system¶

Kernel base type definitions, FPC compiler runtime stubs, and system initialisation.

Overview¶

This is the Free Pascal system unit for the Asuro kernel. It provides three categories of functionality:

Base types — Fundamental integer, pointer, and composite types used throughout the kernel, including fixed-width integer aliases (uint8 through uint64), signed counterparts, bit-width-constrained subrange types (UBit1–UBit31), and types required by the FPC 3.2.2 compiler internals (RTTI, exception frames, GUIDs, file records).
FPC compiler runtime stubs — Implementations of the compilerproc procedures that FPC generates calls to when compiling Pascal code: program initialisation and exit, error handling, range and overflow checks, I/O checks, exception handling (setjmp/longjmp), short string assignment, 64-bit integer arithmetic (multiply, divide, modulo for both signed and unsigned), and memory allocation. All stubs are implemented for a bare-metal environment with no operating system beneath them.
System initialisation — Captures the kernel start and end addresses from linker-defined symbols, and initialises FPC runtime globals.

Constants¶

KERNEL_VIRTUAL_BASE¶

$C0000000 — The virtual address at which the kernel is linked (higher-half at 3 GiB). Physical addresses are converted to kernel virtual addresses by adding this offset.

KERNEL_PAGE_NUMBER¶

KERNEL_VIRTUAL_BASE shr 22 = 768 — The page directory index corresponding to the start of the kernel virtual address space.

BSOD_ENABLE¶

true — Compile-time flag enabling the BSOD (Blue Screen of Death) panic display.

TRACER_ENABLE¶

true — Compile-time flag enabling the call tracer (debug.tracer).

Types¶

Integer and Cardinal Types¶

Type	Underlying Type	Description
cardinal	0..$FFFFFFFF	Unsigned 32-bit integer
hresult	longint	COM-style result code
dword	cardinal	Alias for cardinal
integer	longint	Signed 32-bit integer
uint8	BYTE	Unsigned 8-bit integer
uint16	WORD	Unsigned 16-bit integer
uint32	DWORD	Unsigned 32-bit integer
uint64	QWORD	Unsigned 64-bit integer
sInt8	shortint	Signed 8-bit integer
sInt16	smallint	Signed 16-bit integer
sInt32	integer	Signed 32-bit integer
sInt64	int64	Signed 64-bit integer
Float	Single	32-bit floating-point

uint128¶

A 128-bit value as a variant record with four views: two 64-bit halves (Hi, Lo), four 32-bit dwords, eight 16-bit words, and sixteen bytes.

Pointer Types¶

PuByte, PuInt8, PuInt16, PuInt32, PuInt64, PuInt128, PsInt8, PsInt16, PsInt32, PsInt64, PFloat, PDouble — Pointer variants of the corresponding numeric types.

Void¶

Void = ^uInt32 — Generic untyped pointer used as the parameter type for ISR hook callbacks.

Coordinate Types¶

yord, xord — uint8 aliases for Y and X screen coordinates. zord — uint16 alias for a combined coordinate.

UBit Types¶

UBit1 through UBit31 — Subrange types constrained to N-bit unsigned values (0 to 2^N - 1). Used in bitpacked records for hardware register mapping. Note: UBit8 is absent (use uint8 directly).

TBitMask / PBitMask¶

A bitpacked record of 8 Boolean fields (b0–b7), providing individual bit access to a byte.

TMask / PMask¶

A bitpacked array of 8 Booleans.

FPC Compiler-Required Types¶

Type	Description
SizeInt	Signed pointer-sized integer (longint on 32-bit)
SizeUInt	Unsigned pointer-sized integer (cardinal on 32-bit)
PtrInt	Signed pointer-to-integer type
PtrUInt	Unsigned pointer-to-integer type
TTypeKind	RTTI type kind enumeration
jmp_buf	Non-local jump buffer for i386 (EBX, ESI, EDI, BP, SP, PC)
TExceptAddr	Exception address stack entry (buf, next, frametype)
TGuid	16-byte GUID/UUID in three variant layouts
FileRec	Internal file descriptor record
TextRec	Internal text file record with buffer

Variables¶

AK_START / AK_END¶

var AK_START : uint32; external name 'kernel_start';
var AK_END   : uint32; external name 'kernel_end';

Linker-defined symbols marking the physical start and end of the kernel image.

ASURO_KERNEL_START / ASURO_KERNEL_END / ASURO_KERNEL_SIZE¶

Set by init to the virtual addresses corresponding to AK_START and AK_END, and the size in bytes.

FPC Runtime Globals¶

ExceptAddrStack, ExitCode, ErrorAddr, ErrorCode, ExitProc, StackBottom, StackLength, RandSeed — Standard FPC system unit globals initialised to safe defaults by init.

Functions and Procedures¶

init¶

procedure init();

Initialises kernel size variables from linker symbols and resets all FPC runtime globals to their zero/nil defaults.

move¶

procedure move(const source; var dest; count: SizeInt);

Copies count bytes from source to dest. Handles overlapping regions by copying backward when dest is at a higher address than source. Alias: FPC_MOVE.

Trunc¶

function Trunc(d: Double): int64;

Converts a double-precision floating-point value to a 64-bit integer by truncating toward zero using the x87 FPU. Alias: FPC_TRUNC.

Round¶

function Round(d: Double): int64;

Converts a double-precision floating-point value to a 64-bit integer using the FPU default round-to-nearest mode. Alias: FPC_ROUND.

FPC Compiler Procedure Stubs¶

All compilerproc stubs are mapped to their standard FPC_* aliases:

Procedure	Alias	Description
fpc_initializeunits	FPC_INITIALIZEUNITS	No-op; boot sequence handles unit init
fpc_do_exit	FPC_DO_EXIT	CLI + HLT
fpc_handleerror	FPC_HANDLEERROR	Sets ErrorCode, CLI + HLT
fpc_rangeerror	FPC_RANGEERROR	HandleErrorInternal(201)
fpc_overflow	FPC_OVERFLOW	HandleErrorInternal(215)
fpc_divbyzero	FPC_DIVBYZERO	HandleErrorInternal(200)
fpc_objecterror	FPC_OBJECTERROR	HandleErrorInternal(210)
fpc_abstracterror	FPC_ABSTRACTERROR	HandleErrorInternal(211)
fpc_stackcheck	FPC_STACKCHECK	No-op
fpc_iocheck	FPC_IOCHECK	No-op
fpc_pushexceptaddr	FPC_PUSHEXCEPTADDR	Pushes onto ExceptAddrStack
fpc_popaddrstack	FPC_POPADDRSTACK	Pops from ExceptAddrStack
fpc_setjmp	FPC_SETJMP	Saves EBX/ESI/EDI/EBP/ESP/EIP into jmp_buf
fpc_longjmp	FPC_LONGJMP	Restores registers from jmp_buf
fpc_shortstr_assign	FPC_SHORTSTR_ASSIGN	Copies short string with length clamping
fpc_mul_int64	FPC_MUL_INT64	64×64→64 multiply via three 32-bit MUL instructions
fpc_div_int64	FPC_DIV_INT64	Signed 64÷64 division with fast 32-bit path
fpc_mod_int64	FPC_MOD_INT64	Signed 64 modulo with fast 32-bit path
fpc_div_qword	FPC_DIV_QWORD	Unsigned 64÷64 division with fast 32-bit path
fpc_mod_qword	FPC_MOD_QWORD	Unsigned 64 modulo with fast 32-bit path
fpc_getmem	FPC_GETMEM	Delegates to kernel_kalloc
fpc_freemem	FPC_FREEMEM	Delegates to kernel_kfree
fpc_reraise	FPC_RERAISE	HandleErrorInternal(217)
fpc_raiseexception	FPC_RAISEEXCEPTION	HandleErrorInternal(217)
fpc_catches	FPC_CATCHES	Returns nil
fpc_doneexception	FPC_DONEEXCEPTION	No-op

Notes¶

The 64-bit software division (udiv64) uses a shift-subtract binary long division algorithm. A fast path using the hardware 32-bit DIV instruction is taken when the divisor fits in 32 bits.
The serial debug helpers (serial_putch, serial_puthex8, serial_puthex32, serial_putstr) are internal and not exported. They write directly to COM1 (port $3F8) without any dependency on the kernel I/O subsystem, making them safe for very early boot diagnostics.
fpc_getmem and fpc_freemem are linked to external C-convention symbols kernel_kalloc and kernel_kfree defined in the heap allocator unit. This allows the FPC New/Dispose operators and dynamic arrays to use the kernel heap.