WIP use long for addresses

aeneas/src/arm64/Arm64AddrPatcher.v3

@@ -14,7 +14,7 @@ class Arm64AddrPatcher {
 		// TODO
 	}
 
-	def patch(kind: Arm64PatchKind, posAddr: int, absAddr: int) {
+	def patch(kind: Arm64PatchKind, posAddr: long, absAddr: long) {
 		def pos = mw.pos;
 		match (kind) {
 			ABS_IMM16 => asm.patch_ABS_i16(pos, i16.view(absAddr));

aeneas/src/exe/MachO.v3

@@ -83,8 +83,8 @@ class MachO_LoadCmd {
 }
 // A load segment command
 class MachO_SegmentLoad(segname: string) extends MachO_LoadCmd {
-	var vmaddr: int;
-	var vmsize: int;
+	var vmaddr: long;
+	var vmsize: long;
 	var fileoff: int;
 	var filesize: int;
 	var maxprot: int;
@@ -106,8 +106,8 @@ class MachO_SegmentLoad(segname: string) extends MachO_LoadCmd {
 			w.put_b64(fileoff);
 			w.put_b64(filesize);
 		} else {
-			w.put_b32(vmaddr);
-			w.put_b32(vmsize);
+			w.put_b32(int.!(vmaddr));
+			w.put_b32(int.!(vmsize));
 			w.put_b32(fileoff);
 			w.put_b32(filesize);
 		}

aeneas/src/mach/MachBackend.v3

@@ -743,17 +743,20 @@ class SsaMachGen(context: SsaContext, mach: MachProgram, regSet: MachRegSet, w:
 	//======================================================================
 	// Helper routines for emitting instructions.
 	//======================================================================
+	def matchImm(i: SsaInstr) -> bool {
+		return SsaConst.?(i); // may be overridden in subclasses
+	}
 	def matchAddImm(i: SsaInstr, disp: Val) -> (SsaInstr, Val) {
 		// TODO: check covering condition: in same block and single use
 		if (!SsaApplyOp.?(i)) return (i, disp);
 		var apply = SsaApplyOp.!(i);
 		if (Opcode.IntAdd.?(apply.op.opcode) || Opcode.PtrAdd.?(apply.op.opcode)) {
 			var x = apply.input0(), y = apply.input1();
-			if (SsaConst.?(x)) {
+			if (matchImm(x)) {
 				var disp2 = tryAddConsts(disp, SsaConst.!(x).val);
 				if (disp2 != null) return matchAddImm(y, disp2);
 			}
-			if (SsaConst.?(y)) {
+			if (matchImm(y)) {
 				var disp2 = tryAddConsts(disp, SsaConst.!(y).val);
 				if (disp2 != null) return matchAddImm(x, disp2);
 			}

aeneas/src/mach/MachDataWriter.v3

@@ -18,9 +18,9 @@ enum PatchKind {
 class MachDataWriter extends DataWriter {
 	def pageAlign: Alignment;
 	var startPos: int; // start position from which offset() is calculated
-	var patches: List<(PatchKind, Addr, /*posAddr:*/int, /*pos:*/int)>;
+	var patches: List<(PatchKind, Addr, /*posAddr:*/long, /*pos:*/int)>;
 
-	new(pageAlign, startAddr: int, initialSize: int) {
+	new(pageAlign, startAddr: long, initialSize: int) {
 		this.startAddr = u64.!(startAddr);
 		grow(initialSize);
 	}
@@ -49,21 +49,21 @@ class MachDataWriter extends DataWriter {
 			at(pos);
 			match (t.0) {
 				REL_32 => {
-					put_b32(absAddr - posAddr);
+					put_b32(int.!(absAddr - posAddr));
 				}
 				RELA_32 => {
 					var delta = i8.view(data[pos + 1]);
-					put_b32(absAddr - (posAddr + delta));
+					put_b32(int.!(absAddr - (posAddr + delta)));
 				}
 				ABS_32 => {
-					put_b32(absAddr);
+					put_b32(int.!(absAddr));
 				}
 				ABS_64 => {
 					put_b64(absAddr);
 
 				}
 				ABS_SLEB32 => {
-					overwrite_sleb32(absAddr);
+					overwrite_sleb32(int.!(absAddr));
 				}
 				_ => mach.prog.ERROR.fail1("unimplemented patch code: %s", kind.name);
 			}
@@ -76,8 +76,8 @@ class MachDataWriter extends DataWriter {
 	def skipPage() {
 		skipAddr(pageAlign.size);
 	}
-	def endPageAddr() -> int {
-		return pageAlign.alignDown_i32(addr_end());
+	def endPageAddr() -> long {
+		return pageAlign.alignDown_i64(addr_end());
 	}
 	def putmd(w: MachDataWriter) {
 		var p = this.pos;

aeneas/src/mach/MachProgram.v3

@@ -476,11 +476,11 @@ class MachProgram extends TargetProgram {
 		}
 		encodeAbs(w, space, abs);
 	}
-	def encodeAbs(w: DataWriter, space: AddressSpace, abs: int) {
+	def encodeAbs(w: DataWriter, space: AddressSpace, abs: long) {
 		match (space.addressSize) {
-			1 => w.putb(abs);
-			2 => w.put_b16(abs);
-			4 => w.put_b32(abs);
+			1 => w.putb(u8.!(abs));
+			2 => w.put_b16(u16.!(abs));
+			4 => w.put_b32(int.!(abs));
 			8 => w.put_b64(abs);
 		}
 	}
@@ -519,7 +519,7 @@ class MachProgram extends TargetProgram {
 		if (a == null) methodMap[m] = a = Address.new(codeRegion, m);
 		return a;
 	}
-	def absolute(a: Addr) -> int {
+	def absolute(a: Addr) -> long {
 		if (a == null) return 0;
 		if (a.base != null) return absolute(a.base) + a.offset;
 		if (CiRuntime_Address.?(a)) return runtime.getAddr(CiRuntime_Address.!(a));
@@ -682,7 +682,7 @@ class MachRtPageTable(
 
 	def encode(rt: MachRuntime, w: MachDataWriter) {
 		var pageTableAddr = rt.bindAddr(pageTable, w);
-		var maxPage = rt.codePages(), maxEntry = entries.length;
+		var maxPage = int.!(rt.codePages()), maxEntry = entries.length;
 		var tableSize = 4 * (maxPage + 1);
 		rt.setAddr(table, pageTableAddr + tableSize);
 		// encode the page table offsets

aeneas/src/mach/MachRtSrcTables.v3

@@ -164,7 +164,7 @@ class MachRtSrcTables(mach: MachProgram, rt: MachRuntime) {
 		return off;
 	}
 	// set absolute addresses for addresses in ex region and return the size
-	def layoutExRegion(startAddr: int) -> int {
+	def layoutExRegion(startAddr: long) -> int {
 		for (l = exDests; l != null; l = l.tail) {
 			l.head.absolute = startAddr + l.head.offset;
 		}

aeneas/src/mach/MachRuntime.v3

@@ -4,7 +4,7 @@
 // Represents all runtime data structures exposed to the Virgil runtime code.
 class MachRuntime(mach: MachProgram) {
 	def codeRegion = mach.codeRegion;
-	def addrs = Array<int>.new(CiRuntimeModule.max); // CiRuntime absolute addresses
+	def addrs = Array<long>.new(CiRuntimeModule.max); // CiRuntime absolute addresses
 	var typeCon: CiRuntime_TypeCon;
 
 	// Indexes in the IrModule's roots vector
@@ -38,41 +38,41 @@ class MachRuntime(mach: MachProgram) {
 		}
 	}
 
-	def recordStackStart(addr: int) {
+	def recordStackStart(addr: long) {
 		setAddr(CiRuntimeModule.STACK_START, addr);
 	}
-	def recordStackEnd(addr: int) {
+	def recordStackEnd(addr: long) {
 		setAddr(CiRuntimeModule.STACK_END, addr);
 	}
-	def recordCodeStart(addr: int) {
+	def recordCodeStart(addr: long) {
 		setAddr(CiRuntimeModule.CODE_START, addr);
 	}
-	def recordReservedCode(startAddr: int, size: int, fileOffset: int) {
+	def recordReservedCode(startAddr: long, size: long, fileOffset: int) {
 		setAddr(CiRuntimeModule.RESERVED_CODE_START, startAddr);
 		setAddr(CiRuntimeModule.RESERVED_CODE_END, startAddr + size);
 		setAddr(CiRuntimeModule.RESERVED_CODE_FILE_OFFSET, fileOffset);
 	}
-	def recordRuntimeCode(startAddr: int, size: int) {
+	def recordRuntimeCode(startAddr: long, size: long) {
 		setAddr(CiRuntimeModule.RUNTIME_CODE_START, startAddr);
 		setAddr(CiRuntimeModule.RUNTIME_CODE_END, startAddr + size);
 	}
-	def recordCodeEnd(addr: int) {
+	def recordCodeEnd(addr: long) {
 		setAddr(CiRuntimeModule.CODE_END, addr);
 	}
-	def recordDataStart(addr: int) {
+	def recordDataStart(addr: long) {
 		setAddr(CiRuntimeModule.DATA_START, addr);
 	}
-	def recordDataEnd(addr: int) {
+	def recordDataEnd(addr: long) {
 		setAddr(CiRuntimeModule.DATA_END, addr);
 	}
-	def setAddr(c: CiRuntime_Address, addr: int) {
+	def setAddr(c: CiRuntime_Address, addr: long) {
 		if (CLOptions.PRINT_BIN.get()) Terminal.put2("CiRuntime.%s=%x\n", c.name, addr);
 		addrs[c.index] = addr;
 	}
-	def getAddr(c: CiRuntime_Address) -> int {
+	def getAddr(c: CiRuntime_Address) -> long {
 		return addrs[c.index];
 	}
-	def bindAddr(c: CiRuntime_Address, w: DataWriter) -> int {
+	def bindAddr(c: CiRuntime_Address, w: DataWriter) -> long {
 		var addr = w.addr_end();
 		setAddr(c, addr);
 		return addr;
@@ -109,26 +109,26 @@ class MachRuntime(mach: MachProgram) {
 		bindAddr(ptr, w);
 		w.skipN(mach.data.addressSize);
 	}
-	def setPtr(w: MachDataWriter, ptr: CiRuntime_Address, addr: int) {
+	def setPtr(w: MachDataWriter, ptr: CiRuntime_Address, addr: long) {
 		var abs = addrs[ptr.index];
 		if (CLOptions.PRINT_BIN.get()) {
 			Terminal.put3("CiRuntime.%s @ %x = %x\n", ptr.name, abs, addr);
 		}
 		w.at(int.!(u64.view(abs) - w.startAddr)); // TODO(addr64)
 		match (mach.data.addressSize) {
-			1 => w.putb(addr);
-			2 => w.put_b16(addr);
-			4 => w.put_b32(addr);
+			1 => w.putb(u8.!(addr));
+			2 => w.put_b16(u16.!(addr));
+			4 => w.put_b32(int.!(addr));
 			8 => w.put_b64(addr);
 			_ => { return V3.fail("invalid data address size"); }
 		}
 	}
-	def codePages() -> int {
+	def codePages() -> long { // TODO(addr64) int is probably sufficient
 		var end = getAddr(CiRuntimeModule.CODE_END);
 		if (end == 0) mach.prog.ERROR.fail("code end not set yet");
 		var size = end - getAddr(CiRuntimeModule.CODE_START);
 		var align = codeRegion.space.pageAlign;
-		return align.alignUp_i32(size) >> align.shift_u5;
+		return align.alignUp_i64(size) >> align.shift_u6;
 	}
 	def encode(w: MachDataWriter) {
 		if (gc != null) gc.encode(w);

aeneas/src/mach/Memory.v3

@@ -2,7 +2,7 @@
 // See LICENSE for details of Apache 2.0 license.
 
 def pname(width: int) -> string {
-	return Strings.builderOf("$p").putd(width).toString();
+	return Strings.builderOf("#ptr").putd(width).toString();
 }
 class AddressSpace {
 	def name: string;		// name of address space
@@ -22,6 +22,7 @@ class AddressSpace {
 // the offsets for each allocation for later debugging.
 class Region(name: string, space: AddressSpace) {
 	def contents = Vector<(Addr, int, int)>.new();
+	var refSize = space.addressSize;
 	var mark: int;		// the current end of the allocations
 
 	// align the current allocation mark to the specified size
@@ -39,7 +40,7 @@ class Region(name: string, space: AddressSpace) {
 	def apply(f: (Addr, int, int) -> void) {
 		contents.apply(f);
 	}
-	def setStartAddr(startAddr: int) -> int {
+	def setStartAddr(startAddr: long) -> int {
 		// set all addresses
 		for (i < contents.length) {
 			var t = contents[i];
@@ -50,8 +51,8 @@ class Region(name: string, space: AddressSpace) {
 }
 // Addr represents any kind of address
 class Addr(region: Region, base: Addr, offset: int) extends Val {
-	var absolute: int = -1;	// absolute address if this is a base address
-	var size: int;
+	var absolute: long = -1;	// absolute address if this is a base address
+	var size: long;
 
 	def equals(val: Val) -> bool {
 		if (Addr.?(val)) {

aeneas/src/os/Linux.v3

@@ -50,6 +50,9 @@ class LinuxTarget extends Target {
 	}
 	def configureProgram(prog: Program) {
 		var mach = MachProgram.new(prog, space, space, intNorm);
+		var vaddr_start = long.!(CLOptions.VM_START_ADDR.get()); // TODO(addr64)
+		if (vaddr_start == 0) vaddr_start = DEFAULT_VADDR_START;
+		if (vaddr_start < 3_000_000_000L) mach.codeRegion.refSize = 4;
 		prog.tprog = mach;
 		var rt = MachRuntime.new(mach);
 		mach.runtime = rt;
@@ -77,7 +80,7 @@ class LinuxTarget extends Target {
 
 		var pageAlign = mach.code.pageAlign;
 
-		var vaddr_start = int.view(CLOptions.VM_START_ADDR.get());
+		var vaddr_start = long.!(CLOptions.VM_START_ADDR.get()); // TODO(addr64)
 		if (vaddr_start == 0) vaddr_start = DEFAULT_VADDR_START;
 		var w = MachDataWriter.new(mach.code.pageAlign, vaddr_start, 300);
 
@@ -91,9 +94,9 @@ class LinuxTarget extends Target {
 			stack.p_memsz = stackSize;
 			stack.p_flags = ElfConst.PF_RW;
 			stack.p_vaddr = long.!(w.startAddr);		// TODO(addr64)
-			rt.recordStackStart(int.!(w.startAddr));	// TODO(addr64)
+			rt.recordStackStart(long.!(w.startAddr));	// TODO(addr64)
 			w.skipAddr(stackSize);				// TODO(addr64)
-			rt.recordStackEnd(int.!(w.startAddr));		// TODO(addr64)
+			rt.recordStackEnd(long.!(w.startAddr));		// TODO(addr64)
 		}
 
 		// fill out the code section
@@ -133,7 +136,7 @@ class LinuxTarget extends Target {
 
 		if (rtexe != null) {
 			// TODO: add .runtime_code symbol
-			var startAddr = pageAlign.alignUp_i32(w.addr_end());
+			var startAddr = pageAlign.alignUp_i64(w.addr_end());
 			rt.recordRuntimeCode(startAddr, runtimeCodeSize);
 			rtexe.p_type = ElfPhType.PT_LOAD;
 			rtexe.p_filesz = 1; // file size cannot be zero, apparently
@@ -156,7 +159,7 @@ class LinuxTarget extends Target {
 		if (ex != null) {
 			w.skipPage();
 			var exStart = w.endPageAddr();
-			var exSize = pageAlign.alignUp_i32(rt.src.layoutExRegion(exStart));
+			var exSize = pageAlign.alignUp_i64(rt.src.layoutExRegion(exStart));
 			w.skipAddr(exSize);
 			ex.p_vaddr = exStart;
 			ex.p_memsz = exSize;

aeneas/src/ssa/SsaOptimizer.v3

@@ -1157,6 +1157,7 @@ class SsaInstrReducer(context: SsaContext) extends SsaInstrMatcher {
 				nullcheck(i);
 				state.kill();
 			}
+			// XXX: PtrAdd, PtrSub, PtrLt
 			PtrLoad => {
 				var ptr = nullcheck(i);
 				// XXX: use Fact.O_FOLDABLE to indicate a load of an immutable value

aeneas/src/util/Alignment.v3

@@ -12,6 +12,9 @@ class Alignment(size: u31) {
 	def alignDown_i32(i: int) -> int {
 		return i & int.view(mask_u32);
 	}
+	def alignDown_i64(i: long) -> long {
+		return i & long.view(mask_u64);
+	}
 	def alignUp_i32(i: int) -> int {
 		return (i + add) & int.view(mask_u32);
 	}

aeneas/src/wasm/WasmCodeGen.v3

@@ -1328,7 +1328,7 @@ class WasmCodeGen extends SsaMachGen {
 							}
 						}
 						x: Addr => {
-							var abs = x.absolute;
+							var abs = int.!(x.absolute);
 							if (abs < 0) {
 								context.fail1("pushValue() no global for Addr of type %q", t.render);
 							} else {
@@ -1786,7 +1786,7 @@ class WasmCodeGen extends SsaMachGen {
 							}
 						}
 						x: Addr => {
-							var abs = x.absolute;
+							var abs = int.!(x.absolute);
 							if (abs < 0) {
 								context.fail1("genLoadConst() no global for Addr of type %q", t.render);
 							} else {

aeneas/src/wasm/WasmGcTarget.v3

@@ -1241,7 +1241,7 @@ class WasmGcTarget extends Target {
 		return size;
 	}
 	def patchAddress(mach: MachProgram, w: MachDataWriter, a: Addr, posAddr: int) {
-		var abs = mach.absolute(a);
+		var abs = int.!(mach.absolute(a));
 		if (CLOptions.PRINT_BIN.get()) {
 			TerminalBuffer.new()
 				.puts("patch-address @+")

aeneas/src/wasm/WasmTarget.v3

@@ -366,21 +366,6 @@ class WasmTarget extends Target {
 		System.fileClose(fd);
 		return out;
 	}
-	def patchAddress(mach: MachProgram, w: MachDataWriter, kind: PatchKind, a: Addr, posAddr: int) {
-		var abs = mach.absolute(a);
-		if (CLOptions.PRINT_BIN.get()) {
-			TerminalBuffer.new()
-				.puts("patch-address @+")
-				.putd(w.pos)
-				.puts(" <- ")
-				.putcv(a, null)
-				.puts(" = ")
-				.putx(abs)
-				.outln();
-		}
-		// TODO,XXX: address immediates are in i32.const, which is a signed LEB.
-		w.overwrite_sleb32(abs);
-	}
 	def emitTestWrappers(context: SsaContext, rt: MachRuntime, w: DataWriter, e: ExecuteTestCase) {
 		var prog = context.prog;
 		var fileName = context.compiler.getOutputFileName(prog.name(), "expect.js");