friend class BAN::UniqPtr<SharedMemoryObject>;
+	};
+
 }

kernel/include/kernel/Networking/ARPTable.h

@@ -4,6 +4,8 @@
 #include <BAN/HashMap.h>
 #include <BAN/UniqPtr.h>
 #include <kernel/Networking/NetworkInterface.h>
+#include <kernel/Thread.h>
+#include <kernel/ThreadBlocker.h>
 
 namespace Kernel
 {

kernel/include/kernel/Networking/IPv4Layer.h

@@ -11,6 +11,7 @@
 #include <kernel/Networking/NetworkInterface.h>
 #include <kernel/Networking/NetworkLayer.h>
 #include <kernel/Networking/NetworkSocket.h>
+#include <kernel/Thread.h>
 
 namespace Kernel
 {

kernel/include/kernel/Networking/Loopback.h

@@ -1,7 +1,6 @@
 #pragma once
 
 #include <kernel/Networking/NetworkInterface.h>
-#include <kernel/Memory/VirtualRange.h>
 
 namespace Kernel
 {

kernel/include/kernel/Networking/TCPSocket.h

@@ -7,6 +7,7 @@
 #include <kernel/Memory/ByteRingBuffer.h>
 #include <kernel/Networking/NetworkInterface.h>
 #include <kernel/Networking/NetworkSocket.h>
+#include <kernel/Thread.h>
 #include <kernel/ThreadBlocker.h>
 
 namespace Kernel

kernel/include/kernel/Networking/UNIX/Socket.h

@@ -6,7 +6,6 @@
 #include <kernel/FS/Socket.h>
 #include <kernel/FS/TmpFS/Inode.h>
 #include <kernel/FS/VirtualFileSystem.h>
-#include <kernel/Memory/VirtualRange.h>
 #include <kernel/OpenFileDescriptorSet.h>
 
 namespace Kernel

kernel/include/kernel/OpenFileDescriptorSet.h

@@ -34,7 +34,6 @@ namespace Kernel
 		BAN::ErrorOr<int> dup2(int, int);
 
 		BAN::ErrorOr<int> fcntl(int fd, int cmd, uintptr_t extra);
-		BAN::ErrorOr<long> ioctl(int fd, unsigned long request, void* arg);
 
 		BAN::ErrorOr<off_t> seek(int fd, off_t offset, int whence);
 		BAN::ErrorOr<off_t> tell(int) const;

kernel/include/kernel/PCI.h

@@ -3,8 +3,6 @@
 #include <BAN/UniqPtr.h>
 #include <BAN/Vector.h>
 #include <kernel/ACPI/AML/Node.h>
-#include <kernel/InterruptNumbers.h>
-#include <kernel/Interruptable.h>
 #include <kernel/Memory/Types.h>
 
 #include <sys/types.h>

kernel/include/kernel/PIC.h

@@ -10,10 +10,10 @@ namespace Kernel
 	{
 	public:
 		virtual void eoi(uint8_t) override;
-		virtual void enable_irq(uint8_t, bool level_triggered) override;
+		virtual void enable_irq(uint8_t) override;
 		virtual bool is_in_service(uint8_t) override;
 
-		virtual BAN::ErrorOr<void> reserve_irq(uint8_t irq, bool shared) override;
+		virtual BAN::ErrorOr<void> reserve_irq(uint8_t irq) override;
 		virtual BAN::Optional<uint8_t> get_free_irq() override;
 
 		virtual void initialize_multiprocessor() override;
@@ -29,8 +29,7 @@ namespace Kernel
 
 	private:
 		SpinLock m_lock;
-		uint16_t m_used_irqs { 1u << 2 };
-		uint16_t m_excl_irqs { 1u << 2 };
+		uint16_t m_reserved_irqs { 1u << 2 };
 
 		friend class InterruptController;
 	};

kernel/include/kernel/Process.h

@@ -19,7 +19,6 @@
 
 #include <poll.h>
 #include <sys/banan-os.h>
-#include <sys/epoll.h>
 #include <sys/mman.h>
 #include <sys/select.h>
 #include <sys/socket.h>
@@ -27,6 +26,8 @@
 #include <sys/time.h>
 #include <termios.h>
 
+struct epoll_event;
+
 namespace Kernel
 {
 
@@ -178,10 +179,9 @@ namespace Kernel
 		BAN::ErrorOr<long> sys_mprotect(void* addr, size_t len, int prot);
 		BAN::ErrorOr<long> sys_msync(void* addr, size_t len, int flags);
 
-		BAN::ErrorOr<long> sys_shmget(key_t key, size_t size, int shmflg);
-		BAN::ErrorOr<long> sys_shmctl(int shmid, int cmd, struct shmid_ds* buf);
-		BAN::ErrorOr<long> sys_shmat(int shmid, const void* shmaddr, int shmflg);
-		BAN::ErrorOr<long> sys_shmdt(const void* shmaddr);
+		BAN::ErrorOr<long> sys_smo_create(size_t len, int prot);
+		BAN::ErrorOr<long> sys_smo_delete(SharedMemoryObjectManager::Key);
+		BAN::ErrorOr<long> sys_smo_map(SharedMemoryObjectManager::Key);
 
 		BAN::ErrorOr<long> sys_ttyname(int fildes, char* name, size_t namesize);
 		BAN::ErrorOr<long> sys_posix_openpt(int flags);
@@ -255,8 +255,6 @@ namespace Kernel
 		const VirtualFileSystem::File& working_directory() const { return m_working_directory; }
 		const VirtualFileSystem::File& root_file() const { return m_root_file; }
 
-		BAN::ErrorOr<AddressRange> find_free_address_range(size_t size);
-
 	private:
 		Process(const Credentials&, pid_t pid, pid_t parent, pid_t sid, pid_t pgrp);
 		static Process* create_process(const Credentials&, pid_t parent, pid_t sid = 0, pid_t pgrp = 0);
@@ -282,6 +280,8 @@ namespace Kernel
 		// You must hold reader end of m_mapped_region_lock when calling this.
 		size_t find_mapped_region(vaddr_t) const;
 
+		BAN::ErrorOr<AddressRange> find_free_address_range(size_t size);
+
 		BAN::ErrorOr<VirtualFileSystem::File> find_file(int fd, const char* path, int flags) const;
 		BAN::ErrorOr<FileParent> find_parent_file(int fd, const char* path, int flags) const;
 		BAN::ErrorOr<VirtualFileSystem::File> find_relative_parent(int fd, const char* path) const;

kernel/include/kernel/Processor.h

@@ -1,14 +1,17 @@
 #pragma once
 
-#include <BAN/Array.h>
 #include <BAN/Atomic.h>
-#include <BAN/NoCopyMove.h>
+#include <BAN/ForwardList.h>
 #include <BAN/Math.h>
 
 #include <kernel/API/SharedPage.h>
 #include <kernel/Arch.h>
+#include <kernel/GDT.h>
+#include <kernel/IDT.h>
+#include <kernel/InterruptStack.h>
 #include <kernel/Memory/Types.h>
 #include <kernel/ProcessorID.h>
+#include <kernel/Scheduler.h>
 
 namespace Kernel
 {
@@ -19,12 +22,6 @@ namespace Kernel
 		Enabled,
 	};
 
-	class GDT;
-	class IDT;
-	class Scheduler;
-	class SchedulerThreadNode;
-	class Thread;
-
 #if ARCH(x86_64) || ARCH(i686)
 	class Processor
 	{
@@ -54,8 +51,8 @@ namespace Kernel
 			union
 			{
 				TLBEntry flush_tlb;
-				SchedulerThreadNode* new_thread;
-				SchedulerThreadNode* unblock_thread;
+				SchedulerQueue::Node* new_thread;
+				SchedulerQueue::Node* unblock_thread;
 				bool dummy;
 			};
 		};
@@ -107,17 +104,10 @@ namespace Kernel
 		static void pause()
 		{
 			__builtin_ia32_pause();
-			if (is_smp_enabled() && !smp_messages_disabled())
+			if (is_smp_enabled())
 				handle_smp_messages();
 		}
 
-		static bool smp_messages_disabled() { return read_gs_sized<bool>(offsetof(Processor, m_smp_messages_disabled)); }
-		static void set_disable_smp_messages(bool disabled)
-		{
-			ASSERT(smp_messages_disabled() != disabled);
-			write_gs_sized<bool>(offsetof(Processor, m_smp_messages_disabled), disabled);
-		}
-
 		vaddr_t stack_top_vaddr() const { return m_stack_vaddr + s_stack_size; }
 		paddr_t stack_top_paddr() const { return m_stack_paddr + s_stack_size; }
 
@@ -182,23 +172,17 @@ namespace Kernel
 		}
 
 		template<typename T>
-		static T read_gs_sized(uintptr_t offset) requires(sizeof(T) <= sizeof(uintptr_t) && BAN::Math::is_power_of_two(sizeof(T)))
+		static T read_gs_sized(uintptr_t offset) requires(sizeof(T) <= 8 && BAN::Math::is_power_of_two(sizeof(T)))
 		{
 			T value;
-			if constexpr (sizeof(T) == 1 && ARCH(i686))
-				asm volatile("mov %%gs:%a[offset], %[value]" : [value]"=q"(value) : [offset]"ir"(offset) : "memory");
-			else
-				asm volatile("mov %%gs:%a[offset], %[value]" : [value]"=r"(value) : [offset]"ir"(offset) : "memory");
+			asm volatile("mov %%gs:%a[offset], %[value]" : [value]"=r"(value) : [offset]"ir"(offset));
 			return value;
 		}
 
 		template<typename T>
-		static void write_gs_sized(uintptr_t offset, T value) requires(sizeof(T) <= sizeof(uintptr_t) && BAN::Math::is_power_of_two(sizeof(T)))
+		static void write_gs_sized(uintptr_t offset, T value) requires(sizeof(T) <= 8 && BAN::Math::is_power_of_two(sizeof(T)))
 		{
-			if constexpr (sizeof(T) == 1 && ARCH(i686))
-				asm volatile("mov %[value], %%gs:%a[offset]" :: [value]"q"(value), [offset]"ir"(offset) : "memory");
-			else
-				asm volatile("mov %[value], %%gs:%a[offset]" :: [value]"r"(value), [offset]"ir"(offset) : "memory");
+			asm volatile("mov %[value], %%gs:%a[offset]" :: [value]"r"(value), [offset]"ir"(offset) : "memory");
 		}
 
 		void lock_tlb_lock();
@@ -214,8 +198,6 @@ namespace Kernel
 		ProcessorID m_id { 0 };
 		uint8_t m_index { 0 };
 
-		bool m_smp_messages_disabled { false };
-
 		vaddr_t m_thread_syscall_stack;
 
 		Thread* m_sse_thread { nullptr };

kernel/include/kernel/Scheduler.h

@@ -5,7 +5,6 @@
 #include <BAN/NoCopyMove.h>
 #include <kernel/InterruptStack.h>
 #include <kernel/ProcessorID.h>
-#include <kernel/SchedulerThreadNode.h>
 
 #include <sys/types.h>
 
@@ -15,6 +14,29 @@ namespace Kernel
 	class BaseMutex;
 	class Thread;
 	class ThreadBlocker;
+	struct SchedulerQueueNode;
+
+	class SchedulerQueue
+	{
+	public:
+		using Node = SchedulerQueueNode;
+
+	public:
+		void add_thread_to_back(Node*);
+		bool add_thread_with_wake_time(Node*); // return true if node was inserted as the first element
+		template<typename F>
+		Node* remove_with_condition(F callback);
+		void remove_node(Node*);
+		Node* front();
+		Node* pop_front();
+
+		bool empty() const { return m_head == nullptr; }
+
+	private:
+		Node* m_head { nullptr };
+		Node* m_tail { nullptr };
+	};
+
 	class Scheduler
 	{
 		BAN_NON_COPYABLE(Scheduler);
@@ -23,12 +45,12 @@ namespace Kernel
 	public:
 		struct NewThreadRequest
 		{
-			SchedulerThreadNode* node;
+			SchedulerQueue::Node* node;
 		};
 
 		struct UnblockRequest
 		{
-			SchedulerThreadNode* node;
+			SchedulerQueue::Node* node;
 		};
 
 	public:
@@ -36,7 +58,7 @@ namespace Kernel
 		BAN::ErrorOr<void> initialize();
 
 		void reschedule(YieldRegisters*);
-		void reschedule_if_needed();
+		void reschedule_if_idle();
 
 		void on_timer_interrupt();
 		void on_yield(YieldRegisters*);
@@ -57,9 +79,9 @@ namespace Kernel
 	private:
 		Scheduler() = default;
 
-		void add_current_to_most_loaded(void* target_list);
-		void update_most_loaded_node_list(SchedulerThreadNode*, void* target_list);
-		void remove_node_from_most_loaded(SchedulerThreadNode*);
+		void add_current_to_most_loaded(SchedulerQueue* target_queue);
+		void update_most_loaded_node_queue(SchedulerQueue::Node*, SchedulerQueue* target_queue);
+		void remove_node_from_most_loaded(SchedulerQueue::Node*);
 
 		void update_wake_up_deadline();
 		void wake_up_sleeping_threads();
@@ -68,13 +90,13 @@ namespace Kernel
 
 		class ProcessorID find_least_loaded_processor() const;
 
-		void add_thread(SchedulerThreadNode*);
-		void unblock_thread(SchedulerThreadNode*);
+		void add_thread(SchedulerQueue::Node*);
+		void unblock_thread(SchedulerQueue::Node*);
 
 	private:
-		SchedulerQueue m_run_list;
-		SchedulerHeap m_block_list;
-		SchedulerThreadNode* m_current { nullptr };
+		SchedulerQueue m_run_queue;
+		SchedulerQueue m_block_queue;
+		SchedulerQueue::Node* m_current { nullptr };
 
 		uint32_t m_thread_count { 0 };
 
@@ -86,8 +108,8 @@ namespace Kernel
 
 		struct ThreadInfo
 		{
-			void*                list { nullptr };
-			SchedulerThreadNode* node { nullptr };
+			SchedulerQueue*       queue { nullptr };
+			SchedulerQueue::Node* node  { nullptr };
 		};
 		BAN::Array<ThreadInfo, 10> m_most_loaded_threads;
 
@@ -96,8 +118,6 @@ namespace Kernel
 
 		bool m_should_calculate_max_load_threads { true };
 
-		bool m_has_pending_reschedule { false };
-
 		Thread* m_idle_thread { nullptr };
 
 		friend class ThreadBlocker;

kernel/include/kernel/SchedulerQueueNode.h

@@ -0,0 +1,36 @@
+#pragma once
+
+#include <kernel/ProcessorID.h>
+#include <kernel/Lock/SpinLock.h>
+
+namespace Kernel
+{
+
+	class Thread;
+	class ThreadBlocker;
+
+	struct SchedulerQueueNode
+	{
+		SchedulerQueueNode(Thread* thread)
+			: thread(thread)
+		{}
+
+		Thread* const thread;
+
+		SchedulerQueueNode* next { nullptr };
+		SchedulerQueueNode* prev { nullptr };
+
+		uint64_t wake_time_ns { static_cast<uint64_t>(-1) };
+
+		BAN::Atomic<ThreadBlocker*> blocker { nullptr };
+		SchedulerQueueNode* block_chain_prev { nullptr };
+		SchedulerQueueNode* block_chain_next { nullptr };
+
+		ProcessorID processor_id { PROCESSOR_NONE };
+		bool blocked { false };
+
+		uint64_t last_start_ns { 0 };
+		uint64_t time_used_ns  { 0 };
+	};
+
+}

kernel/include/kernel/SchedulerThreadNode.h

@@ -1,96 +0,0 @@
-#pragma once
-
-#include <BAN/Atomic.h>
-#include <BAN/NoCopyMove.h>
-#include <kernel/ProcessorID.h>
-
-namespace Kernel
-{
-
-	class Thread;
-	class ThreadBlocker;
-
-	struct SchedulerThreadNode
-	{
-		SchedulerThreadNode(Thread* thread)
-			: thread(thread)
-			, heap({ nullptr, nullptr, nullptr })
-		{}
-
-		Thread* const thread;
-
-		union
-		{
-			struct
-			{
-				SchedulerThreadNode* next;
-				SchedulerThreadNode* prev;
-			} queue;
-			struct
-			{
-				SchedulerThreadNode* parent;
-				SchedulerThreadNode* lchild;
-				SchedulerThreadNode* rchild;
-			} heap;
-		};
-
-		uint64_t wake_time_ns { static_cast<uint64_t>(-1) };
-
-		BAN::Atomic<ThreadBlocker*> blocker { nullptr };
-		SchedulerThreadNode* block_chain_prev { nullptr };
-		SchedulerThreadNode* block_chain_next { nullptr };
-
-		ProcessorID processor_id { PROCESSOR_NONE };
-		bool blocked { false };
-
-		uint64_t last_start_ns { 0 };
-		uint64_t time_used_ns  { 0 };
-	};
-
-	class SchedulerQueue
-	{
-		BAN_NON_COPYABLE(SchedulerQueue);
-		BAN_NON_MOVABLE(SchedulerQueue);
-	public:
-		SchedulerQueue() = default;
-
-		SchedulerThreadNode* front();
-		SchedulerThreadNode* pop_front();
-
-		void push(SchedulerThreadNode*);
-		void pop(SchedulerThreadNode*);
-
-		void walk(void (*)(const SchedulerThreadNode*, void*), void*) const;
-		bool empty() const { return m_head == nullptr; }
-
-	private:
-		SchedulerThreadNode* m_head { nullptr };
-		SchedulerThreadNode* m_tail { nullptr };
-	};
-
-	class SchedulerHeap
-	{
-		BAN_NON_COPYABLE(SchedulerHeap);
-		BAN_NON_MOVABLE(SchedulerHeap);
-	public:
-		SchedulerHeap() = default;
-
-		SchedulerThreadNode* front();
-		SchedulerThreadNode* pop_front();
-
-		void push(SchedulerThreadNode*);
-		void pop(SchedulerThreadNode*);
-
-		void walk(void (*)(const SchedulerThreadNode*, void*), void*) const;
-		bool empty() const { return m_root == nullptr; }
-
-	private:
-		void walk_impl(void (*)(const SchedulerThreadNode*, void*), void*, const SchedulerThreadNode*) const;
-		void swap_nodes(SchedulerThreadNode*, SchedulerThreadNode*);
-
-	private:
-		SchedulerThreadNode* m_root { nullptr };
-		SchedulerThreadNode* m_last { nullptr };
-	};
-
-}

kernel/include/kernel/Storage/ATA/AHCI/Controller.h

@@ -22,7 +22,6 @@ namespace Kernel
 
 		virtual void handle_irq() override;
 
-		bool supports_64bit() const { return m_supports_64bit; }
 		uint32_t command_slot_count() const { return m_command_slot_count; }
 
 	private:
@@ -39,8 +38,9 @@ namespace Kernel
 
 		BAN::Array<AHCIDevice*, 32> m_devices;
 
-		bool m_supports_64bit { false };
 		uint32_t m_command_slot_count { 0 };
+
+		friend class ATAController;
 	};
 
 }

kernel/include/kernel/Storage/ATA/AHCI/Definitions.h

@@ -32,6 +32,8 @@
 namespace Kernel
 {
 
+	static constexpr uint32_t s_hba_prdt_count { 8 };
+
 	struct FISRegisterH2D
 	{
 		uint8_t  fis_type;			// FIS_TYPE_REGISTER_H2D
@@ -59,8 +61,7 @@ namespace Kernel
 		uint8_t  control;
 
 		uint8_t  __reserved1[4];
-	};
-	static_assert(sizeof(FISRegisterH2D) == 20);
+	} __attribute__((packed));
 
 	struct FISRegisterD2H
 	{
@@ -89,8 +90,7 @@ namespace Kernel
 		uint8_t  __reserved3[2];
 
 		uint8_t  __reserved4[4];
-	};
-	static_assert(sizeof(FISRegisterD2H) == 20);
+	} __attribute__((packed));
 
 	struct FISDataBI
 	{
@@ -102,8 +102,7 @@ namespace Kernel
 		uint8_t  __reserved1[2];
 
 		uint32_t data[0];			// Payload (1 - 2048 dwords)
-	};
-	static_assert(sizeof(FISDataBI) == 4);
+	} __attribute__((packed));
 
 	struct SetDeviceBitsD2H
 	{
@@ -118,8 +117,7 @@ namespace Kernel
 		uint8_t error;
 
 		uint32_t __reserved1;
-	};
-	static_assert(sizeof(SetDeviceBitsD2H) == 8);
+	} __attribute__((packed));
 
 	struct PIOSetupD2H
 	{
@@ -151,8 +149,7 @@ namespace Kernel
 
 		uint16_t tc;				// Transfer count
 		uint8_t  __reserved4[2];
-	};
-	static_assert(sizeof(PIOSetupD2H) == 20);
+	} __attribute__((packed));
 
 	struct DMASetupBI
 	{
@@ -166,8 +163,8 @@ namespace Kernel
 
 		uint8_t  __reserved1[2];
 
-		uint32_t dma_buffer_id_lo;		// DMA Buffer Identifier. Used to Identify DMA buffer in host memory.
-		uint32_t dma_buffer_id_hi;		// SATA Spec says host specific and not in Spec. Trying AHCI spec might work.
+		uint64_t dma_buffer_id;			// DMA Buffer Identifier. Used to Identify DMA buffer in host memory.
+										// SATA Spec says host specific and not in Spec. Trying AHCI spec might work.
 
 		uint32_t __reserved2;
 
@@ -176,8 +173,7 @@ namespace Kernel
 		uint32_t dma_transfer_count;	// Number of bytes to transfer. Bit 0 must be 0
 
 		uint32_t __reserved3;
-	};
-	static_assert(sizeof(DMASetupBI) == 28);
+	} __attribute__((packed));
 
 	struct HBAPortMemorySpace
 	{
@@ -200,8 +196,7 @@ namespace Kernel
 		uint32_t fbs;		// FIS-based switch control
 		uint32_t __reserved1[11];
 		uint32_t vendor[4];
-	};
-	static_assert(sizeof(HBAPortMemorySpace) == 128);
+	} __attribute__((packed));
 
 	struct HBAGeneralMemorySpace
 	{
@@ -222,8 +217,7 @@ namespace Kernel
 		uint8_t  vendor[0x100-0xA0];
 
 		HBAPortMemorySpace ports[0]; // 1 - 32 ports
-	};
-	static_assert(sizeof(HBAGeneralMemorySpace) == 256);
+	} __attribute__((packed));
 
 	struct ReceivedFIS
 	{
@@ -241,8 +235,7 @@ namespace Kernel
 		uint8_t				ufis[64];
 
 		uint8_t				__reserved[0x100-0xA0];
-	};
-	static_assert(sizeof(ReceivedFIS) == 256);
+	} __attribute__((packed));
 
 	struct HBACommandHeader
 	{
@@ -259,14 +252,13 @@ namespace Kernel
 
 		uint16_t prdtl;		// Physical region descriptor table length in entries
 
-		uint32_t prdbc;		// Physical region descriptor byte count transferred
+		volatile uint32_t prdbc;	// Physical region descriptor byte count transferred
 
 		uint32_t ctba;		// Command table descriptor base address
 		uint32_t ctbau;		// Command table descriptor base address upper 32 bits
 
 		uint32_t __reserved1[4];
-	};
-	static_assert(sizeof(HBACommandHeader) == 32);
+	} __attribute__((packed));
 
 	struct HBAPRDTEntry
 	{
@@ -277,17 +269,15 @@ namespace Kernel
 		uint32_t dbc : 22;		// Byte count, 4M max
 		uint32_t __reserved1 : 9;
 		uint32_t i : 1;			// Interrupt on completion
-	};
-	static_assert(sizeof(HBAPRDTEntry) == 16);
+	} __attribute__((packed));
 
 	struct HBACommandTable
 	{
 		uint8_t cfis[64];
 		uint8_t acmd[16];
 		uint8_t __reserved[48];
-		HBAPRDTEntry prdt_entry[0];
-	};
-	static_assert(sizeof(HBACommandTable) == 128);
+		HBAPRDTEntry prdt_entry[s_hba_prdt_count];
+	} __attribute__((packed));
 
 	enum class AHCIPortType
 	{

kernel/include/kernel/Storage/ATA/AHCI/Device.h

@@ -20,39 +20,30 @@ namespace Kernel
 			, m_port(port)
 		{ }
 		BAN::ErrorOr<void> initialize();
+		BAN::ErrorOr<void> allocate_buffers();
 		BAN::ErrorOr<void> rebase();
 		BAN::ErrorOr<void> read_identify_data();
 
-		paddr_t read_paddr(volatile uint32_t& lo, volatile uint32_t& hi) const;
-		void write_paddr(volatile uint32_t& lo, volatile uint32_t& hi, paddr_t paddr);
-
-		bool can_use_buffer_directly(BAN::ConstByteSpan buffer) const;
-
 		virtual BAN::ErrorOr<void> read_sectors_impl(uint64_t lba, uint64_t sector_count, BAN::ByteSpan) override;
 		virtual BAN::ErrorOr<void> write_sectors_impl(uint64_t lba, uint64_t sector_count, BAN::ConstByteSpan) override;
-		BAN::ErrorOr<size_t> send_command_sync(uint64_t lba, BAN::ConstByteSpan buffer, Command command);
+		BAN::ErrorOr<void> send_command_and_block(uint64_t lba, uint64_t sector_count, Command command);
 
-		BAN::ErrorOr<void> send_command_and_wait(uint32_t slot);
-		BAN::ErrorOr<uint32_t> find_free_command_slot();
+		BAN::Optional<uint32_t> find_free_command_slot();
 
 		void handle_irq();
 
-	private:
-		static constexpr uint32_t m_max_hba_prdt_count { 64 };
+		BAN::ErrorOr<void> block_until_command_completed(uint32_t command_slot);
 
-		BAN::Atomic<uint32_t> m_prev_is { 0 };
-		uint32_t m_free_slots { 0 };
+	private:
+		Mutex m_mutex;
 
 		BAN::RefPtr<AHCIController> m_controller;
 		volatile HBAPortMemorySpace* const m_port;
 
 		BAN::UniqPtr<DMARegion> m_dma_region;
-
-		Mutex m_temp_buffer_mutex;
-		BAN::UniqPtr<DMARegion> m_temp_buffer;
-
-		SpinLock m_command_lock;
-		ThreadBlocker m_command_blocker;
+		// Intermediate read/write buffer
+		// TODO: can we read straight to user buffer?
+		BAN::UniqPtr<DMARegion> m_data_dma_region;
 
 		friend class AHCIController;
 	};

kernel/include/kernel/Terminal/PseudoTerminal.h

@@ -1,7 +1,6 @@
 #pragma once
 
 #include <BAN/WeakPtr.h>
-#include <kernel/Memory/VirtualRange.h>
 #include <kernel/Terminal/TTY.h>
 
 namespace Kernel

kernel/include/kernel/Terminal/VirtualTTY.h

@@ -96,6 +96,8 @@ namespace Kernel
 		uint32_t m_last_cursor_row { static_cast<uint32_t>(-1) };
 		uint32_t m_last_cursor_column { static_cast<uint32_t>(-1) };
 
+		const Palette& m_palette;
+
 		TerminalDriver::Color m_foreground;
 		TerminalDriver::Color m_background;
 		bool m_colors_inverted { false };

kernel/include/kernel/Thread.h

@@ -5,8 +5,8 @@
 #include <BAN/RefPtr.h>
 #include <BAN/UniqPtr.h>
 #include <kernel/InterruptStack.h>
-#include <kernel/Lock/Mutex.h>
 #include <kernel/Memory/VirtualRange.h>
+#include <kernel/ThreadBlocker.h>
 
 #include <LibELF/AuxiliaryVector.h>
 
@@ -18,7 +18,6 @@ namespace Kernel
 
 	class MemoryBackedRegion;
 	class Process;
-	class ThreadBlocker;
 
 	class Thread
 	{
@@ -125,14 +124,10 @@ namespace Kernel
 
 		bool is_userspace() const { return m_is_userspace; }
 
-		uint64_t cpu_time_total_ns() const;
-		void cpu_time_ns(uint64_t& user_ns, uint64_t& system_ns) const;
+		uint64_t cpu_time_ns() const;
 		void set_cpu_time_start();
 		void set_cpu_time_stop();
 
-		void set_is_in_syscall(bool is_in_syscall);
-		bool is_in_syscall() const { return m_is_in_syscall; }
-
 		void update_processor_index_address();
 
 		void set_fsbase(vaddr_t base) { m_fsbase = base; }
@@ -192,7 +187,7 @@ namespace Kernel
 		vaddr_t                    m_fsbase               { 0 };
 		vaddr_t                    m_gsbase               { 0 };
 
-		SchedulerThreadNode*       m_scheduler_node       { nullptr };
+		SchedulerQueue::Node*      m_scheduler_node       { nullptr };
 
 		YieldRegisters             m_yield_registers      { };
 
@@ -205,10 +200,8 @@ namespace Kernel
 		static_assert(_SIGMAX < 64);
 
 		mutable SpinLock           m_cpu_time_lock;
-		uint64_t                   m_cpu_time_user_ns     { 0 };
-		uint64_t                   m_cpu_time_system_ns   { 0 };
+		uint64_t                   m_cpu_time_ns          { 0 };
 		uint64_t                   m_cpu_time_start_ns    { UINT64_MAX };
-		BAN::Atomic<bool>          m_is_in_syscall        { false };
 
 		BAN::Atomic<uint32_t>      m_spinlock_count       { 0 };
 		BAN::Atomic<uint32_t>      m_mutex_count          { 0 };

kernel/include/kernel/ThreadBlocker.h

@@ -1,14 +1,12 @@
 #pragma once
 
 #include <BAN/Math.h>
-#include <kernel/Lock/Mutex.h>
 #include <kernel/Lock/SpinLock.h>
+#include <kernel/Scheduler.h>
 
 namespace Kernel
 {
 
-	class SchedulerThreadNode;
-
 	class ThreadBlocker
 	{
 	public:
@@ -29,11 +27,11 @@ namespace Kernel
 		}
 
 	private:
-		void add_thread_to_block_queue(SchedulerThreadNode*);
-		void remove_thread_from_block_queue(SchedulerThreadNode*);
+		void add_thread_to_block_queue(SchedulerQueue::Node*);
+		void remove_thread_from_block_queue(SchedulerQueue::Node*);
 
 	private:
-		SchedulerThreadNode* m_block_chain { nullptr };
+		SchedulerQueue::Node* m_block_chain { nullptr };
 		SpinLock m_lock;
 
 		friend class Scheduler;

kernel/include/kernel/USB/Device.h

@@ -23,8 +23,6 @@ namespace Kernel
 
 		virtual void handle_stall(uint8_t endpoint_id) = 0;
 		virtual void handle_input_data(size_t byte_count, uint8_t endpoint_id) = 0;
-
-		virtual bool is_hid_driver() const { return false; }
 	};
 
 	class USBDevice
@@ -64,12 +62,6 @@ namespace Kernel
 			uint8_t tt_think_time;
 		};
 
-		struct HIDInfo
-		{
-			BAN::Atomic<uint32_t> led_mask { 0 };
-			BAN::Vector<USBClassDriver*> led_controls;
-		};
-
 	public:
 		USBDevice(USBController& controller, USB::SpeedClass speed_class, uint8_t depth)
 			: m_controller(controller)
@@ -100,8 +92,6 @@ namespace Kernel
 		void register_hub_to_init() { m_controller.register_hub_to_init(m_depth + 1); };
 		void mark_hub_init_done() { m_controller.mark_hub_init_done(m_depth + 1); };
 
-		void update_led_mask(uint32_t led_mask);
-
 	protected:
 		void handle_stall(uint8_t endpoint_id);
 		void handle_input_data(size_t byte_count, uint8_t endpoint_id);
@@ -122,7 +112,6 @@ namespace Kernel
 		BAN::UniqPtr<DMARegion> m_dma_buffer;
 
 		BAN::Vector<BAN::UniqPtr<USBClassDriver>> m_class_drivers;
-		HIDInfo m_hid_info;
 	};
 
 }

kernel/include/kernel/USB/HID/HIDDriver.h

@@ -74,7 +74,6 @@ namespace Kernel
 		struct DeviceReport
 		{
 			BAN::Vector<USBHID::Report> inputs;
-			BAN::Vector<USBHID::Report> outputs;
 			BAN::RefPtr<USBHIDDevice> device;
 		};
 
@@ -82,43 +81,27 @@ namespace Kernel
 		void handle_stall(uint8_t endpoint_id) override;
 		void handle_input_data(size_t byte_count, uint8_t endpoint_id) override;
 
-		bool is_hid_driver() const override { return true; }
-
 		USBDevice& device() { return m_device; }
 		const USBDevice::InterfaceDescriptor& interface() const { return m_interface; }
 
-		bool has_led_control() const { return !m_led_controls.empty(); }
-		void set_leds(uint32_t led_mask);
-
 	private:
 		USBHIDDriver(USBDevice&, const USBDevice::InterfaceDescriptor&);
 		~USBHIDDriver();
 
 		BAN::ErrorOr<void> initialize() override;
 
-		BAN::ErrorOr<void> initializes_device_reports(const BAN::Vector<USBHID::Collection>&);
-
-	private:
-		struct LEDControl
-		{
-			DeviceReport* report;
-			uint32_t report_id;
-			uint32_t report_bits;
-		};
+		BAN::ErrorOr<BAN::Vector<DeviceReport>> initializes_device_reports(const BAN::Vector<USBHID::Collection>&);
 
 	private:
 		USBDevice& m_device;
 		USBDevice::InterfaceDescriptor m_interface;
 
 		bool m_uses_report_id { false };
-		BAN::Vector<DeviceReport> m_device_reports;
+		BAN::Vector<DeviceReport> m_device_inputs;
 
 		uint8_t m_data_endpoint_id = 0;
 		BAN::UniqPtr<DMARegion> m_data_buffer;
 
-		BAN::Vector<LEDControl> m_led_controls;
-		BAN::UniqPtr<DMARegion> m_led_region;
-
 		friend class BAN::UniqPtr<USBHIDDriver>;
 	};
 

kernel/include/kernel/USB/HID/Keyboard.h

@@ -11,6 +11,8 @@ namespace Kernel
 		BAN_NON_MOVABLE(USBKeyboard);
 
 	public:
+		BAN::ErrorOr<void> initialize() override;
+
 		void start_report() override;
 		void stop_report() override;
 
@@ -21,9 +23,12 @@ namespace Kernel
 		void update() override;
 
 	private:
-		USBKeyboard(USBHIDDriver& driver);
+		USBKeyboard(USBHIDDriver& driver, BAN::Vector<USBHID::Report>&& outputs);
 		~USBKeyboard() = default;
 
+		void set_leds(uint16_t mask);
+		void set_leds(uint8_t report_id, uint16_t mask);
+
 	private:
 		USBHIDDriver& m_driver;
 
@@ -33,7 +38,11 @@ namespace Kernel
 		BAN::Array<bool, 0x100> m_keyboard_state      { false };
 		BAN::Array<bool, 0x100> m_keyboard_state_temp { false };
 		uint16_t m_toggle_mask { 0 };
+
 		uint16_t m_led_mask { 0 };
+		BAN::UniqPtr<DMARegion> m_led_region;
+
+		BAN::Vector<USBHID::Report> m_outputs;
 
 		BAN::Optional<uint8_t> m_repeat_scancode;
 		uint8_t m_repeat_modifier { 0 };

kernel/include/kernel/USB/Hub/HubDriver.h

@@ -1,5 +1,6 @@
 #pragma once
 
+#include <kernel/Process.h>
 #include <kernel/USB/Device.h>
 
 namespace Kernel

kernel/include/kernel/USB/XHCI/Definitions.h

@@ -373,33 +373,6 @@ namespace Kernel::XHCI
 				uint32_t slot_id               : 8;
 			} configure_endpoint_command;
 
-			struct
-			{
-				uint32_t             : 32;
-				uint32_t             : 32;
-				uint32_t             : 32;
-				uint32_t cycle_bit   : 1;
-				uint32_t             : 8;
-				uint32_t tsp         : 1;
-				uint32_t trb_type    : 6;
-				uint32_t endpoint_id : 5;
-				uint32_t             : 3;
-				uint32_t slot_id     : 8;
-			} reset_endpoint_command;
-
-			struct
-			{
-				uint64_t new_tr_deque_pointer : 64;
-				uint32_t                      : 16;
-				uint32_t stream_id            : 16;
-				uint32_t cycle_bit            : 1;
-				uint32_t                      : 9;
-				uint32_t trb_type             : 6;
-				uint32_t endpoint_id          : 5;
-				uint32_t                      : 3;
-				uint32_t slot_id              : 8;
-			} set_tr_deque_pointer_command;
-
 			struct
 			{
 				uint64_t ring_segment_ponter     : 64;

kernel/kernel/ACPI/ACPI.cpp

@@ -6,9 +6,9 @@
 #include <kernel/BootInfo.h>
 #include <kernel/InterruptController.h>
 #include <kernel/IO.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/Memory/PageTable.h>
 #include <kernel/Process.h>
-#include <kernel/Scheduler.h>
 #include <kernel/Timer/Timer.h>
 
 #define RSPD_SIZE	20
@@ -790,7 +790,7 @@ acpi_release_global_lock:
 			auto [gpe_path, gpe_obj] = TRY(m_namespace->find_named_object(embedded_controller, TRY(AML::NameString::from_string("_GPE"_sv)), true));
 			if (gpe_obj == nullptr)
 			{
-				dwarnln("EC {} does not have _GPE", embedded_controller);
+				dwarnln("EC {} does have _GPE", embedded_controller);
 				break;
 			}
 
@@ -807,7 +807,7 @@ acpi_release_global_lock:
 		auto [crs_path, crs_obj] = TRY(m_namespace->find_named_object(embedded_controller, TRY(AML::NameString::from_string("_CRS"_sv)), true));
 		if (crs_obj == nullptr)
 		{
-			dwarnln("EC {} does not have _CRS", embedded_controller);
+			dwarnln("EC {} does have _CRS", embedded_controller);
 			return BAN::Error::from_errno(ENOENT);
 		}
 
@@ -935,7 +935,7 @@ acpi_release_global_lock:
 		return false;
 	}
 
-	BAN::ErrorOr<void> ACPI::enter_acpi_mode()
+	BAN::ErrorOr<void> ACPI::enter_acpi_mode(uint8_t mode)
 	{
 		ASSERT(!m_namespace);
 
@@ -1014,7 +1014,7 @@ acpi_release_global_lock:
 
 			AML::Reference arg_ref;
 			arg_ref.node.type = AML::Node::Type::Integer;
-			arg_ref.node.as.integer.value = InterruptController::get().is_using_apic() ? 1 : 0;
+			arg_ref.node.as.integer.value = mode;
 			arg_ref.ref_count = 2;
 
 			BAN::Array<AML::Reference*, 7> arguments(nullptr);
@@ -1022,7 +1022,7 @@ acpi_release_global_lock:
 			TRY(AML::method_call(pic_path, pic_node, BAN::move(arguments)));
 		}
 
-		dprintln("Evaluated \\_PIC({})", InterruptController::get().is_using_apic() ? 1 : 0);
+		dprintln("Evaluated \\_PIC({})", mode);
 
 		uint8_t irq = fadt().sci_int;
 		if (auto ret = InterruptController::get().reserve_irq(irq); ret.is_error())
@@ -1098,17 +1098,14 @@ acpi_release_global_lock:
 
 		dprintln("Initialized ACPI interrupts");
 
-		if (InterruptController::get().is_using_apic())
+		if (auto interrupt_link_devices_or_error = m_namespace->find_device_with_eisa_id("PNP0C0F"_sv); !interrupt_link_devices_or_error.is_error())
 		{
-			if (auto interrupt_link_devices_or_error = m_namespace->find_device_with_eisa_id("PNP0C0F"_sv); !interrupt_link_devices_or_error.is_error())
-			{
-				uint64_t routed_irq_mask = 0;
-				auto interrupt_link_devices = interrupt_link_devices_or_error.release_value();
-				for (const auto& device : interrupt_link_devices)
-					if (auto ret = route_interrupt_link_device(device, routed_irq_mask); ret.is_error())
-						dwarnln("failed to route interrupt link device: {}", ret.error());
-				dprintln("Routed interrupt link devices");
-			}
+			uint64_t routed_irq_mask = 0;
+			auto interrupt_link_devices = interrupt_link_devices_or_error.release_value();
+			for (const auto& device : interrupt_link_devices)
+				if (auto ret = route_interrupt_link_device(device, routed_irq_mask); ret.is_error())
+					dwarnln("failed to route interrupt link device: {}", ret.error());
+			dprintln("Routed interrupt link devices");
 		}
 
 		return {};
@@ -1199,8 +1196,6 @@ acpi_release_global_lock:
 			continue;
 
 handle_event:
-			IO::outw(sts_port, pending);
-
 			if (pending & PM1_EVN_PWRBTN)
 			{
 				dprintln("Power button pressed");
@@ -1211,6 +1206,8 @@ handle_event:
 			{
 				dwarnln("Unhandled ACPI fixed event {H}", pending);
 			}
+
+			IO::outw(sts_port, pending);
 		}
 	}
 

kernel/kernel/ACPI/BatterySystem.cpp

@@ -1,6 +1,5 @@
 #include <kernel/ACPI/BatterySystem.h>
-#include <kernel/Device/Device.h>
-#include <kernel/FS/DevFS/FileSystem.h>
+#include <kernel/FS/DevFS//FileSystem.h>
 #include <kernel/Timer/Timer.h>
 
 namespace Kernel::ACPI
@@ -36,11 +35,34 @@ namespace Kernel::ACPI
 			if (offset < 0)
 				return BAN::Error::from_errno(EINVAL);
 
-			auto target_str = TRY(BAN::String::formatted("{}", TRY(get_value())));
+			if (SystemTimer::get().ms_since_boot() > m_last_read_ms + 1000)
+			{
+				auto [method_path, method_ref] = TRY(m_acpi_namespace.find_named_object(m_battery_path, m_method_name));
+				if (method_ref == nullptr)
+					return BAN::Error::from_errno(EFAULT);
+
+				auto result = TRY(AML::method_call(method_path, method_ref->node, BAN::Array<AML::Reference*, 7>{}));
+				if (result.type != AML::Node::Type::Package || result.as.package->num_elements < m_result_index)
+					return BAN::Error::from_errno(EFAULT);
+
+				auto& target_elem = result.as.package->elements[m_result_index];
+				if (!target_elem.resolved || !target_elem.value.node)
+					return BAN::Error::from_errno(EFAULT);
+
+				auto target_conv = AML::convert_node(TRY(target_elem.value.node->copy()), AML::ConvInteger, sizeof(uint64_t));
+				if (target_conv.is_error())
+					return BAN::Error::from_errno(EFAULT);
+
+				m_last_read_ms = SystemTimer::get().ms_since_boot();
+				m_last_value = target_conv.value().as.integer.value;
+			}
+
+			auto target_str = TRY(BAN::String::formatted("{}", m_last_value.load()));
+
 			if (static_cast<size_t>(offset) >= target_str.size())
 				return 0;
 
-			const size_t ncopy = BAN::Math::min<size_t>(buffer.size(), target_str.size() - offset);
+			const size_t ncopy = BAN::Math::min(buffer.size(), target_str.size() - offset);
 			memcpy(buffer.data(), target_str.data() + offset, ncopy);
 			return ncopy;
 		}
@@ -62,44 +84,14 @@ namespace Kernel::ACPI
 			, m_result_index(index)
 		{ }
 
-		BAN::ErrorOr<uint64_t> get_value()
-		{
-			LockGuard _(m_mutex);
-
-			if (SystemTimer::get().ms_since_boot() < m_last_read_ms + 1000)
-				return m_last_value;
-
-			auto [method_path, method_ref] = TRY(m_acpi_namespace.find_named_object(m_battery_path, m_method_name));
-			if (method_ref == nullptr)
-				return BAN::Error::from_errno(EFAULT);
-
-			auto result = TRY(AML::method_call(method_path, method_ref->node, BAN::Array<AML::Reference*, 7>{}));
-			if (result.type != AML::Node::Type::Package || result.as.package->num_elements < m_result_index)
-				return BAN::Error::from_errno(EFAULT);
-
-			auto& target_elem = result.as.package->elements[m_result_index];
-			if (!target_elem.resolved || !target_elem.value.node)
-				return BAN::Error::from_errno(EFAULT);
-
-			auto target_conv = AML::convert_node(TRY(target_elem.value.node->copy()), AML::ConvInteger, sizeof(uint64_t));
-			if (target_conv.is_error())
-				return BAN::Error::from_errno(EFAULT);
-
-			m_last_read_ms = SystemTimer::get().ms_since_boot();
-			m_last_value = target_conv.value().as.integer.value;
-
-			return m_last_value;
-		}
-
 	private:
 		AML::Namespace& m_acpi_namespace;
 		AML::Scope m_battery_path;
 		AML::NameString m_method_name;
 		size_t m_result_index;
 
-		Mutex m_mutex;
-		uint64_t m_last_read_ms = 0;
-		uint64_t m_last_value = 0;
+		BAN::Atomic<uint64_t> m_last_read_ms = 0;
+		BAN::Atomic<uint64_t> m_last_value = 0;
 	};
 
 	BAN::ErrorOr<void> BatterySystem::initialize(AML::Namespace& acpi_namespace)

kernel/kernel/ACPI/EmbeddedController.cpp

@@ -4,8 +4,6 @@
 #include <kernel/ACPI/EmbeddedController.h>
 #include <kernel/IO.h>
 #include <kernel/Lock/LockGuard.h>
-#include <kernel/Scheduler.h>
-#include <kernel/Thread.h>
 #include <kernel/Timer/Timer.h>
 
 namespace Kernel::ACPI

kernel/kernel/APIC.cpp

@@ -351,17 +351,17 @@ namespace Kernel
 			// give processor upto 100 * 100 us + 200 us to boot
 			PageTable::with_fast_page(ap_init_paddr, [&] {
 				for (int i = 0; i < 100; i++, udelay(100))
-					if (BAN::atomic_load(PageTable::fast_page_as<ap_init_info_t>(8).ready))
+					if (__atomic_load_n(&PageTable::fast_page_as<ap_init_info_t>(8).ready, __ATOMIC_SEQ_CST))
 						break;
 			});
 
 			initialized_aps++;
 		}
 
-		BAN::atomic_store(g_ap_startup_done[0], 1);
+		__atomic_store_n(&g_ap_startup_done[0], 1, __ATOMIC_SEQ_CST);
 
 		const size_t timeout_ms = SystemTimer::get().ms_since_boot() + 500;
-		while (BAN::atomic_load(g_ap_running_count[0]) < initialized_aps)
+		while (__atomic_load_n(&g_ap_running_count[0], __ATOMIC_SEQ_CST) < initialized_aps)
 		{
 			if (SystemTimer::get().ms_since_boot() >= timeout_ms)
 				Kernel::panic("Could not start all APs ({}/{} started)", g_ap_running_count[0], initialized_aps);
@@ -487,12 +487,17 @@ namespace Kernel
 		write_to_local_apic(LAPIC_EIO_REG, 0);
 	}
 
-	void APIC::enable_irq(uint8_t irq, bool level_triggered)
+	void APIC::enable_irq(uint8_t irq)
 	{
 		SpinLockGuard _(m_lock);
 
 		const uint32_t gsi = m_irq_overrides[irq];
-		ASSERT(m_used_gsis[gsi / 8] & (1 << (gsi % 8)));
+
+		{
+			int byte = gsi / 8;
+			int bit  = gsi % 8;
+			ASSERT(m_reserved_gsis[byte] & (1 << bit));
+		}
 
 		IOAPIC* ioapic = nullptr;
 		for (IOAPIC& io : m_io_apics)
@@ -510,30 +515,27 @@ namespace Kernel
 		RedirectionEntry redir;
 		redir.lo_dword = ioapic->read(IOAPIC_REDIRS + pin * 2);
 		redir.hi_dword = ioapic->read(IOAPIC_REDIRS + pin * 2 + 1);
+		ASSERT(redir.mask); // TODO: handle overlapping interrupts
 
 		redir.vector = IRQ_VECTOR_BASE + irq;
-		redir.delivery_mode = 0; // fixed
-		redir.pin_polarity  = 0; // active high
-		redir.trigger_mode = level_triggered;
 		redir.mask = 0;
-		redir.destination_mode = 0; // physical
 		// FIXME: distribute IRQs more evenly?
 		redir.destination = Kernel::Processor::bsp_id().as_u32();
 
-		ioapic->write(IOAPIC_REDIRS + pin * 2,     redir.lo_dword);
-		ioapic->write(IOAPIC_REDIRS + pin * 2 + 1, redir.hi_dword);
+		ioapic->write(IOAPIC_REDIRS + pin * 2,		redir.lo_dword);
+		ioapic->write(IOAPIC_REDIRS + pin * 2 + 1,	redir.hi_dword);
 	}
 
 	bool APIC::is_in_service(uint8_t irq)
 	{
-		const uint32_t dword = (irq + IRQ_VECTOR_BASE) / 32;
-		const uint32_t bit   = (irq + IRQ_VECTOR_BASE) % 32;
+		uint32_t dword = (irq + IRQ_VECTOR_BASE) / 32;
+		uint32_t bit = (irq + IRQ_VECTOR_BASE) % 32;
 
-		const uint32_t isr = read_from_local_apic(LAPIC_IS_REG + dword * 0x10);
+		uint32_t isr = read_from_local_apic(LAPIC_IS_REG + dword * 0x10);
 		return isr & (1 << bit);
 	}
 
-	BAN::ErrorOr<void> APIC::reserve_irq(uint8_t irq, bool shared)
+	BAN::ErrorOr<void> APIC::reserve_irq(uint8_t irq)
 	{
 		SpinLockGuard _(m_lock);
 
@@ -555,26 +557,20 @@ namespace Kernel
 			return BAN::Error::from_errno(EINVAL);
 		}
 
-		const uint8_t byte =       gsi / 8;
-		const uint8_t mask = 1 << (gsi % 8);
-
-		if (!shared && (m_excl_gsis[byte] & mask))
+		int byte = gsi / 8;
+		int bit  = gsi % 8;
+		if (m_reserved_gsis[byte] & (1 << bit))
 		{
-			dwarnln("GSI {} is already reserved as exclusive", gsi);
-			return BAN::Error::from_errno(EINVAL);
+			dwarnln("GSI {} is already reserved (IRQ {})", gsi, irq);
+			return BAN::Error::from_errno(EFAULT);
 		}
-
-		m_used_gsis[byte] |= mask;
-
-		if (!shared)
-			m_excl_gsis[byte] |= mask;
-
+		m_reserved_gsis[byte] |= 1 << bit;
 		return {};
 	}
 
 	// FIXME: rewrite gsi and vector reserving
 	//        this is a hack to allow direct GSI reservation
-	BAN::ErrorOr<uint8_t> APIC::reserve_gsi(uint32_t gsi, bool shared)
+	BAN::ErrorOr<uint8_t> APIC::reserve_gsi(uint32_t gsi)
 	{
 		size_t irq = 0;
 		for (; irq < 0x100; irq++)
@@ -587,7 +583,7 @@ namespace Kernel
 			return BAN::Error::from_errno(ENOTSUP);
 		}
 
-		TRY(reserve_irq(irq, shared));
+		TRY(reserve_irq(irq));
 
 		return irq;
 	}
@@ -595,7 +591,6 @@ namespace Kernel
 	BAN::Optional<uint8_t> APIC::get_free_irq()
 	{
 		SpinLockGuard _(m_lock);
-
 		for (uint8_t irq = 0; irq < m_irq_count; irq++)
 		{
 			const uint8_t gsi = m_irq_overrides[irq];
@@ -613,17 +608,13 @@ namespace Kernel
 			if (!ioapic)
 				continue;
 
-			const uint8_t byte =       gsi / 8;
-			const uint8_t mask = 1 << (gsi % 8);
-
-			if (m_used_gsis[byte] & mask)
+			const uint8_t byte = gsi / 8;
+			const uint8_t bit  = gsi % 8;
+			if (m_reserved_gsis[byte] & (1 << bit))
 				continue;
-			m_used_gsis[byte] |= mask;
-			m_excl_gsis[byte] |= mask;
-
+			m_reserved_gsis[byte] |= 1 << bit;
 			return irq;
 		}
-
 		return {};
 	}
 

kernel/kernel/Audio/AC97/Controller.cpp

@@ -182,22 +182,13 @@ namespace Kernel
 		m_bdl_region = TRY(DMARegion::create(bdl_size + buffer_size * m_used_bdl_entries));
 		memset(reinterpret_cast<void*>(m_bdl_region->vaddr()), 0x00, m_bdl_region->size());
 
-		if (m_bdl_region->paddr() + m_bdl_region->size() > BAN::numeric_limits<uint32_t>::max())
-		{
-			dwarnln("Could not allocate 32 bit dma buffer");
-			return BAN::Error::from_errno(EFAULT);
-		}
-
-		auto* bdl_entries = reinterpret_cast<AC97::BufferDescriptorListEntry*>(m_bdl_region->vaddr());
 		for (size_t i = 0; i < m_bdl_entries; i++)
 		{
-			bdl_entries[i] = {
-				.address = static_cast<uint32_t>(m_bdl_region->paddr() + bdl_size + (i % m_used_bdl_entries) * buffer_size),
-				.samples = 0,
-				.flags   = 1 << 15,
-			};
+			auto& entry = reinterpret_cast<AC97::BufferDescriptorListEntry*>(m_bdl_region->vaddr())[i];
+			entry.address = m_bdl_region->paddr() + bdl_size + (i % m_used_bdl_entries) * buffer_size;
+			entry.samples = 0;
+			entry.flags = 0;
 		}
-		bdl_entries[m_bdl_entries - 1].flags |= 1 << 14;
 
 		m_bus_master->write32(BusMasterRegister::PO_BDBAR, m_bdl_region->paddr());
 
@@ -224,12 +215,7 @@ namespace Kernel
 	BAN::ErrorOr<void> AC97AudioController::set_volume_mdB(int32_t mdB)
 	{
 		m_volume_info.mdB = BAN::Math::clamp(mdB, m_volume_info.min_mdB, m_volume_info.max_mdB);
-
-		const uint32_t volume_data = get_volume_data();
-		m_mixer->write16(AudioMixerRegister::MasterVolume, volume_data);
-
-		m_volume_info.mdB = -(volume_data & 0xFF) * m_volume_info.step_mdB;
-
+		m_mixer->write16(AudioMixerRegister::MasterVolume, get_volume_data());
 		return {};
 	}
 
@@ -259,6 +245,7 @@ namespace Kernel
 
 			auto& entry = reinterpret_cast<AC97::BufferDescriptorListEntry*>(m_bdl_region->vaddr())[m_bdl_head];
 			entry.samples = sample_frames * get_channels();
+			entry.flags = (1 << 15);
 			memcpy(
 				reinterpret_cast<void*>(m_bdl_region->paddr_to_vaddr(entry.address)),
 				m_sample_data->get_data().data(),

kernel/kernel/Audio/Controller.cpp

@@ -3,8 +3,7 @@
 #include <kernel/Audio/HDAudio/Controller.h>
 #include <kernel/Device/DeviceNumbers.h>
 #include <kernel/FS/DevFS/FileSystem.h>
-#include <kernel/Lock/BlockableSpinLock.h>
-#include <kernel/Thread.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 
 #include <sys/ioctl.h>
 #include <sys/sysmacros.h>
@@ -66,8 +65,8 @@ namespace Kernel
 
 		while (m_sample_data->full())
 		{
-			BlockableSpinLock block(m_spinlock);
-			TRY(Thread::current().block_or_eintr_indefinite(m_sample_data_blocker, &block));
+			SpinLockGuardAsMutex smutex(lock_guard);
+			TRY(Thread::current().block_or_eintr_indefinite(m_sample_data_blocker, &smutex));
 		}
 
 		const size_t to_copy = BAN::Math::min(buffer.size(), m_sample_data->free());

kernel/kernel/Audio/HDAudio/AudioFunctionGroup.cpp

@@ -139,27 +139,31 @@ namespace Kernel
 
 	BAN::ErrorOr<void> HDAudioFunctionGroup::set_volume_mdB(int32_t mdB)
 	{
-		if (m_amplifier_idx == SIZE_MAX)
-			return BAN::Error::from_errno(ENOTSUP);
-
 		mdB = BAN::Math::clamp(mdB, m_volume_info.min_mdB, m_volume_info.max_mdB);
 
-		const auto* node = m_output_paths[m_output_path_index][m_amplifier_idx];
+		const auto& path = m_output_paths[m_output_path_index];
+		for (size_t i = 0; i < path.size(); i++)
+		{
+			if (!path[i]->output_amplifier.has_value())
+				continue;
 
-		const int32_t step_round = (mdB >= 0)
-			? +m_volume_info.step_mdB / 2
-			: -m_volume_info.step_mdB / 2;
-		const uint32_t step = (mdB + step_round) / m_volume_info.step_mdB + node->output_amplifier->offset;
-		const uint32_t volume = 0b1'0'1'1'0000'0'0000000 | step;
+			const int32_t step_round = (mdB >= 0)
+				? +m_volume_info.step_mdB / 2
+				: -m_volume_info.step_mdB / 2;
+			const uint32_t step = (mdB + step_round) / m_volume_info.step_mdB + path[i]->output_amplifier->offset;
+			const uint32_t volume = 0b1'0'1'1'0000'0'0000000 | step;
 
-		TRY(m_controller->send_command({
-			.data = static_cast<uint8_t>(volume & 0xFF),
-			.command = static_cast<uint16_t>(0x300 | (volume >> 8)),
-			.node_index = node->id,
-			.codec_address = m_cid,
-		}));
+			TRY(m_controller->send_command({
+				.data = static_cast<uint8_t>(volume & 0xFF),
+				.command = static_cast<uint16_t>(0x300 | (volume >> 8)),
+				.node_index = path[i]->id,
+				.codec_address = m_cid,
+			}));
 
-		m_volume_info.mdB = (mdB + step_round) / m_volume_info.step_mdB * m_volume_info.step_mdB;
+			break;
+		}
+
+		m_volume_info.mdB = mdB;
 
 		return {};
 	}
@@ -327,8 +331,8 @@ namespace Kernel
 	uint16_t HDAudioFunctionGroup::get_format_data() const
 	{
 		// TODO: don't hardcode this
-		// format: PCM, 48 kHz, 16 bit
-		return 0b0'0'000'000'0'001'0000 | (get_channels() - 1);
+		// format: PCM, 48 kHz, 16 bit, 2 channels
+		return 0b0'0'000'000'0'001'0001;
 	}
 
 	BAN::ErrorOr<void> HDAudioFunctionGroup::enable_output_path(uint8_t index)
@@ -342,7 +346,6 @@ namespace Kernel
 			{
 				using HDAudio::AFGWidget;
 				case AFGWidget::Type::OutputConverter:
-				case AFGWidget::Type::Mixer:
 				case AFGWidget::Type::PinComplex:
 					break;
 				default:
@@ -364,7 +367,7 @@ namespace Kernel
 			}));
 
 			// set connection index
-			if (i + 1 < path.size() && path[i]->connections.size() > 1 && path[i]->type != HDAudio::AFGWidget::Type::Mixer)
+			if (i + 1 < path.size() && path[i]->connections.size() > 1)
 			{
 				uint8_t index = 0;
 				for (; index < path[i]->connections.size(); index++)
@@ -404,13 +407,6 @@ namespace Kernel
 						.node_index = path[i]->id,
 						.codec_address = m_cid,
 					}));
-					// set channel count
-					TRY(m_controller->send_command({
-						.data = static_cast<uint8_t>(get_channels() - 1),
-						.command = 0x72D,
-						.node_index = path[i]->id,
-						.codec_address = m_cid,
-					}));
 					// set format
 					TRY(m_controller->send_command({
 						.data = static_cast<uint8_t>(format & 0xFF),
@@ -420,25 +416,6 @@ namespace Kernel
 					}));
 					break;
 
-				case AFGWidget::Type::Mixer:
-					if (path[i]->input_amplifier.has_value())
-					{
-						for (size_t idx = 0; idx < path[i]->connections.size(); idx++)
-						{
-							const uint8_t step = (path[i]->connections[idx] == path[i + 1]->id)
-								? path[i]->input_amplifier->offset
-								: 0b1'0000000;
-							const uint32_t volume = 0b0'1'1'1'0000'0'0000000 | (idx << 8) | step;
-							TRY(m_controller->send_command({
-								.data = static_cast<uint8_t>(volume & 0xFF),
-								.command = static_cast<uint16_t>(0x300 | (volume >> 8)),
-								.node_index = path[i]->id,
-								.codec_address = m_cid,
-							}));
-						}
-					}
-					break;
-
 				case AFGWidget::Type::PinComplex:
 					// enable output and H-Phn
 					TRY(m_controller->send_command({
@@ -461,25 +438,15 @@ namespace Kernel
 			}
 		}
 
-		int32_t max_steps { 0 };
-		m_amplifier_idx = SIZE_MAX;
-
+		// update volume info to this path
+		m_volume_info.min_mdB = 0;
+		m_volume_info.max_mdB = 0;
+		m_volume_info.step_mdB = 0;
 		for (size_t i = 0; i < path.size(); i++)
 		{
 			if (!path[i]->output_amplifier.has_value())
 				continue;
-			if (auto steps = path[i]->output_amplifier->num_steps; steps > max_steps)
-			{
-				max_steps = steps;
-				m_amplifier_idx = i;
-			}
-		}
-
-		if (m_amplifier_idx == SIZE_MAX)
-			m_volume_info = {};
-		else
-		{
-			const auto& amp = path[m_amplifier_idx]->output_amplifier.value();
+			const auto& amp = path[i]->output_amplifier.value();
 
 			const int32_t step_mdB = (amp.step_size + 1) * 250;
 			m_volume_info.step_mdB = step_mdB;
@@ -496,11 +463,16 @@ namespace Kernel
 			TRY(m_controller->send_command({
 				.data = static_cast<uint8_t>(volume & 0xFF),
 				.command = static_cast<uint16_t>(0x300 | (volume >> 8)),
-				.node_index = path[m_amplifier_idx]->id,
+				.node_index = path[i]->id,
 				.codec_address = m_cid,
 			}));
+
+			break;
 		}
 
+		if (m_volume_info.min_mdB == 0 && m_volume_info.max_mdB == 0)
+			m_volume_info.mdB = 0;
+
 		return {};
 	}
 
@@ -524,7 +496,6 @@ namespace Kernel
 				using HDAudio::AFGWidget;
 
 				case AFGWidget::Type::OutputConverter:
-				case AFGWidget::Type::Mixer:
 					break;
 
 				case AFGWidget::Type::PinComplex:
@@ -664,8 +635,11 @@ namespace Kernel
 			m_bdl_tail = (m_bdl_tail + 1) % m_bdl_entry_count;
 			if (m_bdl_tail == m_bdl_head)
 			{
-				bar.write8(base + Regs::SDCTL, bar.read8(base + Regs::SDCTL) & 0xFD);
-				m_stream_running = false;
+				if (auto ret = reset_stream(); ret.is_error())
+				{
+					dwarnln("failed to reset HDA stream: {}", ret.error());
+					return;
+				}
 			}
 
 			queue_bdl_data();

kernel/kernel/Audio/HDAudio/Controller.cpp

@@ -1,8 +1,8 @@
 #include <kernel/Audio/HDAudio/AudioFunctionGroup.h>
 #include <kernel/Audio/HDAudio/Controller.h>
 #include <kernel/Audio/HDAudio/Registers.h>
-#include <kernel/Lock/BlockableSpinLock.h>
 #include <kernel/Lock/LockGuard.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/MMIO.h>
 #include <kernel/Timer/Timer.h>
 
@@ -333,20 +333,6 @@ namespace Kernel
 			};
 		}
 
-		if (const uint32_t in_amp_cap = send_command_or_zero(0xF00, 0x0D))
-		{
-			const uint8_t offset    = (in_amp_cap >>  0) & 0x7F;
-			const uint8_t num_steps = (in_amp_cap >>  8) & 0x7F;
-			const uint8_t step_size = (in_amp_cap >> 16) & 0x7F;
-			const bool    mute      = (in_amp_cap >> 31);
-			result.input_amplifier = HDAudio::AFGWidget::Amplifier {
-				.offset = offset,
-				.num_steps = num_steps,
-				.step_size = step_size,
-				.mute = mute,
-			};
-		}
-
 		const uint8_t connection_info = send_command_or_zero(0xF00, 0x0E);
 		const uint8_t conn_width = (connection_info & 0x80) ? 2 : 1;
 		const uint8_t conn_count = connection_info & 0x3F;
@@ -383,8 +369,8 @@ namespace Kernel
 			{
 				if (SystemTimer::get().ms_since_boot() > waketime_ms)
 					return BAN::Error::from_errno(ETIMEDOUT);
-				BlockableSpinLock block(m_rb_lock);
-				m_rb_blocker.block_with_timeout_ms(10, &block);
+				SpinLockGuardAsMutex smutex(sguard);
+				m_rb_blocker.block_with_timeout_ms(10, &smutex);
 			}
 
 			const size_t offset = 2 * m_rirb.index * sizeof(uint32_t);

kernel/kernel/Banos.cpp

@@ -14,8 +14,6 @@
 #include <kernel/Lock/SpinLock.h>
 #include <kernel/UserCopy.h>
 
-#if ARCH(x86_64)
-
 using namespace LibELF;
 using namespace Kernel;
 
@@ -206,7 +204,7 @@ BAN::ErrorOr<size_t> Banos::load_driver_from_image(const char* u_image) {
 // NOTE: should be more than plenty ;)
 extern char g_drv_builtin_begin[];
 extern char g_drv_builtin_end[];
-void Banos::initialize_initial_drivers() {
+void Banos::initialize_initial_drivers(void) {
 	import_symbols(g_banos_export, g_banos_export_end - g_banos_export);
 	char* head = g_drv_builtin_begin;
 	while(head < g_drv_builtin_end) {
@@ -215,14 +213,3 @@ void Banos::initialize_initial_drivers() {
 		head += drv->driver_size;
 	}
 }
-
-#else
-void Banos::initialize_initial_drivers()
-{
-}
-BAN::ErrorOr<size_t> Banos::load_driver_from_image(const char* u_image)
-{
-	(void)u_image;
-	return BAN::Error::from_errno(ENOTSUP);
-}
-#endif

kernel/kernel/Epoll.cpp

@@ -1,7 +1,6 @@
 #include <kernel/Epoll.h>
-#include <kernel/Lock/BlockableSpinLock.h>
 #include <kernel/Lock/LockGuard.h>
-#include <kernel/Thread.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/Timer/Timer.h>
 
 namespace Kernel
@@ -223,8 +222,8 @@ namespace Kernel
 			if (!m_ready_events.empty())
 				continue;
 
-			BlockableSpinLock block(m_ready_lock);
-			TRY(Thread::current().block_or_eintr_or_waketime_ns(m_thread_blocker, waketime_ns, false, &block));
+			SpinLockGuardAsMutex smutex(guard);
+			TRY(Thread::current().block_or_eintr_or_waketime_ns(m_thread_blocker, waketime_ns, false, &smutex));
 		}
 
 		return event_count;

kernel/kernel/FS/DevFS/FileSystem.cpp

@@ -7,8 +7,8 @@
 #include <kernel/FS/DevFS/FileSystem.h>
 #include <kernel/FS/TmpFS/Inode.h>
 #include <kernel/Input/InputDevice.h>
-#include <kernel/Lock/BlockableSpinLock.h>
 #include <kernel/Lock/LockGuard.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/Process.h>
 #include <kernel/Scheduler.h>
 #include <kernel/Storage/StorageDevice.h>
@@ -77,8 +77,8 @@ namespace Kernel
 						bool expected = true;
 						if (!devfs->m_should_drop_disk_cache.compare_exchange(expected, false))
 						{
-							BlockableSpinLock block(devfs->m_disk_cache_lock);
-							devfs->m_disk_cache_thread_blocker.block_indefinite(&block);
+							SpinLockGuardAsMutex smutex(guard);
+							devfs->m_disk_cache_thread_blocker.block_indefinite(&smutex);
 							continue;
 						}
 					}

kernel/kernel/FS/EventFD.cpp

@@ -1,6 +1,5 @@
 #include <kernel/FS/EventFD.h>
 #include <kernel/Lock/LockGuard.h>
-#include <kernel/Thread.h>
 
 #include <sys/epoll.h>
 
@@ -44,7 +43,7 @@ namespace Kernel
 		while (m_value == 0)
 			TRY(Thread::current().block_or_eintr_indefinite(m_thread_blocker, &m_mutex));
 
-		const uint64_t read_value = m_is_semaphore ? 1 : m_value;
+		const uint64_t read_value = m_is_semaphore ? 1 : m_value.load();
 		m_value -= read_value;
 
 		buffer.as<uint64_t>() = read_value;
@@ -80,16 +79,4 @@ namespace Kernel
 		return sizeof(uint64_t);
 	}
 
-	bool EventFD::can_read_impl() const
-	{
-		LockGuard _(m_mutex);
-		return m_value > 0;
-	}
-
-	bool EventFD::can_write_impl() const
-	{
-		LockGuard _(m_mutex);
-		return m_value < UINT64_MAX - 1;
-	}
-
 }

kernel/kernel/FS/Ext2/FileSystem.cpp

@@ -2,7 +2,6 @@
 #include <BAN/Sort.h>
 #include <kernel/FS/Ext2/FileSystem.h>
 #include <kernel/Lock/LockGuard.h>
-#include <kernel/Thread.h>
 
 #define EXT2_DEBUG_PRINT 0
 #define EXT2_VERIFY_INODE 0
@@ -625,7 +624,7 @@ namespace Kernel
 				continue;
 			buffer.used = true;
 			return Ext2FS::BlockBufferWrapper {
-				buffer.buffer,
+				buffer.buffer.span(),
 				[](void* self, const uint8_t* buffer) { static_cast<BlockBufferManager*>(self)->destroy_callback(buffer); },
 				this
 			};
@@ -636,23 +635,11 @@ namespace Kernel
 
 	BAN::ErrorOr<void> Ext2FS::BlockBufferManager::initialize(size_t block_size)
 	{
-		m_buffer_data = TRY(VirtualRange::create_to_vaddr_range(
-			PageTable::kernel(),
-			{ KERNEL_OFFSET, BAN::numeric_limits<vaddr_t>::max() },
-			block_size * m_buffers.size(),
-			PageTable::ReadWrite | PageTable::Present,
-			false
-		));
-
-		auto buffer_data_span = BAN::ByteSpan { reinterpret_cast<uint8_t*>(m_buffer_data->vaddr()), m_buffer_data->size() };
-		for (size_t i = 0; i < m_buffers.size(); i++)
+		for (auto& buffer : m_buffers)
 		{
-			m_buffers[i] = {
-				.buffer = buffer_data_span.slice(i * block_size, block_size),
-				.used = false,
-			};
+			TRY(buffer.buffer.resize(block_size));
+			buffer.used = false;
 		}
-
 		return {};
 	}
 

kernel/kernel/FS/FAT/FileSystem.cpp

@@ -1,4 +1,3 @@
-#include <kernel/Device/Device.h>
 #include <kernel/FS/FAT/FileSystem.h>
 #include <kernel/Lock/LockGuard.h>
 
@@ -124,6 +123,18 @@ namespace Kernel
 	{
 		LockGuard _(m_mutex);
 
+		uint32_t block_count = 0;
+		{
+			uint32_t cluster = entry.first_cluster_lo;
+			if (m_type == Type::FAT32)
+				cluster |= static_cast<uint32_t>(entry.first_cluster_hi) << 16;
+			while (cluster >= 2 && cluster < cluster_count())
+			{
+				block_count++;
+				cluster = TRY(get_next_cluster(cluster));
+			}
+		}
+
 		uint32_t entry_cluster;
 		switch (m_type)
 		{
@@ -132,34 +143,25 @@ namespace Kernel
 				if (parent == m_root_inode)
 					entry_cluster = 1;
 				else
+				{
 					entry_cluster = parent->entry().first_cluster_lo;
+					for (uint32_t i = 0; i < cluster_index; i++)
+						entry_cluster = TRY(get_next_cluster(entry_cluster));
+				}
 				break;
 			case Type::FAT32:
 				if (parent == m_root_inode)
 					entry_cluster = m_bpb.ext_32.root_cluster;
 				else
 					entry_cluster = (static_cast<uint32_t>(parent->entry().first_cluster_hi) << 16) | parent->entry().first_cluster_lo;
+				for (uint32_t i = 0; i < cluster_index; i++)
+					entry_cluster = TRY(get_next_cluster(entry_cluster));
 				break;
 			default:
 				ASSERT_NOT_REACHED();
 		}
 
-		uint32_t block_count = 0;
-		for (uint32_t i = 0; i < cluster_index; i++)
-		{
-			block_count++;
-			entry_cluster = TRY(get_next_cluster(entry_cluster));
-		}
-
-		const uint32_t dirent_per_cluster = m_bpb.bytes_per_sector * m_bpb.sectors_per_cluster / sizeof(FAT::DirectoryEntry);
-		ASSERT(BAN::Math::is_power_of_two(dirent_per_cluster));
-		ASSERT(entry_index < dirent_per_cluster);
-
-		const uint32_t ino_cluster_shift = BAN::Math::ctz(dirent_per_cluster);
-		const ino_t ino = (static_cast<ino_t>(entry_cluster) << ino_cluster_shift) | entry_index;
-		if (ino >> ino_cluster_shift != entry_cluster)
-			dwarnln("FAT ino mapping not unique");
-
+		const ino_t ino = (static_cast<ino_t>(entry_cluster) << 32) | entry_index;
 		auto it = m_inode_cache.find(ino);
 		if (it != m_inode_cache.end())
 		{

kernel/kernel/FS/FAT/Inode.cpp

@@ -1,6 +1,5 @@
 #include <BAN/Time.h>
 
-#include <kernel/Device/Device.h>
 #include <kernel/FS/FAT/FileSystem.h>
 #include <kernel/FS/FAT/Inode.h>
 

kernel/kernel/FS/Inode.cpp

@@ -11,18 +11,18 @@
 namespace Kernel
 {
 
-	bool Inode::can_access(uid_t uid, gid_t gid, mode_t mode, const Credentials& credentials, int flags)
+	bool Inode::can_access(const Credentials& credentials, int flags) const
 	{
 		if (credentials.is_superuser())
 			return true;
 
 		if (flags & O_RDONLY)
 		{
-			if (mode & S_IROTH)
+			if (mode().mode & S_IROTH)
 			{ }
-			else if ((mode & S_IRUSR) && credentials.euid() == uid)
+			else if ((mode().mode & S_IRUSR) && credentials.euid() == uid())
 			{ }
-			else if ((mode & S_IRGRP) && credentials.has_egid(gid))
+			else if ((mode().mode & S_IRGRP) && credentials.has_egid(gid()))
 			{ }
 			else
 			{
@@ -32,11 +32,11 @@ namespace Kernel
 
 		if (flags & O_WRONLY)
 		{
-			if (mode & S_IWOTH)
+			if (mode().mode & S_IWOTH)
 			{ }
-			else if ((mode & S_IWUSR) && credentials.euid() == uid)
+			else if ((mode().mode & S_IWUSR) && credentials.euid() == uid())
 			{ }
-			else if ((mode & S_IWGRP) && credentials.has_egid(gid))
+			else if ((mode().mode & S_IWGRP) && credentials.has_egid(gid()))
 			{ }
 			else
 			{
@@ -44,13 +44,13 @@ namespace Kernel
 			}
 		}
 
-		if ((flags & O_EXEC) || (Mode { mode }.ifdir() && (flags & O_SEARCH)))
+		if ((flags & O_EXEC) || (mode().ifdir() && (flags & O_SEARCH)))
 		{
-			if (mode & S_IXOTH)
+			if (mode().mode & S_IXOTH)
 			{ }
-			else if ((mode & S_IXUSR) && credentials.euid() == uid)
+			else if ((mode().mode & S_IXUSR) && credentials.euid() == uid())
 			{ }
-			else if ((mode & S_IXGRP) && credentials.has_egid(gid))
+			else if ((mode().mode & S_IXGRP) && credentials.has_egid(gid()))
 			{ }
 			else
 			{
@@ -61,11 +61,6 @@ namespace Kernel
 		return true;
 	}
 
-	bool Inode::can_access(const Credentials& credentials, int flags) const
-	{
-		return can_access(uid(), gid(), mode().mode, credentials, flags);
-	}
-
 	BAN::ErrorOr<BAN::RefPtr<Inode>> Inode::find_inode(BAN::StringView name)
 	{
 		if (!mode().ifdir())

kernel/kernel/FS/ProcFS/FileSystem.cpp

@@ -1,7 +1,6 @@
 #include <kernel/BootInfo.h>
 #include <kernel/FS/ProcFS/FileSystem.h>
 #include <kernel/FS/ProcFS/Inode.h>
-#include <kernel/Process.h>
 
 namespace Kernel
 {

kernel/kernel/FS/ProcFS/Inode.cpp

@@ -1,5 +1,4 @@
 #include <kernel/FS/ProcFS/Inode.h>
-#include <kernel/Process.h>
 
 #include <ctype.h>
 

kernel/kernel/IDT.cpp

@@ -1,9 +1,7 @@
 #include <BAN/Array.h>
 #include <BAN/Errors.h>
-#include <kernel/GDT.h>
 #include <kernel/IDT.h>
 #include <kernel/InterruptController.h>
-#include <kernel/InterruptNumbers.h>
 #include <kernel/InterruptStack.h>
 #include <kernel/Memory/kmalloc.h>
 #include <kernel/Panic.h>
@@ -415,14 +413,10 @@ namespace Kernel
 
 	extern "C" void cpp_ipi_handler()
 	{
-		if (!InterruptController::get().is_in_service(IRQ_IPI - IRQ_VECTOR_BASE))
-			return;
-
-		Processor::handle_ipi();
-
+		ASSERT(InterruptController::get().is_in_service(IRQ_IPI - IRQ_VECTOR_BASE));
 		InterruptController::get().eoi(IRQ_IPI - IRQ_VECTOR_BASE);
-
-		Processor::scheduler().reschedule_if_needed();
+		Processor::handle_ipi();
+		Processor::scheduler().reschedule_if_idle();
 	}
 
 	extern "C" void cpp_timer_handler()
@@ -435,17 +429,13 @@ namespace Kernel
 			asm volatile("cli; 1: hlt; jmp 1b");
 		}
 
-		if (!InterruptController::get().is_in_service(IRQ_TIMER - IRQ_VECTOR_BASE))
-			return;
+		ASSERT(InterruptController::get().is_in_service(IRQ_TIMER - IRQ_VECTOR_BASE));
+		InterruptController::get().eoi(IRQ_TIMER - IRQ_VECTOR_BASE);
 
 		if (Processor::current_is_bsp())
 			Process::update_alarm_queue();
 
 		Processor::scheduler().on_timer_interrupt();
-
-		InterruptController::get().eoi(IRQ_TIMER - IRQ_VECTOR_BASE);
-
-		Processor::scheduler().reschedule_if_needed();
 	}
 
 	extern "C" void cpp_irq_handler(uint32_t irq)
@@ -463,14 +453,15 @@ namespace Kernel
 		if (!InterruptController::get().is_in_service(irq))
 			return;
 
+		InterruptController::get().eoi(irq);
 		if (auto* handler = s_interruptables[irq])
 			handler->handle_irq();
 		else
 			dprintln("no handler for irq 0x{2H}", irq);
 
-		InterruptController::get().eoi(irq);
+		Processor::scheduler().reschedule_if_idle();
 
-		Processor::scheduler().reschedule_if_needed();
+		ASSERT(Thread::current().state() != Thread::State::Terminated);
 	}
 
 	extern "C" void cpp_check_signal()

kernel/kernel/Input/InputDevice.cpp

@@ -3,10 +3,8 @@
 #include <kernel/Device/DeviceNumbers.h>
 #include <kernel/FS/DevFS/FileSystem.h>
 #include <kernel/Input/InputDevice.h>
-#include <kernel/Lock/BlockableSpinLock.h>
-#include <kernel/Scheduler.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/Terminal/TTY.h>
-#include <kernel/Thread.h>
 
 #include <LibInput/Joystick.h>
 #include <LibInput/KeyEvent.h>
@@ -247,8 +245,8 @@ namespace Kernel
 		while (m_event_count == 0)
 		{
 			// FIXME: should m_mutex be unlocked?
-			BlockableSpinLock block(m_event_lock);
-			TRY(Thread::current().block_or_eintr_indefinite(m_event_thread_blocker, &block));
+			SpinLockGuardAsMutex smutex(guard);
+			TRY(Thread::current().block_or_eintr_indefinite(m_event_thread_blocker, &smutex));
 		}
 
 		memcpy(buffer.data(), &m_event_buffer[m_event_tail * m_event_size], m_event_size);
@@ -291,8 +289,8 @@ namespace Kernel
 
 				if (s_tty_keyboard_events.empty())
 				{
-					BlockableSpinLock block(s_tty_keyboard_event_lock);
-					s_tty_keyboard_event_blocker.block_indefinite(&block);
+					SpinLockGuardAsMutex smutex(guard);
+					s_tty_keyboard_event_blocker.block_indefinite(&smutex);
 					continue;
 				}
 
@@ -352,8 +350,8 @@ namespace Kernel
 					return bytes;
 			}
 
-			BlockableSpinLock block(s_keyboard_lock);
-			TRY(Thread::current().block_or_eintr_indefinite(m_thread_blocker, &block));
+			SpinLockGuardAsMutex smutex(keyboard_guard);
+			TRY(Thread::current().block_or_eintr_indefinite(m_thread_blocker, &smutex));
 		}
 	}
 
@@ -409,8 +407,8 @@ namespace Kernel
 					return bytes;
 			}
 
-			BlockableSpinLock block(s_mouse_lock);
-			TRY(Thread::current().block_or_eintr_indefinite(m_thread_blocker, &block));
+			SpinLockGuardAsMutex smutex(mouse_guard);
+			TRY(Thread::current().block_or_eintr_indefinite(m_thread_blocker, &smutex));
 		}
 	}
 

kernel/kernel/Input/PS2/Controller.cpp

@@ -8,8 +8,6 @@
 #include <kernel/Input/PS2/Keyboard.h>
 #include <kernel/Input/PS2/Mouse.h>
 #include <kernel/IO.h>
-#include <kernel/Scheduler.h>
-#include <kernel/Thread.h>
 #include <kernel/Timer/Timer.h>
 
 namespace Kernel::Input
@@ -347,7 +345,7 @@ namespace Kernel::Input
 									result.command_port = data->as.fixed_io_port.range_base;
 								break;
 							case ACPI::ResourceData::Type::IRQ:
-								if (BAN::Math::popcount(data->as.irq.irq_mask) != 1)
+								if (__builtin_popcount(data->as.irq.irq_mask) != 1)
 									break;
 								for (int i = 0; i < 16; i++)
 									if (data->as.irq.irq_mask & (1 << i))
@@ -388,11 +386,14 @@ namespace Kernel::Input
 			}
 
 			dprintln("Found {} PS/2 devices from ACPI namespace", acpi_devices.size());
+			if (acpi_devices.empty())
+				return {};
 
 			if (acpi_devices.size() > 2)
 			{
 				dwarnln("TODO: over 2 PS/2 devices");
-				MUST(acpi_devices.resize(2));
+				while (acpi_devices.size() > 2)
+					acpi_devices.pop_back();
 			}
 
 			if (acpi_devices.size() == 2)
@@ -418,43 +419,42 @@ namespace Kernel::Input
 				}
 			}
 
-			if (!acpi_devices.empty())
-			{
-				BAN::Optional<uint16_t> command_port;
-				command_port = acpi_devices[0].command_port;
-				if (!command_port.has_value() && acpi_devices.size() >= 2)
-					command_port = acpi_devices[1].command_port;
-				if (command_port.has_value())
-					m_command_port = command_port.value();
+			BAN::Optional<uint16_t> command_port;
+			command_port = acpi_devices[0].command_port;
+			if (!command_port.has_value() && acpi_devices.size() >= 2)
+				command_port = acpi_devices[1].command_port;
+			if (command_port.has_value())
+				m_command_port = command_port.value();
 
-				BAN::Optional<uint16_t> data_port;
-				data_port = acpi_devices[0].data_port;
-				if (!data_port.has_value() && acpi_devices.size() >= 2)
-					data_port = acpi_devices[1].data_port;
-				if (data_port.has_value())
-					m_data_port = data_port.value();
-			}
+			BAN::Optional<uint16_t> data_port;
+			data_port = acpi_devices[0].data_port;
+			if (!data_port.has_value() && acpi_devices.size() >= 2)
+				data_port = acpi_devices[1].data_port;
+			if (data_port.has_value())
+				m_data_port = data_port.value();
 
-			for (size_t i = 0; i < acpi_devices.size(); i++)
+			devices[0] = {
+				.type      = acpi_devices[0].type,
+				.interrupt = acpi_devices[0].irq.value_or(PS2::INTERRUPT_FIRST_PORT)
+			};
+
+			if (acpi_devices.size() > 1)
 			{
-				devices[i] = {
-					.type      = acpi_devices[i].type,
-					.interrupt = acpi_devices[i].irq.value_or(i == 0 ? PS2::IRQ::DEVICE0 : PS2::IRQ::DEVICE1),
+				devices[1] = {
+					.type      = acpi_devices[1].type,
+					.interrupt = acpi_devices[1].irq.value_or(PS2::INTERRUPT_SECOND_PORT)
 				};
 			}
 		}
-
-		if (devices[0].type != DeviceType::None)
-			;
-		else if (scancode_set || has_legacy_8042())
+		else if (has_legacy_8042())
 		{
 			devices[0] = {
 				.type = DeviceType::Unknown,
-				.interrupt = PS2::IRQ::DEVICE0,
+				.interrupt = PS2::INTERRUPT_FIRST_PORT,
 			};
 			devices[1] = {
 				.type = DeviceType::Unknown,
-				.interrupt = PS2::IRQ::DEVICE1,
+				.interrupt = PS2::INTERRUPT_SECOND_PORT,
 			};
 		}
 		else

kernel/kernel/Input/PS2/Keyboard.cpp

@@ -53,9 +53,8 @@ namespace Kernel::Input
 
 		if (command_data[0] == Command::CONFIG_SCANCODE_SET && m_scancode_set >= 0xFE)
 		{
-			dwarnln("Could not detect scancode set, assuming 1");
-			m_scancode_set_uncertain = true;
-			m_scancode_set = 1;
+			dwarnln("Could not detect scancode set, assuming 2");
+			m_scancode_set = 2;
 			m_keymap.initialize(m_scancode_set);
 			append_command_queue(PS2::DeviceCommand::ENABLE_SCANNING, 0);
 		}
@@ -90,7 +89,6 @@ namespace Kernel::Input
 			else
 			{
 				dwarnln("Could not detect scancode set, assuming 1");
-				m_scancode_set_uncertain = true;
 				m_scancode_set = 1;
 			}
 			m_keymap.initialize(m_scancode_set);
@@ -105,17 +103,6 @@ namespace Kernel::Input
 			return;
 		}
 
-		// If we could not detect scancode set, we assume it to be 1
-		// If we get byte 0xF0 which is indicates release in scancode set 2
-		// and nothing in scancode set 1, switch to scancode set 2
-		if (m_scancode_set_uncertain && byte == 0xF0)
-		{
-			dprintln("Switching to scancode set 2");
-			m_scancode_set_uncertain = false;
-			m_scancode_set = 2;
-			m_keymap.initialize(m_scancode_set);
-		}
-
 		m_byte_buffer[m_byte_index++] = byte;
 		if (byte == 0xE0)
 			return;

kernel/kernel/Lock/Mutex.cpp

@@ -1,134 +0,0 @@
-#include <kernel/Lock/Mutex.h>
-#include <kernel/Thread.h>
-
-namespace Kernel
-{
-
-	bool Mutex::try_lock()
-	{
-		const auto tid = Thread::current_tid();
-		if (tid == m_locker)
-			ASSERT(m_lock_depth > 0);
-		else
-		{
-			pid_t expected = -1;
-			if (!m_locker.compare_exchange(expected, tid))
-				return false;
-			ASSERT(m_lock_depth == 0);
-			if (tid)
-				Thread::current().add_mutex();
-		}
-		m_lock_depth++;
-		return true;
-	}
-
-	void Mutex::lock()
-	{
-		const auto tid = Thread::current_tid();
-		if (tid == m_locker)
-			ASSERT(m_lock_depth > 0);
-		else
-		{
-			ASSERT(!tid || !Thread::current().has_spinlock());
-			pid_t expected = -1;
-			while (!m_locker.compare_exchange(expected, tid))
-			{
-				ASSERT(Processor::get_interrupt_state() == InterruptState::Enabled);
-				Processor::yield();
-				expected = -1;
-			}
-			ASSERT(m_lock_depth == 0);
-			if (tid)
-				Thread::current().add_mutex();
-		}
-		m_lock_depth++;
-	}
-
-	void Mutex::unlock()
-	{
-		const auto tid = Thread::current_tid();
-		ASSERT(m_locker == tid);
-		ASSERT(m_lock_depth > 0);
-		if (--m_lock_depth == 0)
-		{
-			m_locker = -1;
-			if (tid)
-				Thread::current().remove_mutex();
-		}
-	}
-
-	bool Mutex::is_locked_by_current_thread() const
-	{
-		return m_locker == Thread::current_tid();
-	}
-
-	bool PriorityMutex::try_lock()
-	{
-		const auto tid = Thread::current_tid();
-
-		if (tid == m_locker)
-			ASSERT(m_lock_depth > 0);
-		else
-		{
-			bool has_priority = tid ? !Thread::current().is_userspace() : true;
-			pid_t expected = -1;
-			if (!(has_priority || m_queue_length == 0) || !m_locker.compare_exchange(expected, tid))
-				return false;
-			if (has_priority)
-				m_queue_length++;
-			ASSERT(m_lock_depth == 0);
-			if (tid)
-				Thread::current().add_mutex();
-		}
-		m_lock_depth++;
-		return true;
-	}
-
-	void PriorityMutex::lock()
-	{
-		const auto tid = Thread::current_tid();
-
-		if (tid == m_locker)
-			ASSERT(m_lock_depth > 0);
-		else
-		{
-			ASSERT(!tid || !Thread::current().has_spinlock());
-			bool has_priority = tid ? !Thread::current().is_userspace() : true;
-			if (has_priority)
-				m_queue_length++;
-			pid_t expected = -1;
-			while (!(has_priority || m_queue_length == 0) || !m_locker.compare_exchange(expected, tid))
-			{
-				ASSERT(Processor::get_interrupt_state() == InterruptState::Enabled);
-				Processor::yield();
-				expected = -1;
-			}
-			ASSERT(m_lock_depth == 0);
-			if (tid)
-				Thread::current().add_mutex();
-		}
-		m_lock_depth++;
-	}
-
-	void PriorityMutex::unlock()
-	{
-		const auto tid = Thread::current_tid();
-		ASSERT(m_locker == tid);
-		ASSERT(m_lock_depth > 0);
-		if (--m_lock_depth == 0)
-		{
-			bool has_priority = tid ? !Thread::current().is_userspace() : true;
-			if (has_priority)
-				m_queue_length--;
-			m_locker = -1;
-			if (tid)
-				Thread::current().remove_mutex();
-		}
-	}
-
-	bool PriorityMutex::is_locked_by_current_thread() const
-	{
-		return m_locker == Thread::current_tid();
-	}
-
-}

kernel/kernel/Lock/RWLock.cpp

@@ -1,58 +0,0 @@
-
-#include <kernel/Lock/BlockableSpinLock.h>
-#include <kernel/Lock/RWLock.h>
-#include <kernel/Thread.h>
-
-namespace Kernel
-{
-
-	void RWLock::rd_lock()
-	{
-		SpinLockGuard _(m_lock);
-		while (m_writers_waiting > 0 || m_writer != -1)
-		{
-			BlockableSpinLock block(m_lock);
-			m_thread_blocker.block_indefinite(&block);
-		}
-		m_readers_active++;
-	}
-
-	void RWLock::rd_unlock()
-	{
-		SpinLockGuard _(m_lock);
-		if (--m_readers_active == 0)
-			m_thread_blocker.unblock();
-	}
-
-	void RWLock::wr_lock()
-	{
-		if (m_writer == Thread::current_tid())
-		{
-			m_writer_depth++;
-			return;
-		}
-
-		SpinLockGuard _(m_lock);
-
-		m_writers_waiting++;
-		while (m_readers_active > 0 || m_writer != -1)
-		{
-			BlockableSpinLock block(m_lock);
-			m_thread_blocker.block_indefinite(&block);
-		}
-		m_writers_waiting--;
-
-		m_writer = Thread::current_tid();
-		m_writer_depth = 1;
-	}
-
-	void RWLock::wr_unlock()
-	{
-		if (--m_writer_depth != 0)
-			return;
-		SpinLockGuard _(m_lock);
-		m_writer = -1;
-		m_thread_blocker.unblock();
-	}
-
-}

kernel/kernel/Memory/FileBackedRegion.cpp

@@ -115,7 +115,7 @@ namespace Kernel
 			return {};
 
 		const vaddr_t first_page = BAN::Math::max(m_vaddr, address) & PAGE_ADDR_MASK;
-		const vaddr_t last_page = BAN::Math::div_round_up<vaddr_t>(BAN::Math::min(m_vaddr + m_size, address + size), PAGE_SIZE) * PAGE_SIZE;
+		const vaddr_t last_page = BAN::Math::div_round_up(BAN::Math::min(m_vaddr + m_size, address + size), PAGE_SIZE) * PAGE_SIZE;
 
 		RWLockRDGuard _(m_shared_data->rw_lock);
 		for (vaddr_t page_addr = first_page; page_addr < last_page; page_addr += PAGE_SIZE)

kernel/kernel/Memory/PhysicalRange.cpp

@@ -43,10 +43,11 @@ namespace Kernel
 			PageTable::with_per_cpu_fast_page(current_paddr, [&page_matched_bit](void* addr) {
 				for (size_t j = 0; j < PAGE_SIZE / sizeof(size_t); j++)
 				{
+					static_assert(sizeof(size_t) == sizeof(long));
 					auto& current = static_cast<size_t*>(addr)[j];
 					if (current == BAN::numeric_limits<size_t>::max())
 						continue;
-					const int ctz = BAN::Math::ctz(~current);
+					const int ctz = __builtin_ctzl(~current);
 					current |= static_cast<size_t>(1) << ctz;
 					page_matched_bit = j * sizeof(size_t) * 8 + ctz;
 					return;

kernel/kernel/Memory/SharedMemoryObject.cpp

@@ -1,44 +1,13 @@
 #include <kernel/Lock/LockGuard.h>
 #include <kernel/Memory/Heap.h>
 #include <kernel/Memory/SharedMemoryObject.h>
-#include <kernel/Process.h>
 #include <kernel/Random.h>
-#include <kernel/Timer/Timer.h>
-#include <kernel/UserCopy.h>
-
-#include <fcntl.h>
 
 namespace Kernel
 {
 
 	static BAN::UniqPtr<SharedMemoryObjectManager> s_instance;
 
-	class SharedMemoryObject : public MemoryRegion
-	{
-		BAN_NON_COPYABLE(SharedMemoryObject);
-		BAN_NON_MOVABLE(SharedMemoryObject);
-
-	public:
-		static BAN::ErrorOr<BAN::UniqPtr<SharedMemoryObject>> create(BAN::RefPtr<SharedMemoryObjectManager::Object>, PageTable&, AddressRange, int status_flags);
-		~SharedMemoryObject();
-
-		BAN::ErrorOr<BAN::UniqPtr<MemoryRegion>> clone(PageTable& new_page_table) override;
-		BAN::ErrorOr<BAN::UniqPtr<MemoryRegion>> split(size_t offset) override;
-
-		BAN::ErrorOr<void> msync(vaddr_t, size_t, int) override { return {}; }
-
-	protected:
-		BAN::ErrorOr<bool> allocate_page_containing_impl(vaddr_t vaddr, bool wants_write) override;
-
-	private:
-		SharedMemoryObject(BAN::RefPtr<SharedMemoryObjectManager::Object>, PageTable& page_table, int status_flags);
-
-	private:
-		BAN::RefPtr<SharedMemoryObjectManager::Object> m_object;
-
-		friend class BAN::UniqPtr<SharedMemoryObject>;
-	};
-
 	BAN::ErrorOr<void> SharedMemoryObjectManager::initialize()
 	{
 		ASSERT(!s_instance);
@@ -59,193 +28,70 @@ namespace Kernel
 				Heap::get().release_page(paddr);
 	}
 
-	BAN::ErrorOr<int> SharedMemoryObjectManager::shmget(key_t key, size_t size, int shmflg)
+	BAN::ErrorOr<SharedMemoryObjectManager::Key> SharedMemoryObjectManager::create_object(size_t size, PageTable::flags_t flags)
+	{
+		ASSERT(size % PAGE_SIZE == 0);
+
+		auto object = TRY(BAN::RefPtr<Object>::create());
+		object->size = size;
+		object->flags = flags;
+		TRY(object->paddrs.resize(size / PAGE_SIZE, 0));
+
+		LockGuard _(m_mutex);
+
+		// NOTE: don't set the top bit so cast to signed is not negative
+		auto generate_key = []() { return Random::get<Key>() & (~(Key)0 >> 1); };
+
+		Key key = generate_key();
+		while (m_objects.contains(key))
+			key = generate_key();
+		object->key = key;
+
+		TRY(m_objects.insert(key, object));
+		return key;
+	}
+
+	BAN::ErrorOr<void> SharedMemoryObjectManager::delete_object(Key key)
 	{
 		LockGuard _(m_mutex);
 
-		if (key != IPC_PRIVATE)
-		{
-			if (auto it = m_ids.find(key); it != m_ids.end())
-			{
-				if (shmflg & IPC_EXCL)
-					return BAN::Error::from_errno(EEXIST);
-
-				const auto& object = m_objects[it->value];
-				if (object->info.shm_segsz < size)
-					return BAN::Error::from_errno(EINVAL);
-
-				const int flags =
-					(shmflg & S_IRUSR ? O_RDONLY : 0) |
-					(shmflg & S_IWUSR ? O_WRONLY : 0);
-				if (!object->can_current_process_access(flags))
-					return BAN::Error::from_errno(EACCES);
-
-				return it->value;
-			}
-
-			if (!(shmflg & IPC_CREAT))
-				return BAN::Error::from_errno(ENOENT);
-		}
-
-		const auto& process = Process::current();
-		const uid_t uid = process.credentials().euid();
-		const gid_t gid = process.credentials().egid();
-		const pid_t pid = process.pid();
-		const mode_t mode = shmflg & 0777;
-
-		const int shmid = ({
-			int id;
-			do {
-				id = Random::get<unsigned>() & BAN::numeric_limits<int>::max();
-			} while (m_ids.contains(shmid));
-			id;
-		});
-
-		auto object = TRY(BAN::RefPtr<Object>::create(key, shmid, shmid_ds {
-				.shm_perm = {
-					.uid = uid,
-					.gid = gid,
-					.cuid = uid,
-					.cgid = uid,
-					.mode = mode,
-				},
-				.shm_segsz = size,
-				.shm_lpid = 0,
-				.shm_cpid = pid,
-				.shm_nattch = 0,
-				.shm_atime = 0,
-				.shm_dtime = 0,
-				.shm_ctime = SystemTimer::get().real_time().tv_sec,
-		}));
-		TRY(object->paddrs.resize(BAN::Math::div_round_up<size_t>(size, PAGE_SIZE), 0));
-
-		if (key != IPC_PRIVATE)
-			TRY(m_ids.insert(key, shmid));
-
-		if (auto ret = m_objects.insert(shmid, object); ret.is_error())
-		{
-			if (key != IPC_PRIVATE)
-				m_ids.remove(key);
-			return ret.release_error();
-		}
-
-		return shmid;
-	}
-
-	BAN::ErrorOr<void> SharedMemoryObjectManager::shmctl(int shmid, int cmd, struct shmid_ds* user_buf)
-	{
-		LockGuard _0(m_mutex);
-
-		auto it = m_objects.find(shmid);
-		if (it == m_objects.end())
-			return BAN::Error::from_errno(EINVAL);
-
-		LockGuard _1(it->value->mutex);
-
-		switch (cmd)
-		{
-			case IPC_RMID:
-				if (!it->value->can_current_process_access(O_WRONLY))
-					return BAN::Error::from_errno(EACCES);
-				if (it->key != IPC_PRIVATE)
-					m_ids.remove(it->key);
-				if (it->value->info.shm_nattch == 0)
-					m_objects.remove(it);
-				else
-					it->value->marked_for_deletion = true;
-				break;
-			case IPC_STAT:
-				TRY(write_to_user(user_buf, &it->value->info, sizeof(shmid_ds)));
-				break;
-			case IPC_SET:
-			{
-				if (!it->value->can_current_process_access(O_WRONLY))
-					return BAN::Error::from_errno(EACCES);
-				shmid_ds buf;
-				TRY(read_from_user(user_buf, &buf, sizeof(shmid_ds)));
-				it->value->info.shm_perm.uid  = buf.shm_perm.uid;
-				it->value->info.shm_perm.gid  = buf.shm_perm.gid;
-				it->value->info.shm_perm.mode = buf.shm_perm.mode & 0777;
-				break;
-			}
-			default:
-				return BAN::Error::from_errno(EINVAL);
-		}
-
-		return {};
-	}
-
-	BAN::ErrorOr<BAN::UniqPtr<MemoryRegion>> SharedMemoryObjectManager::shmat(int shmid, const void* shmaddr, int shmflg)
-	{
-		LockGuard _(m_mutex);
-
-		auto it = m_objects.find(shmid);
+		auto it = m_objects.find(key);
 		if (it == m_objects.end())
 			return BAN::Error::from_errno(ENOENT);
 
-		const int status_flags = (shmflg & SHM_RDONLY) ? O_RDONLY : O_RDWR;
-		if (!it->value->can_current_process_access(status_flags))
-			return BAN::Error::from_errno(EACCES);
-
-		const AddressRange address_range = (shmaddr == nullptr)
-			? AddressRange { .start = 0x400000, .end = USERSPACE_END }
-			: TRY(Process::current().find_free_address_range(it->value->info.shm_segsz));
-
-		auto region = TRY(SharedMemoryObject::create(it->value, Process::current().page_table(), address_range, status_flags));
-		return BAN::UniqPtr<MemoryRegion>(BAN::move(region));
+		m_objects.remove(it);
+		return {};
 	}
 
-	bool SharedMemoryObjectManager::Object::can_current_process_access(int flags) const
+	BAN::ErrorOr<BAN::UniqPtr<SharedMemoryObject>> SharedMemoryObjectManager::map_object(Key key, PageTable& page_table, AddressRange address_range)
 	{
-		const auto& creds = Process::current().credentials();
-		if (Inode::can_access(info.shm_perm.uid,  info.shm_perm.gid,  info.shm_perm.mode, creds, flags))
-			return true;
-		if (Inode::can_access(info.shm_perm.cuid, info.shm_perm.cgid, info.shm_perm.mode, creds, flags))
-			return true;
-		return false;
+		LockGuard _(m_mutex);
+
+		auto it = m_objects.find(key);
+		if (it == m_objects.end())
+			return BAN::Error::from_errno(ENOENT);
+
+		return TRY(SharedMemoryObject::create(it->value, page_table, address_range));
 	}
 
-	BAN::ErrorOr<BAN::UniqPtr<SharedMemoryObject>> SharedMemoryObject::create(BAN::RefPtr<SharedMemoryObjectManager::Object> object, PageTable& page_table, AddressRange address_range, int status_flags)
+	BAN::ErrorOr<BAN::UniqPtr<SharedMemoryObject>> SharedMemoryObject::create(BAN::RefPtr<SharedMemoryObjectManager::Object> object, PageTable& page_table, AddressRange address_range)
 	{
-		auto shared_memory_object = TRY(BAN::UniqPtr<SharedMemoryObject>::create(object, page_table, status_flags));
-		TRY(shared_memory_object->initialize(address_range));
-		return shared_memory_object;
-	}
-
-	SharedMemoryObject::SharedMemoryObject(BAN::RefPtr<SharedMemoryObjectManager::Object> object, PageTable& page_table, int status_flags)
-		: MemoryRegion(
-			page_table,
-			object->info.shm_segsz,
-			MemoryRegion::Type::SHARED,
-			PageTable::UserSupervisor | ((status_flags & O_WRONLY) ? PageTable::ReadWrite : 0) | PageTable::Present,
-			status_flags
-		), m_object(object)
-	{
-		LockGuard _(m_object->mutex);
-		m_object->info.shm_nattch++;
-		m_object->info.shm_atime = SystemTimer::get().real_time().tv_sec;
-		m_object->info.shm_lpid = Process::current().pid();
-	}
-
-	SharedMemoryObject::~SharedMemoryObject()
-	{
-		LockGuard _(m_object->mutex);
-		if (--m_object->info.shm_nattch == 0 && m_object->marked_for_deletion)
-			SharedMemoryObjectManager::get().m_objects.remove(m_object->id);
-		m_object->info.shm_dtime = SystemTimer::get().real_time().tv_sec;
-		m_object->info.shm_lpid = Process::current().pid();
+		auto smo = TRY(BAN::UniqPtr<SharedMemoryObject>::create(object, page_table));
+		TRY(smo->initialize(address_range));
+		return BAN::move(smo);
 	}
 
 	BAN::ErrorOr<BAN::UniqPtr<MemoryRegion>> SharedMemoryObject::clone(PageTable& new_page_table)
 	{
-		auto region = TRY(SharedMemoryObject::create(m_object, new_page_table, { .start = vaddr(), .end = vaddr() + size() }, m_status_flags));
+		auto region = TRY(SharedMemoryObject::create(m_object, new_page_table, { .start = vaddr(), .end = vaddr() + size() }));
 		return BAN::UniqPtr<MemoryRegion>(BAN::move(region));
 	}
 
-	BAN::ErrorOr<BAN::UniqPtr<MemoryRegion>> SharedMemoryObject::split(size_t)
+	BAN::ErrorOr<BAN::UniqPtr<MemoryRegion>> SharedMemoryObject::split(size_t offset)
 	{
-		derrorln("SharedMemoryObjects are not splittable");
-		return BAN::Error::from_errno(EINVAL);
+		(void)offset;
+		dwarnln("TODO: SharedMemoryObject::split");
+		return BAN::Error::from_errno(ENOTSUP);
 	}
 
 	BAN::ErrorOr<bool> SharedMemoryObject::allocate_page_containing_impl(vaddr_t address, bool wants_write)
@@ -253,11 +99,12 @@ namespace Kernel
 		ASSERT(contains(address));
 		(void)wants_write;
 
-		const vaddr_t vaddr = address & PAGE_ADDR_MASK;
+		// Check if address is already mapped
+		vaddr_t vaddr = address & PAGE_ADDR_MASK;
 		if (m_page_table.physical_address_of(vaddr) != 0)
 			return false;
 
-		LockGuard _(m_object->mutex);
+		SpinLockGuard _(m_object->spin_lock);
 
 		paddr_t paddr = m_object->paddrs[(vaddr - m_vaddr) / PAGE_SIZE];
 		if (paddr == 0)
@@ -270,7 +117,6 @@ namespace Kernel
 			});
 			m_object->paddrs[(vaddr - m_vaddr) / PAGE_SIZE] = paddr;
 		}
-
 		m_page_table.map_page_at(paddr, vaddr, m_flags);
 
 		return true;

kernel/kernel/Memory/kmalloc.cpp

@@ -125,6 +125,8 @@ struct BitmapAllocator
 		// NOTE: We could optimize other bitmap functions than this
 		//       but this one is the bottle neck so it doesn't matter
 
+		static_assert(sizeof(unsigned long long) == sizeof(uint64_t));
+
 		if (index >= total_chunks)
 			return index;
 
@@ -132,7 +134,7 @@ struct BitmapAllocator
 		{
 			const uint64_t qword = *reinterpret_cast<const uint64_t*>(base + (index - rem) / 8) >> rem;
 			if (qword != (1ull << (64 - rem)) - 1)
-				return index + BAN::Math::ctz(~qword);
+				return index + __builtin_ctzll(~qword);
 			index += 64 - rem;
 		}
 
@@ -140,7 +142,7 @@ struct BitmapAllocator
 		{
 			const uint64_t qword = *reinterpret_cast<const uint64_t*>(base + index / 8);
 			if (qword != UINT64_MAX)
-				return index + BAN::Math::ctz(~qword);
+				return index + __builtin_ctzll(~qword);
 			index += 64;
 		}
 

kernel/kernel/Networking/ARPTable.cpp

@@ -1,3 +1,4 @@
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/Networking/ARPTable.h>
 #include <kernel/Scheduler.h>
 #include <kernel/Timer/Timer.h>

kernel/kernel/Networking/E1000/E1000.cpp

@@ -1,13 +1,11 @@
 #include <kernel/IDT.h>
 #include <kernel/InterruptController.h>
 #include <kernel/IO.h>
-#include <kernel/Lock/BlockableSpinLock.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/Memory/PageTable.h>
 #include <kernel/MMIO.h>
 #include <kernel/Networking/E1000/E1000.h>
 #include <kernel/Networking/NetworkManager.h>
-#include <kernel/Scheduler.h>
-#include <kernel/Thread.h>
 
 namespace Kernel
 {
@@ -382,8 +380,8 @@ namespace Kernel
 			if (rx_current != rx_tail)
 				write32(REG_RDT0, rx_tail);
 
-			BlockableSpinLock block(m_rx_lock);
-			m_rx_blocker.block_indefinite(&block);
+			SpinLockGuardAsMutex smutex(guard);
+			m_rx_blocker.block_indefinite(&smutex);
 		}
 
 		m_thread_is_dead = true;

kernel/kernel/Networking/IPv4Layer.cpp

@@ -1,6 +1,6 @@
-#include <kernel/Lock/SpinLock.h>
 #include <kernel/Memory/Heap.h>
 #include <kernel/Memory/PageTable.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/Networking/ICMP.h>
 #include <kernel/Networking/IPv4Layer.h>
 #include <kernel/Networking/NetworkManager.h>

kernel/kernel/Networking/Loopback.cpp

@@ -1,8 +1,6 @@
-#include <kernel/Lock/BlockableSpinLock.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/Networking/Loopback.h>
 #include <kernel/Networking/NetworkManager.h>
-#include <kernel/Scheduler.h>
-#include <kernel/Thread.h>
 
 namespace Kernel
 {
@@ -74,8 +72,8 @@ namespace Kernel
 						descriptor.state = 1;
 						return descriptor;
 					}
-					BlockableSpinLock block(m_buffer_lock);
-					m_thread_blocker.block_indefinite(&block);
+					SpinLockGuardAsMutex smutex(guard);
+					m_thread_blocker.block_indefinite(&smutex);
 				}
 			}();
 
@@ -120,8 +118,8 @@ namespace Kernel
 				m_thread_blocker.unblock();
 			}
 
-			BlockableSpinLock block(m_buffer_lock);
-			m_thread_blocker.block_indefinite(&block);
+			SpinLockGuardAsMutex smutex(guard);
+			m_thread_blocker.block_indefinite(&smutex);
 		}
 
 		m_thread_is_dead = true;

kernel/kernel/Networking/RTL8169/RTL8169.cpp

@@ -1,9 +1,7 @@
-#include <kernel/Lock/BlockableSpinLock.h>
+#include <kernel/Lock/SpinLockAsMutex.h>
 #include <kernel/Networking/NetworkManager.h>
 #include <kernel/Networking/RTL8169/Definitions.h>
 #include <kernel/Networking/RTL8169/RTL8169.h>
-#include <kernel/Scheduler.h>
-#include <kernel/Thread.h>
 #include <kernel/Timer/Timer.h>
 
 namespace Kernel
@@ -220,8 +218,8 @@ namespace Kernel
 			SpinLockGuard guard(m_tx_lock);
 			while (descriptor.command & RTL8169_DESC_CMD_OWN)
 			{
-				BlockableSpinLock block(m_tx_lock);
-				m_tx_blocker.block_indefinite(&block);
+				SpinLockGuardAsMutex smutex(guard);
+				m_tx_blocker.block_indefinite(&smutex);
 			}
 		}
 
@@ -310,8 +308,8 @@ namespace Kernel
 				m_rx_head = (m_rx_head + 1) % m_rx_descriptor_count;
 			}
 
-			BlockableSpinLock block(m_rx_lock);
-			m_rx_blocker.block_indefinite(&block);
+			SpinLockGuardAsMutex smutex(rx_lock_guard);
+			m_rx_blocker.block_indefinite(&smutex);
 		}
 
 		m_rx_thread_is_dead = true;

kernel/kernel/Networking/TCPSocket.cpp

@@ -3,7 +3,6 @@
 #include <kernel/Networking/TCPSocket.h>
 #include <kernel/Process.h>
 #include <kernel/Random.h>
-#include <kernel/Scheduler.h>
 #include <kernel/Timer/Timer.h>
 
 #include <fcntl.h>