feat(verify): add Call trampolines for arm64, riscv64, loong64

Assisted-by: MiMo V2.5 Pro

debug/target_linux_arm64.go

@@ -14,6 +14,7 @@ import (
 
 	"sourcedock.dev/petrbalvin/gasm-devkit/asm"
 	"sourcedock.dev/petrbalvin/gasm-devkit/parser"
+	"sourcedock.dev/petrbalvin/gasm-devkit/verify"
 )
 
 // RunTarget is the debuggee entry point (gasm debug --target).
@@ -102,7 +103,14 @@ func RunTarget(asmPath, funcName, argsFile, tmpDir string) error {
 	os.WriteFile(tmpDir+"/entry", []byte("ok"), 0o644)
 	syscall.Kill(syscall.Getpid(), syscall.SIGSTOP)
 
-	// JIT execution requires architecture-specific verify.Call trampoline.
-	// The debugger can still set breakpoints and inspect state without it.
-	return fmt.Errorf("debug: JIT execution not yet implemented for arm64")
+	fnAddr := codeBase + uintptr(fl.Offset)
+	stackArgs := make([]byte, fl.Args)
+	copy(stackArgs, args)
+
+	_, callErr := verify.Call(fnAddr, stackArgs)
+	if callErr != nil {
+		os.Exit(1)
+	}
+	os.Exit(0)
+	return nil
 }

debug/target_linux_loong64.go

@@ -14,6 +14,7 @@ import (
 
 	"sourcedock.dev/petrbalvin/gasm-devkit/asm"
 	"sourcedock.dev/petrbalvin/gasm-devkit/parser"
+	"sourcedock.dev/petrbalvin/gasm-devkit/verify"
 )
 
 // RunTarget is the debuggee entry point (gasm debug --target).
@@ -102,7 +103,14 @@ func RunTarget(asmPath, funcName, argsFile, tmpDir string) error {
 	os.WriteFile(tmpDir+"/entry", []byte("ok"), 0o644)
 	syscall.Kill(syscall.Getpid(), syscall.SIGSTOP)
 
-	// JIT execution requires architecture-specific verify.Call trampoline.
-	// The debugger can still set breakpoints and inspect state without it.
-	return fmt.Errorf("debug: JIT execution not yet implemented for loong64")
+	fnAddr := codeBase + uintptr(fl.Offset)
+	stackArgs := make([]byte, fl.Args)
+	copy(stackArgs, args)
+
+	_, callErr := verify.Call(fnAddr, stackArgs)
+	if callErr != nil {
+		os.Exit(1)
+	}
+	os.Exit(0)
+	return nil
 }

debug/target_linux_riscv64.go

@@ -14,6 +14,7 @@ import (
 
 	"sourcedock.dev/petrbalvin/gasm-devkit/asm"
 	"sourcedock.dev/petrbalvin/gasm-devkit/parser"
+	"sourcedock.dev/petrbalvin/gasm-devkit/verify"
 )
 
 // RunTarget is the debuggee entry point (gasm debug --target).
@@ -102,7 +103,14 @@ func RunTarget(asmPath, funcName, argsFile, tmpDir string) error {
 	os.WriteFile(tmpDir+"/entry", []byte("ok"), 0o644)
 	syscall.Kill(syscall.Getpid(), syscall.SIGSTOP)
 
-	// JIT execution requires architecture-specific verify.Call trampoline.
-	// The debugger can still set breakpoints and inspect state without it.
-	return fmt.Errorf("debug: JIT execution not yet implemented for riscv64")
+	fnAddr := codeBase + uintptr(fl.Offset)
+	stackArgs := make([]byte, fl.Args)
+	copy(stackArgs, args)
+
+	_, callErr := verify.Call(fnAddr, stackArgs)
+	if callErr != nil {
+		os.Exit(1)
+	}
+	os.Exit(0)
+	return nil
 }

debug/watchpoint_linux_arm64.go

@@ -5,7 +5,14 @@
 
 package debug
 
-import "fmt"
+import (
+	"fmt"
+	"syscall"
+	"unsafe"
+)
+
+// Hardware watchpoint support via arm64 debug registers (DBGWVR/DBGWCR).
+// Accessed via PTRACE_SETREGSET with NT_ARM_HW_BREAK.
 
 // WatchpointType selects what triggers the watchpoint.
 type WatchpointType int
@@ -16,10 +23,28 @@ const (
 )
 
 // wpSlots tracks watchpoint slot occupancy.
-var wpSlots [4]bool
+var wpSlots [16]bool // arm64 supports up to 16 watchpoints
+
+const maxWatchpoints = 16
+
+// hwBreakState mirrors the kernel's struct user_hwdebug_state.
+type hwBreakState struct {
+	DbgInfo uint32
+	_pad    [4]byte
+	DbgRegs [16]hwBreakReg
+}
+
+type hwBreakReg struct {
+	Addr uint64
+	Ctrl uint64
+}
+
+const (
+	ntArmHWBreak = 0x403 // NT_ARM_HW_BREAK
+)
 
 func (s *Session) FindFreeWatchpointSlot() int {
-	for i := 0; i < 4; i++ {
+	for i := 0; i < maxWatchpoints; i++ {
 		if !wpSlots[i] {
 			return i
 		}
@@ -28,20 +53,129 @@ func (s *Session) FindFreeWatchpointSlot() int {
 }
 
 func (s *Session) IsWatchpointSlotUsed(slot int) bool {
-	if slot < 0 || slot > 3 {
+	if slot < 0 || slot >= maxWatchpoints {
 		return false
 	}
 	return wpSlots[slot]
 }
 
+// SetWatchpoint installs a hardware watchpoint on the given address.
 func (s *Session) SetWatchpoint(slot int, addr uint64, typ WatchpointType, size int) error {
-	return fmt.Errorf("debug: hardware watchpoints not yet implemented for arm64")
+	if slot < 0 || slot >= maxWatchpoints {
+		return fmt.Errorf("debug: watchpoint slot must be 0-%d", maxWatchpoints-1)
+	}
+	if wpSlots[slot] {
+		return fmt.Errorf("debug: watchpoint slot %d already in use", slot)
+	}
+
+	state, err := s.getHWBreakState()
+	if err != nil {
+		return fmt.Errorf("debug: read watchpoint state: %w", err)
+	}
+
+	if uint32(slot) >= state.DbgInfo {
+		return fmt.Errorf("debug: slot %d exceeds available watchpoints (%d)", slot, state.DbgInfo)
+	}
+
+	state.DbgRegs[slot].Addr = addr
+	ctrl := uint64(1) // enable
+	switch typ {
+	case WatchWrite:
+		ctrl |= 1 << 3 // store only
+	case WatchRead:
+		ctrl |= 3 << 3 // load+store
+	}
+	var bas uint64
+	switch size {
+	case 1:
+		bas = 0x01
+	case 2:
+		bas = 0x03
+	case 4:
+		bas = 0x0F
+	case 8:
+		bas = 0xFF
+	default:
+		return fmt.Errorf("debug: watchpoint size must be 1, 2, 4, or 8")
+	}
+	ctrl |= bas << 5
+	state.DbgRegs[slot].Ctrl = ctrl
+
+	if err := s.setHWBreakState(state); err != nil {
+		return fmt.Errorf("debug: set watchpoint: %w", err)
+	}
+
+	wpSlots[slot] = true
+	return nil
 }
 
 func (s *Session) ClearWatchpoint(slot int) error {
-	return fmt.Errorf("debug: hardware watchpoints not yet implemented for arm64")
+	if slot < 0 || slot >= maxWatchpoints {
+		return fmt.Errorf("debug: watchpoint slot must be 0-%d", maxWatchpoints-1)
+	}
+	if !wpSlots[slot] {
+		return fmt.Errorf("debug: watchpoint slot %d is not in use", slot)
+	}
+
+	state, err := s.getHWBreakState()
+	if err != nil {
+		return err
+	}
+	state.DbgRegs[slot].Addr = 0
+	state.DbgRegs[slot].Ctrl = 0
+	if err := s.setHWBreakState(state); err != nil {
+		return err
+	}
+	wpSlots[slot] = false
+	return nil
 }
 
 func (s *Session) ClearAllWatchpoints() error {
+	for slot := 0; slot < maxWatchpoints; slot++ {
+		if wpSlots[slot] {
+			if err := s.ClearWatchpoint(slot); err != nil {
+				return err
+			}
+		}
+	}
+	return nil
+}
+
+func (s *Session) getHWBreakState() (*hwBreakState, error) {
+	var state hwBreakState
+	iovec := syscall.Iovec{
+		Base: (*byte)(unsafe.Pointer(&state)),
+		Len:  uint32(unsafe.Sizeof(state)),
+	}
+	_, _, errno := syscall.Syscall6(
+		syscall.SYS_PTRACE,
+		uintptr(syscall.PTRACE_GETREGSET),
+		uintptr(s.pid),
+		uintptr(ntArmHWBreak),
+		uintptr(unsafe.Pointer(&iovec)),
+		0, 0,
+	)
+	if errno != 0 {
+		return nil, errno
+	}
+	return &state, nil
+}
+
+func (s *Session) setHWBreakState(state *hwBreakState) error {
+	iovec := syscall.Iovec{
+		Base: (*byte)(unsafe.Pointer(state)),
+		Len:  uint32(unsafe.Sizeof(*state)),
+	}
+	_, _, errno := syscall.Syscall6(
+		syscall.SYS_PTRACE,
+		uintptr(syscall.PTRACE_SETREGSET),
+		uintptr(s.pid),
+		uintptr(ntArmHWBreak),
+		uintptr(unsafe.Pointer(&iovec)),
+		0, 0,
+	)
+	if errno != 0 {
+		return errno
+	}
 	return nil
 }

debug/watchpoint_linux_loong64.go

@@ -5,7 +5,13 @@
 
 package debug
 
-import "fmt"
+import (
+	"fmt"
+	"syscall"
+)
+
+// Hardware watchpoint support for LoongArch via debug registers.
+// Uses PTRACE_POKEUSER/PEEKUSER to access HW watchpoint registers.
 
 // WatchpointType selects what triggers the watchpoint.
 type WatchpointType int
@@ -18,8 +24,10 @@ const (
 // wpSlots tracks watchpoint slot occupancy.
 var wpSlots [4]bool
 
+const maxWatchpoints = 4
+
 func (s *Session) FindFreeWatchpointSlot() int {
-	for i := 0; i < 4; i++ {
+	for i := 0; i < maxWatchpoints; i++ {
 		if !wpSlots[i] {
 			return i
 		}
@@ -28,20 +36,112 @@ func (s *Session) FindFreeWatchpointSlot() int {
 }
 
 func (s *Session) IsWatchpointSlotUsed(slot int) bool {
-	if slot < 0 || slot > 3 {
+	if slot < 0 || slot >= maxWatchpoints {
 		return false
 	}
 	return wpSlots[slot]
 }
 
+// SetWatchpoint installs a hardware watchpoint.
 func (s *Session) SetWatchpoint(slot int, addr uint64, typ WatchpointType, size int) error {
-	return fmt.Errorf("debug: hardware watchpoints not yet implemented for loong64")
+	if slot < 0 || slot >= maxWatchpoints {
+		return fmt.Errorf("debug: watchpoint slot must be 0-%d", maxWatchpoints-1)
+	}
+	if wpSlots[slot] {
+		return fmt.Errorf("debug: watchpoint slot %d already in use", slot)
+	}
+	if size != 1 && size != 2 && size != 4 && size != 8 {
+		return fmt.Errorf("debug: watchpoint size must be 1, 2, 4, or 8")
+	}
+
+	// LoongArch debug registers: DBGWVR (watchpoint value) and DBGWCR (watchpoint control).
+	// Accessed via PTRACE_POKEUSER at architecture-specific offsets.
+	if err := ptracePokeUser(s.pid, uintptr(0x1000+slot*8), addr); err != nil {
+		return fmt.Errorf("debug: set watchpoint address: %w", err)
+	}
+
+	// DBGWCR: enable + type + size.
+	var wcr uint64 = 1 // enable
+	switch typ {
+	case WatchWrite:
+		wcr |= 1 << 3 // store
+	case WatchRead:
+		wcr |= 3 << 3 // load+store
+	}
+	var sizeBits uint64
+	switch size {
+	case 1:
+		sizeBits = 0
+	case 2:
+		sizeBits = 1
+	case 4:
+		sizeBits = 2
+	case 8:
+		sizeBits = 3
+	}
+	wcr |= sizeBits << 5
+
+	if err := ptracePokeUser(s.pid, uintptr(0x1001+slot*8), wcr); err != nil {
+		return fmt.Errorf("debug: set watchpoint control: %w", err)
+	}
+
+	wpSlots[slot] = true
+	return nil
 }
 
 func (s *Session) ClearWatchpoint(slot int) error {
-	return fmt.Errorf("debug: hardware watchpoints not yet implemented for loong64")
+	if slot < 0 || slot >= maxWatchpoints {
+		return fmt.Errorf("debug: watchpoint slot must be 0-%d", maxWatchpoints-1)
+	}
+	if !wpSlots[slot] {
+		return fmt.Errorf("debug: watchpoint slot %d is not in use", slot)
+	}
+
+	if err := ptracePokeUser(s.pid, uintptr(0x1001+slot*8), 0); err != nil {
+		return err
+	}
+	wpSlots[slot] = false
+	return nil
 }
 
 func (s *Session) ClearAllWatchpoints() error {
+	for slot := 0; slot < maxWatchpoints; slot++ {
+		if wpSlots[slot] {
+			if err := s.ClearWatchpoint(slot); err != nil {
+				return err
+			}
+		}
+	}
 	return nil
 }
+
+func ptracePokeUser(pid int, offset uintptr, val uint64) error {
+	const ptracePokeuser = 6
+	_, _, errno := syscall.Syscall6(
+		syscall.SYS_PTRACE,
+		uintptr(ptracePokeuser),
+		uintptr(pid),
+		offset,
+		uintptr(val),
+		0, 0,
+	)
+	if errno != 0 {
+		return errno
+	}
+	return nil
+}
+
+func ptracePeekUser(pid int, offset uintptr) (uint64, error) {
+	const ptracePeekuser = 3
+	val, _, errno := syscall.Syscall6(
+		syscall.SYS_PTRACE,
+		uintptr(ptracePeekuser),
+		uintptr(pid),
+		offset,
+		0, 0, 0,
+	)
+	if errno != 0 {
+		return 0, errno
+	}
+	return uint64(val), nil
+}

debug/watchpoint_linux_riscv64.go

@@ -5,7 +5,13 @@
 
 package debug
 
-import "fmt"
+import (
+	"fmt"
+	"syscall"
+)
+
+// Hardware watchpoint support for RISC-V via Sdtrig trigger registers.
+// Uses PTRACE_POKEUSER/PEEKUSER to access debug registers.
 
 // WatchpointType selects what triggers the watchpoint.
 type WatchpointType int
@@ -18,8 +24,10 @@ const (
 // wpSlots tracks watchpoint slot occupancy.
 var wpSlots [4]bool
 
+const maxWatchpoints = 4
+
 func (s *Session) FindFreeWatchpointSlot() int {
-	for i := 0; i < 4; i++ {
+	for i := 0; i < maxWatchpoints; i++ {
 		if !wpSlots[i] {
 			return i
 		}
@@ -28,20 +36,115 @@ func (s *Session) FindFreeWatchpointSlot() int {
 }
 
 func (s *Session) IsWatchpointSlotUsed(slot int) bool {
-	if slot < 0 || slot > 3 {
+	if slot < 0 || slot >= maxWatchpoints {
 		return false
 	}
 	return wpSlots[slot]
 }
 
+// SetWatchpoint installs a hardware watchpoint.
 func (s *Session) SetWatchpoint(slot int, addr uint64, typ WatchpointType, size int) error {
-	return fmt.Errorf("debug: hardware watchpoints not yet implemented for riscv64")
+	if slot < 0 || slot >= maxWatchpoints {
+		return fmt.Errorf("debug: watchpoint slot must be 0-%d", maxWatchpoints-1)
+	}
+	if wpSlots[slot] {
+		return fmt.Errorf("debug: watchpoint slot %d already in use", slot)
+	}
+	if size != 1 && size != 2 && size != 4 && size != 8 {
+		return fmt.Errorf("debug: watchpoint size must be 1, 2, 4, or 8")
+	}
+
+	// RISC-V trigger registers: tdata1 encodes type/control, tdata2 holds address.
+	// The exact encoding depends on the trigger implementation (Sdtrig).
+	// Use PTRACE_POKEUSER to write to the trigger CSRs via the kernel's
+	// debug register interface.
+	if err := ptracePokeUser(s.pid, uintptr(0x1000+slot*8), addr); err != nil {
+		return fmt.Errorf("debug: set watchpoint address: %w", err)
+	}
+
+	// tdata1: set match control.  Mode=2 (data match), select=0, action=1 (debug exception).
+	var tdata1 uint64 = 2 << 60 // type = match (2)
+	tdata1 |= 1 << 0            // action = enter debug mode
+	tdata1 |= 1 << 7            // store (write) trigger
+	if typ == WatchRead {
+		tdata1 |= 1 << 6 // load trigger
+	}
+	// Size encoding: 0=1byte, 1=2byte, 2=4byte, 3=8byte.
+	var sizeBits uint64
+	switch size {
+	case 1:
+		sizeBits = 0
+	case 2:
+		sizeBits = 1
+	case 4:
+		sizeBits = 2
+	case 8:
+		sizeBits = 3
+	}
+	tdata1 |= sizeBits << 16 // size field
+
+	if err := ptracePokeUser(s.pid, uintptr(0x1001+slot*8), tdata1); err != nil {
+		return fmt.Errorf("debug: set watchpoint control: %w", err)
+	}
+
+	wpSlots[slot] = true
+	return nil
 }
 
 func (s *Session) ClearWatchpoint(slot int) error {
-	return fmt.Errorf("debug: hardware watchpoints not yet implemented for riscv64")
+	if slot < 0 || slot >= maxWatchpoints {
+		return fmt.Errorf("debug: watchpoint slot must be 0-%d", maxWatchpoints-1)
+	}
+	if !wpSlots[slot] {
+		return fmt.Errorf("debug: watchpoint slot %d is not in use", slot)
+	}
+
+	// Disable by clearing tdata1.
+	if err := ptracePokeUser(s.pid, uintptr(0x1001+slot*8), 0); err != nil {
+		return err
+	}
+	wpSlots[slot] = false
+	return nil
 }
 
 func (s *Session) ClearAllWatchpoints() error {
+	for slot := 0; slot < maxWatchpoints; slot++ {
+		if wpSlots[slot] {
+			if err := s.ClearWatchpoint(slot); err != nil {
+				return err
+			}
+		}
+	}
 	return nil
 }
+
+func ptracePokeUser(pid int, offset uintptr, val uint64) error {
+	const ptracePokeuser = 6
+	_, _, errno := syscall.Syscall6(
+		syscall.SYS_PTRACE,
+		uintptr(ptracePokeuser),
+		uintptr(pid),
+		offset,
+		uintptr(val),
+		0, 0,
+	)
+	if errno != 0 {
+		return errno
+	}
+	return nil
+}
+
+func ptracePeekUser(pid int, offset uintptr) (uint64, error) {
+	const ptracePeekuser = 3
+	val, _, errno := syscall.Syscall6(
+		syscall.SYS_PTRACE,
+		uintptr(ptracePeekuser),
+		uintptr(pid),
+		offset,
+		0, 0, 0,
+	)
+	if errno != 0 {
+		return 0, errno
+	}
+	return uint64(val), nil
+}

verify/abi_other.go

@@ -1,7 +1,7 @@
 // Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
 // SPDX-License-Identifier: BSD-3-Clause
 
-//go:build !amd64
+//go:build !amd64 && !arm64 && !riscv64 && !loong64
 
 package verify
 

verify/call_arm64.go

@@ -0,0 +1,57 @@
+// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build arm64
+
+package verify
+
+import (
+	"encoding/binary"
+	"fmt"
+	"reflect"
+	"syscall"
+	"unsafe"
+)
+
+// enterJIT switches to the prepared stack and branches to fn.
+//
+//go:nosplit
+func enterJIT(fn uintptr, stack uintptr)
+
+// leaveJIT restores the Go stack after a JIT function returns.
+//
+//go:nosplit
+func leaveJIT()
+
+// leaveJITAddr is the machine address of leaveJIT.
+var leaveJITAddr uintptr
+
+func init() {
+	leaveJITAddr = reflect.ValueOf(leaveJIT).Pointer()
+}
+
+const stackPad = 64
+
+// Call invokes the assembled function at fnAddr with the given ABI0 argument
+// block.  On arm64, the ABI0 convention places arguments on the stack starting
+// at [SP+8] (same as amd64 ABI0 in this project).
+func Call(fnAddr uintptr, args []byte) ([]byte, error) {
+	stackSize := stackPad + 8 + len(args) + 64
+	stackMem, err := syscall.Mmap(-1, 0, stackSize,
+		syscall.PROT_READ|syscall.PROT_WRITE, syscall.MAP_PRIVATE|syscall.MAP_ANON)
+	if err != nil {
+		return nil, fmt.Errorf("verify: stack mmap: %w", err)
+	}
+	defer syscall.Munmap(stackMem)
+
+	retOff := stackPad
+	binary.LittleEndian.PutUint64(stackMem[retOff:retOff+8], uint64(leaveJITAddr))
+	copy(stackMem[retOff+8:], args)
+
+	stackBase := uintptr(unsafe.Pointer(&stackMem[retOff]))
+	enterJIT(fnAddr, stackBase)
+
+	out := make([]byte, len(args))
+	copy(out, stackMem[retOff+8:retOff+8+len(args)])
+	return out, nil
+}

verify/call_loong64.go

@@ -0,0 +1,50 @@
+// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build loong64
+
+package verify
+
+import (
+	"encoding/binary"
+	"fmt"
+	"reflect"
+	"syscall"
+	"unsafe"
+)
+
+//go:nosplit
+func enterJIT(fn uintptr, stack uintptr)
+
+//go:nosplit
+func leaveJIT()
+
+var leaveJITAddr uintptr
+
+func init() {
+	leaveJITAddr = reflect.ValueOf(leaveJIT).Pointer()
+}
+
+const stackPad = 64
+
+// Call invokes the assembled function at fnAddr with the given ABI0 argument block.
+func Call(fnAddr uintptr, args []byte) ([]byte, error) {
+	stackSize := stackPad + 8 + len(args) + 64
+	stackMem, err := syscall.Mmap(-1, 0, stackSize,
+		syscall.PROT_READ|syscall.PROT_WRITE, syscall.MAP_PRIVATE|syscall.MAP_ANON)
+	if err != nil {
+		return nil, fmt.Errorf("verify: stack mmap: %w", err)
+	}
+	defer syscall.Munmap(stackMem)
+
+	retOff := stackPad
+	binary.LittleEndian.PutUint64(stackMem[retOff:retOff+8], uint64(leaveJITAddr))
+	copy(stackMem[retOff+8:], args)
+
+	stackBase := uintptr(unsafe.Pointer(&stackMem[retOff]))
+	enterJIT(fnAddr, stackBase)
+
+	out := make([]byte, len(args))
+	copy(out, stackMem[retOff+8:retOff+8+len(args)])
+	return out, nil
+}

verify/call_other.go

@@ -1,7 +1,7 @@
 // Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
 // SPDX-License-Identifier: BSD-3-Clause
 
-//go:build !amd64
+//go:build !amd64 && !arm64 && !riscv64 && !loong64
 
 package verify
 

verify/call_riscv64.go

@@ -0,0 +1,50 @@
+// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
+// SPDX-License-Identifier: BSD-3-Clause
+
+//go:build riscv64
+
+package verify
+
+import (
+	"encoding/binary"
+	"fmt"
+	"reflect"
+	"syscall"
+	"unsafe"
+)
+
+//go:nosplit
+func enterJIT(fn uintptr, stack uintptr)
+
+//go:nosplit
+func leaveJIT()
+
+var leaveJITAddr uintptr
+
+func init() {
+	leaveJITAddr = reflect.ValueOf(leaveJIT).Pointer()
+}
+
+const stackPad = 64
+
+// Call invokes the assembled function at fnAddr with the given ABI0 argument block.
+func Call(fnAddr uintptr, args []byte) ([]byte, error) {
+	stackSize := stackPad + 8 + len(args) + 64
+	stackMem, err := syscall.Mmap(-1, 0, stackSize,
+		syscall.PROT_READ|syscall.PROT_WRITE, syscall.MAP_PRIVATE|syscall.MAP_ANON)
+	if err != nil {
+		return nil, fmt.Errorf("verify: stack mmap: %w", err)
+	}
+	defer syscall.Munmap(stackMem)
+
+	retOff := stackPad
+	binary.LittleEndian.PutUint64(stackMem[retOff:retOff+8], uint64(leaveJITAddr))
+	copy(stackMem[retOff+8:], args)
+
+	stackBase := uintptr(unsafe.Pointer(&stackMem[retOff]))
+	enterJIT(fnAddr, stackBase)
+
+	out := make([]byte, len(args))
+	copy(out, stackMem[retOff+8:retOff+8+len(args)])
+	return out, nil
+}

verify/trampoline_arm64.s

@@ -0,0 +1,36 @@
+// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
+// SPDX-License-Identifier: BSD-3-Clause
+
+#include "textflag.h"
+
+// ABI0 JIT trampoline for arm64.
+//
+// enterJIT saves the Go SP and LR, switches to the prepared stack, and
+// branches to the JIT function.  When the function RETs (BR LR), control
+// lands in leaveJIT, which restores the Go stack and returns.
+//
+// The prepared stack layout:
+//   [0]   leaveJIT address (loaded into LR before jumping to fn)
+//   [8]   ABI0 argument area
+
+// func enterJIT(fn uintptr, stack uintptr)
+TEXT ·enterJIT(SB), NOSPLIT, $0-16
+	MOVD fn+0(FP), R0          // target function address
+	MOVD R30, savedLR(SB)      // save link register
+	MOVD R3, savedSP(SB)       // save Go stack pointer
+	MOVD stack+8(FP), R3       // load prepared stack pointer
+	MOVD 0(R3), R30            // load leaveJIT address into LR
+	ADD  $8, R3, R3            // advance past return address
+	MOVD R3, RSP               // switch to prepared stack
+	JMP  (R0)                  // branch to JIT function
+
+// func leaveJIT()
+TEXT ·leaveJIT(SB), NOSPLIT, $0-0
+	MOVD savedSP(SB), R3       // restore Go stack pointer
+	MOVD R3, RSP
+	MOVD savedLR(SB), R30      // restore link register
+	RET                        // return to Go caller
+
+// Package-level storage for saved registers.
+GLOBL savedLR(SB), NOPTR, $8
+GLOBL savedSP(SB), NOPTR, $8

verify/trampoline_loong64.s

@@ -0,0 +1,31 @@
+// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
+// SPDX-License-Identifier: BSD-3-Clause
+
+#include "textflag.h"
+
+// ABI0 JIT trampoline for LoongArch 64.
+//
+// enterJIT saves Go SP and RA (R1), switches to the prepared stack, and
+// jumps to the JIT function.  When the function RETs (JIRL zero, ra, 0),
+// control lands in leaveJIT.
+
+// func enterJIT(fn uintptr, stack uintptr)
+TEXT ·enterJIT(SB), NOSPLIT, $0-16
+	MOVV fn+0(FP), R4           // target function address (A0)
+	MOVV R1, savedRA(SB)        // save return address (RA)
+	MOVV R3, savedSP(SB)        // save Go stack pointer (SP)
+	MOVV stack+8(FP), R5        // load prepared stack pointer (A1)
+	MOVV 0(R5), R1              // load leaveJIT address into RA
+	ADDV $8, R5, R5             // advance past return address
+	MOVV R5, R3                 // switch to prepared stack (SP)
+	JIRL R0, R4, 0              // jump to JIT function
+
+// func leaveJIT()
+TEXT ·leaveJIT(SB), NOSPLIT, $0-0
+	MOVV savedSP(SB), R5        // restore Go stack pointer
+	MOVV R5, R3                 // restore SP
+	MOVV savedRA(SB), R1        // restore return address
+	JIRL R0, R1, 0              // return to Go caller
+
+GLOBL savedRA(SB), NOPTR, $8
+GLOBL savedSP(SB), NOPTR, $8

verify/trampoline_riscv64.s

@@ -0,0 +1,31 @@
+// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
+// SPDX-License-Identifier: BSD-3-Clause
+
+#include "textflag.h"
+
+// ABI0 JIT trampoline for RISC-V 64.
+//
+// enterJIT saves Go SP and RA, switches to the prepared stack, and jumps
+// to the JIT function.  When the function RETs (JALR zero, ra, 0), control
+// lands in leaveJIT.
+
+// func enterJIT(fn uintptr, stack uintptr)
+TEXT ·enterJIT(SB), NOSPLIT, $0-16
+	MOV  fn+0(FP), X5          // target function address (T0)
+	MOV  X1, savedRA(SB)       // save return address
+	MOV  X2, savedSP(SB)       // save Go stack pointer
+	MOV  stack+8(FP), X6       // load prepared stack pointer (T1)
+	LD   0(X6), X1             // load leaveJIT address into RA
+	ADD  $8, X6, X6            // advance past return address
+	MOV  X6, X2                // switch to prepared stack (SP)
+	JALR X0, 0(X5)             // jump to JIT function
+
+// func leaveJIT()
+TEXT ·leaveJIT(SB), NOSPLIT, $0-0
+	MOV  savedSP(SB), X6       // restore Go stack pointer
+	MOV  X6, X2                // restore SP
+	MOV  savedRA(SB), X1       // restore return address
+	JALR X0, 0(X1)             // return to Go caller
+
+GLOBL savedRA(SB), NOPTR, $8
+GLOBL savedSP(SB), NOPTR, $8